From 1b434b9fff746c2a56d4186fe787fadf7571c603 Mon Sep 17 00:00:00 2001 From: "Nicolas \"Pixel\" Noble" Date: Thu, 31 Jul 2014 00:39:59 -0700 Subject: Changing the regex library for one that isn't GPL. --- win32/regex/msvc-regex.c | 4949 ---------------------------------------------- win32/regex/regex.h | 559 +----- 2 files changed, 63 insertions(+), 5445 deletions(-) delete mode 100644 win32/regex/msvc-regex.c (limited to 'win32/regex') diff --git a/win32/regex/msvc-regex.c b/win32/regex/msvc-regex.c deleted file mode 100644 index 2d74eae..0000000 --- a/win32/regex/msvc-regex.c +++ /dev/null @@ -1,4949 +0,0 @@ -#ifndef __STDC__ -#define __STDC__ 1 -#endif -#define STDC_HEADERS 1 -#define HAVE_STRING_H 1 -#define REGEX_MALLOC 1 - -/* Extended regular expression matching and search library, - version 0.12. - (Implements POSIX draft P10003.2/D11.2, except for - internationalization features.) - - Copyright (C) 1993 Free Software Foundation, Inc. - - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2, or (at your option) - any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ - -/* AIX requires this to be the first thing in the file. */ -#if defined (_AIX) && !defined (REGEX_MALLOC) - #pragma alloca -#endif - -#define _GNU_SOURCE - -/* We need this for `regex.h', and perhaps for the Emacs include files. */ -#include - -#ifdef HAVE_CONFIG_H -#include "config.h" -#endif - -/* The `emacs' switch turns on certain matching commands - that make sense only in Emacs. */ -#ifdef emacs - -#include "lisp.h" -#include "buffer.h" -#include "syntax.h" - -/* Emacs uses `NULL' as a predicate. */ -#undef NULL - -#else /* not emacs */ - -/* We used to test for `BSTRING' here, but only GCC and Emacs define - `BSTRING', as far as I know, and neither of them use this code. */ -#if HAVE_STRING_H || STDC_HEADERS -#include -#ifndef bcmp -#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n)) -#endif -#ifndef bcopy -#define bcopy(s, d, n) memcpy ((d), (s), (n)) -#endif -#ifndef bzero -#define bzero(s, n) memset ((s), 0, (n)) -#endif -#else -#include -#endif - -#ifdef STDC_HEADERS -#include -#else -char *malloc (); -char *realloc (); -#endif - - -/* Define the syntax stuff for \<, \>, etc. */ - -/* This must be nonzero for the wordchar and notwordchar pattern - commands in re_match_2. */ -#ifndef Sword -#define Sword 1 -#endif - -#ifdef SYNTAX_TABLE - -extern char *re_syntax_table; - -#else /* not SYNTAX_TABLE */ - -/* How many characters in the character set. */ -#define CHAR_SET_SIZE 256 - -static char re_syntax_table[CHAR_SET_SIZE]; - -static void -init_syntax_once () -{ - register int c; - static int done = 0; - - if (done) - return; - - bzero (re_syntax_table, sizeof re_syntax_table); - - for (c = 'a'; c <= 'z'; c++) - re_syntax_table[c] = Sword; - - for (c = 'A'; c <= 'Z'; c++) - re_syntax_table[c] = Sword; - - for (c = '0'; c <= '9'; c++) - re_syntax_table[c] = Sword; - - re_syntax_table['_'] = Sword; - - done = 1; -} - -#endif /* not SYNTAX_TABLE */ - -#define SYNTAX(c) re_syntax_table[c] - -#endif /* not emacs */ - -/* Get the interface, including the syntax bits. */ -#include "regex.h" - -/* isalpha etc. are used for the character classes. */ -#include - -#ifndef isascii -#define isascii(c) 1 -#endif - -#ifdef isblank -#define ISBLANK(c) (isascii (c) && isblank (c)) -#else -#define ISBLANK(c) ((c) == ' ' || (c) == '\t') -#endif -#ifdef isgraph -#define ISGRAPH(c) (isascii (c) && isgraph (c)) -#else -#define ISGRAPH(c) (isascii (c) && isprint (c) && !isspace (c)) -#endif - -#define ISPRINT(c) (isascii (c) && isprint (c)) -#define ISDIGIT(c) (isascii (c) && isdigit (c)) -#define ISALNUM(c) (isascii (c) && isalnum (c)) -#define ISALPHA(c) (isascii (c) && isalpha (c)) -#define ISCNTRL(c) (isascii (c) && iscntrl (c)) -#define ISLOWER(c) (isascii (c) && islower (c)) -#define ISPUNCT(c) (isascii (c) && ispunct (c)) -#define ISSPACE(c) (isascii (c) && isspace (c)) -#define ISUPPER(c) (isascii (c) && isupper (c)) -#define ISXDIGIT(c) (isascii (c) && isxdigit (c)) - -#ifndef NULL -#define NULL 0 -#endif - -/* We remove any previous definition of `SIGN_EXTEND_CHAR', - since ours (we hope) works properly with all combinations of - machines, compilers, `char' and `unsigned char' argument types. - (Per Bothner suggested the basic approach.) */ -#undef SIGN_EXTEND_CHAR -#if __STDC__ -#define SIGN_EXTEND_CHAR(c) ((signed char) (c)) -#else /* not __STDC__ */ -/* As in Harbison and Steele. */ -#define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) -#endif - -/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we - use `alloca' instead of `malloc'. This is because using malloc in - re_search* or re_match* could cause memory leaks when C-g is used in - Emacs; also, malloc is slower and causes storage fragmentation. On - the other hand, malloc is more portable, and easier to debug. - - Because we sometimes use alloca, some routines have to be macros, - not functions -- `alloca'-allocated space disappears at the end of the - function it is called in. */ - -#ifdef REGEX_MALLOC - -#define REGEX_ALLOCATE malloc -#define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize) - -#else /* not REGEX_MALLOC */ - -/* Emacs already defines alloca, sometimes. */ -#ifndef alloca - -/* Make alloca work the best possible way. */ -#ifdef __GNUC__ -#define alloca __builtin_alloca -#else /* not __GNUC__ */ -#if HAVE_ALLOCA_H -#include -#else /* not __GNUC__ or HAVE_ALLOCA_H */ -#ifndef _AIX /* Already did AIX, up at the top. */ -char *alloca (); -#endif /* not _AIX */ -#endif /* not HAVE_ALLOCA_H */ -#endif /* not __GNUC__ */ - -#endif /* not alloca */ - -#define REGEX_ALLOCATE alloca - -/* Assumes a `char *destination' variable. */ -#define REGEX_REALLOCATE(source, osize, nsize) \ - (destination = (char *) alloca (nsize), \ - bcopy (source, destination, osize), \ - destination) - -#endif /* not REGEX_MALLOC */ - - -/* True if `size1' is non-NULL and PTR is pointing anywhere inside - `string1' or just past its end. This works if PTR is NULL, which is - a good thing. */ -#define FIRST_STRING_P(ptr) \ - (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) - -/* (Re)Allocate N items of type T using malloc, or fail. */ -#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) -#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) -#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) - -#define BYTEWIDTH 8 /* In bits. */ - -#define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) - -#define MAX(a, b) ((a) > (b) ? (a) : (b)) -#define MIN(a, b) ((a) < (b) ? (a) : (b)) - -typedef char boolean; -#define false 0 -#define true 1 - -/* These are the command codes that appear in compiled regular - expressions. Some opcodes are followed by argument bytes. A - command code can specify any interpretation whatsoever for its - arguments. Zero bytes may appear in the compiled regular expression. - - The value of `exactn' is needed in search.c (search_buffer) in Emacs. - So regex.h defines a symbol `RE_EXACTN_VALUE' to be 1; the value of - `exactn' we use here must also be 1. */ - -typedef enum -{ - no_op = 0, - - /* Followed by one byte giving n, then by n literal bytes. */ - exactn = 1, - - /* Matches any (more or less) character. */ - anychar, - - /* Matches any one char belonging to specified set. First - following byte is number of bitmap bytes. Then come bytes - for a bitmap saying which chars are in. Bits in each byte - are ordered low-bit-first. A character is in the set if its - bit is 1. A character too large to have a bit in the map is - automatically not in the set. */ - charset, - - /* Same parameters as charset, but match any character that is - not one of those specified. */ - charset_not, - - /* Start remembering the text that is matched, for storing in a - register. Followed by one byte with the register number, in - the range 0 to one less than the pattern buffer's re_nsub - field. Then followed by one byte with the number of groups - inner to this one. (This last has to be part of the - start_memory only because we need it in the on_failure_jump - of re_match_2.) */ - start_memory, - - /* Stop remembering the text that is matched and store it in a - memory register. Followed by one byte with the register - number, in the range 0 to one less than `re_nsub' in the - pattern buffer, and one byte with the number of inner groups, - just like `start_memory'. (We need the number of inner - groups here because we don't have any easy way of finding the - corresponding start_memory when we're at a stop_memory.) */ - stop_memory, - - /* Match a duplicate of something remembered. Followed by one - byte containing the register number. */ - duplicate, - - /* Fail unless at beginning of line. */ - begline, - - /* Fail unless at end of line. */ - endline, - - /* Succeeds if at beginning of buffer (if emacs) or at beginning - of string to be matched (if not). */ - begbuf, - - /* Analogously, for end of buffer/string. */ - endbuf, - - /* Followed by two byte relative address to which to jump. */ - jump, - - /* Same as jump, but marks the end of an alternative. */ - jump_past_alt, - - /* Followed by two-byte relative address of place to resume at - in case of failure. */ - on_failure_jump, - - /* Like on_failure_jump, but pushes a placeholder instead of the - current string position when executed. */ - on_failure_keep_string_jump, - - /* Throw away latest failure point and then jump to following - two-byte relative address. */ - pop_failure_jump, - - /* Change to pop_failure_jump if know won't have to backtrack to - match; otherwise change to jump. This is used to jump - back to the beginning of a repeat. If what follows this jump - clearly won't match what the repeat does, such that we can be - sure that there is no use backtracking out of repetitions - already matched, then we change it to a pop_failure_jump. - Followed by two-byte address. */ - maybe_pop_jump, - - /* Jump to following two-byte address, and push a dummy failure - point. This failure point will be thrown away if an attempt - is made to use it for a failure. A `+' construct makes this - before the first repeat. Also used as an intermediary kind - of jump when compiling an alternative. */ - dummy_failure_jump, - - /* Push a dummy failure point and continue. Used at the end of - alternatives. */ - push_dummy_failure, - - /* Followed by two-byte relative address and two-byte number n. - After matching N times, jump to the address upon failure. */ - succeed_n, - - /* Followed by two-byte relative address, and two-byte number n. - Jump to the address N times, then fail. */ - jump_n, - - /* Set the following two-byte relative address to the - subsequent two-byte number. The address *includes* the two - bytes of number. */ - set_number_at, - - wordchar, /* Matches any word-constituent character. */ - notwordchar, /* Matches any char that is not a word-constituent. */ - - wordbeg, /* Succeeds if at word beginning. */ - wordend, /* Succeeds if at word end. */ - - wordbound, /* Succeeds if at a word boundary. */ - notwordbound /* Succeeds if not at a word boundary. */ - -#ifdef emacs - ,before_dot, /* Succeeds if before point. */ - at_dot, /* Succeeds if at point. */ - after_dot, /* Succeeds if after point. */ - - /* Matches any character whose syntax is specified. Followed by - a byte which contains a syntax code, e.g., Sword. */ - syntaxspec, - - /* Matches any character whose syntax is not that specified. */ - notsyntaxspec -#endif /* emacs */ -} re_opcode_t; - -/* Common operations on the compiled pattern. */ - -/* Store NUMBER in two contiguous bytes starting at DESTINATION. */ - -#define STORE_NUMBER(destination, number) \ - do { \ - (destination)[0] = (number) & 0377; \ - (destination)[1] = (number) >> 8; \ - } while (0) - -/* Same as STORE_NUMBER, except increment DESTINATION to - the byte after where the number is stored. Therefore, DESTINATION - must be an lvalue. */ - -#define STORE_NUMBER_AND_INCR(destination, number) \ - do { \ - STORE_NUMBER (destination, number); \ - (destination) += 2; \ - } while (0) - -/* Put into DESTINATION a number stored in two contiguous bytes starting - at SOURCE. */ - -#define EXTRACT_NUMBER(destination, source) \ - do { \ - (destination) = *(source) & 0377; \ - (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ - } while (0) - -#ifdef DEBUG -static void -extract_number (dest, source) - int *dest; - unsigned char *source; -{ - int temp = SIGN_EXTEND_CHAR (*(source + 1)); - *dest = *source & 0377; - *dest += temp << 8; -} - -#ifndef EXTRACT_MACROS /* To debug the macros. */ -#undef EXTRACT_NUMBER -#define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) -#endif /* not EXTRACT_MACROS */ - -#endif /* DEBUG */ - -/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. - SOURCE must be an lvalue. */ - -#define EXTRACT_NUMBER_AND_INCR(destination, source) \ - do { \ - EXTRACT_NUMBER (destination, source); \ - (source) += 2; \ - } while (0) - -#ifdef DEBUG -static void -extract_number_and_incr (destination, source) - int *destination; - unsigned char **source; -{ - extract_number (destination, *source); - *source += 2; -} - -#ifndef EXTRACT_MACROS -#undef EXTRACT_NUMBER_AND_INCR -#define EXTRACT_NUMBER_AND_INCR(dest, src) \ - extract_number_and_incr (&dest, &src) -#endif /* not EXTRACT_MACROS */ - -#endif /* DEBUG */ - -/* If DEBUG is defined, Regex prints many voluminous messages about what - it is doing (if the variable `debug' is nonzero). If linked with the - main program in `iregex.c', you can enter patterns and strings - interactively. And if linked with the main program in `main.c' and - the other test files, you can run the already-written tests. */ - -#ifdef DEBUG - -/* We use standard I/O for debugging. */ -#include - -/* It is useful to test things that ``must'' be true when debugging. */ -#include - -static int debug = 0; - -#define DEBUG_STATEMENT(e) e -#define DEBUG_PRINT1(x) if (debug) printf (x) -#define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) -#define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) -#define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) -#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ - if (debug) print_partial_compiled_pattern (s, e) -#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ - if (debug) print_double_string (w, s1, sz1, s2, sz2) - - -extern void printchar (); - -/* Print the fastmap in human-readable form. */ - -void -print_fastmap (fastmap) - char *fastmap; -{ - unsigned was_a_range = 0; - unsigned i = 0; - - while (i < (1 << BYTEWIDTH)) - { - if (fastmap[i++]) - { - was_a_range = 0; - printchar (i - 1); - while (i < (1 << BYTEWIDTH) && fastmap[i]) - { - was_a_range = 1; - i++; - } - if (was_a_range) - { - printf ("-"); - printchar (i - 1); - } - } - } - putchar ('\n'); -} - - -/* Print a compiled pattern string in human-readable form, starting at - the START pointer into it and ending just before the pointer END. */ - -void -print_partial_compiled_pattern (start, end) - unsigned char *start; - unsigned char *end; -{ - int mcnt, mcnt2; - unsigned char *p = start; - unsigned char *pend = end; - - if (start == NULL) - { - printf ("(null)\n"); - return; - } - - /* Loop over pattern commands. */ - while (p < pend) - { - switch ((re_opcode_t) *p++) - { - case no_op: - printf ("/no_op"); - break; - - case exactn: - mcnt = *p++; - printf ("/exactn/%d", mcnt); - do - { - putchar ('/'); - printchar (*p++); - } - while (--mcnt); - break; - - case start_memory: - mcnt = *p++; - printf ("/start_memory/%d/%d", mcnt, *p++); - break; - - case stop_memory: - mcnt = *p++; - printf ("/stop_memory/%d/%d", mcnt, *p++); - break; - - case duplicate: - printf ("/duplicate/%d", *p++); - break; - - case anychar: - printf ("/anychar"); - break; - - case charset: - case charset_not: - { - register int c; - - printf ("/charset%s", - (re_opcode_t) *(p - 1) == charset_not ? "_not" : ""); - - assert (p + *p < pend); - - for (c = 0; c < *p; c++) - { - unsigned bit; - unsigned char map_byte = p[1 + c]; - - putchar ('/'); - - for (bit = 0; bit < BYTEWIDTH; bit++) - if (map_byte & (1 << bit)) - printchar (c * BYTEWIDTH + bit); - } - p += 1 + *p; - break; - } - - case begline: - printf ("/begline"); - break; - - case endline: - printf ("/endline"); - break; - - case on_failure_jump: - extract_number_and_incr (&mcnt, &p); - printf ("/on_failure_jump/0/%d", mcnt); - break; - - case on_failure_keep_string_jump: - extract_number_and_incr (&mcnt, &p); - printf ("/on_failure_keep_string_jump/0/%d", mcnt); - break; - - case dummy_failure_jump: - extract_number_and_incr (&mcnt, &p); - printf ("/dummy_failure_jump/0/%d", mcnt); - break; - - case push_dummy_failure: - printf ("/push_dummy_failure"); - break; - - case maybe_pop_jump: - extract_number_and_incr (&mcnt, &p); - printf ("/maybe_pop_jump/0/%d", mcnt); - break; - - case pop_failure_jump: - extract_number_and_incr (&mcnt, &p); - printf ("/pop_failure_jump/0/%d", mcnt); - break; - - case jump_past_alt: - extract_number_and_incr (&mcnt, &p); - printf ("/jump_past_alt/0/%d", mcnt); - break; - - case jump: - extract_number_and_incr (&mcnt, &p); - printf ("/jump/0/%d", mcnt); - break; - - case succeed_n: - extract_number_and_incr (&mcnt, &p); - extract_number_and_incr (&mcnt2, &p); - printf ("/succeed_n/0/%d/0/%d", mcnt, mcnt2); - break; - - case jump_n: - extract_number_and_incr (&mcnt, &p); - extract_number_and_incr (&mcnt2, &p); - printf ("/jump_n/0/%d/0/%d", mcnt, mcnt2); - break; - - case set_number_at: - extract_number_and_incr (&mcnt, &p); - extract_number_and_incr (&mcnt2, &p); - printf ("/set_number_at/0/%d/0/%d", mcnt, mcnt2); - break; - - case wordbound: - printf ("/wordbound"); - break; - - case notwordbound: - printf ("/notwordbound"); - break; - - case wordbeg: - printf ("/wordbeg"); - break; - - case wordend: - printf ("/wordend"); - -#ifdef emacs - case before_dot: - printf ("/before_dot"); - break; - - case at_dot: - printf ("/at_dot"); - break; - - case after_dot: - printf ("/after_dot"); - break; - - case syntaxspec: - printf ("/syntaxspec"); - mcnt = *p++; - printf ("/%d", mcnt); - break; - - case notsyntaxspec: - printf ("/notsyntaxspec"); - mcnt = *p++; - printf ("/%d", mcnt); - break; -#endif /* emacs */ - - case wordchar: - printf ("/wordchar"); - break; - - case notwordchar: - printf ("/notwordchar"); - break; - - case begbuf: - printf ("/begbuf"); - break; - - case endbuf: - printf ("/endbuf"); - break; - - default: - printf ("?%d", *(p-1)); - } - } - printf ("/\n"); -} - - -void -print_compiled_pattern (bufp) - struct re_pattern_buffer *bufp; -{ - unsigned char *buffer = bufp->buffer; - - print_partial_compiled_pattern (buffer, buffer + bufp->used); - printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated); - - if (bufp->fastmap_accurate && bufp->fastmap) - { - printf ("fastmap: "); - print_fastmap (bufp->fastmap); - } - - printf ("re_nsub: %d\t", bufp->re_nsub); - printf ("regs_alloc: %d\t", bufp->regs_allocated); - printf ("can_be_null: %d\t", bufp->can_be_null); - printf ("newline_anchor: %d\n", bufp->newline_anchor); - printf ("no_sub: %d\t", bufp->no_sub); - printf ("not_bol: %d\t", bufp->not_bol); - printf ("not_eol: %d\t", bufp->not_eol); - printf ("syntax: %d\n", bufp->syntax); - /* Perhaps we should print the translate table? */ -} - - -void -print_double_string (where, string1, size1, string2, size2) - const char *where; - const char *string1; - const char *string2; - int size1; - int size2; -{ - unsigned this_char; - - if (where == NULL) - printf ("(null)"); - else - { - if (FIRST_STRING_P (where)) - { - for (this_char = where - string1; this_char < size1; this_char++) - printchar (string1[this_char]); - - where = string2; - } - - for (this_char = where - string2; this_char < size2; this_char++) - printchar (string2[this_char]); - } -} - -#else /* not DEBUG */ - -#undef assert -#define assert(e) - -#define DEBUG_STATEMENT(e) -#define DEBUG_PRINT1(x) -#define DEBUG_PRINT2(x1, x2) -#define DEBUG_PRINT3(x1, x2, x3) -#define DEBUG_PRINT4(x1, x2, x3, x4) -#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) -#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) - -#endif /* not DEBUG */ - -/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can - also be assigned to arbitrarily: each pattern buffer stores its own - syntax, so it can be changed between regex compilations. */ -reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS; - - -/* Specify the precise syntax of regexps for compilation. This provides - for compatibility for various utilities which historically have - different, incompatible syntaxes. - - The argument SYNTAX is a bit mask comprised of the various bits - defined in regex.h. We return the old syntax. */ - -reg_syntax_t -re_set_syntax (syntax) - reg_syntax_t syntax; -{ - reg_syntax_t ret = re_syntax_options; - - re_syntax_options = syntax; - return ret; -} - -/* This table gives an error message for each of the error codes listed - in regex.h. Obviously the order here has to be same as there. */ - -static const char *re_error_msg[] = - { NULL, /* REG_NOERROR */ - "No match", /* REG_NOMATCH */ - "Invalid regular expression", /* REG_BADPAT */ - "Invalid collation character", /* REG_ECOLLATE */ - "Invalid character class name", /* REG_ECTYPE */ - "Trailing backslash", /* REG_EESCAPE */ - "Invalid back reference", /* REG_ESUBREG */ - "Unmatched [ or [^", /* REG_EBRACK */ - "Unmatched ( or \\(", /* REG_EPAREN */ - "Unmatched \\{", /* REG_EBRACE */ - "Invalid content of \\{\\}", /* REG_BADBR */ - "Invalid range end", /* REG_ERANGE */ - "Memory exhausted", /* REG_ESPACE */ - "Invalid preceding regular expression", /* REG_BADRPT */ - "Premature end of regular expression", /* REG_EEND */ - "Regular expression too big", /* REG_ESIZE */ - "Unmatched ) or \\)", /* REG_ERPAREN */ - }; - -/* Subroutine declarations and macros for regex_compile. */ - -static void store_op1 (), store_op2 (); -static void insert_op1 (), insert_op2 (); -static boolean at_begline_loc_p (), at_endline_loc_p (); -static boolean group_in_compile_stack (); -static reg_errcode_t compile_range (); - -/* Fetch the next character in the uncompiled pattern---translating it - if necessary. Also cast from a signed character in the constant - string passed to us by the user to an unsigned char that we can use - as an array index (in, e.g., `translate'). */ -#define PATFETCH(c) \ - do {if (p == pend) return REG_EEND; \ - c = (unsigned char) *p++; \ - if (translate) c = translate[c]; \ - } while (0) - -/* Fetch the next character in the uncompiled pattern, with no - translation. */ -#define PATFETCH_RAW(c) \ - do {if (p == pend) return REG_EEND; \ - c = (unsigned char) *p++; \ - } while (0) - -/* Go backwards one character in the pattern. */ -#define PATUNFETCH p-- - - -/* If `translate' is non-null, return translate[D], else just D. We - cast the subscript to translate because some data is declared as - `char *', to avoid warnings when a string constant is passed. But - when we use a character as a subscript we must make it unsigned. */ -#define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d)) - - -/* Macros for outputting the compiled pattern into `buffer'. */ - -/* If the buffer isn't allocated when it comes in, use this. */ -#define INIT_BUF_SIZE 32 - -/* Make sure we have at least N more bytes of space in buffer. */ -#define GET_BUFFER_SPACE(n) \ - while (b - bufp->buffer + (n) > bufp->allocated) \ - EXTEND_BUFFER () - -/* Make sure we have one more byte of buffer space and then add C to it. */ -#define BUF_PUSH(c) \ - do { \ - GET_BUFFER_SPACE (1); \ - *b++ = (unsigned char) (c); \ - } while (0) - - -/* Ensure we have two more bytes of buffer space and then append C1 and C2. */ -#define BUF_PUSH_2(c1, c2) \ - do { \ - GET_BUFFER_SPACE (2); \ - *b++ = (unsigned char) (c1); \ - *b++ = (unsigned char) (c2); \ - } while (0) - - -/* As with BUF_PUSH_2, except for three bytes. */ -#define BUF_PUSH_3(c1, c2, c3) \ - do { \ - GET_BUFFER_SPACE (3); \ - *b++ = (unsigned char) (c1); \ - *b++ = (unsigned char) (c2); \ - *b++ = (unsigned char) (c3); \ - } while (0) - - -/* Store a jump with opcode OP at LOC to location TO. We store a - relative address offset by the three bytes the jump itself occupies. */ -#define STORE_JUMP(op, loc, to) \ - store_op1 (op, loc, (to) - (loc) - 3) - -/* Likewise, for a two-argument jump. */ -#define STORE_JUMP2(op, loc, to, arg) \ - store_op2 (op, loc, (to) - (loc) - 3, arg) - -/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ -#define INSERT_JUMP(op, loc, to) \ - insert_op1 (op, loc, (to) - (loc) - 3, b) - -/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ -#define INSERT_JUMP2(op, loc, to, arg) \ - insert_op2 (op, loc, (to) - (loc) - 3, arg, b) - - -/* This is not an arbitrary limit: the arguments which represent offsets - into the pattern are two bytes long. So if 2^16 bytes turns out to - be too small, many things would have to change. */ -#define MAX_BUF_SIZE (1L << 16) - - -/* Extend the buffer by twice its current size via realloc and - reset the pointers that pointed into the old block to point to the - correct places in the new one. If extending the buffer results in it - being larger than MAX_BUF_SIZE, then flag memory exhausted. */ -#define EXTEND_BUFFER() \ - do { \ - unsigned char *old_buffer = bufp->buffer; \ - if (bufp->allocated == MAX_BUF_SIZE) \ - return REG_ESIZE; \ - bufp->allocated <<= 1; \ - if (bufp->allocated > MAX_BUF_SIZE) \ - bufp->allocated = MAX_BUF_SIZE; \ - bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\ - if (bufp->buffer == NULL) \ - return REG_ESPACE; \ - /* If the buffer moved, move all the pointers into it. */ \ - if (old_buffer != bufp->buffer) \ - { \ - b = (b - old_buffer) + bufp->buffer; \ - begalt = (begalt - old_buffer) + bufp->buffer; \ - if (fixup_alt_jump) \ - fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ - if (laststart) \ - laststart = (laststart - old_buffer) + bufp->buffer; \ - if (pending_exact) \ - pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ - } \ - } while (0) - - -/* Since we have one byte reserved for the register number argument to - {start,stop}_memory, the maximum number of groups we can report - things about is what fits in that byte. */ -#define MAX_REGNUM 255 - -/* But patterns can have more than `MAX_REGNUM' registers. We just - ignore the excess. */ -typedef unsigned regnum_t; - - -/* Macros for the compile stack. */ - -/* Since offsets can go either forwards or backwards, this type needs to - be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ -typedef int pattern_offset_t; - -typedef struct -{ - pattern_offset_t begalt_offset; - pattern_offset_t fixup_alt_jump; - pattern_offset_t inner_group_offset; - pattern_offset_t laststart_offset; - regnum_t regnum; -} compile_stack_elt_t; - - -typedef struct -{ - compile_stack_elt_t *stack; - unsigned size; - unsigned avail; /* Offset of next open position. */ -} compile_stack_type; - - -#define INIT_COMPILE_STACK_SIZE 32 - -#define COMPILE_STACK_EMPTY (compile_stack.avail == 0) -#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) - -/* The next available element. */ -#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) - - -/* Set the bit for character C in a list. */ -#define SET_LIST_BIT(c) \ - (b[((unsigned char) (c)) / BYTEWIDTH] \ - |= 1 << (((unsigned char) c) % BYTEWIDTH)) - - -/* Get the next unsigned number in the uncompiled pattern. */ -#define GET_UNSIGNED_NUMBER(num) \ - { if (p != pend) \ - { \ - PATFETCH (c); \ - while (ISDIGIT (c)) \ - { \ - if (num < 0) \ - num = 0; \ - num = num * 10 + c - '0'; \ - if (p == pend) \ - break; \ - PATFETCH (c); \ - } \ - } \ - } - -#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ - -#define IS_CHAR_CLASS(string) \ - (STREQ (string, "alpha") || STREQ (string, "upper") \ - || STREQ (string, "lower") || STREQ (string, "digit") \ - || STREQ (string, "alnum") || STREQ (string, "xdigit") \ - || STREQ (string, "space") || STREQ (string, "print") \ - || STREQ (string, "punct") || STREQ (string, "graph") \ - || STREQ (string, "cntrl") || STREQ (string, "blank")) - -/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. - Returns one of error codes defined in `regex.h', or zero for success. - - Assumes the `allocated' (and perhaps `buffer') and `translate' - fields are set in BUFP on entry. - - If it succeeds, results are put in BUFP (if it returns an error, the - contents of BUFP are undefined): - `buffer' is the compiled pattern; - `syntax' is set to SYNTAX; - `used' is set to the length of the compiled pattern; - `fastmap_accurate' is zero; - `re_nsub' is the number of subexpressions in PATTERN; - `not_bol' and `not_eol' are zero; - - The `fastmap' and `newline_anchor' fields are neither - examined nor set. */ - -static reg_errcode_t -regex_compile (pattern, size, syntax, bufp) - const char *pattern; - int size; - reg_syntax_t syntax; - struct re_pattern_buffer *bufp; -{ - /* We fetch characters from PATTERN here. Even though PATTERN is - `char *' (i.e., signed), we declare these variables as unsigned, so - they can be reliably used as array indices. */ - register unsigned char c, c1; - - /* A random tempory spot in PATTERN. */ - const char *p1; - - /* Points to the end of the buffer, where we should append. */ - register unsigned char *b; - - /* Keeps track of unclosed groups. */ - compile_stack_type compile_stack; - - /* Points to the current (ending) position in the pattern. */ - const char *p = pattern; - const char *pend = pattern + size; - - /* How to translate the characters in the pattern. */ - char *translate = bufp->translate; - - /* Address of the count-byte of the most recently inserted `exactn' - command. This makes it possible to tell if a new exact-match - character can be added to that command or if the character requires - a new `exactn' command. */ - unsigned char *pending_exact = 0; - - /* Address of start of the most recently finished expression. - This tells, e.g., postfix * where to find the start of its - operand. Reset at the beginning of groups and alternatives. */ - unsigned char *laststart = 0; - - /* Address of beginning of regexp, or inside of last group. */ - unsigned char *begalt; - - /* Place in the uncompiled pattern (i.e., the {) to - which to go back if the interval is invalid. */ - const char *beg_interval; - - /* Address of the place where a forward jump should go to the end of - the containing expression. Each alternative of an `or' -- except the - last -- ends with a forward jump of this sort. */ - unsigned char *fixup_alt_jump = 0; - - /* Counts open-groups as they are encountered. Remembered for the - matching close-group on the compile stack, so the same register - number is put in the stop_memory as the start_memory. */ - regnum_t regnum = 0; - -#ifdef DEBUG - DEBUG_PRINT1 ("\nCompiling pattern: "); - if (debug) - { - unsigned debug_count; - - for (debug_count = 0; debug_count < size; debug_count++) - printchar (pattern[debug_count]); - putchar ('\n'); - } -#endif /* DEBUG */ - - /* Initialize the compile stack. */ - compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); - if (compile_stack.stack == NULL) - return REG_ESPACE; - - compile_stack.size = INIT_COMPILE_STACK_SIZE; - compile_stack.avail = 0; - - /* Initialize the pattern buffer. */ - bufp->syntax = syntax; - bufp->fastmap_accurate = 0; - bufp->not_bol = bufp->not_eol = 0; - - /* Set `used' to zero, so that if we return an error, the pattern - printer (for debugging) will think there's no pattern. We reset it - at the end. */ - bufp->used = 0; - - /* Always count groups, whether or not bufp->no_sub is set. */ - bufp->re_nsub = 0; - -#if !defined (emacs) && !defined (SYNTAX_TABLE) - /* Initialize the syntax table. */ - init_syntax_once (); -#endif - - if (bufp->allocated == 0) - { - if (bufp->buffer) - { /* If zero allocated, but buffer is non-null, try to realloc - enough space. This loses if buffer's address is bogus, but - that is the user's responsibility. */ - RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); - } - else - { /* Caller did not allocate a buffer. Do it for them. */ - bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); - } - if (!bufp->buffer) return REG_ESPACE; - - bufp->allocated = INIT_BUF_SIZE; - } - - begalt = b = bufp->buffer; - - /* Loop through the uncompiled pattern until we're at the end. */ - while (p != pend) - { - PATFETCH (c); - - switch (c) - { - case '^': - { - if ( /* If at start of pattern, it's an operator. */ - p == pattern + 1 - /* If context independent, it's an operator. */ - || syntax & RE_CONTEXT_INDEP_ANCHORS - /* Otherwise, depends on what's come before. */ - || at_begline_loc_p (pattern, p, syntax)) - BUF_PUSH (begline); - else - goto normal_char; - } - break; - - - case '$': - { - if ( /* If at end of pattern, it's an operator. */ - p == pend - /* If context independent, it's an operator. */ - || syntax & RE_CONTEXT_INDEP_ANCHORS - /* Otherwise, depends on what's next. */ - || at_endline_loc_p (p, pend, syntax)) - BUF_PUSH (endline); - else - goto normal_char; - } - break; - - - case '+': - case '?': - if ((syntax & RE_BK_PLUS_QM) - || (syntax & RE_LIMITED_OPS)) - goto normal_char; - handle_plus: - case '*': - /* If there is no previous pattern... */ - if (!laststart) - { - if (syntax & RE_CONTEXT_INVALID_OPS) - return REG_BADRPT; - else if (!(syntax & RE_CONTEXT_INDEP_OPS)) - goto normal_char; - } - - { - /* Are we optimizing this jump? */ - boolean keep_string_p = false; - - /* 1 means zero (many) matches is allowed. */ - char zero_times_ok = 0, many_times_ok = 0; - - /* If there is a sequence of repetition chars, collapse it - down to just one (the right one). We can't combine - interval operators with these because of, e.g., `a{2}*', - which should only match an even number of `a's. */ - - for (;;) - { - zero_times_ok |= c != '+'; - many_times_ok |= c != '?'; - - if (p == pend) - break; - - PATFETCH (c); - - if (c == '*' - || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) - ; - - else if (syntax & RE_BK_PLUS_QM && c == '\\') - { - if (p == pend) return REG_EESCAPE; - - PATFETCH (c1); - if (!(c1 == '+' || c1 == '?')) - { - PATUNFETCH; - PATUNFETCH; - break; - } - - c = c1; - } - else - { - PATUNFETCH; - break; - } - - /* If we get here, we found another repeat character. */ - } - - /* Star, etc. applied to an empty pattern is equivalent - to an empty pattern. */ - if (!laststart) - break; - - /* Now we know whether or not zero matches is allowed - and also whether or not two or more matches is allowed. */ - if (many_times_ok) - { /* More than one repetition is allowed, so put in at the - end a backward relative jump from `b' to before the next - jump we're going to put in below (which jumps from - laststart to after this jump). - - But if we are at the `*' in the exact sequence `.*\n', - insert an unconditional jump backwards to the ., - instead of the beginning of the loop. This way we only - push a failure point once, instead of every time - through the loop. */ - assert (p - 1 > pattern); - - /* Allocate the space for the jump. */ - GET_BUFFER_SPACE (3); - - /* We know we are not at the first character of the pattern, - because laststart was nonzero. And we've already - incremented `p', by the way, to be the character after - the `*'. Do we have to do something analogous here - for null bytes, because of RE_DOT_NOT_NULL? */ - if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') - && zero_times_ok - && p < pend && TRANSLATE (*p) == TRANSLATE ('\n') - && !(syntax & RE_DOT_NEWLINE)) - { /* We have .*\n. */ - STORE_JUMP (jump, b, laststart); - keep_string_p = true; - } - else - /* Anything else. */ - STORE_JUMP (maybe_pop_jump, b, laststart - 3); - - /* We've added more stuff to the buffer. */ - b += 3; - } - - /* On failure, jump from laststart to b + 3, which will be the - end of the buffer after this jump is inserted. */ - GET_BUFFER_SPACE (3); - INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump - : on_failure_jump, - laststart, b + 3); - pending_exact = 0; - b += 3; - - if (!zero_times_ok) - { - /* At least one repetition is required, so insert a - `dummy_failure_jump' before the initial - `on_failure_jump' instruction of the loop. This - effects a skip over that instruction the first time - we hit that loop. */ - GET_BUFFER_SPACE (3); - INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); - b += 3; - } - } - break; - - - case '.': - laststart = b; - BUF_PUSH (anychar); - break; - - - case '[': - { - boolean had_char_class = false; - - if (p == pend) return REG_EBRACK; - - /* Ensure that we have enough space to push a charset: the - opcode, the length count, and the bitset; 34 bytes in all. */ - GET_BUFFER_SPACE (34); - - laststart = b; - - /* We test `*p == '^' twice, instead of using an if - statement, so we only need one BUF_PUSH. */ - BUF_PUSH (*p == '^' ? charset_not : charset); - if (*p == '^') - p++; - - /* Remember the first position in the bracket expression. */ - p1 = p; - - /* Push the number of bytes in the bitmap. */ - BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); - - /* Clear the whole map. */ - bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); - - /* charset_not matches newline according to a syntax bit. */ - if ((re_opcode_t) b[-2] == charset_not - && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) - SET_LIST_BIT ('\n'); - - /* Read in characters and ranges, setting map bits. */ - for (;;) - { - if (p == pend) return REG_EBRACK; - - PATFETCH (c); - - /* \ might escape characters inside [...] and [^...]. */ - if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') - { - if (p == pend) return REG_EESCAPE; - - PATFETCH (c1); - SET_LIST_BIT (c1); - continue; - } - - /* Could be the end of the bracket expression. If it's - not (i.e., when the bracket expression is `[]' so - far), the ']' character bit gets set way below. */ - if (c == ']' && p != p1 + 1) - break; - - /* Look ahead to see if it's a range when the last thing - was a character class. */ - if (had_char_class && c == '-' && *p != ']') - return REG_ERANGE; - - /* Look ahead to see if it's a range when the last thing - was a character: if this is a hyphen not at the - beginning or the end of a list, then it's the range - operator. */ - if (c == '-' - && !(p - 2 >= pattern && p[-2] == '[') - && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') - && *p != ']') - { - reg_errcode_t ret - = compile_range (&p, pend, translate, syntax, b); - if (ret != REG_NOERROR) return ret; - } - - else if (p[0] == '-' && p[1] != ']') - { /* This handles ranges made up of characters only. */ - reg_errcode_t ret; - - /* Move past the `-'. */ - PATFETCH (c1); - - ret = compile_range (&p, pend, translate, syntax, b); - if (ret != REG_NOERROR) return ret; - } - - /* See if we're at the beginning of a possible character - class. */ - - else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') - { /* Leave room for the null. */ - char str[CHAR_CLASS_MAX_LENGTH + 1]; - - PATFETCH (c); - c1 = 0; - - /* If pattern is `[[:'. */ - if (p == pend) return REG_EBRACK; - - for (;;) - { - PATFETCH (c); - if (c == ':' || c == ']' || p == pend - || c1 == CHAR_CLASS_MAX_LENGTH) - break; - str[c1++] = c; - } - str[c1] = '\0'; - - /* If isn't a word bracketed by `[:' and:`]': - undo the ending character, the letters, and leave - the leading `:' and `[' (but set bits for them). */ - if (c == ':' && *p == ']') - { - int ch; - boolean is_alnum = STREQ (str, "alnum"); - boolean is_alpha = STREQ (str, "alpha"); - boolean is_blank = STREQ (str, "blank"); - boolean is_cntrl = STREQ (str, "cntrl"); - boolean is_digit = STREQ (str, "digit"); - boolean is_graph = STREQ (str, "graph"); - boolean is_lower = STREQ (str, "lower"); - boolean is_print = STREQ (str, "print"); - boolean is_punct = STREQ (str, "punct"); - boolean is_space = STREQ (str, "space"); - boolean is_upper = STREQ (str, "upper"); - boolean is_xdigit = STREQ (str, "xdigit"); - - if (!IS_CHAR_CLASS (str)) return REG_ECTYPE; - - /* Throw away the ] at the end of the character - class. */ - PATFETCH (c); - - if (p == pend) return REG_EBRACK; - - for (ch = 0; ch < 1 << BYTEWIDTH; ch++) - { - if ( (is_alnum && ISALNUM (ch)) - || (is_alpha && ISALPHA (ch)) - || (is_blank && ISBLANK (ch)) - || (is_cntrl && ISCNTRL (ch)) - || (is_digit && ISDIGIT (ch)) - || (is_graph && ISGRAPH (ch)) - || (is_lower && ISLOWER (ch)) - || (is_print && ISPRINT (ch)) - || (is_punct && ISPUNCT (ch)) - || (is_space && ISSPACE (ch)) - || (is_upper && ISUPPER (ch)) - || (is_xdigit && ISXDIGIT (ch))) - SET_LIST_BIT (ch); - } - had_char_class = true; - } - else - { - c1++; - while (c1--) - PATUNFETCH; - SET_LIST_BIT ('['); - SET_LIST_BIT (':'); - had_char_class = false; - } - } - else - { - had_char_class = false; - SET_LIST_BIT (c); - } - } - - /* Discard any (non)matching list bytes that are all 0 at the - end of the map. Decrease the map-length byte too. */ - while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) - b[-1]--; - b += b[-1]; - } - break; - - - case '(': - if (syntax & RE_NO_BK_PARENS) - goto handle_open; - else - goto normal_char; - - - case ')': - if (syntax & RE_NO_BK_PARENS) - goto handle_close; - else - goto normal_char; - - - case '\n': - if (syntax & RE_NEWLINE_ALT) - goto handle_alt; - else - goto normal_char; - - - case '|': - if (syntax & RE_NO_BK_VBAR) - goto handle_alt; - else - goto normal_char; - - - case '{': - if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) - goto handle_interval; - else - goto normal_char; - - - case '\\': - if (p == pend) return REG_EESCAPE; - - /* Do not translate the character after the \, so that we can - distinguish, e.g., \B from \b, even if we normally would - translate, e.g., B to b. */ - PATFETCH_RAW (c); - - switch (c) - { - case '(': - if (syntax & RE_NO_BK_PARENS) - goto normal_backslash; - - handle_open: - bufp->re_nsub++; - regnum++; - - if (COMPILE_STACK_FULL) - { - RETALLOC (compile_stack.stack, compile_stack.size << 1, - compile_stack_elt_t); - if (compile_stack.stack == NULL) return REG_ESPACE; - - compile_stack.size <<= 1; - } - - /* These are the values to restore when we hit end of this - group. They are all relative offsets, so that if the - whole pattern moves because of realloc, they will still - be valid. */ - COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; - COMPILE_STACK_TOP.fixup_alt_jump - = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; - COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; - COMPILE_STACK_TOP.regnum = regnum; - - /* We will eventually replace the 0 with the number of - groups inner to this one. But do not push a - start_memory for groups beyond the last one we can - represent in the compiled pattern. */ - if (regnum <= MAX_REGNUM) - { - COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; - BUF_PUSH_3 (start_memory, regnum, 0); - } - - compile_stack.avail++; - - fixup_alt_jump = 0; - laststart = 0; - begalt = b; - /* If we've reached MAX_REGNUM groups, then this open - won't actually generate any code, so we'll have to - clear pending_exact explicitly. */ - pending_exact = 0; - break; - - - case ')': - if (syntax & RE_NO_BK_PARENS) goto normal_backslash; - - if (COMPILE_STACK_EMPTY) - if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) - goto normal_backslash; - else - return REG_ERPAREN; - - handle_close: - if (fixup_alt_jump) - { /* Push a dummy failure point at the end of the - alternative for a possible future - `pop_failure_jump' to pop. See comments at - `push_dummy_failure' in `re_match_2'. */ - BUF_PUSH (push_dummy_failure); - - /* We allocated space for this jump when we assigned - to `fixup_alt_jump', in the `handle_alt' case below. */ - STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); - } - - /* See similar code for backslashed left paren above. */ - if (COMPILE_STACK_EMPTY) - if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) - goto normal_char; - else - return REG_ERPAREN; - - /* Since we just checked for an empty stack above, this - ``can't happen''. */ - assert (compile_stack.avail != 0); - { - /* We don't just want to restore into `regnum', because - later groups should continue to be numbered higher, - as in `(ab)c(de)' -- the second group is #2. */ - regnum_t this_group_regnum; - - compile_stack.avail--; - begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; - fixup_alt_jump - = COMPILE_STACK_TOP.fixup_alt_jump - ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 - : 0; - laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; - this_group_regnum = COMPILE_STACK_TOP.regnum; - /* If we've reached MAX_REGNUM groups, then this open - won't actually generate any code, so we'll have to - clear pending_exact explicitly. */ - pending_exact = 0; - - /* We're at the end of the group, so now we know how many - groups were inside this one. */ - if (this_group_regnum <= MAX_REGNUM) - { - unsigned char *inner_group_loc - = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; - - *inner_group_loc = regnum - this_group_regnum; - BUF_PUSH_3 (stop_memory, this_group_regnum, - regnum - this_group_regnum); - } - } - break; - - - case '|': /* `\|'. */ - if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) - goto normal_backslash; - handle_alt: - if (syntax & RE_LIMITED_OPS) - goto normal_char; - - /* Insert before the previous alternative a jump which - jumps to this alternative if the former fails. */ - GET_BUFFER_SPACE (3); - INSERT_JUMP (on_failure_jump, begalt, b + 6); - pending_exact = 0; - b += 3; - - /* The alternative before this one has a jump after it - which gets executed if it gets matched. Adjust that - jump so it will jump to this alternative's analogous - jump (put in below, which in turn will jump to the next - (if any) alternative's such jump, etc.). The last such - jump jumps to the correct final destination. A picture: - _____ _____ - | | | | - | v | v - a | b | c - - If we are at `b', then fixup_alt_jump right now points to a - three-byte space after `a'. We'll put in the jump, set - fixup_alt_jump to right after `b', and leave behind three - bytes which we'll fill in when we get to after `c'. */ - - if (fixup_alt_jump) - STORE_JUMP (jump_past_alt, fixup_alt_jump, b); - - /* Mark and leave space for a jump after this alternative, - to be filled in later either by next alternative or - when know we're at the end of a series of alternatives. */ - fixup_alt_jump = b; - GET_BUFFER_SPACE (3); - b += 3; - - laststart = 0; - begalt = b; - break; - - - case '{': - /* If \{ is a literal. */ - if (!(syntax & RE_INTERVALS) - /* If we're at `\{' and it's not the open-interval - operator. */ - || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) - || (p - 2 == pattern && p == pend)) - goto normal_backslash; - - handle_interval: - { - /* If got here, then the syntax allows intervals. */ - - /* At least (most) this many matches must be made. */ - int lower_bound = -1, upper_bound = -1; - - beg_interval = p - 1; - - if (p == pend) - { - if (syntax & RE_NO_BK_BRACES) - goto unfetch_interval; - else - return REG_EBRACE; - } - - GET_UNSIGNED_NUMBER (lower_bound); - - if (c == ',') - { - GET_UNSIGNED_NUMBER (upper_bound); - if (upper_bound < 0) upper_bound = RE_DUP_MAX; - } - else - /* Interval such as `{1}' => match exactly once. */ - upper_bound = lower_bound; - - if (lower_bound < 0 || upper_bound > RE_DUP_MAX - || lower_bound > upper_bound) - { - if (syntax & RE_NO_BK_BRACES) - goto unfetch_interval; - else - return REG_BADBR; - } - - if (!(syntax & RE_NO_BK_BRACES)) - { - if (c != '\\') return REG_EBRACE; - - PATFETCH (c); - } - - if (c != '}') - { - if (syntax & RE_NO_BK_BRACES) - goto unfetch_interval; - else - return REG_BADBR; - } - - /* We just parsed a valid interval. */ - - /* If it's invalid to have no preceding re. */ - if (!laststart) - { - if (syntax & RE_CONTEXT_INVALID_OPS) - return REG_BADRPT; - else if (syntax & RE_CONTEXT_INDEP_OPS) - laststart = b; - else - goto unfetch_interval; - } - - /* If the upper bound is zero, don't want to succeed at - all; jump from `laststart' to `b + 3', which will be - the end of the buffer after we insert the jump. */ - if (upper_bound == 0) - { - GET_BUFFER_SPACE (3); - INSERT_JUMP (jump, laststart, b + 3); - b += 3; - } - - /* Otherwise, we have a nontrivial interval. When - we're all done, the pattern will look like: - set_number_at - set_number_at - succeed_n - - jump_n - (The upper bound and `jump_n' are omitted if - `upper_bound' is 1, though.) */ - else - { /* If the upper bound is > 1, we need to insert - more at the end of the loop. */ - unsigned nbytes = 10 + (upper_bound > 1) * 10; - - GET_BUFFER_SPACE (nbytes); - - /* Initialize lower bound of the `succeed_n', even - though it will be set during matching by its - attendant `set_number_at' (inserted next), - because `re_compile_fastmap' needs to know. - Jump to the `jump_n' we might insert below. */ - INSERT_JUMP2 (succeed_n, laststart, - b + 5 + (upper_bound > 1) * 5, - lower_bound); - b += 5; - - /* Code to initialize the lower bound. Insert - before the `succeed_n'. The `5' is the last two - bytes of this `set_number_at', plus 3 bytes of - the following `succeed_n'. */ - insert_op2 (set_number_at, laststart, 5, lower_bound, b); - b += 5; - - if (upper_bound > 1) - { /* More than one repetition is allowed, so - append a backward jump to the `succeed_n' - that starts this interval. - - When we've reached this during matching, - we'll have matched the interval once, so - jump back only `upper_bound - 1' times. */ - STORE_JUMP2 (jump_n, b, laststart + 5, - upper_bound - 1); - b += 5; - - /* The location we want to set is the second - parameter of the `jump_n'; that is `b-2' as - an absolute address. `laststart' will be - the `set_number_at' we're about to insert; - `laststart+3' the number to set, the source - for the relative address. But we are - inserting into the middle of the pattern -- - so everything is getting moved up by 5. - Conclusion: (b - 2) - (laststart + 3) + 5, - i.e., b - laststart. - - We insert this at the beginning of the loop - so that if we fail during matching, we'll - reinitialize the bounds. */ - insert_op2 (set_number_at, laststart, b - laststart, - upper_bound - 1, b); - b += 5; - } - } - pending_exact = 0; - beg_interval = NULL; - } - break; - - unfetch_interval: - /* If an invalid interval, match the characters as literals. */ - assert (beg_interval); - p = beg_interval; - beg_interval = NULL; - - /* normal_char and normal_backslash need `c'. */ - PATFETCH (c); - - if (!(syntax & RE_NO_BK_BRACES)) - { - if (p > pattern && p[-1] == '\\') - goto normal_backslash; - } - goto normal_char; - -#ifdef emacs - /* There is no way to specify the before_dot and after_dot - operators. rms says this is ok. --karl */ - case '=': - BUF_PUSH (at_dot); - break; - - case 's': - laststart = b; - PATFETCH (c); - BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); - break; - - case 'S': - laststart = b; - PATFETCH (c); - BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); - break; -#endif /* emacs */ - - - case 'w': - laststart = b; - BUF_PUSH (wordchar); - break; - - - case 'W': - laststart = b; - BUF_PUSH (notwordchar); - break; - - - case '<': - BUF_PUSH (wordbeg); - break; - - case '>': - BUF_PUSH (wordend); - break; - - case 'b': - BUF_PUSH (wordbound); - break; - - case 'B': - BUF_PUSH (notwordbound); - break; - - case '`': - BUF_PUSH (begbuf); - break; - - case '\'': - BUF_PUSH (endbuf); - break; - - case '1': case '2': case '3': case '4': case '5': - case '6': case '7': case '8': case '9': - if (syntax & RE_NO_BK_REFS) - goto normal_char; - - c1 = c - '0'; - - if (c1 > regnum) - return REG_ESUBREG; - - /* Can't back reference to a subexpression if inside of it. */ - if (group_in_compile_stack (compile_stack, c1)) - goto normal_char; - - laststart = b; - BUF_PUSH_2 (duplicate, c1); - break; - - - case '+': - case '?': - if (syntax & RE_BK_PLUS_QM) - goto handle_plus; - else - goto normal_backslash; - - default: - normal_backslash: - /* You might think it would be useful for \ to mean - not to translate; but if we don't translate it - it will never match anything. */ - c = TRANSLATE (c); - goto normal_char; - } - break; - - - default: - /* Expects the character in `c'. */ - normal_char: - /* If no exactn currently being built. */ - if (!pending_exact - - /* If last exactn not at current position. */ - || pending_exact + *pending_exact + 1 != b - - /* We have only one byte following the exactn for the count. */ - || *pending_exact == (1 << BYTEWIDTH) - 1 - - /* If followed by a repetition operator. */ - || *p == '*' || *p == '^' - || ((syntax & RE_BK_PLUS_QM) - ? *p == '\\' && (p[1] == '+' || p[1] == '?') - : (*p == '+' || *p == '?')) - || ((syntax & RE_INTERVALS) - && ((syntax & RE_NO_BK_BRACES) - ? *p == '{' - : (p[0] == '\\' && p[1] == '{')))) - { - /* Start building a new exactn. */ - - laststart = b; - - BUF_PUSH_2 (exactn, 0); - pending_exact = b - 1; - } - - BUF_PUSH (c); - (*pending_exact)++; - break; - } /* switch (c) */ - } /* while p != pend */ - - - /* Through the pattern now. */ - - if (fixup_alt_jump) - STORE_JUMP (jump_past_alt, fixup_alt_jump, b); - - if (!COMPILE_STACK_EMPTY) - return REG_EPAREN; - - free (compile_stack.stack); - - /* We have succeeded; set the length of the buffer. */ - bufp->used = b - bufp->buffer; - -#ifdef DEBUG - if (debug) - { - DEBUG_PRINT1 ("\nCompiled pattern: "); - print_compiled_pattern (bufp); - } -#endif /* DEBUG */ - - return REG_NOERROR; -} /* regex_compile */ - -/* Subroutines for `regex_compile'. */ - -/* Store OP at LOC followed by two-byte integer parameter ARG. */ - -static void -store_op1 (op, loc, arg) - re_opcode_t op; - unsigned char *loc; - int arg; -{ - *loc = (unsigned char) op; - STORE_NUMBER (loc + 1, arg); -} - - -/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ - -static void -store_op2 (op, loc, arg1, arg2) - re_opcode_t op; - unsigned char *loc; - int arg1, arg2; -{ - *loc = (unsigned char) op; - STORE_NUMBER (loc + 1, arg1); - STORE_NUMBER (loc + 3, arg2); -} - - -/* Copy the bytes from LOC to END to open up three bytes of space at LOC - for OP followed by two-byte integer parameter ARG. */ - -static void -insert_op1 (op, loc, arg, end) - re_opcode_t op; - unsigned char *loc; - int arg; - unsigned char *end; -{ - register unsigned char *pfrom = end; - register unsigned char *pto = end + 3; - - while (pfrom != loc) - *--pto = *--pfrom; - - store_op1 (op, loc, arg); -} - - -/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ - -static void -insert_op2 (op, loc, arg1, arg2, end) - re_opcode_t op; - unsigned char *loc; - int arg1, arg2; - unsigned char *end; -{ - register unsigned char *pfrom = end; - register unsigned char *pto = end + 5; - - while (pfrom != loc) - *--pto = *--pfrom; - - store_op2 (op, loc, arg1, arg2); -} - - -/* P points to just after a ^ in PATTERN. Return true if that ^ comes - after an alternative or a begin-subexpression. We assume there is at - least one character before the ^. */ - -static boolean -at_begline_loc_p (pattern, p, syntax) - const char *pattern, *p; - reg_syntax_t syntax; -{ - const char *prev = p - 2; - boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; - - return - /* After a subexpression? */ - (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) - /* After an alternative? */ - || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); -} - - -/* The dual of at_begline_loc_p. This one is for $. We assume there is - at least one character after the $, i.e., `P < PEND'. */ - -static boolean -at_endline_loc_p (p, pend, syntax) - const char *p, *pend; - int syntax; -{ - const char *next = p; - boolean next_backslash = *next == '\\'; - const char *next_next = p + 1 < pend ? p + 1 : NULL; - - return - /* Before a subexpression? */ - (syntax & RE_NO_BK_PARENS ? *next == ')' - : next_backslash && next_next && *next_next == ')') - /* Before an alternative? */ - || (syntax & RE_NO_BK_VBAR ? *next == '|' - : next_backslash && next_next && *next_next == '|'); -} - - -/* Returns true if REGNUM is in one of COMPILE_STACK's elements and - false if it's not. */ - -static boolean -group_in_compile_stack (compile_stack, regnum) - compile_stack_type compile_stack; - regnum_t regnum; -{ - int this_element; - - for (this_element = compile_stack.avail - 1; - this_element >= 0; - this_element--) - if (compile_stack.stack[this_element].regnum == regnum) - return true; - - return false; -} - - -/* Read the ending character of a range (in a bracket expression) from the - uncompiled pattern *P_PTR (which ends at PEND). We assume the - starting character is in `P[-2]'. (`P[-1]' is the character `-'.) - Then we set the translation of all bits between the starting and - ending characters (inclusive) in the compiled pattern B. - - Return an error code. - - We use these short variable names so we can use the same macros as - `regex_compile' itself. */ - -static reg_errcode_t -compile_range (p_ptr, pend, translate, syntax, b) - const char **p_ptr, *pend; - char *translate; - reg_syntax_t syntax; - unsigned char *b; -{ - unsigned this_char; - - const char *p = *p_ptr; - int range_start, range_end; - - if (p == pend) - return REG_ERANGE; - - /* Even though the pattern is a signed `char *', we need to fetch - with unsigned char *'s; if the high bit of the pattern character - is set, the range endpoints will be negative if we fetch using a - signed char *. - - We also want to fetch the endpoints without translating them; the - appropriate translation is done in the bit-setting loop below. */ - range_start = ((unsigned char *) p)[-2]; - range_end = ((unsigned char *) p)[0]; - - /* Have to increment the pointer into the pattern string, so the - caller isn't still at the ending character. */ - (*p_ptr)++; - - /* If the start is after the end, the range is empty. */ - if (range_start > range_end) - return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; - - /* Here we see why `this_char' has to be larger than an `unsigned - char' -- the range is inclusive, so if `range_end' == 0xff - (assuming 8-bit characters), we would otherwise go into an infinite - loop, since all characters <= 0xff. */ - for (this_char = range_start; this_char <= range_end; this_char++) - { - SET_LIST_BIT (TRANSLATE (this_char)); - } - - return REG_NOERROR; -} - -/* Failure stack declarations and macros; both re_compile_fastmap and - re_match_2 use a failure stack. These have to be macros because of - REGEX_ALLOCATE. */ - - -/* Number of failure points for which to initially allocate space - when matching. If this number is exceeded, we allocate more - space, so it is not a hard limit. */ -#ifndef INIT_FAILURE_ALLOC -#define INIT_FAILURE_ALLOC 5 -#endif - -/* Roughly the maximum number of failure points on the stack. Would be - exactly that if always used MAX_FAILURE_SPACE each time we failed. - This is a variable only so users of regex can assign to it; we never - change it ourselves. */ -int re_max_failures = 2000; - -typedef const unsigned char *fail_stack_elt_t; - -typedef struct -{ - fail_stack_elt_t *stack; - unsigned size; - unsigned avail; /* Offset of next open position. */ -} fail_stack_type; - -#define FAIL_STACK_EMPTY() (fail_stack.avail == 0) -#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) -#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) -#define FAIL_STACK_TOP() (fail_stack.stack[fail_stack.avail]) - - -/* Initialize `fail_stack'. Do `return -2' if the alloc fails. */ - -#define INIT_FAIL_STACK() \ - do { \ - fail_stack.stack = (fail_stack_elt_t *) \ - REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \ - \ - if (fail_stack.stack == NULL) \ - return -2; \ - \ - fail_stack.size = INIT_FAILURE_ALLOC; \ - fail_stack.avail = 0; \ - } while (0) - - -/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. - - Return 1 if succeeds, and 0 if either ran out of memory - allocating space for it or it was already too large. - - REGEX_REALLOCATE requires `destination' be declared. */ - -#define DOUBLE_FAIL_STACK(fail_stack) \ - ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ - ? 0 \ - : ((fail_stack).stack = (fail_stack_elt_t *) \ - REGEX_REALLOCATE ((fail_stack).stack, \ - (fail_stack).size * sizeof (fail_stack_elt_t), \ - ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ - \ - (fail_stack).stack == NULL \ - ? 0 \ - : ((fail_stack).size <<= 1, \ - 1))) - - -/* Push PATTERN_OP on FAIL_STACK. - - Return 1 if was able to do so and 0 if ran out of memory allocating - space to do so. */ -#define PUSH_PATTERN_OP(pattern_op, fail_stack) \ - ((FAIL_STACK_FULL () \ - && !DOUBLE_FAIL_STACK (fail_stack)) \ - ? 0 \ - : ((fail_stack).stack[(fail_stack).avail++] = pattern_op, \ - 1)) - -/* This pushes an item onto the failure stack. Must be a four-byte - value. Assumes the variable `fail_stack'. Probably should only - be called from within `PUSH_FAILURE_POINT'. */ -#define PUSH_FAILURE_ITEM(item) \ - fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item - -/* The complement operation. Assumes `fail_stack' is nonempty. */ -#define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail] - -/* Used to omit pushing failure point id's when we're not debugging. */ -#ifdef DEBUG -#define DEBUG_PUSH PUSH_FAILURE_ITEM -#define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM () -#else -#define DEBUG_PUSH(item) -#define DEBUG_POP(item_addr) -#endif - - -/* Push the information about the state we will need - if we ever fail back to it. - - Requires variables fail_stack, regstart, regend, reg_info, and - num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be - declared. - - Does `return FAILURE_CODE' if runs out of memory. */ - -#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ - do { \ - char *destination; \ - /* Must be int, so when we don't save any registers, the arithmetic \ - of 0 + -1 isn't done as unsigned. */ \ - int this_reg; \ - \ - DEBUG_STATEMENT (failure_id++); \ - DEBUG_STATEMENT (nfailure_points_pushed++); \ - DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ - DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ - DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ - \ - DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ - DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ - \ - /* Ensure we have enough space allocated for what we will push. */ \ - while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ - { \ - if (!DOUBLE_FAIL_STACK (fail_stack)) \ - return failure_code; \ - \ - DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ - (fail_stack).size); \ - DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ - } \ - \ - /* Push the info, starting with the registers. */ \ - DEBUG_PRINT1 ("\n"); \ - \ - for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ - this_reg++) \ - { \ - DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \ - DEBUG_STATEMENT (num_regs_pushed++); \ - \ - DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ - PUSH_FAILURE_ITEM (regstart[this_reg]); \ - \ - DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ - PUSH_FAILURE_ITEM (regend[this_reg]); \ - \ - DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \ - DEBUG_PRINT2 (" match_null=%d", \ - REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ - DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ - DEBUG_PRINT2 (" matched_something=%d", \ - MATCHED_SOMETHING (reg_info[this_reg])); \ - DEBUG_PRINT2 (" ever_matched=%d", \ - EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ - DEBUG_PRINT1 ("\n"); \ - PUSH_FAILURE_ITEM (reg_info[this_reg].word); \ - } \ - \ - DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\ - PUSH_FAILURE_ITEM (lowest_active_reg); \ - \ - DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\ - PUSH_FAILURE_ITEM (highest_active_reg); \ - \ - DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \ - DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ - PUSH_FAILURE_ITEM (pattern_place); \ - \ - DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \ - DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ - size2); \ - DEBUG_PRINT1 ("'\n"); \ - PUSH_FAILURE_ITEM (string_place); \ - \ - DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ - DEBUG_PUSH (failure_id); \ - } while (0) - -/* This is the number of items that are pushed and popped on the stack - for each register. */ -#define NUM_REG_ITEMS 3 - -/* Individual items aside from the registers. */ -#ifdef DEBUG -#define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ -#else -#define NUM_NONREG_ITEMS 4 -#endif - -/* We push at most this many items on the stack. */ -#define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS) - -/* We actually push this many items. */ -#define NUM_FAILURE_ITEMS \ - ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \ - + NUM_NONREG_ITEMS) - -/* How many items can still be added to the stack without overflowing it. */ -#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) - - -/* Pops what PUSH_FAIL_STACK pushes. - - We restore into the parameters, all of which should be lvalues: - STR -- the saved data position. - PAT -- the saved pattern position. - LOW_REG, HIGH_REG -- the highest and lowest active registers. - REGSTART, REGEND -- arrays of string positions. - REG_INFO -- array of information about each subexpression. - - Also assumes the variables `fail_stack' and (if debugging), `bufp', - `pend', `string1', `size1', `string2', and `size2'. */ - -#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ -{ \ - DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \ - int this_reg; \ - const unsigned char *string_temp; \ - \ - assert (!FAIL_STACK_EMPTY ()); \ - \ - /* Remove failure points and point to how many regs pushed. */ \ - DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ - DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ - DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ - \ - assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ - \ - DEBUG_POP (&failure_id); \ - DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ - \ - /* If the saved string location is NULL, it came from an \ - on_failure_keep_string_jump opcode, and we want to throw away the \ - saved NULL, thus retaining our current position in the string. */ \ - string_temp = POP_FAILURE_ITEM (); \ - if (string_temp != NULL) \ - str = (const char *) string_temp; \ - \ - DEBUG_PRINT2 (" Popping string 0x%x: `", str); \ - DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ - DEBUG_PRINT1 ("'\n"); \ - \ - pat = (unsigned char *) POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \ - DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ - \ - /* Restore register info. */ \ - high_reg = (unsigned) POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \ - \ - low_reg = (unsigned) POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ - \ - for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ - { \ - DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ - \ - reg_info[this_reg].word = POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \ - \ - regend[this_reg] = (const char *) POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ - \ - regstart[this_reg] = (const char *) POP_FAILURE_ITEM (); \ - DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ - } \ - \ - DEBUG_STATEMENT (nfailure_points_popped++); \ -} /* POP_FAILURE_POINT */ - -/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in - BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible - characters can start a string that matches the pattern. This fastmap - is used by re_search to skip quickly over impossible starting points. - - The caller must supply the address of a (1 << BYTEWIDTH)-byte data - area as BUFP->fastmap. - - We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in - the pattern buffer. - - Returns 0 if we succeed, -2 if an internal error. */ - -int -re_compile_fastmap (bufp) - struct re_pattern_buffer *bufp; -{ - int j, k; - fail_stack_type fail_stack; -#ifndef REGEX_MALLOC - char *destination; -#endif - /* We don't push any register information onto the failure stack. */ - unsigned num_regs = 0; - - register char *fastmap = bufp->fastmap; - unsigned char *pattern = bufp->buffer; - unsigned long size = bufp->used; - const unsigned char *p = pattern; - register unsigned char *pend = pattern + size; - - /* Assume that each path through the pattern can be null until - proven otherwise. We set this false at the bottom of switch - statement, to which we get only if a particular path doesn't - match the empty string. */ - boolean path_can_be_null = true; - - /* We aren't doing a `succeed_n' to begin with. */ - boolean succeed_n_p = false; - - assert (fastmap != NULL && p != NULL); - - INIT_FAIL_STACK (); - bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ - bufp->fastmap_accurate = 1; /* It will be when we're done. */ - bufp->can_be_null = 0; - - while (p != pend || !FAIL_STACK_EMPTY ()) - { - if (p == pend) - { - bufp->can_be_null |= path_can_be_null; - - /* Reset for next path. */ - path_can_be_null = true; - - p = fail_stack.stack[--fail_stack.avail]; - } - - /* We should never be about to go beyond the end of the pattern. */ - assert (p < pend); - -#ifdef SWITCH_ENUM_BUG - switch ((int) ((re_opcode_t) *p++)) -#else - switch ((re_opcode_t) *p++) -#endif - { - - /* I guess the idea here is to simply not bother with a fastmap - if a backreference is used, since it's too hard to figure out - the fastmap for the corresponding group. Setting - `can_be_null' stops `re_search_2' from using the fastmap, so - that is all we do. */ - case duplicate: - bufp->can_be_null = 1; - return 0; - - - /* Following are the cases which match a character. These end - with `break'. */ - - case exactn: - fastmap[p[1]] = 1; - break; - - - case charset: - for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) - if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) - fastmap[j] = 1; - break; - - - case charset_not: - /* Chars beyond end of map must be allowed. */ - for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) - fastmap[j] = 1; - - for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) - if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) - fastmap[j] = 1; - break; - - - case wordchar: - for (j = 0; j < (1 << BYTEWIDTH); j++) - if (SYNTAX (j) == Sword) - fastmap[j] = 1; - break; - - - case notwordchar: - for (j = 0; j < (1 << BYTEWIDTH); j++) - if (SYNTAX (j) != Sword) - fastmap[j] = 1; - break; - - - case anychar: - /* `.' matches anything ... */ - for (j = 0; j < (1 << BYTEWIDTH); j++) - fastmap[j] = 1; - - /* ... except perhaps newline. */ - if (!(bufp->syntax & RE_DOT_NEWLINE)) - fastmap['\n'] = 0; - - /* Return if we have already set `can_be_null'; if we have, - then the fastmap is irrelevant. Something's wrong here. */ - else if (bufp->can_be_null) - return 0; - - /* Otherwise, have to check alternative paths. */ - break; - - -#ifdef emacs - case syntaxspec: - k = *p++; - for (j = 0; j < (1 << BYTEWIDTH); j++) - if (SYNTAX (j) == (enum syntaxcode) k) - fastmap[j] = 1; - break; - - - case notsyntaxspec: - k = *p++; - for (j = 0; j < (1 << BYTEWIDTH); j++) - if (SYNTAX (j) != (enum syntaxcode) k) - fastmap[j] = 1; - break; - - - /* All cases after this match the empty string. These end with - `continue'. */ - - - case before_dot: - case at_dot: - case after_dot: - continue; -#endif /* not emacs */ - - - case no_op: - case begline: - case endline: - case begbuf: - case endbuf: - case wordbound: - case notwordbound: - case wordbeg: - case wordend: - case push_dummy_failure: - continue; - - - case jump_n: - case pop_failure_jump: - case maybe_pop_jump: - case jump: - case jump_past_alt: - case dummy_failure_jump: - EXTRACT_NUMBER_AND_INCR (j, p); - p += j; - if (j > 0) - continue; - - /* Jump backward implies we just went through the body of a - loop and matched nothing. Opcode jumped to should be - `on_failure_jump' or `succeed_n'. Just treat it like an - ordinary jump. For a * loop, it has pushed its failure - point already; if so, discard that as redundant. */ - if ((re_opcode_t) *p != on_failure_jump - && (re_opcode_t) *p != succeed_n) - continue; - - p++; - EXTRACT_NUMBER_AND_INCR (j, p); - p += j; - - /* If what's on the stack is where we are now, pop it. */ - if (!FAIL_STACK_EMPTY () - && fail_stack.stack[fail_stack.avail - 1] == p) - fail_stack.avail--; - - continue; - - - case on_failure_jump: - case on_failure_keep_string_jump: - handle_on_failure_jump: - EXTRACT_NUMBER_AND_INCR (j, p); - - /* For some patterns, e.g., `(a?)?', `p+j' here points to the - end of the pattern. We don't want to push such a point, - since when we restore it above, entering the switch will - increment `p' past the end of the pattern. We don't need - to push such a point since we obviously won't find any more - fastmap entries beyond `pend'. Such a pattern can match - the null string, though. */ - if (p + j < pend) - { - if (!PUSH_PATTERN_OP (p + j, fail_stack)) - return -2; - } - else - bufp->can_be_null = 1; - - if (succeed_n_p) - { - EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ - succeed_n_p = false; - } - - continue; - - - case succeed_n: - /* Get to the number of times to succeed. */ - p += 2; - - /* Increment p past the n for when k != 0. */ - EXTRACT_NUMBER_AND_INCR (k, p); - if (k == 0) - { - p -= 4; - succeed_n_p = true; /* Spaghetti code alert. */ - goto handle_on_failure_jump; - } - continue; - - - case set_number_at: - p += 4; - continue; - - - case start_memory: - case stop_memory: - p += 2; - continue; - - - default: - abort (); /* We have listed all the cases. */ - } /* switch *p++ */ - - /* Getting here means we have found the possible starting - characters for one path of the pattern -- and that the empty - string does not match. We need not follow this path further. - Instead, look at the next alternative (remembered on the - stack), or quit if no more. The test at the top of the loop - does these things. */ - path_can_be_null = false; - p = pend; - } /* while p */ - - /* Set `can_be_null' for the last path (also the first path, if the - pattern is empty). */ - bufp->can_be_null |= path_can_be_null; - return 0; -} /* re_compile_fastmap */ - -/* Set REGS to hold NUM_REGS registers, storing them in STARTS and - ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use - this memory for recording register information. STARTS and ENDS - must be allocated using the malloc library routine, and must each - be at least NUM_REGS * sizeof (regoff_t) bytes long. - - If NUM_REGS == 0, then subsequent matches should allocate their own - register data. - - Unless this function is called, the first search or match using - PATTERN_BUFFER will allocate its own register data, without - freeing the old data. */ - -void -re_set_registers (bufp, regs, num_regs, starts, ends) - struct re_pattern_buffer *bufp; - struct re_registers *regs; - unsigned num_regs; - regoff_t *starts, *ends; -{ - if (num_regs) - { - bufp->regs_allocated = REGS_REALLOCATE; - regs->num_regs = num_regs; - regs->start = starts; - regs->end = ends; - } - else - { - bufp->regs_allocated = REGS_UNALLOCATED; - regs->num_regs = 0; - regs->start = regs->end = (regoff_t) 0; - } -} - -/* Searching routines. */ - -/* Like re_search_2, below, but only one string is specified, and - doesn't let you say where to stop matching. */ - -int -re_search (bufp, string, size, startpos, range, regs) - struct re_pattern_buffer *bufp; - const char *string; - int size, startpos, range; - struct re_registers *regs; -{ - return re_search_2 (bufp, NULL, 0, string, size, startpos, range, - regs, size); -} - - -/* Using the compiled pattern in BUFP->buffer, first tries to match the - virtual concatenation of STRING1 and STRING2, starting first at index - STARTPOS, then at STARTPOS + 1, and so on. - - STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. - - RANGE is how far to scan while trying to match. RANGE = 0 means try - only at STARTPOS; in general, the last start tried is STARTPOS + - RANGE. - - In REGS, return the indices of the virtual concatenation of STRING1 - and STRING2 that matched the entire BUFP->buffer and its contained - subexpressions. - - Do not consider matching one past the index STOP in the virtual - concatenation of STRING1 and STRING2. - - We return either the position in the strings at which the match was - found, -1 if no match, or -2 if error (such as failure - stack overflow). */ - -int -re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop) - struct re_pattern_buffer *bufp; - const char *string1, *string2; - int size1, size2; - int startpos; - int range; - struct re_registers *regs; - int stop; -{ - int val; - register char *fastmap = bufp->fastmap; - register char *translate = bufp->translate; - int total_size = size1 + size2; - int endpos = startpos + range; - - /* Check for out-of-range STARTPOS. */ - if (startpos < 0 || startpos > total_size) - return -1; - - /* Fix up RANGE if it might eventually take us outside - the virtual concatenation of STRING1 and STRING2. */ - if (endpos < -1) - range = -1 - startpos; - else if (endpos > total_size) - range = total_size - startpos; - - /* If the search isn't to be a backwards one, don't waste time in a - search for a pattern that must be anchored. */ - if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0) - { - if (startpos > 0) - return -1; - else - range = 1; - } - - /* Update the fastmap now if not correct already. */ - if (fastmap && !bufp->fastmap_accurate) - if (re_compile_fastmap (bufp) == -2) - return -2; - - /* Loop through the string, looking for a place to start matching. */ - for (;;) - { - /* If a fastmap is supplied, skip quickly over characters that - cannot be the start of a match. If the pattern can match the - null string, however, we don't need to skip characters; we want - the first null string. */ - if (fastmap && startpos < total_size && !bufp->can_be_null) - { - if (range > 0) /* Searching forwards. */ - { - register const char *d; - register int lim = 0; - int irange = range; - - if (startpos < size1 && startpos + range >= size1) - lim = range - (size1 - startpos); - - d = (startpos >= size1 ? string2 - size1 : string1) + startpos; - - /* Written out as an if-else to avoid testing `translate' - inside the loop. */ - if (translate) - while (range > lim - && !fastmap[(unsigned char) - translate[(unsigned char) *d++]]) - range--; - else - while (range > lim && !fastmap[(unsigned char) *d++]) - range--; - - startpos += irange - range; - } - else /* Searching backwards. */ - { - register char c = (size1 == 0 || startpos >= size1 - ? string2[startpos - size1] - : string1[startpos]); - - if (!fastmap[(unsigned char) TRANSLATE (c)]) - goto advance; - } - } - - /* If can't match the null string, and that's all we have left, fail. */ - if (range >= 0 && startpos == total_size && fastmap - && !bufp->can_be_null) - return -1; - - val = re_match_2 (bufp, string1, size1, string2, size2, - startpos, regs, stop); - if (val >= 0) - return startpos; - - if (val == -2) - return -2; - - advance: - if (!range) - break; - else if (range > 0) - { - range--; - startpos++; - } - else - { - range++; - startpos--; - } - } - return -1; -} /* re_search_2 */ - -/* Declarations and macros for re_match_2. */ - -static int bcmp_translate (); -static boolean alt_match_null_string_p (), - common_op_match_null_string_p (), - group_match_null_string_p (); - -/* Structure for per-register (a.k.a. per-group) information. - This must not be longer than one word, because we push this value - onto the failure stack. Other register information, such as the - starting and ending positions (which are addresses), and the list of - inner groups (which is a bits list) are maintained in separate - variables. - - We are making a (strictly speaking) nonportable assumption here: that - the compiler will pack our bit fields into something that fits into - the type of `word', i.e., is something that fits into one item on the - failure stack. */ -typedef union -{ - fail_stack_elt_t word; - struct - { - /* This field is one if this group can match the empty string, - zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ -#define MATCH_NULL_UNSET_VALUE 3 - unsigned match_null_string_p : 2; - unsigned is_active : 1; - unsigned matched_something : 1; - unsigned ever_matched_something : 1; - } bits; -} register_info_type; - -#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) -#define IS_ACTIVE(R) ((R).bits.is_active) -#define MATCHED_SOMETHING(R) ((R).bits.matched_something) -#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) - - -/* Call this when have matched a real character; it sets `matched' flags - for the subexpressions which we are currently inside. Also records - that those subexprs have matched. */ -#define SET_REGS_MATCHED() \ - do \ - { \ - unsigned r; \ - for (r = lowest_active_reg; r <= highest_active_reg; r++) \ - { \ - MATCHED_SOMETHING (reg_info[r]) \ - = EVER_MATCHED_SOMETHING (reg_info[r]) \ - = 1; \ - } \ - } \ - while (0) - - -/* This converts PTR, a pointer into one of the search strings `string1' - and `string2' into an offset from the beginning of that string. */ -#define POINTER_TO_OFFSET(ptr) \ - (FIRST_STRING_P (ptr) ? (ptr) - string1 : (ptr) - string2 + size1) - -/* Registers are set to a sentinel when they haven't yet matched. */ -#define REG_UNSET_VALUE ((char *) -1) -#define REG_UNSET(e) ((e) == REG_UNSET_VALUE) - - -/* Macros for dealing with the split strings in re_match_2. */ - -#define MATCHING_IN_FIRST_STRING (dend == end_match_1) - -/* Call before fetching a character with *d. This switches over to - string2 if necessary. */ -#define PREFETCH() \ - while (d == dend) \ - { \ - /* End of string2 => fail. */ \ - if (dend == end_match_2) \ - goto fail; \ - /* End of string1 => advance to string2. */ \ - d = string2; \ - dend = end_match_2; \ - } - - -/* Test if at very beginning or at very end of the virtual concatenation - of `string1' and `string2'. If only one string, it's `string2'. */ -#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) -#define AT_STRINGS_END(d) ((d) == end2) - - -/* Test if D points to a character which is word-constituent. We have - two special cases to check for: if past the end of string1, look at - the first character in string2; and if before the beginning of - string2, look at the last character in string1. */ -#define WORDCHAR_P(d) \ - (SYNTAX ((d) == end1 ? *string2 \ - : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ - == Sword) - -/* Test if the character before D and the one at D differ with respect - to being word-constituent. */ -#define AT_WORD_BOUNDARY(d) \ - (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \ - || WORDCHAR_P (d - 1) != WORDCHAR_P (d)) - - -/* Free everything we malloc. */ -#ifdef REGEX_MALLOC -#define FREE_VAR(var) if (var) free (var); var = NULL -#define FREE_VARIABLES() \ - do { \ - FREE_VAR (fail_stack.stack); \ - FREE_VAR (regstart); \ - FREE_VAR (regend); \ - FREE_VAR (old_regstart); \ - FREE_VAR (old_regend); \ - FREE_VAR (best_regstart); \ - FREE_VAR (best_regend); \ - FREE_VAR (reg_info); \ - FREE_VAR (reg_dummy); \ - FREE_VAR (reg_info_dummy); \ - } while (0) -#else /* not REGEX_MALLOC */ -/* Some MIPS systems (at least) want this to free alloca'd storage. */ -#define FREE_VARIABLES() alloca (0) -#endif /* not REGEX_MALLOC */ - - -/* These values must meet several constraints. They must not be valid - register values; since we have a limit of 255 registers (because - we use only one byte in the pattern for the register number), we can - use numbers larger than 255. They must differ by 1, because of - NUM_FAILURE_ITEMS above. And the value for the lowest register must - be larger than the value for the highest register, so we do not try - to actually save any registers when none are active. */ -#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) -#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) - -/* Matching routines. */ - -#ifndef emacs /* Emacs never uses this. */ -/* re_match is like re_match_2 except it takes only a single string. */ - -int -re_match (bufp, string, size, pos, regs) - struct re_pattern_buffer *bufp; - const char *string; - int size, pos; - struct re_registers *regs; - { - return re_match_2 (bufp, NULL, 0, string, size, pos, regs, size); -} -#endif /* not emacs */ - - -/* re_match_2 matches the compiled pattern in BUFP against the - the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 - and SIZE2, respectively). We start matching at POS, and stop - matching at STOP. - - If REGS is non-null and the `no_sub' field of BUFP is nonzero, we - store offsets for the substring each group matched in REGS. See the - documentation for exactly how many groups we fill. - - We return -1 if no match, -2 if an internal error (such as the - failure stack overflowing). Otherwise, we return the length of the - matched substring. */ - -int -re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) - struct re_pattern_buffer *bufp; - const char *string1, *string2; - int size1, size2; - int pos; - struct re_registers *regs; - int stop; -{ - /* General temporaries. */ - int mcnt; - unsigned char *p1; - - /* Just past the end of the corresponding string. */ - const char *end1, *end2; - - /* Pointers into string1 and string2, just past the last characters in - each to consider matching. */ - const char *end_match_1, *end_match_2; - - /* Where we are in the data, and the end of the current string. */ - const char *d, *dend; - - /* Where we are in the pattern, and the end of the pattern. */ - unsigned char *p = bufp->buffer; - register unsigned char *pend = p + bufp->used; - - /* We use this to map every character in the string. */ - char *translate = bufp->translate; - - /* Failure point stack. Each place that can handle a failure further - down the line pushes a failure point on this stack. It consists of - restart, regend, and reg_info for all registers corresponding to - the subexpressions we're currently inside, plus the number of such - registers, and, finally, two char *'s. The first char * is where - to resume scanning the pattern; the second one is where to resume - scanning the strings. If the latter is zero, the failure point is - a ``dummy''; if a failure happens and the failure point is a dummy, - it gets discarded and the next next one is tried. */ - fail_stack_type fail_stack; -#ifdef DEBUG - static unsigned failure_id = 0; - unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; -#endif - - /* We fill all the registers internally, independent of what we - return, for use in backreferences. The number here includes - an element for register zero. */ - unsigned num_regs = bufp->re_nsub + 1; - - /* The currently active registers. */ - unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; - unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG; - - /* Information on the contents of registers. These are pointers into - the input strings; they record just what was matched (on this - attempt) by a subexpression part of the pattern, that is, the - regnum-th regstart pointer points to where in the pattern we began - matching and the regnum-th regend points to right after where we - stopped matching the regnum-th subexpression. (The zeroth register - keeps track of what the whole pattern matches.) */ - const char **regstart, **regend; - - /* If a group that's operated upon by a repetition operator fails to - match anything, then the register for its start will need to be - restored because it will have been set to wherever in the string we - are when we last see its open-group operator. Similarly for a - register's end. */ - const char **old_regstart, **old_regend; - - /* The is_active field of reg_info helps us keep track of which (possibly - nested) subexpressions we are currently in. The matched_something - field of reg_info[reg_num] helps us tell whether or not we have - matched any of the pattern so far this time through the reg_num-th - subexpression. These two fields get reset each time through any - loop their register is in. */ - register_info_type *reg_info; - - /* The following record the register info as found in the above - variables when we find a match better than any we've seen before. - This happens as we backtrack through the failure points, which in - turn happens only if we have not yet matched the entire string. */ - unsigned best_regs_set = false; - const char **best_regstart, **best_regend; - - /* Logically, this is `best_regend[0]'. But we don't want to have to - allocate space for that if we're not allocating space for anything - else (see below). Also, we never need info about register 0 for - any of the other register vectors, and it seems rather a kludge to - treat `best_regend' differently than the rest. So we keep track of - the end of the best match so far in a separate variable. We - initialize this to NULL so that when we backtrack the first time - and need to test it, it's not garbage. */ - const char *match_end = NULL; - - /* Used when we pop values we don't care about. */ - const char **reg_dummy; - register_info_type *reg_info_dummy; - -#ifdef DEBUG - /* Counts the total number of registers pushed. */ - unsigned num_regs_pushed = 0; -#endif - - DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); - - INIT_FAIL_STACK (); - - /* Do not bother to initialize all the register variables if there are - no groups in the pattern, as it takes a fair amount of time. If - there are groups, we include space for register 0 (the whole - pattern), even though we never use it, since it simplifies the - array indexing. We should fix this. */ - if (bufp->re_nsub) - { - regstart = REGEX_TALLOC (num_regs, const char *); - regend = REGEX_TALLOC (num_regs, const char *); - old_regstart = REGEX_TALLOC (num_regs, const char *); - old_regend = REGEX_TALLOC (num_regs, const char *); - best_regstart = REGEX_TALLOC (num_regs, const char *); - best_regend = REGEX_TALLOC (num_regs, const char *); - reg_info = REGEX_TALLOC (num_regs, register_info_type); - reg_dummy = REGEX_TALLOC (num_regs, const char *); - reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); - - if (!(regstart && regend && old_regstart && old_regend && reg_info - && best_regstart && best_regend && reg_dummy && reg_info_dummy)) - { - FREE_VARIABLES (); - return -2; - } - } -#ifdef REGEX_MALLOC - else - { - /* We must initialize all our variables to NULL, so that - `FREE_VARIABLES' doesn't try to free them. */ - regstart = regend = old_regstart = old_regend = best_regstart - = best_regend = reg_dummy = NULL; - reg_info = reg_info_dummy = (register_info_type *) NULL; - } -#endif /* REGEX_MALLOC */ - - /* The starting position is bogus. */ - if (pos < 0 || pos > size1 + size2) - { - FREE_VARIABLES (); - return -1; - } - - /* Initialize subexpression text positions to -1 to mark ones that no - start_memory/stop_memory has been seen for. Also initialize the - register information struct. */ - for (mcnt = 1; mcnt < num_regs; mcnt++) - { - regstart[mcnt] = regend[mcnt] - = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; - - REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; - IS_ACTIVE (reg_info[mcnt]) = 0; - MATCHED_SOMETHING (reg_info[mcnt]) = 0; - EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; - } - - /* We move `string1' into `string2' if the latter's empty -- but not if - `string1' is null. */ - if (size2 == 0 && string1 != NULL) - { - string2 = string1; - size2 = size1; - string1 = 0; - size1 = 0; - } - end1 = string1 + size1; - end2 = string2 + size2; - - /* Compute where to stop matching, within the two strings. */ - if (stop <= size1) - { - end_match_1 = string1 + stop; - end_match_2 = string2; - } - else - { - end_match_1 = end1; - end_match_2 = string2 + stop - size1; - } - - /* `p' scans through the pattern as `d' scans through the data. - `dend' is the end of the input string that `d' points within. `d' - is advanced into the following input string whenever necessary, but - this happens before fetching; therefore, at the beginning of the - loop, `d' can be pointing at the end of a string, but it cannot - equal `string2'. */ - if (size1 > 0 && pos <= size1) - { - d = string1 + pos; - dend = end_match_1; - } - else - { - d = string2 + pos - size1; - dend = end_match_2; - } - - DEBUG_PRINT1 ("The compiled pattern is: "); - DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); - DEBUG_PRINT1 ("The string to match is: `"); - DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); - DEBUG_PRINT1 ("'\n"); - - /* This loops over pattern commands. It exits by returning from the - function if the match is complete, or it drops through if the match - fails at this starting point in the input data. */ - for (;;) - { - DEBUG_PRINT2 ("\n0x%x: ", p); - - if (p == pend) - { /* End of pattern means we might have succeeded. */ - DEBUG_PRINT1 ("end of pattern ... "); - - /* If we haven't matched the entire string, and we want the - longest match, try backtracking. */ - if (d != end_match_2) - { - DEBUG_PRINT1 ("backtracking.\n"); - - if (!FAIL_STACK_EMPTY ()) - { /* More failure points to try. */ - boolean same_str_p = (FIRST_STRING_P (match_end) - == MATCHING_IN_FIRST_STRING); - - /* If exceeds best match so far, save it. */ - if (!best_regs_set - || (same_str_p && d > match_end) - || (!same_str_p && !MATCHING_IN_FIRST_STRING)) - { - best_regs_set = true; - match_end = d; - - DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); - - for (mcnt = 1; mcnt < num_regs; mcnt++) - { - best_regstart[mcnt] = regstart[mcnt]; - best_regend[mcnt] = regend[mcnt]; - } - } - goto fail; - } - - /* If no failure points, don't restore garbage. */ - else if (best_regs_set) - { - restore_best_regs: - /* Restore best match. It may happen that `dend == - end_match_1' while the restored d is in string2. - For example, the pattern `x.*y.*z' against the - strings `x-' and `y-z-', if the two strings are - not consecutive in memory. */ - DEBUG_PRINT1 ("Restoring best registers.\n"); - - d = match_end; - dend = ((d >= string1 && d <= end1) - ? end_match_1 : end_match_2); - - for (mcnt = 1; mcnt < num_regs; mcnt++) - { - regstart[mcnt] = best_regstart[mcnt]; - regend[mcnt] = best_regend[mcnt]; - } - } - } /* d != end_match_2 */ - - DEBUG_PRINT1 ("Accepting match.\n"); - - /* If caller wants register contents data back, do it. */ - if (regs && !bufp->no_sub) - { - /* Have the register data arrays been allocated? */ - if (bufp->regs_allocated == REGS_UNALLOCATED) - { /* No. So allocate them with malloc. We need one - extra element beyond `num_regs' for the `-1' marker - GNU code uses. */ - regs->num_regs = MAX (RE_NREGS, num_regs + 1); - regs->start = TALLOC (regs->num_regs, regoff_t); - regs->end = TALLOC (regs->num_regs, regoff_t); - if (regs->start == NULL || regs->end == NULL) - return -2; - bufp->regs_allocated = REGS_REALLOCATE; - } - else if (bufp->regs_allocated == REGS_REALLOCATE) - { /* Yes. If we need more elements than were already - allocated, reallocate them. If we need fewer, just - leave it alone. */ - if (regs->num_regs < num_regs + 1) - { - regs->num_regs = num_regs + 1; - RETALLOC (regs->start, regs->num_regs, regoff_t); - RETALLOC (regs->end, regs->num_regs, regoff_t); - if (regs->start == NULL || regs->end == NULL) - return -2; - } - } - else - assert (bufp->regs_allocated == REGS_FIXED); - - /* Convert the pointer data in `regstart' and `regend' to - indices. Register zero has to be set differently, - since we haven't kept track of any info for it. */ - if (regs->num_regs > 0) - { - regs->start[0] = pos; - regs->end[0] = (MATCHING_IN_FIRST_STRING ? d - string1 - : d - string2 + size1); - } - - /* Go through the first `min (num_regs, regs->num_regs)' - registers, since that is all we initialized. */ - for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) - { - if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) - regs->start[mcnt] = regs->end[mcnt] = -1; - else - { - regs->start[mcnt] = POINTER_TO_OFFSET (regstart[mcnt]); - regs->end[mcnt] = POINTER_TO_OFFSET (regend[mcnt]); - } - } - - /* If the regs structure we return has more elements than - were in the pattern, set the extra elements to -1. If - we (re)allocated the registers, this is the case, - because we always allocate enough to have at least one - -1 at the end. */ - for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++) - regs->start[mcnt] = regs->end[mcnt] = -1; - } /* regs && !bufp->no_sub */ - - FREE_VARIABLES (); - DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", - nfailure_points_pushed, nfailure_points_popped, - nfailure_points_pushed - nfailure_points_popped); - DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); - - mcnt = d - pos - (MATCHING_IN_FIRST_STRING - ? string1 - : string2 - size1); - - DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); - - return mcnt; - } - - /* Otherwise match next pattern command. */ -#ifdef SWITCH_ENUM_BUG - switch ((int) ((re_opcode_t) *p++)) -#else - switch ((re_opcode_t) *p++) -#endif - { - /* Ignore these. Used to ignore the n of succeed_n's which - currently have n == 0. */ - case no_op: - DEBUG_PRINT1 ("EXECUTING no_op.\n"); - break; - - - /* Match the next n pattern characters exactly. The following - byte in the pattern defines n, and the n bytes after that - are the characters to match. */ - case exactn: - mcnt = *p++; - DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); - - /* This is written out as an if-else so we don't waste time - testing `translate' inside the loop. */ - if (translate) - { - do - { - PREFETCH (); - if (translate[(unsigned char) *d++] != (char) *p++) - goto fail; - } - while (--mcnt); - } - else - { - do - { - PREFETCH (); - if (*d++ != (char) *p++) goto fail; - } - while (--mcnt); - } - SET_REGS_MATCHED (); - break; - - - /* Match any character except possibly a newline or a null. */ - case anychar: - DEBUG_PRINT1 ("EXECUTING anychar.\n"); - - PREFETCH (); - - if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') - || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) - goto fail; - - SET_REGS_MATCHED (); - DEBUG_PRINT2 (" Matched `%d'.\n", *d); - d++; - break; - - - case charset: - case charset_not: - { - register unsigned char c; - boolean not = (re_opcode_t) *(p - 1) == charset_not; - - DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); - - PREFETCH (); - c = TRANSLATE (*d); /* The character to match. */ - - /* Cast to `unsigned' instead of `unsigned char' in case the - bit list is a full 32 bytes long. */ - if (c < (unsigned) (*p * BYTEWIDTH) - && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) - not = !not; - - p += 1 + *p; - - if (!not) goto fail; - - SET_REGS_MATCHED (); - d++; - break; - } - - - /* The beginning of a group is represented by start_memory. - The arguments are the register number in the next byte, and the - number of groups inner to this one in the next. The text - matched within the group is recorded (in the internal - registers data structure) under the register number. */ - case start_memory: - DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); - - /* Find out if this group can match the empty string. */ - p1 = p; /* To send to group_match_null_string_p. */ - - if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) - REG_MATCH_NULL_STRING_P (reg_info[*p]) - = group_match_null_string_p (&p1, pend, reg_info); - - /* Save the position in the string where we were the last time - we were at this open-group operator in case the group is - operated upon by a repetition operator, e.g., with `(a*)*b' - against `ab'; then we want to ignore where we are now in - the string in case this attempt to match fails. */ - old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) - ? REG_UNSET (regstart[*p]) ? d : regstart[*p] - : regstart[*p]; - DEBUG_PRINT2 (" old_regstart: %d\n", - POINTER_TO_OFFSET (old_regstart[*p])); - - regstart[*p] = d; - DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); - - IS_ACTIVE (reg_info[*p]) = 1; - MATCHED_SOMETHING (reg_info[*p]) = 0; - - /* This is the new highest active register. */ - highest_active_reg = *p; - - /* If nothing was active before, this is the new lowest active - register. */ - if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) - lowest_active_reg = *p; - - /* Move past the register number and inner group count. */ - p += 2; - break; - - - /* The stop_memory opcode represents the end of a group. Its - arguments are the same as start_memory's: the register - number, and the number of inner groups. */ - case stop_memory: - DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); - - /* We need to save the string position the last time we were at - this close-group operator in case the group is operated - upon by a repetition operator, e.g., with `((a*)*(b*)*)*' - against `aba'; then we want to ignore where we are now in - the string in case this attempt to match fails. */ - old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) - ? REG_UNSET (regend[*p]) ? d : regend[*p] - : regend[*p]; - DEBUG_PRINT2 (" old_regend: %d\n", - POINTER_TO_OFFSET (old_regend[*p])); - - regend[*p] = d; - DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); - - /* This register isn't active anymore. */ - IS_ACTIVE (reg_info[*p]) = 0; - - /* If this was the only register active, nothing is active - anymore. */ - if (lowest_active_reg == highest_active_reg) - { - lowest_active_reg = NO_LOWEST_ACTIVE_REG; - highest_active_reg = NO_HIGHEST_ACTIVE_REG; - } - else - { /* We must scan for the new highest active register, since - it isn't necessarily one less than now: consider - (a(b)c(d(e)f)g). When group 3 ends, after the f), the - new highest active register is 1. */ - unsigned char r = *p - 1; - while (r > 0 && !IS_ACTIVE (reg_info[r])) - r--; - - /* If we end up at register zero, that means that we saved - the registers as the result of an `on_failure_jump', not - a `start_memory', and we jumped to past the innermost - `stop_memory'. For example, in ((.)*) we save - registers 1 and 2 as a result of the *, but when we pop - back to the second ), we are at the stop_memory 1. - Thus, nothing is active. */ - if (r == 0) - { - lowest_active_reg = NO_LOWEST_ACTIVE_REG; - highest_active_reg = NO_HIGHEST_ACTIVE_REG; - } - else - highest_active_reg = r; - } - - /* If just failed to match something this time around with a - group that's operated on by a repetition operator, try to - force exit from the ``loop'', and restore the register - information for this group that we had before trying this - last match. */ - if ((!MATCHED_SOMETHING (reg_info[*p]) - || (re_opcode_t) p[-3] == start_memory) - && (p + 2) < pend) - { - boolean is_a_jump_n = false; - - p1 = p + 2; - mcnt = 0; - switch ((re_opcode_t) *p1++) - { - case jump_n: - is_a_jump_n = true; - case pop_failure_jump: - case maybe_pop_jump: - case jump: - case dummy_failure_jump: - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - if (is_a_jump_n) - p1 += 2; - break; - - default: - /* do nothing */ ; - } - p1 += mcnt; - - /* If the next operation is a jump backwards in the pattern - to an on_failure_jump right before the start_memory - corresponding to this stop_memory, exit from the loop - by forcing a failure after pushing on the stack the - on_failure_jump's jump in the pattern, and d. */ - if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump - && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) - { - /* If this group ever matched anything, then restore - what its registers were before trying this last - failed match, e.g., with `(a*)*b' against `ab' for - regstart[1], and, e.g., with `((a*)*(b*)*)*' - against `aba' for regend[3]. - - Also restore the registers for inner groups for, - e.g., `((a*)(b*))*' against `aba' (register 3 would - otherwise get trashed). */ - - if (EVER_MATCHED_SOMETHING (reg_info[*p])) - { - unsigned r; - - EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; - - /* Restore this and inner groups' (if any) registers. */ - for (r = *p; r < *p + *(p + 1); r++) - { - regstart[r] = old_regstart[r]; - - /* xx why this test? */ - if ((int) old_regend[r] >= (int) regstart[r]) - regend[r] = old_regend[r]; - } - } - p1++; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - PUSH_FAILURE_POINT (p1 + mcnt, d, -2); - - goto fail; - } - } - - /* Move past the register number and the inner group count. */ - p += 2; - break; - - - /* \ has been turned into a `duplicate' command which is - followed by the numeric value of as the register number. */ - case duplicate: - { - register const char *d2, *dend2; - int regno = *p++; /* Get which register to match against. */ - DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); - - /* Can't back reference a group which we've never matched. */ - if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) - goto fail; - - /* Where in input to try to start matching. */ - d2 = regstart[regno]; - - /* Where to stop matching; if both the place to start and - the place to stop matching are in the same string, then - set to the place to stop, otherwise, for now have to use - the end of the first string. */ - - dend2 = ((FIRST_STRING_P (regstart[regno]) - == FIRST_STRING_P (regend[regno])) - ? regend[regno] : end_match_1); - for (;;) - { - /* If necessary, advance to next segment in register - contents. */ - while (d2 == dend2) - { - if (dend2 == end_match_2) break; - if (dend2 == regend[regno]) break; - - /* End of string1 => advance to string2. */ - d2 = string2; - dend2 = regend[regno]; - } - /* At end of register contents => success */ - if (d2 == dend2) break; - - /* If necessary, advance to next segment in data. */ - PREFETCH (); - - /* How many characters left in this segment to match. */ - mcnt = dend - d; - - /* Want how many consecutive characters we can match in - one shot, so, if necessary, adjust the count. */ - if (mcnt > dend2 - d2) - mcnt = dend2 - d2; - - /* Compare that many; failure if mismatch, else move - past them. */ - if (translate - ? bcmp_translate (d, d2, mcnt, translate) - : bcmp (d, d2, mcnt)) - goto fail; - d += mcnt, d2 += mcnt; - } - } - break; - - - /* begline matches the empty string at the beginning of the string - (unless `not_bol' is set in `bufp'), and, if - `newline_anchor' is set, after newlines. */ - case begline: - DEBUG_PRINT1 ("EXECUTING begline.\n"); - - if (AT_STRINGS_BEG (d)) - { - if (!bufp->not_bol) break; - } - else if (d[-1] == '\n' && bufp->newline_anchor) - { - break; - } - /* In all other cases, we fail. */ - goto fail; - - - /* endline is the dual of begline. */ - case endline: - DEBUG_PRINT1 ("EXECUTING endline.\n"); - - if (AT_STRINGS_END (d)) - { - if (!bufp->not_eol) break; - } - - /* We have to ``prefetch'' the next character. */ - else if ((d == end1 ? *string2 : *d) == '\n' - && bufp->newline_anchor) - { - break; - } - goto fail; - - - /* Match at the very beginning of the data. */ - case begbuf: - DEBUG_PRINT1 ("EXECUTING begbuf.\n"); - if (AT_STRINGS_BEG (d)) - break; - goto fail; - - - /* Match at the very end of the data. */ - case endbuf: - DEBUG_PRINT1 ("EXECUTING endbuf.\n"); - if (AT_STRINGS_END (d)) - break; - goto fail; - - - /* on_failure_keep_string_jump is used to optimize `.*\n'. It - pushes NULL as the value for the string on the stack. Then - `pop_failure_point' will keep the current value for the - string, instead of restoring it. To see why, consider - matching `foo\nbar' against `.*\n'. The .* matches the foo; - then the . fails against the \n. But the next thing we want - to do is match the \n against the \n; if we restored the - string value, we would be back at the foo. - - Because this is used only in specific cases, we don't need to - check all the things that `on_failure_jump' does, to make - sure the right things get saved on the stack. Hence we don't - share its code. The only reason to push anything on the - stack at all is that otherwise we would have to change - `anychar's code to do something besides goto fail in this - case; that seems worse than this. */ - case on_failure_keep_string_jump: - DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); - - EXTRACT_NUMBER_AND_INCR (mcnt, p); - DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); - - PUSH_FAILURE_POINT (p + mcnt, NULL, -2); - break; - - - /* Uses of on_failure_jump: - - Each alternative starts with an on_failure_jump that points - to the beginning of the next alternative. Each alternative - except the last ends with a jump that in effect jumps past - the rest of the alternatives. (They really jump to the - ending jump of the following alternative, because tensioning - these jumps is a hassle.) - - Repeats start with an on_failure_jump that points past both - the repetition text and either the following jump or - pop_failure_jump back to this on_failure_jump. */ - case on_failure_jump: - on_failure: - DEBUG_PRINT1 ("EXECUTING on_failure_jump"); - - EXTRACT_NUMBER_AND_INCR (mcnt, p); - DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); - - /* If this on_failure_jump comes right before a group (i.e., - the original * applied to a group), save the information - for that group and all inner ones, so that if we fail back - to this point, the group's information will be correct. - For example, in \(a*\)*\1, we need the preceding group, - and in \(\(a*\)b*\)\2, we need the inner group. */ - - /* We can't use `p' to check ahead because we push - a failure point to `p + mcnt' after we do this. */ - p1 = p; - - /* We need to skip no_op's before we look for the - start_memory in case this on_failure_jump is happening as - the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 - against aba. */ - while (p1 < pend && (re_opcode_t) *p1 == no_op) - p1++; - - if (p1 < pend && (re_opcode_t) *p1 == start_memory) - { - /* We have a new highest active register now. This will - get reset at the start_memory we are about to get to, - but we will have saved all the registers relevant to - this repetition op, as described above. */ - highest_active_reg = *(p1 + 1) + *(p1 + 2); - if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) - lowest_active_reg = *(p1 + 1); - } - - DEBUG_PRINT1 (":\n"); - PUSH_FAILURE_POINT (p + mcnt, d, -2); - break; - - - /* A smart repeat ends with `maybe_pop_jump'. - We change it to either `pop_failure_jump' or `jump'. */ - case maybe_pop_jump: - EXTRACT_NUMBER_AND_INCR (mcnt, p); - DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); - { - register unsigned char *p2 = p; - - /* Compare the beginning of the repeat with what in the - pattern follows its end. If we can establish that there - is nothing that they would both match, i.e., that we - would have to backtrack because of (as in, e.g., `a*a') - then we can change to pop_failure_jump, because we'll - never have to backtrack. - - This is not true in the case of alternatives: in - `(a|ab)*' we do need to backtrack to the `ab' alternative - (e.g., if the string was `ab'). But instead of trying to - detect that here, the alternative has put on a dummy - failure point which is what we will end up popping. */ - - /* Skip over open/close-group commands. */ - while (p2 + 2 < pend - && ((re_opcode_t) *p2 == stop_memory - || (re_opcode_t) *p2 == start_memory)) - p2 += 3; /* Skip over args, too. */ - - /* If we're at the end of the pattern, we can change. */ - if (p2 == pend) - { - /* Consider what happens when matching ":\(.*\)" - against ":/". I don't really understand this code - yet. */ - p[-3] = (unsigned char) pop_failure_jump; - DEBUG_PRINT1 - (" End of pattern: change to `pop_failure_jump'.\n"); - } - - else if ((re_opcode_t) *p2 == exactn - || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) - { - register unsigned char c - = *p2 == (unsigned char) endline ? '\n' : p2[2]; - p1 = p + mcnt; - - /* p1[0] ... p1[2] are the `on_failure_jump' corresponding - to the `maybe_finalize_jump' of this case. Examine what - follows. */ - if ((re_opcode_t) p1[3] == exactn && p1[5] != c) - { - p[-3] = (unsigned char) pop_failure_jump; - DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", - c, p1[5]); - } - - else if ((re_opcode_t) p1[3] == charset - || (re_opcode_t) p1[3] == charset_not) - { - int not = (re_opcode_t) p1[3] == charset_not; - - if (c < (unsigned char) (p1[4] * BYTEWIDTH) - && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) - not = !not; - - /* `not' is equal to 1 if c would match, which means - that we can't change to pop_failure_jump. */ - if (!not) - { - p[-3] = (unsigned char) pop_failure_jump; - DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); - } - } - } - } - p -= 2; /* Point at relative address again. */ - if ((re_opcode_t) p[-1] != pop_failure_jump) - { - p[-1] = (unsigned char) jump; - DEBUG_PRINT1 (" Match => jump.\n"); - goto unconditional_jump; - } - /* Note fall through. */ - - - /* The end of a simple repeat has a pop_failure_jump back to - its matching on_failure_jump, where the latter will push a - failure point. The pop_failure_jump takes off failure - points put on by this pop_failure_jump's matching - on_failure_jump; we got through the pattern to here from the - matching on_failure_jump, so didn't fail. */ - case pop_failure_jump: - { - /* We need to pass separate storage for the lowest and - highest registers, even though we don't care about the - actual values. Otherwise, we will restore only one - register from the stack, since lowest will == highest in - `pop_failure_point'. */ - unsigned dummy_low_reg, dummy_high_reg; - unsigned char *pdummy; - const char *sdummy = NULL; - - DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); - POP_FAILURE_POINT (sdummy, pdummy, - dummy_low_reg, dummy_high_reg, - reg_dummy, reg_dummy, reg_info_dummy); - } - /* Note fall through. */ - - - /* Unconditionally jump (without popping any failure points). */ - case jump: - unconditional_jump: - EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ - DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); - p += mcnt; /* Do the jump. */ - DEBUG_PRINT2 ("(to 0x%x).\n", p); - break; - - - /* We need this opcode so we can detect where alternatives end - in `group_match_null_string_p' et al. */ - case jump_past_alt: - DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); - goto unconditional_jump; - - - /* Normally, the on_failure_jump pushes a failure point, which - then gets popped at pop_failure_jump. We will end up at - pop_failure_jump, also, and with a pattern of, say, `a+', we - are skipping over the on_failure_jump, so we have to push - something meaningless for pop_failure_jump to pop. */ - case dummy_failure_jump: - DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); - /* It doesn't matter what we push for the string here. What - the code at `fail' tests is the value for the pattern. */ - PUSH_FAILURE_POINT (0, 0, -2); - goto unconditional_jump; - - - /* At the end of an alternative, we need to push a dummy failure - point in case we are followed by a `pop_failure_jump', because - we don't want the failure point for the alternative to be - popped. For example, matching `(a|ab)*' against `aab' - requires that we match the `ab' alternative. */ - case push_dummy_failure: - DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); - /* See comments just above at `dummy_failure_jump' about the - two zeroes. */ - PUSH_FAILURE_POINT (0, 0, -2); - break; - - /* Have to succeed matching what follows at least n times. - After that, handle like `on_failure_jump'. */ - case succeed_n: - EXTRACT_NUMBER (mcnt, p + 2); - DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); - - assert (mcnt >= 0); - /* Originally, this is how many times we HAVE to succeed. */ - if (mcnt > 0) - { - mcnt--; - p += 2; - STORE_NUMBER_AND_INCR (p, mcnt); - DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt); - } - else if (mcnt == 0) - { - DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); - p[2] = (unsigned char) no_op; - p[3] = (unsigned char) no_op; - goto on_failure; - } - break; - - case jump_n: - EXTRACT_NUMBER (mcnt, p + 2); - DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); - - /* Originally, this is how many times we CAN jump. */ - if (mcnt) - { - mcnt--; - STORE_NUMBER (p + 2, mcnt); - goto unconditional_jump; - } - /* If don't have to jump any more, skip over the rest of command. */ - else - p += 4; - break; - - case set_number_at: - { - DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); - - EXTRACT_NUMBER_AND_INCR (mcnt, p); - p1 = p + mcnt; - EXTRACT_NUMBER_AND_INCR (mcnt, p); - DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); - STORE_NUMBER (p1, mcnt); - break; - } - - case wordbound: - DEBUG_PRINT1 ("EXECUTING wordbound.\n"); - if (AT_WORD_BOUNDARY (d)) - break; - goto fail; - - case notwordbound: - DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); - if (AT_WORD_BOUNDARY (d)) - goto fail; - break; - - case wordbeg: - DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); - if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1))) - break; - goto fail; - - case wordend: - DEBUG_PRINT1 ("EXECUTING wordend.\n"); - if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1) - && (!WORDCHAR_P (d) || AT_STRINGS_END (d))) - break; - goto fail; - -#ifdef emacs -#ifdef emacs19 - case before_dot: - DEBUG_PRINT1 ("EXECUTING before_dot.\n"); - if (PTR_CHAR_POS ((unsigned char *) d) >= point) - goto fail; - break; - - case at_dot: - DEBUG_PRINT1 ("EXECUTING at_dot.\n"); - if (PTR_CHAR_POS ((unsigned char *) d) != point) - goto fail; - break; - - case after_dot: - DEBUG_PRINT1 ("EXECUTING after_dot.\n"); - if (PTR_CHAR_POS ((unsigned char *) d) <= point) - goto fail; - break; -#else /* not emacs19 */ - case at_dot: - DEBUG_PRINT1 ("EXECUTING at_dot.\n"); - if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point) - goto fail; - break; -#endif /* not emacs19 */ - - case syntaxspec: - DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); - mcnt = *p++; - goto matchsyntax; - - case wordchar: - DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); - mcnt = (int) Sword; - matchsyntax: - PREFETCH (); - if (SYNTAX (*d++) != (enum syntaxcode) mcnt) - goto fail; - SET_REGS_MATCHED (); - break; - - case notsyntaxspec: - DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); - mcnt = *p++; - goto matchnotsyntax; - - case notwordchar: - DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); - mcnt = (int) Sword; - matchnotsyntax: - PREFETCH (); - if (SYNTAX (*d++) == (enum syntaxcode) mcnt) - goto fail; - SET_REGS_MATCHED (); - break; - -#else /* not emacs */ - case wordchar: - DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); - PREFETCH (); - if (!WORDCHAR_P (d)) - goto fail; - SET_REGS_MATCHED (); - d++; - break; - - case notwordchar: - DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); - PREFETCH (); - if (WORDCHAR_P (d)) - goto fail; - SET_REGS_MATCHED (); - d++; - break; -#endif /* not emacs */ - - default: - abort (); - } - continue; /* Successfully executed one pattern command; keep going. */ - - - /* We goto here if a matching operation fails. */ - fail: - if (!FAIL_STACK_EMPTY ()) - { /* A restart point is known. Restore to that state. */ - DEBUG_PRINT1 ("\nFAIL:\n"); - POP_FAILURE_POINT (d, p, - lowest_active_reg, highest_active_reg, - regstart, regend, reg_info); - - /* If this failure point is a dummy, try the next one. */ - if (!p) - goto fail; - - /* If we failed to the end of the pattern, don't examine *p. */ - assert (p <= pend); - if (p < pend) - { - boolean is_a_jump_n = false; - - /* If failed to a backwards jump that's part of a repetition - loop, need to pop this failure point and use the next one. */ - switch ((re_opcode_t) *p) - { - case jump_n: - is_a_jump_n = true; - case maybe_pop_jump: - case pop_failure_jump: - case jump: - p1 = p + 1; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - p1 += mcnt; - - if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) - || (!is_a_jump_n - && (re_opcode_t) *p1 == on_failure_jump)) - goto fail; - break; - default: - /* do nothing */ ; - } - } - - if (d >= string1 && d <= end1) - dend = end_match_1; - } - else - break; /* Matching at this starting point really fails. */ - } /* for (;;) */ - - if (best_regs_set) - goto restore_best_regs; - - FREE_VARIABLES (); - - return -1; /* Failure to match. */ -} /* re_match_2 */ - -/* Subroutine definitions for re_match_2. */ - - -/* We are passed P pointing to a register number after a start_memory. - - Return true if the pattern up to the corresponding stop_memory can - match the empty string, and false otherwise. - - If we find the matching stop_memory, sets P to point to one past its number. - Otherwise, sets P to an undefined byte less than or equal to END. - - We don't handle duplicates properly (yet). */ - -static boolean -group_match_null_string_p (p, end, reg_info) - unsigned char **p, *end; - register_info_type *reg_info; -{ - int mcnt; - /* Point to after the args to the start_memory. */ - unsigned char *p1 = *p + 2; - - while (p1 < end) - { - /* Skip over opcodes that can match nothing, and return true or - false, as appropriate, when we get to one that can't, or to the - matching stop_memory. */ - - switch ((re_opcode_t) *p1) - { - /* Could be either a loop or a series of alternatives. */ - case on_failure_jump: - p1++; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - - /* If the next operation is not a jump backwards in the - pattern. */ - - if (mcnt >= 0) - { - /* Go through the on_failure_jumps of the alternatives, - seeing if any of the alternatives cannot match nothing. - The last alternative starts with only a jump, - whereas the rest start with on_failure_jump and end - with a jump, e.g., here is the pattern for `a|b|c': - - /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 - /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 - /exactn/1/c - - So, we have to first go through the first (n-1) - alternatives and then deal with the last one separately. */ - - - /* Deal with the first (n-1) alternatives, which start - with an on_failure_jump (see above) that jumps to right - past a jump_past_alt. */ - - while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) - { - /* `mcnt' holds how many bytes long the alternative - is, including the ending `jump_past_alt' and - its number. */ - - if (!alt_match_null_string_p (p1, p1 + mcnt - 3, - reg_info)) - return false; - - /* Move to right after this alternative, including the - jump_past_alt. */ - p1 += mcnt; - - /* Break if it's the beginning of an n-th alternative - that doesn't begin with an on_failure_jump. */ - if ((re_opcode_t) *p1 != on_failure_jump) - break; - - /* Still have to check that it's not an n-th - alternative that starts with an on_failure_jump. */ - p1++; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) - { - /* Get to the beginning of the n-th alternative. */ - p1 -= 3; - break; - } - } - - /* Deal with the last alternative: go back and get number - of the `jump_past_alt' just before it. `mcnt' contains - the length of the alternative. */ - EXTRACT_NUMBER (mcnt, p1 - 2); - - if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) - return false; - - p1 += mcnt; /* Get past the n-th alternative. */ - } /* if mcnt > 0 */ - break; - - - case stop_memory: - assert (p1[1] == **p); - *p = p1 + 2; - return true; - - - default: - if (!common_op_match_null_string_p (&p1, end, reg_info)) - return false; - } - } /* while p1 < end */ - - return false; -} /* group_match_null_string_p */ - - -/* Similar to group_match_null_string_p, but doesn't deal with alternatives: - It expects P to be the first byte of a single alternative and END one - byte past the last. The alternative can contain groups. */ - -static boolean -alt_match_null_string_p (p, end, reg_info) - unsigned char *p, *end; - register_info_type *reg_info; -{ - int mcnt; - unsigned char *p1 = p; - - while (p1 < end) - { - /* Skip over opcodes that can match nothing, and break when we get - to one that can't. */ - - switch ((re_opcode_t) *p1) - { - /* It's a loop. */ - case on_failure_jump: - p1++; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - p1 += mcnt; - break; - - default: - if (!common_op_match_null_string_p (&p1, end, reg_info)) - return false; - } - } /* while p1 < end */ - - return true; -} /* alt_match_null_string_p */ - - -/* Deals with the ops common to group_match_null_string_p and - alt_match_null_string_p. - - Sets P to one after the op and its arguments, if any. */ - -static boolean -common_op_match_null_string_p (p, end, reg_info) - unsigned char **p, *end; - register_info_type *reg_info; -{ - int mcnt; - boolean ret; - int reg_no; - unsigned char *p1 = *p; - - switch ((re_opcode_t) *p1++) - { - case no_op: - case begline: - case endline: - case begbuf: - case endbuf: - case wordbeg: - case wordend: - case wordbound: - case notwordbound: -#ifdef emacs - case before_dot: - case at_dot: - case after_dot: -#endif - break; - - case start_memory: - reg_no = *p1; - assert (reg_no > 0 && reg_no <= MAX_REGNUM); - ret = group_match_null_string_p (&p1, end, reg_info); - - /* Have to set this here in case we're checking a group which - contains a group and a back reference to it. */ - - if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) - REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; - - if (!ret) - return false; - break; - - /* If this is an optimized succeed_n for zero times, make the jump. */ - case jump: - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - if (mcnt >= 0) - p1 += mcnt; - else - return false; - break; - - case succeed_n: - /* Get to the number of times to succeed. */ - p1 += 2; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - - if (mcnt == 0) - { - p1 -= 4; - EXTRACT_NUMBER_AND_INCR (mcnt, p1); - p1 += mcnt; - } - else - return false; - break; - - case duplicate: - if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) - return false; - break; - - case set_number_at: - p1 += 4; - - default: - /* All other opcodes mean we cannot match the empty string. */ - return false; - } - - *p = p1; - return true; -} /* common_op_match_null_string_p */ - - -/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN - bytes; nonzero otherwise. */ - -static int -bcmp_translate (s1, s2, len, translate) - unsigned char *s1, *s2; - register int len; - char *translate; -{ - register unsigned char *p1 = s1, *p2 = s2; - while (len) - { - if (translate[*p1++] != translate[*p2++]) return 1; - len--; - } - return 0; -} - -/* Entry points for GNU code. */ - -/* re_compile_pattern is the GNU regular expression compiler: it - compiles PATTERN (of length SIZE) and puts the result in BUFP. - Returns 0 if the pattern was valid, otherwise an error string. - - Assumes the `allocated' (and perhaps `buffer') and `translate' fields - are set in BUFP on entry. - - We call regex_compile to do the actual compilation. */ - -const char * -re_compile_pattern (pattern, length, bufp) - const char *pattern; - int length; - struct re_pattern_buffer *bufp; -{ - reg_errcode_t ret; - - /* GNU code is written to assume at least RE_NREGS registers will be set - (and at least one extra will be -1). */ - bufp->regs_allocated = REGS_UNALLOCATED; - - /* And GNU code determines whether or not to get register information - by passing null for the REGS argument to re_match, etc., not by - setting no_sub. */ - bufp->no_sub = 0; - - /* Match anchors at newline. */ - bufp->newline_anchor = 1; - - ret = regex_compile (pattern, length, re_syntax_options, bufp); - - return re_error_msg[(int) ret]; -} - -/* Entry points compatible with 4.2 BSD regex library. We don't define - them if this is an Emacs or POSIX compilation. */ - -#if !defined (emacs) && !defined (_POSIX_SOURCE) - -/* BSD has one and only one pattern buffer. */ -static struct re_pattern_buffer re_comp_buf; - -char * -re_comp (s) - const char *s; -{ - reg_errcode_t ret; - - if (!s) - { - if (!re_comp_buf.buffer) - return "No previous regular expression"; - return 0; - } - - if (!re_comp_buf.buffer) - { - re_comp_buf.buffer = (unsigned char *) malloc (200); - if (re_comp_buf.buffer == NULL) - return "Memory exhausted"; - re_comp_buf.allocated = 200; - - re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH); - if (re_comp_buf.fastmap == NULL) - return "Memory exhausted"; - } - - /* Since `re_exec' always passes NULL for the `regs' argument, we - don't need to initialize the pattern buffer fields which affect it. */ - - /* Match anchors at newlines. */ - re_comp_buf.newline_anchor = 1; - - ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); - - /* Yes, we're discarding `const' here. */ - return (char *) re_error_msg[(int) ret]; -} - - -int -re_exec (s) - const char *s; -{ - const int len = strlen (s); - return - 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); -} -#endif /* not emacs and not _POSIX_SOURCE */ - -/* POSIX.2 functions. Don't define these for Emacs. */ - -#ifndef emacs - -/* regcomp takes a regular expression as a string and compiles it. - - PREG is a regex_t *. We do not expect any fields to be initialized, - since POSIX says we shouldn't. Thus, we set - - `buffer' to the compiled pattern; - `used' to the length of the compiled pattern; - `syntax' to RE_SYNTAX_POSIX_EXTENDED if the - REG_EXTENDED bit in CFLAGS is set; otherwise, to - RE_SYNTAX_POSIX_BASIC; - `newline_anchor' to REG_NEWLINE being set in CFLAGS; - `fastmap' and `fastmap_accurate' to zero; - `re_nsub' to the number of subexpressions in PATTERN. - - PATTERN is the address of the pattern string. - - CFLAGS is a series of bits which affect compilation. - - If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we - use POSIX basic syntax. - - If REG_NEWLINE is set, then . and [^...] don't match newline. - Also, regexec will try a match beginning after every newline. - - If REG_ICASE is set, then we considers upper- and lowercase - versions of letters to be equivalent when matching. - - If REG_NOSUB is set, then when PREG is passed to regexec, that - routine will report only success or failure, and nothing about the - registers. - - It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for - the return codes and their meanings.) */ - -int -regcomp (preg, pattern, cflags) - regex_t *preg; - const char *pattern; - int cflags; -{ - reg_errcode_t ret; - unsigned syntax - = (cflags & REG_EXTENDED) ? - RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; - - /* regex_compile will allocate the space for the compiled pattern. */ - preg->buffer = 0; - preg->allocated = 0; - - /* Don't bother to use a fastmap when searching. This simplifies the - REG_NEWLINE case: if we used a fastmap, we'd have to put all the - characters after newlines into the fastmap. This way, we just try - every character. */ - preg->fastmap = 0; - - if (cflags & REG_ICASE) - { - unsigned i; - - preg->translate = (char *) malloc (CHAR_SET_SIZE); - if (preg->translate == NULL) - return (int) REG_ESPACE; - - /* Map uppercase characters to corresponding lowercase ones. */ - for (i = 0; i < CHAR_SET_SIZE; i++) - preg->translate[i] = ISUPPER (i) ? tolower (i) : i; - } - else - preg->translate = NULL; - - /* If REG_NEWLINE is set, newlines are treated differently. */ - if (cflags & REG_NEWLINE) - { /* REG_NEWLINE implies neither . nor [^...] match newline. */ - syntax &= ~RE_DOT_NEWLINE; - syntax |= RE_HAT_LISTS_NOT_NEWLINE; - /* It also changes the matching behavior. */ - preg->newline_anchor = 1; - } - else - preg->newline_anchor = 0; - - preg->no_sub = !!(cflags & REG_NOSUB); - - /* POSIX says a null character in the pattern terminates it, so we - can use strlen here in compiling the pattern. */ - ret = regex_compile (pattern, strlen (pattern), syntax, preg); - - /* POSIX doesn't distinguish between an unmatched open-group and an - unmatched close-group: both are REG_EPAREN. */ - if (ret == REG_ERPAREN) ret = REG_EPAREN; - - return (int) ret; -} - - -/* regexec searches for a given pattern, specified by PREG, in the - string STRING. - - If NMATCH is zero or REG_NOSUB was set in the cflags argument to - `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at - least NMATCH elements, and we set them to the offsets of the - corresponding matched substrings. - - EFLAGS specifies `execution flags' which affect matching: if - REG_NOTBOL is set, then ^ does not match at the beginning of the - string; if REG_NOTEOL is set, then $ does not match at the end. - - We return 0 if we find a match and REG_NOMATCH if not. */ - -int -regexec (preg, string, nmatch, pmatch, eflags) - const regex_t *preg; - const char *string; - size_t nmatch; - regmatch_t pmatch[]; - int eflags; -{ - int ret; - struct re_registers regs; - regex_t private_preg; - int len = strlen (string); - boolean want_reg_info = !preg->no_sub && nmatch > 0; - - private_preg = *preg; - - private_preg.not_bol = !!(eflags & REG_NOTBOL); - private_preg.not_eol = !!(eflags & REG_NOTEOL); - - /* The user has told us exactly how many registers to return - information about, via `nmatch'. We have to pass that on to the - matching routines. */ - private_preg.regs_allocated = REGS_FIXED; - - if (want_reg_info) - { - regs.num_regs = nmatch; - regs.start = TALLOC (nmatch, regoff_t); - regs.end = TALLOC (nmatch, regoff_t); - if (regs.start == NULL || regs.end == NULL) - return (int) REG_NOMATCH; - } - - /* Perform the searching operation. */ - ret = re_search (&private_preg, string, len, - /* start: */ 0, /* range: */ len, - want_reg_info ? ®s : (struct re_registers *) 0); - - /* Copy the register information to the POSIX structure. */ - if (want_reg_info) - { - if (ret >= 0) - { - unsigned r; - - for (r = 0; r < nmatch; r++) - { - pmatch[r].rm_so = regs.start[r]; - pmatch[r].rm_eo = regs.end[r]; - } - } - - /* If we needed the temporary register info, free the space now. */ - free (regs.start); - free (regs.end); - } - - /* We want zero return to mean success, unlike `re_search'. */ - return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; -} - - -/* Returns a message corresponding to an error code, ERRCODE, returned - from either regcomp or regexec. We don't use PREG here. */ - -size_t regerror (int errcode, const regex_t * preg, char * errbuf, size_t errbuf_size) -{ - const char *msg; - size_t msg_size; - - if (errcode < 0 - || errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0]))) - /* Only error codes returned by the rest of the code should be passed - to this routine. If we are given anything else, or if other regex - code generates an invalid error code, then the program has a bug. - Dump core so we can fix it. */ - abort (); - - msg = re_error_msg[errcode]; - - /* POSIX doesn't require that we do anything in this case, but why - not be nice. */ - if (! msg) - msg = "Success"; - - msg_size = strlen (msg) + 1; /* Includes the null. */ - - if (errbuf_size != 0) - { - if (msg_size > errbuf_size) - { - strncpy (errbuf, msg, errbuf_size - 1); - errbuf[errbuf_size - 1] = 0; - } - else - strcpy (errbuf, msg); - } - - return msg_size; -} - -/* Free dynamically allocated space used by PREG. */ - -void -regfree (preg) - regex_t *preg; -{ - if (preg->buffer != NULL) - free (preg->buffer); - preg->buffer = NULL; - - preg->allocated = 0; - preg->used = 0; - - if (preg->fastmap != NULL) - free (preg->fastmap); - preg->fastmap = NULL; - preg->fastmap_accurate = 0; - - if (preg->translate != NULL) - free (preg->translate); - preg->translate = NULL; -} - -#endif /* not emacs */ - -/* -Local variables: -make-backup-files: t -version-control: t -trim-versions-without-asking: nil -End: -*/ diff --git a/win32/regex/regex.h b/win32/regex/regex.h index 264ed5f..d094d07 100644 --- a/win32/regex/regex.h +++ b/win32/regex/regex.h @@ -1,507 +1,74 @@ -#ifndef __STDC__ -#define __STDC__ 1 -#define no__STDC__ -#endif - -/* Definitions for data structures and routines for the regular - expression library, version 0.12. - - Copyright (C) 1985, 1989, 1990, 1991, 1992, 1993 Free Software Foundation, Inc. - - This program is free software; you can redistribute it and/or modify - it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2, or (at your option) - any later version. - - This program is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - GNU General Public License for more details. - - You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ - -#ifndef __REGEXP_LIBRARY_H__ -#define __REGEXP_LIBRARY_H__ - -/* POSIX says that must be included (by the caller) before - . */ - -#ifdef VMS -/* VMS doesn't have `size_t' in , even though POSIX says it - should be there. */ -#include -#endif - - -/* The following bits are used to determine the regexp syntax we - recognize. The set/not-set meanings are chosen so that Emacs syntax - remains the value 0. The bits are given in alphabetical order, and - the definitions shifted by one from the previous bit; thus, when we - add or remove a bit, only one other definition need change. */ -typedef unsigned reg_syntax_t; - -/* If this bit is not set, then \ inside a bracket expression is literal. - If set, then such a \ quotes the following character. */ -#define RE_BACKSLASH_ESCAPE_IN_LISTS (1) - -/* If this bit is not set, then + and ? are operators, and \+ and \? are - literals. - If set, then \+ and \? are operators and + and ? are literals. */ -#define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) - -/* If this bit is set, then character classes are supported. They are: - [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], - [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. - If not set, then character classes are not supported. */ -#define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) - -/* If this bit is set, then ^ and $ are always anchors (outside bracket - expressions, of course). - If this bit is not set, then it depends: - ^ is an anchor if it is at the beginning of a regular - expression or after an open-group or an alternation operator; - $ is an anchor if it is at the end of a regular expression, or - before a close-group or an alternation operator. - - This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because - POSIX draft 11.2 says that * etc. in leading positions is undefined. - We already implemented a previous draft which made those constructs - invalid, though, so we haven't changed the code back. */ -#define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) - -/* If this bit is set, then special characters are always special - regardless of where they are in the pattern. - If this bit is not set, then special characters are special only in - some contexts; otherwise they are ordinary. Specifically, - * + ? and intervals are only special when not after the beginning, - open-group, or alternation operator. */ -#define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) - -/* If this bit is set, then *, +, ?, and { cannot be first in an re or - immediately after an alternation or begin-group operator. */ -#define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) - -/* If this bit is set, then . matches newline. - If not set, then it doesn't. */ -#define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) - -/* If this bit is set, then . doesn't match NUL. - If not set, then it does. */ -#define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) - -/* If this bit is set, nonmatching lists [^...] do not match newline. - If not set, they do. */ -#define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) - -/* If this bit is set, either \{...\} or {...} defines an - interval, depending on RE_NO_BK_BRACES. - If not set, \{, \}, {, and } are literals. */ -#define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) - -/* If this bit is set, +, ? and | aren't recognized as operators. - If not set, they are. */ -#define RE_LIMITED_OPS (RE_INTERVALS << 1) - -/* If this bit is set, newline is an alternation operator. - If not set, newline is literal. */ -#define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) - -/* If this bit is set, then `{...}' defines an interval, and \{ and \} - are literals. - If not set, then `\{...\}' defines an interval. */ -#define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) - -/* If this bit is set, (...) defines a group, and \( and \) are literals. - If not set, \(...\) defines a group, and ( and ) are literals. */ -#define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) - -/* If this bit is set, then \ matches . - If not set, then \ is a back-reference. */ -#define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) - -/* If this bit is set, then | is an alternation operator, and \| is literal. - If not set, then \| is an alternation operator, and | is literal. */ -#define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) - -/* If this bit is set, then an ending range point collating higher - than the starting range point, as in [z-a], is invalid. - If not set, then when ending range point collates higher than the - starting range point, the range is ignored. */ -#define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) - -/* If this bit is set, then an unmatched ) is ordinary. - If not set, then an unmatched ) is invalid. */ -#define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) - -/* This global variable defines the particular regexp syntax to use (for - some interfaces). When a regexp is compiled, the syntax used is - stored in the pattern buffer, so changing this does not affect - already-compiled regexps. */ -extern reg_syntax_t re_syntax_options; - -/* Define combinations of the above bits for the standard possibilities. - (The [[[ comments delimit what gets put into the Texinfo file, so - don't delete them!) */ -/* [[[begin syntaxes]]] */ -#define RE_SYNTAX_EMACS 0 - -#define RE_SYNTAX_AWK \ - (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ - | RE_NO_BK_PARENS | RE_NO_BK_REFS \ - | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ - | RE_UNMATCHED_RIGHT_PAREN_ORD) - -#define RE_SYNTAX_POSIX_AWK \ - (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS) - -#define RE_SYNTAX_GREP \ - (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ - | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ - | RE_NEWLINE_ALT) - -#define RE_SYNTAX_EGREP \ - (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ - | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ - | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ - | RE_NO_BK_VBAR) - -#define RE_SYNTAX_POSIX_EGREP \ - (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES) - -/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ -#define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC - -#define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC - -/* Syntax bits common to both basic and extended POSIX regex syntax. */ -#define _RE_SYNTAX_POSIX_COMMON \ - (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ - | RE_INTERVALS | RE_NO_EMPTY_RANGES) - -#define RE_SYNTAX_POSIX_BASIC \ - (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM) - -/* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes - RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this - isn't minimal, since other operators, such as \`, aren't disabled. */ -#define RE_SYNTAX_POSIX_MINIMAL_BASIC \ - (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) - -#define RE_SYNTAX_POSIX_EXTENDED \ - (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ - | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ - | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ - | RE_UNMATCHED_RIGHT_PAREN_ORD) - -/* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INVALID_OPS - replaces RE_CONTEXT_INDEP_OPS and RE_NO_BK_REFS is added. */ -#define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ - (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ - | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ - | RE_NO_BK_PARENS | RE_NO_BK_REFS \ - | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) -/* [[[end syntaxes]]] */ - -/* Maximum number of duplicates an interval can allow. Some systems - (erroneously) define this in other header files, but we want our - value, so remove any previous define. */ -#ifdef RE_DUP_MAX -#undef RE_DUP_MAX -#endif -#define RE_DUP_MAX ((1 << 15) - 1) - - -/* POSIX `cflags' bits (i.e., information for `regcomp'). */ - -/* If this bit is set, then use extended regular expression syntax. - If not set, then use basic regular expression syntax. */ -#define REG_EXTENDED 1 - -/* If this bit is set, then ignore case when matching. - If not set, then case is significant. */ -#define REG_ICASE (REG_EXTENDED << 1) - -/* If this bit is set, then anchors do not match at newline - characters in the string. - If not set, then anchors do match at newlines. */ -#define REG_NEWLINE (REG_ICASE << 1) - -/* If this bit is set, then report only success or fail in regexec. - If not set, then returns differ between not matching and errors. */ -#define REG_NOSUB (REG_NEWLINE << 1) - - -/* POSIX `eflags' bits (i.e., information for regexec). */ - -/* If this bit is set, then the beginning-of-line operator doesn't match - the beginning of the string (presumably because it's not the - beginning of a line). - If not set, then the beginning-of-line operator does match the - beginning of the string. */ -#define REG_NOTBOL 1 - -/* Like REG_NOTBOL, except for the end-of-line. */ -#define REG_NOTEOL (1 << 1) - - -/* If any error codes are removed, changed, or added, update the - `re_error_msg' table in regex.c. */ -typedef enum -{ - REG_NOERROR = 0, /* Success. */ - REG_NOMATCH, /* Didn't find a match (for regexec). */ - - /* POSIX regcomp return error codes. (In the order listed in the - standard.) */ - REG_BADPAT, /* Invalid pattern. */ - REG_ECOLLATE, /* Not implemented. */ - REG_ECTYPE, /* Invalid character class name. */ - REG_EESCAPE, /* Trailing backslash. */ - REG_ESUBREG, /* Invalid back reference. */ - REG_EBRACK, /* Unmatched left bracket. */ - REG_EPAREN, /* Parenthesis imbalance. */ - REG_EBRACE, /* Unmatched \{. */ - REG_BADBR, /* Invalid contents of \{\}. */ - REG_ERANGE, /* Invalid range end. */ - REG_ESPACE, /* Ran out of memory. */ - REG_BADRPT, /* No preceding re for repetition op. */ - - /* Error codes we've added. */ - REG_EEND, /* Premature end. */ - REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ - REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ -} reg_errcode_t; - -/* This data structure represents a compiled pattern. Before calling - the pattern compiler, the fields `buffer', `allocated', `fastmap', - `translate', and `no_sub' can be set. After the pattern has been - compiled, the `re_nsub' field is available. All other fields are - private to the regex routines. */ - -struct re_pattern_buffer -{ -/* [[[begin pattern_buffer]]] */ - /* Space that holds the compiled pattern. It is declared as - `unsigned char *' because its elements are - sometimes used as array indexes. */ - unsigned char *buffer; - - /* Number of bytes to which `buffer' points. */ - unsigned long allocated; - - /* Number of bytes actually used in `buffer'. */ - unsigned long used; - - /* Syntax setting with which the pattern was compiled. */ - reg_syntax_t syntax; - - /* Pointer to a fastmap, if any, otherwise zero. re_search uses - the fastmap, if there is one, to skip over impossible - starting points for matches. */ - char *fastmap; - - /* Either a translate table to apply to all characters before - comparing them, or zero for no translation. The translation - is applied to a pattern when it is compiled and to a string - when it is matched. */ - char *translate; - - /* Number of subexpressions found by the compiler. */ - size_t re_nsub; - - /* Zero if this pattern cannot match the empty string, one else. - Well, in truth it's used only in `re_search_2', to see - whether or not we should use the fastmap, so we don't set - this absolutely perfectly; see `re_compile_fastmap' (the - `duplicate' case). */ - unsigned can_be_null : 1; - - /* If REGS_UNALLOCATED, allocate space in the `regs' structure - for `max (RE_NREGS, re_nsub + 1)' groups. - If REGS_REALLOCATE, reallocate space if necessary. - If REGS_FIXED, use what's there. */ -#define REGS_UNALLOCATED 0 -#define REGS_REALLOCATE 1 -#define REGS_FIXED 2 - unsigned regs_allocated : 2; - - /* Set to zero when `regex_compile' compiles a pattern; set to one - by `re_compile_fastmap' if it updates the fastmap. */ - unsigned fastmap_accurate : 1; - - /* If set, `re_match_2' does not return information about - subexpressions. */ - unsigned no_sub : 1; - - /* If set, a beginning-of-line anchor doesn't match at the - beginning of the string. */ - unsigned not_bol : 1; - - /* Similarly for an end-of-line anchor. */ - unsigned not_eol : 1; - - /* If true, an anchor at a newline matches. */ - unsigned newline_anchor : 1; - -/* [[[end pattern_buffer]]] */ -}; - -typedef struct re_pattern_buffer regex_t; - - -/* search.c (search_buffer) in Emacs needs this one opcode value. It is - defined both in `regex.c' and here. */ -#define RE_EXACTN_VALUE 1 - -/* Type for byte offsets within the string. POSIX mandates this. */ -typedef int regoff_t; - - -/* This is the structure we store register match data in. See - regex.texinfo for a full description of what registers match. */ -struct re_registers -{ - unsigned num_regs; - regoff_t *start; - regoff_t *end; -}; - - -/* If `regs_allocated' is REGS_UNALLOCATED in the pattern buffer, - `re_match_2' returns information about at least this many registers - the first time a `regs' structure is passed. */ -#ifndef RE_NREGS -#define RE_NREGS 30 -#endif - - -/* POSIX specification for registers. Aside from the different names than - `re_registers', POSIX uses an array of structures, instead of a - structure of arrays. */ -typedef struct -{ - regoff_t rm_so; /* Byte offset from string's start to substring's start. */ - regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ -} regmatch_t; - -/* Declarations for routines. */ - -/* To avoid duplicating every routine declaration -- once with a - prototype (if we are ANSI), and once without (if we aren't) -- we - use the following macro to declare argument types. This - unfortunately clutters up the declarations a bit, but I think it's - worth it. */ - -#if __STDC__ - -#define _RE_ARGS(args) args - -#else /* not __STDC__ */ - -#define _RE_ARGS(args) () - -#endif /* not __STDC__ */ - +#ifndef _REGEX_H_ +#define _REGEX_H_ /* never again */ +/* ========= begin header generated by ./mkh ========= */ #ifdef __cplusplus extern "C" { #endif -/* Sets the current default syntax to SYNTAX, and return the old syntax. - You can also simply assign to the `re_syntax_options' variable. */ -extern reg_syntax_t re_set_syntax _RE_ARGS ((reg_syntax_t syntax)); - -/* Compile the regular expression PATTERN, with length LENGTH - and syntax given by the global `re_syntax_options', into the buffer - BUFFER. Return NULL if successful, and an error string if not. */ -extern const char *re_compile_pattern - _RE_ARGS ((const char *pattern, int length, - struct re_pattern_buffer *buffer)); - - -/* Compile a fastmap for the compiled pattern in BUFFER; used to - accelerate searches. Return 0 if successful and -2 if was an - internal error. */ -extern int re_compile_fastmap _RE_ARGS ((struct re_pattern_buffer *buffer)); - - -/* Search in the string STRING (with length LENGTH) for the pattern - compiled into BUFFER. Start searching at position START, for RANGE - characters. Return the starting position of the match, -1 for no - match, or -2 for an internal error. Also return register - information in REGS (if REGS and BUFFER->no_sub are nonzero). */ -extern int re_search - _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string, - int length, int start, int range, struct re_registers *regs)); - - -/* Like `re_search', but search in the concatenation of STRING1 and - STRING2. Also, stop searching at index START + STOP. */ -extern int re_search_2 - _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1, - int length1, const char *string2, int length2, - int start, int range, struct re_registers *regs, int stop)); - - -/* Like `re_search', but return how many characters in STRING the regexp - in BUFFER matched, starting at position START. */ -extern int re_match - _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string, - int length, int start, struct re_registers *regs)); - - -/* Relates to `re_match' as `re_search_2' relates to `re_search'. */ -extern int re_match_2 - _RE_ARGS ((struct re_pattern_buffer *buffer, const char *string1, - int length1, const char *string2, int length2, - int start, struct re_registers *regs, int stop)); - - -/* Set REGS to hold NUM_REGS registers, storing them in STARTS and - ENDS. Subsequent matches using BUFFER and REGS will use this memory - for recording register information. STARTS and ENDS must be - allocated with malloc, and must each be at least `NUM_REGS * sizeof - (regoff_t)' bytes long. - - If NUM_REGS == 0, then subsequent matches should allocate their own - register data. - - Unless this function is called, the first search or match using - PATTERN_BUFFER will allocate its own register data, without - freeing the old data. */ -extern void re_set_registers - _RE_ARGS ((struct re_pattern_buffer *buffer, struct re_registers *regs, - unsigned num_regs, regoff_t *starts, regoff_t *ends)); +/* === regex2.h === */ +typedef off_t regoff_t; +typedef struct { + int re_magic; + size_t re_nsub; /* number of parenthesized subexpressions */ + const char *re_endp; /* end pointer for REG_PEND */ + struct re_guts *re_g; /* none of your business :-) */ +} regex_t; +typedef struct { + regoff_t rm_so; /* start of match */ + regoff_t rm_eo; /* end of match */ +} regmatch_t; -/* 4.2 bsd compatibility. */ -extern char *re_comp _RE_ARGS ((const char *)); -extern int re_exec _RE_ARGS ((const char *)); -/* POSIX compatibility. */ -extern int regcomp _RE_ARGS ((regex_t *preg, const char *pattern, int cflags)); -extern int regexec - _RE_ARGS ((const regex_t *preg, const char *string, size_t nmatch, - regmatch_t pmatch[], int eflags)); -extern size_t regerror - _RE_ARGS ((int errcode, const regex_t *preg, char *errbuf, - size_t errbuf_size)); -extern void regfree _RE_ARGS ((regex_t *preg)); +/* === regcomp.c === */ +extern int regcomp(regex_t *, const char *, int); +#define REG_BASIC 0000 +#define REG_EXTENDED 0001 +#define REG_ICASE 0002 +#define REG_NOSUB 0004 +#define REG_NEWLINE 0010 +#define REG_NOSPEC 0020 +#define REG_PEND 0040 +#define REG_DUMP 0200 + + +/* === regerror.c === */ +#define REG_OKAY 0 +#define REG_NOMATCH 1 +#define REG_BADPAT 2 +#define REG_ECOLLATE 3 +#define REG_ECTYPE 4 +#define REG_EESCAPE 5 +#define REG_ESUBREG 6 +#define REG_EBRACK 7 +#define REG_EPAREN 8 +#define REG_EBRACE 9 +#define REG_BADBR 10 +#define REG_ERANGE 11 +#define REG_ESPACE 12 +#define REG_BADRPT 13 +#define REG_EMPTY 14 +#define REG_ASSERT 15 +#define REG_INVARG 16 +#define REG_ATOI 255 /* convert name to number (!) */ +#define REG_ITOA 0400 /* convert number to name (!) */ +extern size_t regerror(int, const regex_t *, char *, size_t); + + +/* === regexec.c === */ +extern int regexec(const regex_t *, const char *, size_t, regmatch_t [], int); +#define REG_NOTBOL 00001 +#define REG_NOTEOL 00002 +#define REG_STARTEND 00004 +#define REG_TRACE 00400 /* tracing of execution */ +#define REG_LARGE 01000 /* force large representation */ +#define REG_BACKR 02000 /* force use of backref code */ + + +/* === regfree.c === */ +extern void regfree(regex_t *); #ifdef __cplusplus } #endif - -#endif /* not __REGEXP_LIBRARY_H__ */ - -/* -Local variables: -make-backup-files: t -version-control: t -trim-versions-without-asking: nil -End: -*/ - -#ifdef no__STDC__ -#undef __STDC__ +/* ========= end header generated by ./mkh ========= */ #endif -- cgit v1.2.3 From 3afee2ec57f662bd06a1da7794ce08f0504da2a0 Mon Sep 17 00:00:00 2001 From: "Nicolas \"Pixel\" Noble" Date: Thu, 31 Jul 2014 00:47:25 -0700 Subject: Forgot to actually add all of these new files for the new regex code. --- win32/regex/cclass.h | 31 + win32/regex/cname.h | 102 +++ win32/regex/engine.c | 1019 ++++++++++++++++++++++++++++++ win32/regex/engine.ih | 35 ++ win32/regex/regcomp.c | 1603 +++++++++++++++++++++++++++++++++++++++++++++++ win32/regex/regcomp.ih | 51 ++ win32/regex/regerror.c | 126 ++++ win32/regex/regerror.ih | 12 + win32/regex/regex2.h | 134 ++++ win32/regex/regexec.c | 138 ++++ win32/regex/regfree.c | 37 ++ win32/regex/split.c | 316 ++++++++++ win32/regex/utils.h | 22 + 13 files changed, 3626 insertions(+) create mode 100644 win32/regex/cclass.h create mode 100644 win32/regex/cname.h create mode 100644 win32/regex/engine.c create mode 100644 win32/regex/engine.ih create mode 100644 win32/regex/regcomp.c create mode 100644 win32/regex/regcomp.ih create mode 100644 win32/regex/regerror.c create mode 100644 win32/regex/regerror.ih create mode 100644 win32/regex/regex2.h create mode 100644 win32/regex/regexec.c create mode 100644 win32/regex/regfree.c create mode 100644 win32/regex/split.c create mode 100644 win32/regex/utils.h (limited to 'win32/regex') diff --git a/win32/regex/cclass.h b/win32/regex/cclass.h new file mode 100644 index 0000000..0c29302 --- /dev/null +++ b/win32/regex/cclass.h @@ -0,0 +1,31 @@ +/* character-class table */ +static struct cclass { + char *name; + char *chars; + char *multis; +} cclasses[] = { + "alnum", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\ +0123456789", "", + "alpha", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz", + "", + "blank", " \t", "", + "cntrl", "\007\b\t\n\v\f\r\1\2\3\4\5\6\16\17\20\21\22\23\24\ +\25\26\27\30\31\32\33\34\35\36\37\177", "", + "digit", "0123456789", "", + "graph", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\ +0123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", + "", + "lower", "abcdefghijklmnopqrstuvwxyz", + "", + "print", "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\ +0123456789!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~ ", + "", + "punct", "!\"#$%&'()*+,-./:;<=>?@[\\]^_`{|}~", + "", + "space", "\t\n\v\f\r ", "", + "upper", "ABCDEFGHIJKLMNOPQRSTUVWXYZ", + "", + "xdigit", "0123456789ABCDEFabcdef", + "", + NULL, 0, "" +}; diff --git a/win32/regex/cname.h b/win32/regex/cname.h new file mode 100644 index 0000000..02e86e9 --- /dev/null +++ b/win32/regex/cname.h @@ -0,0 +1,102 @@ +/* character-name table */ +static struct cname { + char *name; + char code; +} cnames[] = { + "NUL", '\0', + "SOH", '\001', + "STX", '\002', + "ETX", '\003', + "EOT", '\004', + "ENQ", '\005', + "ACK", '\006', + "BEL", '\007', + "alert", '\007', + "BS", '\010', + "backspace", '\b', + "HT", '\011', + "tab", '\t', + "LF", '\012', + "newline", '\n', + "VT", '\013', + "vertical-tab", '\v', + "FF", '\014', + "form-feed", '\f', + "CR", '\015', + "carriage-return", '\r', + "SO", '\016', + "SI", '\017', + "DLE", '\020', + "DC1", '\021', + "DC2", '\022', + "DC3", '\023', + "DC4", '\024', + "NAK", '\025', + "SYN", '\026', + "ETB", '\027', + "CAN", '\030', + "EM", '\031', + "SUB", '\032', + "ESC", '\033', + "IS4", '\034', + "FS", '\034', + "IS3", '\035', + "GS", '\035', + "IS2", '\036', + "RS", '\036', + "IS1", '\037', + "US", '\037', + "space", ' ', + "exclamation-mark", '!', + "quotation-mark", '"', + "number-sign", '#', + "dollar-sign", '$', + "percent-sign", '%', + "ampersand", '&', + "apostrophe", '\'', + "left-parenthesis", '(', + "right-parenthesis", ')', + "asterisk", '*', + "plus-sign", '+', + "comma", ',', + "hyphen", '-', + "hyphen-minus", '-', + "period", '.', + "full-stop", '.', + "slash", '/', + "solidus", '/', + "zero", '0', + "one", '1', + "two", '2', + "three", '3', + "four", '4', + "five", '5', + "six", '6', + "seven", '7', + "eight", '8', + "nine", '9', + "colon", ':', + "semicolon", ';', + "less-than-sign", '<', + "equals-sign", '=', + "greater-than-sign", '>', + "question-mark", '?', + "commercial-at", '@', + "left-square-bracket", '[', + "backslash", '\\', + "reverse-solidus", '\\', + "right-square-bracket", ']', + "circumflex", '^', + "circumflex-accent", '^', + "underscore", '_', + "low-line", '_', + "grave-accent", '`', + "left-brace", '{', + "left-curly-bracket", '{', + "vertical-line", '|', + "right-brace", '}', + "right-curly-bracket", '}', + "tilde", '~', + "DEL", '\177', + NULL, 0, +}; diff --git a/win32/regex/engine.c b/win32/regex/engine.c new file mode 100644 index 0000000..919fe3f --- /dev/null +++ b/win32/regex/engine.c @@ -0,0 +1,1019 @@ +/* + * The matching engine and friends. This file is #included by regexec.c + * after suitable #defines of a variety of macros used herein, so that + * different state representations can be used without duplicating masses + * of code. + */ + +#ifdef SNAMES +#define matcher smatcher +#define fast sfast +#define slow sslow +#define dissect sdissect +#define backref sbackref +#define step sstep +#define print sprint +#define at sat +#define match smat +#endif +#ifdef LNAMES +#define matcher lmatcher +#define fast lfast +#define slow lslow +#define dissect ldissect +#define backref lbackref +#define step lstep +#define print lprint +#define at lat +#define match lmat +#endif + +/* another structure passed up and down to avoid zillions of parameters */ +struct match { + struct re_guts *g; + int eflags; + regmatch_t *pmatch; /* [nsub+1] (0 element unused) */ + char *offp; /* offsets work from here */ + char *beginp; /* start of string -- virtual NUL precedes */ + char *endp; /* end of string -- virtual NUL here */ + char *coldp; /* can be no match starting before here */ + char **lastpos; /* [nplus+1] */ + STATEVARS; + states st; /* current states */ + states fresh; /* states for a fresh start */ + states tmp; /* temporary */ + states empty; /* empty set of states */ +}; + +#include "engine.ih" + +#ifdef REDEBUG +#define SP(t, s, c) print(m, t, s, c, stdout) +#define AT(t, p1, p2, s1, s2) at(m, t, p1, p2, s1, s2) +#define NOTE(str) { if (m->eflags®_TRACE) printf("=%s\n", (str)); } +#else +#define SP(t, s, c) /* nothing */ +#define AT(t, p1, p2, s1, s2) /* nothing */ +#define NOTE(s) /* nothing */ +#endif + +/* + - matcher - the actual matching engine + == static int matcher(register struct re_guts *g, char *string, \ + == size_t nmatch, regmatch_t pmatch[], int eflags); + */ +static int /* 0 success, REG_NOMATCH failure */ +matcher(g, string, nmatch, pmatch, eflags) +register struct re_guts *g; +char *string; +size_t nmatch; +regmatch_t pmatch[]; +int eflags; +{ + register char *endp; + register int i; + struct match mv; + register struct match *m = &mv; + register char *dp; + const register sopno gf = g->firststate+1; /* +1 for OEND */ + const register sopno gl = g->laststate; + char *start; + char *stop; + + /* simplify the situation where possible */ + if (g->cflags®_NOSUB) + nmatch = 0; + if (eflags®_STARTEND) { + start = string + pmatch[0].rm_so; + stop = string + pmatch[0].rm_eo; + } else { + start = string; + stop = start + strlen(start); + } + if (stop < start) + return(REG_INVARG); + + /* prescreening; this does wonders for this rather slow code */ + if (g->must != NULL) { + for (dp = start; dp < stop; dp++) + if (*dp == g->must[0] && stop - dp >= g->mlen && + memcmp(dp, g->must, (size_t)g->mlen) == 0) + break; + if (dp == stop) /* we didn't find g->must */ + return(REG_NOMATCH); + } + + /* match struct setup */ + m->g = g; + m->eflags = eflags; + m->pmatch = NULL; + m->lastpos = NULL; + m->offp = string; + m->beginp = start; + m->endp = stop; + STATESETUP(m, 4); + SETUP(m->st); + SETUP(m->fresh); + SETUP(m->tmp); + SETUP(m->empty); + CLEAR(m->empty); + + /* this loop does only one repetition except for backrefs */ + for (;;) { + endp = fast(m, start, stop, gf, gl); + if (endp == NULL) { /* a miss */ + STATETEARDOWN(m); + return(REG_NOMATCH); + } + if (nmatch == 0 && !g->backrefs) + break; /* no further info needed */ + + /* where? */ + assert(m->coldp != NULL); + for (;;) { + NOTE("finding start"); + endp = slow(m, m->coldp, stop, gf, gl); + if (endp != NULL) + break; + assert(m->coldp < m->endp); + m->coldp++; + } + if (nmatch == 1 && !g->backrefs) + break; /* no further info needed */ + + /* oh my, he wants the subexpressions... */ + if (m->pmatch == NULL) + m->pmatch = (regmatch_t *)malloc((m->g->nsub + 1) * + sizeof(regmatch_t)); + if (m->pmatch == NULL) { + STATETEARDOWN(m); + return(REG_ESPACE); + } + for (i = 1; i <= m->g->nsub; i++) + m->pmatch[i].rm_so = m->pmatch[i].rm_eo = -1; + if (!g->backrefs && !(m->eflags®_BACKR)) { + NOTE("dissecting"); + dp = dissect(m, m->coldp, endp, gf, gl); + } else { + if (g->nplus > 0 && m->lastpos == NULL) + m->lastpos = (char **)malloc((g->nplus+1) * + sizeof(char *)); + if (g->nplus > 0 && m->lastpos == NULL) { + free(m->pmatch); + STATETEARDOWN(m); + return(REG_ESPACE); + } + NOTE("backref dissect"); + dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); + } + if (dp != NULL) + break; + + /* uh-oh... we couldn't find a subexpression-level match */ + assert(g->backrefs); /* must be back references doing it */ + assert(g->nplus == 0 || m->lastpos != NULL); + for (;;) { + if (dp != NULL || endp <= m->coldp) + break; /* defeat */ + NOTE("backoff"); + endp = slow(m, m->coldp, endp-1, gf, gl); + if (endp == NULL) + break; /* defeat */ + /* try it on a shorter possibility */ +#ifndef NDEBUG + for (i = 1; i <= m->g->nsub; i++) { + assert(m->pmatch[i].rm_so == -1); + assert(m->pmatch[i].rm_eo == -1); + } +#endif + NOTE("backoff dissect"); + dp = backref(m, m->coldp, endp, gf, gl, (sopno)0); + } + assert(dp == NULL || dp == endp); + if (dp != NULL) /* found a shorter one */ + break; + + /* despite initial appearances, there is no match here */ + NOTE("false alarm"); + start = m->coldp + 1; /* recycle starting later */ + assert(start <= stop); + } + + /* fill in the details if requested */ + if (nmatch > 0) { + pmatch[0].rm_so = m->coldp - m->offp; + pmatch[0].rm_eo = endp - m->offp; + } + if (nmatch > 1) { + assert(m->pmatch != NULL); + for (i = 1; i < nmatch; i++) + if (i <= m->g->nsub) + pmatch[i] = m->pmatch[i]; + else { + pmatch[i].rm_so = -1; + pmatch[i].rm_eo = -1; + } + } + + if (m->pmatch != NULL) + free((char *)m->pmatch); + if (m->lastpos != NULL) + free((char *)m->lastpos); + STATETEARDOWN(m); + return(0); +} + +/* + - dissect - figure out what matched what, no back references + == static char *dissect(register struct match *m, char *start, \ + == char *stop, sopno startst, sopno stopst); + */ +static char * /* == stop (success) always */ +dissect(m, start, stop, startst, stopst) +register struct match *m; +char *start; +char *stop; +sopno startst; +sopno stopst; +{ + register int i; + register sopno ss; /* start sop of current subRE */ + register sopno es; /* end sop of current subRE */ + register char *sp; /* start of string matched by it */ + register char *stp; /* string matched by it cannot pass here */ + register char *rest; /* start of rest of string */ + register char *tail; /* string unmatched by rest of RE */ + register sopno ssub; /* start sop of subsubRE */ + register sopno esub; /* end sop of subsubRE */ + register char *ssp; /* start of string matched by subsubRE */ + register char *sep; /* end of string matched by subsubRE */ + register char *oldssp; /* previous ssp */ + register char *dp; + + AT("diss", start, stop, startst, stopst); + sp = start; + for (ss = startst; ss < stopst; ss = es) { + /* identify end of subRE */ + es = ss; + switch (OP(m->g->strip[es])) { + case OPLUS_: + case OQUEST_: + es += OPND(m->g->strip[es]); + break; + case OCH_: + while (OP(m->g->strip[es]) != O_CH) + es += OPND(m->g->strip[es]); + break; + } + es++; + + /* figure out what it matched */ + switch (OP(m->g->strip[ss])) { + case OEND: + assert(nope); + break; + case OCHAR: + sp++; + break; + case OBOL: + case OEOL: + case OBOW: + case OEOW: + break; + case OANY: + case OANYOF: + sp++; + break; + case OBACK_: + case O_BACK: + assert(nope); + break; + /* cases where length of match is hard to find */ + case OQUEST_: + stp = stop; + for (;;) { + /* how long could this one be? */ + rest = slow(m, sp, stp, ss, es); + assert(rest != NULL); /* it did match */ + /* could the rest match the rest? */ + tail = slow(m, rest, stop, es, stopst); + if (tail == stop) + break; /* yes! */ + /* no -- try a shorter match for this one */ + stp = rest - 1; + assert(stp >= sp); /* it did work */ + } + ssub = ss + 1; + esub = es - 1; + /* did innards match? */ + if (slow(m, sp, rest, ssub, esub) != NULL) { + dp = dissect(m, sp, rest, ssub, esub); + assert(dp == rest); + } else /* no */ + assert(sp == rest); + sp = rest; + break; + case OPLUS_: + stp = stop; + for (;;) { + /* how long could this one be? */ + rest = slow(m, sp, stp, ss, es); + assert(rest != NULL); /* it did match */ + /* could the rest match the rest? */ + tail = slow(m, rest, stop, es, stopst); + if (tail == stop) + break; /* yes! */ + /* no -- try a shorter match for this one */ + stp = rest - 1; + assert(stp >= sp); /* it did work */ + } + ssub = ss + 1; + esub = es - 1; + ssp = sp; + oldssp = ssp; + for (;;) { /* find last match of innards */ + sep = slow(m, ssp, rest, ssub, esub); + if (sep == NULL || sep == ssp) + break; /* failed or matched null */ + oldssp = ssp; /* on to next try */ + ssp = sep; + } + if (sep == NULL) { + /* last successful match */ + sep = ssp; + ssp = oldssp; + } + assert(sep == rest); /* must exhaust substring */ + assert(slow(m, ssp, sep, ssub, esub) == rest); + dp = dissect(m, ssp, sep, ssub, esub); + assert(dp == sep); + sp = rest; + break; + case OCH_: + stp = stop; + for (;;) { + /* how long could this one be? */ + rest = slow(m, sp, stp, ss, es); + assert(rest != NULL); /* it did match */ + /* could the rest match the rest? */ + tail = slow(m, rest, stop, es, stopst); + if (tail == stop) + break; /* yes! */ + /* no -- try a shorter match for this one */ + stp = rest - 1; + assert(stp >= sp); /* it did work */ + } + ssub = ss + 1; + esub = ss + OPND(m->g->strip[ss]) - 1; + assert(OP(m->g->strip[esub]) == OOR1); + for (;;) { /* find first matching branch */ + if (slow(m, sp, rest, ssub, esub) == rest) + break; /* it matched all of it */ + /* that one missed, try next one */ + assert(OP(m->g->strip[esub]) == OOR1); + esub++; + assert(OP(m->g->strip[esub]) == OOR2); + ssub = esub + 1; + esub += OPND(m->g->strip[esub]); + if (OP(m->g->strip[esub]) == OOR2) + esub--; + else + assert(OP(m->g->strip[esub]) == O_CH); + } + dp = dissect(m, sp, rest, ssub, esub); + assert(dp == rest); + sp = rest; + break; + case O_PLUS: + case O_QUEST: + case OOR1: + case OOR2: + case O_CH: + assert(nope); + break; + case OLPAREN: + i = OPND(m->g->strip[ss]); + assert(0 < i && i <= m->g->nsub); + m->pmatch[i].rm_so = sp - m->offp; + break; + case ORPAREN: + i = OPND(m->g->strip[ss]); + assert(0 < i && i <= m->g->nsub); + m->pmatch[i].rm_eo = sp - m->offp; + break; + default: /* uh oh */ + assert(nope); + break; + } + } + + assert(sp == stop); + return(sp); +} + +/* + - backref - figure out what matched what, figuring in back references + == static char *backref(register struct match *m, char *start, \ + == char *stop, sopno startst, sopno stopst, sopno lev); + */ +static char * /* == stop (success) or NULL (failure) */ +backref(m, start, stop, startst, stopst, lev) +register struct match *m; +char *start; +char *stop; +sopno startst; +sopno stopst; +sopno lev; /* PLUS nesting level */ +{ + register int i; + register sopno ss; /* start sop of current subRE */ + register char *sp; /* start of string matched by it */ + register sopno ssub; /* start sop of subsubRE */ + register sopno esub; /* end sop of subsubRE */ + register char *ssp; /* start of string matched by subsubRE */ + register char *dp; + register size_t len; + register int hard; + register sop s; + register regoff_t offsave; + register cset *cs; + + AT("back", start, stop, startst, stopst); + sp = start; + + /* get as far as we can with easy stuff */ + hard = 0; + for (ss = startst; !hard && ss < stopst; ss++) + switch (OP(s = m->g->strip[ss])) { + case OCHAR: + if (sp == stop || *sp++ != (char)OPND(s)) + return(NULL); + break; + case OANY: + if (sp == stop) + return(NULL); + sp++; + break; + case OANYOF: + cs = &m->g->sets[OPND(s)]; + if (sp == stop || !CHIN(cs, *sp++)) + return(NULL); + break; + case OBOL: + if ( (sp == m->beginp && !(m->eflags®_NOTBOL)) || + (sp < m->endp && *(sp-1) == '\n' && + (m->g->cflags®_NEWLINE)) ) + { /* yes */ } + else + return(NULL); + break; + case OEOL: + if ( (sp == m->endp && !(m->eflags®_NOTEOL)) || + (sp < m->endp && *sp == '\n' && + (m->g->cflags®_NEWLINE)) ) + { /* yes */ } + else + return(NULL); + break; + case OBOW: + if (( (sp == m->beginp && !(m->eflags®_NOTBOL)) || + (sp < m->endp && *(sp-1) == '\n' && + (m->g->cflags®_NEWLINE)) || + (sp > m->beginp && + !ISWORD(*(sp-1))) ) && + (sp < m->endp && ISWORD(*sp)) ) + { /* yes */ } + else + return(NULL); + break; + case OEOW: + if (( (sp == m->endp && !(m->eflags®_NOTEOL)) || + (sp < m->endp && *sp == '\n' && + (m->g->cflags®_NEWLINE)) || + (sp < m->endp && !ISWORD(*sp)) ) && + (sp > m->beginp && ISWORD(*(sp-1))) ) + { /* yes */ } + else + return(NULL); + break; + case O_QUEST: + break; + case OOR1: /* matches null but needs to skip */ + ss++; + s = m->g->strip[ss]; + do { + assert(OP(s) == OOR2); + ss += OPND(s); + } while (OP(s = m->g->strip[ss]) != O_CH); + /* note that the ss++ gets us past the O_CH */ + break; + default: /* have to make a choice */ + hard = 1; + break; + } + if (!hard) { /* that was it! */ + if (sp != stop) + return(NULL); + return(sp); + } + ss--; /* adjust for the for's final increment */ + + /* the hard stuff */ + AT("hard", sp, stop, ss, stopst); + s = m->g->strip[ss]; + switch (OP(s)) { + case OBACK_: /* the vilest depths */ + i = OPND(s); + assert(0 < i && i <= m->g->nsub); + if (m->pmatch[i].rm_eo == -1) + return(NULL); + assert(m->pmatch[i].rm_so != -1); + len = m->pmatch[i].rm_eo - m->pmatch[i].rm_so; + assert(stop - m->beginp >= len); + if (sp > stop - len) + return(NULL); /* not enough left to match */ + ssp = m->offp + m->pmatch[i].rm_so; + if (memcmp(sp, ssp, len) != 0) + return(NULL); + while (m->g->strip[ss] != SOP(O_BACK, i)) + ss++; + return(backref(m, sp+len, stop, ss+1, stopst, lev)); + break; + case OQUEST_: /* to null or not */ + dp = backref(m, sp, stop, ss+1, stopst, lev); + if (dp != NULL) + return(dp); /* not */ + return(backref(m, sp, stop, ss+OPND(s)+1, stopst, lev)); + break; + case OPLUS_: + assert(m->lastpos != NULL); + assert(lev+1 <= m->g->nplus); + m->lastpos[lev+1] = sp; + return(backref(m, sp, stop, ss+1, stopst, lev+1)); + break; + case O_PLUS: + if (sp == m->lastpos[lev]) /* last pass matched null */ + return(backref(m, sp, stop, ss+1, stopst, lev-1)); + /* try another pass */ + m->lastpos[lev] = sp; + dp = backref(m, sp, stop, ss-OPND(s)+1, stopst, lev); + if (dp == NULL) + return(backref(m, sp, stop, ss+1, stopst, lev-1)); + else + return(dp); + break; + case OCH_: /* find the right one, if any */ + ssub = ss + 1; + esub = ss + OPND(s) - 1; + assert(OP(m->g->strip[esub]) == OOR1); + for (;;) { /* find first matching branch */ + dp = backref(m, sp, stop, ssub, esub, lev); + if (dp != NULL) + return(dp); + /* that one missed, try next one */ + if (OP(m->g->strip[esub]) == O_CH) + return(NULL); /* there is none */ + esub++; + assert(OP(m->g->strip[esub]) == OOR2); + ssub = esub + 1; + esub += OPND(m->g->strip[esub]); + if (OP(m->g->strip[esub]) == OOR2) + esub--; + else + assert(OP(m->g->strip[esub]) == O_CH); + } + break; + case OLPAREN: /* must undo assignment if rest fails */ + i = OPND(s); + assert(0 < i && i <= m->g->nsub); + offsave = m->pmatch[i].rm_so; + m->pmatch[i].rm_so = sp - m->offp; + dp = backref(m, sp, stop, ss+1, stopst, lev); + if (dp != NULL) + return(dp); + m->pmatch[i].rm_so = offsave; + return(NULL); + break; + case ORPAREN: /* must undo assignment if rest fails */ + i = OPND(s); + assert(0 < i && i <= m->g->nsub); + offsave = m->pmatch[i].rm_eo; + m->pmatch[i].rm_eo = sp - m->offp; + dp = backref(m, sp, stop, ss+1, stopst, lev); + if (dp != NULL) + return(dp); + m->pmatch[i].rm_eo = offsave; + return(NULL); + break; + default: /* uh oh */ + assert(nope); + break; + } + + /* "can't happen" */ + assert(nope); + /* NOTREACHED */ + return((char *)NULL); /* dummy */ +} + +/* + - fast - step through the string at top speed + == static char *fast(register struct match *m, char *start, \ + == char *stop, sopno startst, sopno stopst); + */ +static char * /* where tentative match ended, or NULL */ +fast(m, start, stop, startst, stopst) +register struct match *m; +char *start; +char *stop; +sopno startst; +sopno stopst; +{ + register states st = m->st; + register states fresh = m->fresh; + register states tmp = m->tmp; + register char *p = start; + register int c = (start == m->beginp) ? OUT : *(start-1); + register int lastc; /* previous c */ + register int flagch; + register int i; + register char *coldp; /* last p after which no match was underway */ + + CLEAR(st); + SET1(st, startst); + st = step(m->g, startst, stopst, st, NOTHING, st); + ASSIGN(fresh, st); + SP("start", st, *p); + coldp = NULL; + for (;;) { + /* next character */ + lastc = c; + c = (p == m->endp) ? OUT : *p; + if (EQ(st, fresh)) + coldp = p; + + /* is there an EOL and/or BOL between lastc and c? */ + flagch = '\0'; + i = 0; + if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || + (lastc == OUT && !(m->eflags®_NOTBOL)) ) { + flagch = BOL; + i = m->g->nbol; + } + if ( (c == '\n' && m->g->cflags®_NEWLINE) || + (c == OUT && !(m->eflags®_NOTEOL)) ) { + flagch = (flagch == BOL) ? BOLEOL : EOL; + i += m->g->neol; + } + if (i != 0) { + for (; i > 0; i--) + st = step(m->g, startst, stopst, st, flagch, st); + SP("boleol", st, c); + } + + /* how about a word boundary? */ + if ( (flagch == BOL || (lastc != OUT && !ISWORD(lastc))) && + (c != OUT && ISWORD(c)) ) { + flagch = BOW; + } + if ( (lastc != OUT && ISWORD(lastc)) && + (flagch == EOL || (c != OUT && !ISWORD(c))) ) { + flagch = EOW; + } + if (flagch == BOW || flagch == EOW) { + st = step(m->g, startst, stopst, st, flagch, st); + SP("boweow", st, c); + } + + /* are we done? */ + if (ISSET(st, stopst) || p == stop) + break; /* NOTE BREAK OUT */ + + /* no, we must deal with this character */ + ASSIGN(tmp, st); + ASSIGN(st, fresh); + assert(c != OUT); + st = step(m->g, startst, stopst, tmp, c, st); + SP("aft", st, c); + assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); + p++; + } + + assert(coldp != NULL); + m->coldp = coldp; + if (ISSET(st, stopst)) + return(p+1); + else + return(NULL); +} + +/* + - slow - step through the string more deliberately + == static char *slow(register struct match *m, char *start, \ + == char *stop, sopno startst, sopno stopst); + */ +static char * /* where it ended */ +slow(m, start, stop, startst, stopst) +register struct match *m; +char *start; +char *stop; +sopno startst; +sopno stopst; +{ + register states st = m->st; + register states empty = m->empty; + register states tmp = m->tmp; + register char *p = start; + register int c = (start == m->beginp) ? OUT : *(start-1); + register int lastc; /* previous c */ + register int flagch; + register int i; + register char *matchp; /* last p at which a match ended */ + + AT("slow", start, stop, startst, stopst); + CLEAR(st); + SET1(st, startst); + SP("sstart", st, *p); + st = step(m->g, startst, stopst, st, NOTHING, st); + matchp = NULL; + for (;;) { + /* next character */ + lastc = c; + c = (p == m->endp) ? OUT : *p; + + /* is there an EOL and/or BOL between lastc and c? */ + flagch = '\0'; + i = 0; + if ( (lastc == '\n' && m->g->cflags®_NEWLINE) || + (lastc == OUT && !(m->eflags®_NOTBOL)) ) { + flagch = BOL; + i = m->g->nbol; + } + if ( (c == '\n' && m->g->cflags®_NEWLINE) || + (c == OUT && !(m->eflags®_NOTEOL)) ) { + flagch = (flagch == BOL) ? BOLEOL : EOL; + i += m->g->neol; + } + if (i != 0) { + for (; i > 0; i--) + st = step(m->g, startst, stopst, st, flagch, st); + SP("sboleol", st, c); + } + + /* how about a word boundary? */ + if ( (flagch == BOL || (lastc != OUT && !ISWORD(lastc))) && + (c != OUT && ISWORD(c)) ) { + flagch = BOW; + } + if ( (lastc != OUT && ISWORD(lastc)) && + (flagch == EOL || (c != OUT && !ISWORD(c))) ) { + flagch = EOW; + } + if (flagch == BOW || flagch == EOW) { + st = step(m->g, startst, stopst, st, flagch, st); + SP("sboweow", st, c); + } + + /* are we done? */ + if (ISSET(st, stopst)) + matchp = p; + if (EQ(st, empty) || p == stop) + break; /* NOTE BREAK OUT */ + + /* no, we must deal with this character */ + ASSIGN(tmp, st); + ASSIGN(st, empty); + assert(c != OUT); + st = step(m->g, startst, stopst, tmp, c, st); + SP("saft", st, c); + assert(EQ(step(m->g, startst, stopst, st, NOTHING, st), st)); + p++; + } + + return(matchp); +} + + +/* + - step - map set of states reachable before char to set reachable after + == static states step(register struct re_guts *g, sopno start, sopno stop, \ + == register states bef, int ch, register states aft); + == #define BOL (OUT+1) + == #define EOL (BOL+1) + == #define BOLEOL (BOL+2) + == #define NOTHING (BOL+3) + == #define BOW (BOL+4) + == #define EOW (BOL+5) + == #define CODEMAX (BOL+5) // highest code used + == #define NONCHAR(c) ((c) > CHAR_MAX) + == #define NNONCHAR (CODEMAX-CHAR_MAX) + */ +static states +step(g, start, stop, bef, ch, aft) +register struct re_guts *g; +sopno start; /* start state within strip */ +sopno stop; /* state after stop state within strip */ +register states bef; /* states reachable before */ +int ch; /* character or NONCHAR code */ +register states aft; /* states already known reachable after */ +{ + register cset *cs; + register sop s; + register sopno pc; + register onestate here; /* note, macros know this name */ + register sopno look; + register long i; + + for (pc = start, INIT(here, pc); pc != stop; pc++, INC(here)) { + s = g->strip[pc]; + switch (OP(s)) { + case OEND: + assert(pc == stop-1); + break; + case OCHAR: + /* only characters can match */ + assert(!NONCHAR(ch) || ch != (char)OPND(s)); + if (ch == (char)OPND(s)) + FWD(aft, bef, 1); + break; + case OBOL: + if (ch == BOL || ch == BOLEOL) + FWD(aft, bef, 1); + break; + case OEOL: + if (ch == EOL || ch == BOLEOL) + FWD(aft, bef, 1); + break; + case OBOW: + if (ch == BOW) + FWD(aft, bef, 1); + break; + case OEOW: + if (ch == EOW) + FWD(aft, bef, 1); + break; + case OANY: + if (!NONCHAR(ch)) + FWD(aft, bef, 1); + break; + case OANYOF: + cs = &g->sets[OPND(s)]; + if (!NONCHAR(ch) && CHIN(cs, ch)) + FWD(aft, bef, 1); + break; + case OBACK_: /* ignored here */ + case O_BACK: + FWD(aft, aft, 1); + break; + case OPLUS_: /* forward, this is just an empty */ + FWD(aft, aft, 1); + break; + case O_PLUS: /* both forward and back */ + FWD(aft, aft, 1); + i = ISSETBACK(aft, OPND(s)); + BACK(aft, aft, OPND(s)); + if (!i && ISSETBACK(aft, OPND(s))) { + /* oho, must reconsider loop body */ + pc -= OPND(s) + 1; + INIT(here, pc); + } + break; + case OQUEST_: /* two branches, both forward */ + FWD(aft, aft, 1); + FWD(aft, aft, OPND(s)); + break; + case O_QUEST: /* just an empty */ + FWD(aft, aft, 1); + break; + case OLPAREN: /* not significant here */ + case ORPAREN: + FWD(aft, aft, 1); + break; + case OCH_: /* mark the first two branches */ + FWD(aft, aft, 1); + assert(OP(g->strip[pc+OPND(s)]) == OOR2); + FWD(aft, aft, OPND(s)); + break; + case OOR1: /* done a branch, find the O_CH */ + if (ISSTATEIN(aft, here)) { + for (look = 1; + OP(s = g->strip[pc+look]) != O_CH; + look += OPND(s)) + assert(OP(s) == OOR2); + FWD(aft, aft, look); + } + break; + case OOR2: /* propagate OCH_'s marking */ + FWD(aft, aft, 1); + if (OP(g->strip[pc+OPND(s)]) != O_CH) { + assert(OP(g->strip[pc+OPND(s)]) == OOR2); + FWD(aft, aft, OPND(s)); + } + break; + case O_CH: /* just empty */ + FWD(aft, aft, 1); + break; + default: /* ooooops... */ + assert(nope); + break; + } + } + + return(aft); +} + +#ifdef REDEBUG +/* + - print - print a set of states + == #ifdef REDEBUG + == static void print(struct match *m, char *caption, states st, \ + == int ch, FILE *d); + == #endif + */ +static void +print(m, caption, st, ch, d) +struct match *m; +char *caption; +states st; +int ch; +FILE *d; +{ + register struct re_guts *g = m->g; + register int i; + register int first = 1; + + if (!(m->eflags®_TRACE)) + return; + + fprintf(d, "%s", caption); + if (ch != '\0') + fprintf(d, " %s", pchar(ch)); + for (i = 0; i < g->nstates; i++) + if (ISSET(st, i)) { + fprintf(d, "%s%d", (first) ? "\t" : ", ", i); + first = 0; + } + fprintf(d, "\n"); +} + +/* + - at - print current situation + == #ifdef REDEBUG + == static void at(struct match *m, char *title, char *start, char *stop, \ + == sopno startst, sopno stopst); + == #endif + */ +static void +at(m, title, start, stop, startst, stopst) +struct match *m; +char *title; +char *start; +char *stop; +sopno startst; +sopno stopst; +{ + if (!(m->eflags®_TRACE)) + return; + + printf("%s %s-", title, pchar(*start)); + printf("%s ", pchar(*stop)); + printf("%ld-%ld\n", (long)startst, (long)stopst); +} + +#ifndef PCHARDONE +#define PCHARDONE /* never again */ +/* + - pchar - make a character printable + == #ifdef REDEBUG + == static char *pchar(int ch); + == #endif + * + * Is this identical to regchar() over in debug.c? Well, yes. But a + * duplicate here avoids having a debugging-capable regexec.o tied to + * a matching debug.o, and this is convenient. It all disappears in + * the non-debug compilation anyway, so it doesn't matter much. + */ +static char * /* -> representation */ +pchar(ch) +int ch; +{ + static char pbuf[10]; + + if (isprint(ch) || ch == ' ') + sprintf(pbuf, "%c", ch); + else + sprintf(pbuf, "\\%o", ch); + return(pbuf); +} +#endif +#endif + +#undef matcher +#undef fast +#undef slow +#undef dissect +#undef backref +#undef step +#undef print +#undef at +#undef match diff --git a/win32/regex/engine.ih b/win32/regex/engine.ih new file mode 100644 index 0000000..cc98334 --- /dev/null +++ b/win32/regex/engine.ih @@ -0,0 +1,35 @@ +/* ========= begin header generated by ./mkh ========= */ +#ifdef __cplusplus +extern "C" { +#endif + +/* === engine.c === */ +static int matcher(register struct re_guts *g, char *string, size_t nmatch, regmatch_t pmatch[], int eflags); +static char *dissect(register struct match *m, char *start, char *stop, sopno startst, sopno stopst); +static char *backref(register struct match *m, char *start, char *stop, sopno startst, sopno stopst, sopno lev); +static char *fast(register struct match *m, char *start, char *stop, sopno startst, sopno stopst); +static char *slow(register struct match *m, char *start, char *stop, sopno startst, sopno stopst); +static states step(register struct re_guts *g, sopno start, sopno stop, register states bef, int ch, register states aft); +#define BOL (OUT+1) +#define EOL (BOL+1) +#define BOLEOL (BOL+2) +#define NOTHING (BOL+3) +#define BOW (BOL+4) +#define EOW (BOL+5) +#define CODEMAX (BOL+5) /* highest code used */ +#define NONCHAR(c) ((c) > CHAR_MAX) +#define NNONCHAR (CODEMAX-CHAR_MAX) +#ifdef REDEBUG +static void print(struct match *m, char *caption, states st, int ch, FILE *d); +#endif +#ifdef REDEBUG +static void at(struct match *m, char *title, char *start, char *stop, sopno startst, sopno stopst); +#endif +#ifdef REDEBUG +static char *pchar(int ch); +#endif + +#ifdef __cplusplus +} +#endif +/* ========= end header generated by ./mkh ========= */ diff --git a/win32/regex/regcomp.c b/win32/regex/regcomp.c new file mode 100644 index 0000000..dc48601 --- /dev/null +++ b/win32/regex/regcomp.c @@ -0,0 +1,1603 @@ +#include +#include +#include +#include +#include +#include +#include + +#include "utils.h" +#include "regex2.h" + +#include "cclass.h" +#include "cname.h" + +/* + * parse structure, passed up and down to avoid global variables and + * other clumsinesses + */ +struct parse { + char *next; /* next character in RE */ + char *end; /* end of string (-> NUL normally) */ + int error; /* has an error been seen? */ + sop *strip; /* malloced strip */ + sopno ssize; /* malloced strip size (allocated) */ + sopno slen; /* malloced strip length (used) */ + int ncsalloc; /* number of csets allocated */ + struct re_guts *g; +# define NPAREN 10 /* we need to remember () 1-9 for back refs */ + sopno pbegin[NPAREN]; /* -> ( ([0] unused) */ + sopno pend[NPAREN]; /* -> ) ([0] unused) */ +}; + +#include "regcomp.ih" + +static char nuls[10]; /* place to point scanner in event of error */ + +/* + * macros for use with parse structure + * BEWARE: these know that the parse structure is named `p' !!! + */ +#define PEEK() (*p->next) +#define PEEK2() (*(p->next+1)) +#define MORE() (p->next < p->end) +#define MORE2() (p->next+1 < p->end) +#define SEE(c) (MORE() && PEEK() == (c)) +#define SEETWO(a, b) (MORE() && MORE2() && PEEK() == (a) && PEEK2() == (b)) +#define EAT(c) ((SEE(c)) ? (NEXT(), 1) : 0) +#define EATTWO(a, b) ((SEETWO(a, b)) ? (NEXT2(), 1) : 0) +#define NEXT() (p->next++) +#define NEXT2() (p->next += 2) +#define NEXTn(n) (p->next += (n)) +#define GETNEXT() (*p->next++) +#define SETERROR(e) seterr(p, (e)) +#define REQUIRE(co, e) ((co) || SETERROR(e)) +#define MUSTSEE(c, e) (REQUIRE(MORE() && PEEK() == (c), e)) +#define MUSTEAT(c, e) (REQUIRE(MORE() && GETNEXT() == (c), e)) +#define MUSTNOTSEE(c, e) (REQUIRE(!MORE() || PEEK() != (c), e)) +#define EMIT(op, sopnd) doemit(p, (sop)(op), (size_t)(sopnd)) +#define INSERT(op, pos) doinsert(p, (sop)(op), HERE()-(pos)+1, pos) +#define AHEAD(pos) dofwd(p, pos, HERE()-(pos)) +#define ASTERN(sop, pos) EMIT(sop, HERE()-pos) +#define HERE() (p->slen) +#define THERE() (p->slen - 1) +#define THERETHERE() (p->slen - 2) +#define DROP(n) (p->slen -= (n)) + +#ifndef NDEBUG +static int never = 0; /* for use in asserts; shuts lint up */ +#else +#define never 0 /* some s have bugs too */ +#endif + +/* + - regcomp - interface for parser and compilation + = extern int regcomp(regex_t *, const char *, int); + = #define REG_BASIC 0000 + = #define REG_EXTENDED 0001 + = #define REG_ICASE 0002 + = #define REG_NOSUB 0004 + = #define REG_NEWLINE 0010 + = #define REG_NOSPEC 0020 + = #define REG_PEND 0040 + = #define REG_DUMP 0200 + */ +int /* 0 success, otherwise REG_something */ +regcomp(preg, pattern, cflags) +regex_t *preg; +const char *pattern; +int cflags; +{ + struct parse pa; + register struct re_guts *g; + register struct parse *p = &pa; + register int i; + register size_t len; +#ifdef REDEBUG +# define GOODFLAGS(f) (f) +#else +# define GOODFLAGS(f) ((f)&~REG_DUMP) +#endif + + cflags = GOODFLAGS(cflags); + if ((cflags®_EXTENDED) && (cflags®_NOSPEC)) + return(REG_INVARG); + + if (cflags®_PEND) { + if (preg->re_endp < pattern) + return(REG_INVARG); + len = preg->re_endp - pattern; + } else + len = strlen((char *)pattern); + + /* do the mallocs early so failure handling is easy */ + g = (struct re_guts *)malloc(sizeof(struct re_guts) + + (NC-1)*sizeof(cat_t)); + if (g == NULL) + return(REG_ESPACE); + p->ssize = len/(size_t)2*(size_t)3 + (size_t)1; /* ugh */ + p->strip = (sop *)malloc(p->ssize * sizeof(sop)); + p->slen = 0; + if (p->strip == NULL) { + free((char *)g); + return(REG_ESPACE); + } + + /* set things up */ + p->g = g; + p->next = (char *)pattern; /* convenience; we do not modify it */ + p->end = p->next + len; + p->error = 0; + p->ncsalloc = 0; + for (i = 0; i < NPAREN; i++) { + p->pbegin[i] = 0; + p->pend[i] = 0; + } + g->csetsize = NC; + g->sets = NULL; + g->setbits = NULL; + g->ncsets = 0; + g->cflags = cflags; + g->iflags = 0; + g->nbol = 0; + g->neol = 0; + g->must = NULL; + g->mlen = 0; + g->nsub = 0; + g->ncategories = 1; /* category 0 is "everything else" */ + g->categories = &g->catspace[-(CHAR_MIN)]; + (void) memset((char *)g->catspace, 0, NC*sizeof(cat_t)); + g->backrefs = 0; + + /* do it */ + EMIT(OEND, 0); + g->firststate = THERE(); + if (cflags®_EXTENDED) + p_ere(p, OUT); + else if (cflags®_NOSPEC) + p_str(p); + else + p_bre(p, OUT, OUT); + EMIT(OEND, 0); + g->laststate = THERE(); + + /* tidy up loose ends and fill things in */ + categorize(p, g); + stripsnug(p, g); + findmust(p, g); + g->nplus = pluscount(p, g); + g->magic = MAGIC2; + preg->re_nsub = g->nsub; + preg->re_g = g; + preg->re_magic = MAGIC1; +#ifndef REDEBUG + /* not debugging, so can't rely on the assert() in regexec() */ + if (g->iflags&BAD) + SETERROR(REG_ASSERT); +#endif + + /* win or lose, we're done */ + if (p->error != 0) /* lose */ + regfree(preg); + return(p->error); +} + +/* + - p_ere - ERE parser top level, concatenation and alternation + == static void p_ere(register struct parse *p, int stop); + */ +static void +p_ere(p, stop) +register struct parse *p; +int stop; /* character this ERE should end at */ +{ + register char c; + register sopno prevback; + register sopno prevfwd; + register sopno conc; + register int first = 1; /* is this the first alternative? */ + + for (;;) { + /* do a bunch of concatenated expressions */ + conc = HERE(); + while (MORE() && (c = PEEK()) != '|' && c != stop) + p_ere_exp(p); + REQUIRE(HERE() != conc, REG_EMPTY); /* require nonempty */ + + if (!EAT('|')) + break; /* NOTE BREAK OUT */ + + if (first) { + INSERT(OCH_, conc); /* offset is wrong */ + prevfwd = conc; + prevback = conc; + first = 0; + } + ASTERN(OOR1, prevback); + prevback = THERE(); + AHEAD(prevfwd); /* fix previous offset */ + prevfwd = HERE(); + EMIT(OOR2, 0); /* offset is very wrong */ + } + + if (!first) { /* tail-end fixups */ + AHEAD(prevfwd); + ASTERN(O_CH, prevback); + } + + assert(!MORE() || SEE(stop)); +} + +/* + - p_ere_exp - parse one subERE, an atom possibly followed by a repetition op + == static void p_ere_exp(register struct parse *p); + */ +static void +p_ere_exp(p) +register struct parse *p; +{ + register char c; + register sopno pos; + register int count; + register int count2; + register sopno subno; + int wascaret = 0; + + assert(MORE()); /* caller should have ensured this */ + c = GETNEXT(); + + pos = HERE(); + switch (c) { + case '(': + REQUIRE(MORE(), REG_EPAREN); + p->g->nsub++; + subno = p->g->nsub; + if (subno < NPAREN) + p->pbegin[subno] = HERE(); + EMIT(OLPAREN, subno); + if (!SEE(')')) + p_ere(p, ')'); + if (subno < NPAREN) { + p->pend[subno] = HERE(); + assert(p->pend[subno] != 0); + } + EMIT(ORPAREN, subno); + MUSTEAT(')', REG_EPAREN); + break; +#ifndef POSIX_MISTAKE + case ')': /* happens only if no current unmatched ( */ + /* + * You may ask, why the ifndef? Because I didn't notice + * this until slightly too late for 1003.2, and none of the + * other 1003.2 regular-expression reviewers noticed it at + * all. So an unmatched ) is legal POSIX, at least until + * we can get it fixed. + */ + SETERROR(REG_EPAREN); + break; +#endif + case '^': + EMIT(OBOL, 0); + p->g->iflags |= USEBOL; + p->g->nbol++; + wascaret = 1; + break; + case '$': + EMIT(OEOL, 0); + p->g->iflags |= USEEOL; + p->g->neol++; + break; + case '|': + SETERROR(REG_EMPTY); + break; + case '*': + case '+': + case '?': + SETERROR(REG_BADRPT); + break; + case '.': + if (p->g->cflags®_NEWLINE) + nonnewline(p); + else + EMIT(OANY, 0); + break; + case '[': + p_bracket(p); + break; + case '\\': + REQUIRE(MORE(), REG_EESCAPE); + c = GETNEXT(); + ordinary(p, c); + break; + case '{': /* okay as ordinary except if digit follows */ + REQUIRE(!MORE() || !isdigit(PEEK()), REG_BADRPT); + /* FALLTHROUGH */ + default: + ordinary(p, c); + break; + } + + if (!MORE()) + return; + c = PEEK(); + /* we call { a repetition if followed by a digit */ + if (!( c == '*' || c == '+' || c == '?' || + (c == '{' && MORE2() && isdigit(PEEK2())) )) + return; /* no repetition, we're done */ + NEXT(); + + REQUIRE(!wascaret, REG_BADRPT); + switch (c) { + case '*': /* implemented as +? */ + /* this case does not require the (y|) trick, noKLUDGE */ + INSERT(OPLUS_, pos); + ASTERN(O_PLUS, pos); + INSERT(OQUEST_, pos); + ASTERN(O_QUEST, pos); + break; + case '+': + INSERT(OPLUS_, pos); + ASTERN(O_PLUS, pos); + break; + case '?': + /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */ + INSERT(OCH_, pos); /* offset slightly wrong */ + ASTERN(OOR1, pos); /* this one's right */ + AHEAD(pos); /* fix the OCH_ */ + EMIT(OOR2, 0); /* offset very wrong... */ + AHEAD(THERE()); /* ...so fix it */ + ASTERN(O_CH, THERETHERE()); + break; + case '{': + count = p_count(p); + if (EAT(',')) { + if (isdigit(PEEK())) { + count2 = p_count(p); + REQUIRE(count <= count2, REG_BADBR); + } else /* single number with comma */ + count2 = INFINITY; + } else /* just a single number */ + count2 = count; + repeat(p, pos, count, count2); + if (!EAT('}')) { /* error heuristics */ + while (MORE() && PEEK() != '}') + NEXT(); + REQUIRE(MORE(), REG_EBRACE); + SETERROR(REG_BADBR); + } + break; + } + + if (!MORE()) + return; + c = PEEK(); + if (!( c == '*' || c == '+' || c == '?' || + (c == '{' && MORE2() && isdigit(PEEK2())) ) ) + return; + SETERROR(REG_BADRPT); +} + +/* + - p_str - string (no metacharacters) "parser" + == static void p_str(register struct parse *p); + */ +static void +p_str(p) +register struct parse *p; +{ + REQUIRE(MORE(), REG_EMPTY); + while (MORE()) + ordinary(p, GETNEXT()); +} + +/* + - p_bre - BRE parser top level, anchoring and concatenation + == static void p_bre(register struct parse *p, register int end1, \ + == register int end2); + * Giving end1 as OUT essentially eliminates the end1/end2 check. + * + * This implementation is a bit of a kludge, in that a trailing $ is first + * taken as an ordinary character and then revised to be an anchor. The + * only undesirable side effect is that '$' gets included as a character + * category in such cases. This is fairly harmless; not worth fixing. + * The amount of lookahead needed to avoid this kludge is excessive. + */ +static void +p_bre(p, end1, end2) +register struct parse *p; +register int end1; /* first terminating character */ +register int end2; /* second terminating character */ +{ + register sopno start = HERE(); + register int first = 1; /* first subexpression? */ + register int wasdollar = 0; + + if (EAT('^')) { + EMIT(OBOL, 0); + p->g->iflags |= USEBOL; + p->g->nbol++; + } + while (MORE() && !SEETWO(end1, end2)) { + wasdollar = p_simp_re(p, first); + first = 0; + } + if (wasdollar) { /* oops, that was a trailing anchor */ + DROP(1); + EMIT(OEOL, 0); + p->g->iflags |= USEEOL; + p->g->neol++; + } + + REQUIRE(HERE() != start, REG_EMPTY); /* require nonempty */ +} + +/* + - p_simp_re - parse a simple RE, an atom possibly followed by a repetition + == static int p_simp_re(register struct parse *p, int starordinary); + */ +static int /* was the simple RE an unbackslashed $? */ +p_simp_re(p, starordinary) +register struct parse *p; +int starordinary; /* is a leading * an ordinary character? */ +{ + register int c; + register int count; + register int count2; + register sopno pos; + register int i; + register sopno subno; +# define BACKSL (1<g->cflags®_NEWLINE) + nonnewline(p); + else + EMIT(OANY, 0); + break; + case '[': + p_bracket(p); + break; + case BACKSL|'{': + SETERROR(REG_BADRPT); + break; + case BACKSL|'(': + p->g->nsub++; + subno = p->g->nsub; + if (subno < NPAREN) + p->pbegin[subno] = HERE(); + EMIT(OLPAREN, subno); + /* the MORE here is an error heuristic */ + if (MORE() && !SEETWO('\\', ')')) + p_bre(p, '\\', ')'); + if (subno < NPAREN) { + p->pend[subno] = HERE(); + assert(p->pend[subno] != 0); + } + EMIT(ORPAREN, subno); + REQUIRE(EATTWO('\\', ')'), REG_EPAREN); + break; + case BACKSL|')': /* should not get here -- must be user */ + case BACKSL|'}': + SETERROR(REG_EPAREN); + break; + case BACKSL|'1': + case BACKSL|'2': + case BACKSL|'3': + case BACKSL|'4': + case BACKSL|'5': + case BACKSL|'6': + case BACKSL|'7': + case BACKSL|'8': + case BACKSL|'9': + i = (c&~BACKSL) - '0'; + assert(i < NPAREN); + if (p->pend[i] != 0) { + assert(i <= p->g->nsub); + EMIT(OBACK_, i); + assert(p->pbegin[i] != 0); + assert(OP(p->strip[p->pbegin[i]]) == OLPAREN); + assert(OP(p->strip[p->pend[i]]) == ORPAREN); + (void) dupl(p, p->pbegin[i]+1, p->pend[i]); + EMIT(O_BACK, i); + } else + SETERROR(REG_ESUBREG); + p->g->backrefs = 1; + break; + case '*': + REQUIRE(starordinary, REG_BADRPT); + /* FALLTHROUGH */ + default: + ordinary(p, (char)c); /* takes off BACKSL, if any */ + break; + } + + if (EAT('*')) { /* implemented as +? */ + /* this case does not require the (y|) trick, noKLUDGE */ + INSERT(OPLUS_, pos); + ASTERN(O_PLUS, pos); + INSERT(OQUEST_, pos); + ASTERN(O_QUEST, pos); + } else if (EATTWO('\\', '{')) { + count = p_count(p); + if (EAT(',')) { + if (MORE() && isdigit(PEEK())) { + count2 = p_count(p); + REQUIRE(count <= count2, REG_BADBR); + } else /* single number with comma */ + count2 = INFINITY; + } else /* just a single number */ + count2 = count; + repeat(p, pos, count, count2); + if (!EATTWO('\\', '}')) { /* error heuristics */ + while (MORE() && !SEETWO('\\', '}')) + NEXT(); + REQUIRE(MORE(), REG_EBRACE); + SETERROR(REG_BADBR); + } + } else if (c == (unsigned char)'$') /* $ (but not \$) ends it */ + return(1); + + return(0); +} + +/* + - p_count - parse a repetition count + == static int p_count(register struct parse *p); + */ +static int /* the value */ +p_count(p) +register struct parse *p; +{ + register int count = 0; + register int ndigits = 0; + + while (MORE() && isdigit(PEEK()) && count <= DUPMAX) { + count = count*10 + (GETNEXT() - '0'); + ndigits++; + } + + REQUIRE(ndigits > 0 && count <= DUPMAX, REG_BADBR); + return(count); +} + +/* + - p_bracket - parse a bracketed character list + == static void p_bracket(register struct parse *p); + * + * Note a significant property of this code: if the allocset() did SETERROR, + * no set operations are done. + */ +static void +p_bracket(p) +register struct parse *p; +{ + register cset *cs = allocset(p); + register int invert = 0; + + /* Dept of Truly Sickening Special-Case Kludges */ + if (p->next + 5 < p->end && strncmp(p->next, "[:<:]]", 6) == 0) { + EMIT(OBOW, 0); + NEXTn(6); + return; + } + if (p->next + 5 < p->end && strncmp(p->next, "[:>:]]", 6) == 0) { + EMIT(OEOW, 0); + NEXTn(6); + return; + } + + if (EAT('^')) + invert++; /* make note to invert set at end */ + if (EAT(']')) + CHadd(cs, ']'); + else if (EAT('-')) + CHadd(cs, '-'); + while (MORE() && PEEK() != ']' && !SEETWO('-', ']')) + p_b_term(p, cs); + if (EAT('-')) + CHadd(cs, '-'); + MUSTEAT(']', REG_EBRACK); + + if (p->error != 0) /* don't mess things up further */ + return; + + if (p->g->cflags®_ICASE) { + register int i; + register int ci; + + for (i = p->g->csetsize - 1; i >= 0; i--) + if (CHIN(cs, i) && isalpha(i)) { + ci = othercase(i); + if (ci != i) + CHadd(cs, ci); + } + if (cs->multis != NULL) + mccase(p, cs); + } + if (invert) { + register int i; + + for (i = p->g->csetsize - 1; i >= 0; i--) + if (CHIN(cs, i)) + CHsub(cs, i); + else + CHadd(cs, i); + if (p->g->cflags®_NEWLINE) + CHsub(cs, '\n'); + if (cs->multis != NULL) + mcinvert(p, cs); + } + + assert(cs->multis == NULL); /* xxx */ + + if (nch(p, cs) == 1) { /* optimize singleton sets */ + ordinary(p, firstch(p, cs)); + freeset(p, cs); + } else + EMIT(OANYOF, freezeset(p, cs)); +} + +/* + - p_b_term - parse one term of a bracketed character list + == static void p_b_term(register struct parse *p, register cset *cs); + */ +static void +p_b_term(p, cs) +register struct parse *p; +register cset *cs; +{ + register char c; + register char start, finish; + register int i; + + /* classify what we've got */ + switch ((MORE()) ? PEEK() : '\0') { + case '[': + c = (MORE2()) ? PEEK2() : '\0'; + break; + case '-': + SETERROR(REG_ERANGE); + return; /* NOTE RETURN */ + break; + default: + c = '\0'; + break; + } + + switch (c) { + case ':': /* character class */ + NEXT2(); + REQUIRE(MORE(), REG_EBRACK); + c = PEEK(); + REQUIRE(c != '-' && c != ']', REG_ECTYPE); + p_b_cclass(p, cs); + REQUIRE(MORE(), REG_EBRACK); + REQUIRE(EATTWO(':', ']'), REG_ECTYPE); + break; + case '=': /* equivalence class */ + NEXT2(); + REQUIRE(MORE(), REG_EBRACK); + c = PEEK(); + REQUIRE(c != '-' && c != ']', REG_ECOLLATE); + p_b_eclass(p, cs); + REQUIRE(MORE(), REG_EBRACK); + REQUIRE(EATTWO('=', ']'), REG_ECOLLATE); + break; + default: /* symbol, ordinary character, or range */ +/* xxx revision needed for multichar stuff */ + start = p_b_symbol(p); + if (SEE('-') && MORE2() && PEEK2() != ']') { + /* range */ + NEXT(); + if (EAT('-')) + finish = '-'; + else + finish = p_b_symbol(p); + } else + finish = start; +/* xxx what about signed chars here... */ + REQUIRE(start <= finish, REG_ERANGE); + for (i = start; i <= finish; i++) + CHadd(cs, i); + break; + } +} + +/* + - p_b_cclass - parse a character-class name and deal with it + == static void p_b_cclass(register struct parse *p, register cset *cs); + */ +static void +p_b_cclass(p, cs) +register struct parse *p; +register cset *cs; +{ + register char *sp = p->next; + register struct cclass *cp; + register size_t len; + register char *u; + register char c; + + while (MORE() && isalpha(PEEK())) + NEXT(); + len = p->next - sp; + for (cp = cclasses; cp->name != NULL; cp++) + if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0') + break; + if (cp->name == NULL) { + /* oops, didn't find it */ + SETERROR(REG_ECTYPE); + return; + } + + u = cp->chars; + while ((c = *u++) != '\0') + CHadd(cs, c); + for (u = cp->multis; *u != '\0'; u += strlen(u) + 1) + MCadd(p, cs, u); +} + +/* + - p_b_eclass - parse an equivalence-class name and deal with it + == static void p_b_eclass(register struct parse *p, register cset *cs); + * + * This implementation is incomplete. xxx + */ +static void +p_b_eclass(p, cs) +register struct parse *p; +register cset *cs; +{ + register char c; + + c = p_b_coll_elem(p, '='); + CHadd(cs, c); +} + +/* + - p_b_symbol - parse a character or [..]ed multicharacter collating symbol + == static char p_b_symbol(register struct parse *p); + */ +static char /* value of symbol */ +p_b_symbol(p) +register struct parse *p; +{ + register char value; + + REQUIRE(MORE(), REG_EBRACK); + if (!EATTWO('[', '.')) + return(GETNEXT()); + + /* collating symbol */ + value = p_b_coll_elem(p, '.'); + REQUIRE(EATTWO('.', ']'), REG_ECOLLATE); + return(value); +} + +/* + - p_b_coll_elem - parse a collating-element name and look it up + == static char p_b_coll_elem(register struct parse *p, int endc); + */ +static char /* value of collating element */ +p_b_coll_elem(p, endc) +register struct parse *p; +int endc; /* name ended by endc,']' */ +{ + register char *sp = p->next; + register struct cname *cp; + register int len; + + while (MORE() && !SEETWO(endc, ']')) + NEXT(); + if (!MORE()) { + SETERROR(REG_EBRACK); + return(0); + } + len = p->next - sp; + for (cp = cnames; cp->name != NULL; cp++) + if (strncmp(cp->name, sp, len) == 0 && cp->name[len] == '\0') + return(cp->code); /* known name */ + if (len == 1) + return(*sp); /* single character */ + SETERROR(REG_ECOLLATE); /* neither */ + return(0); +} + +/* + - othercase - return the case counterpart of an alphabetic + == static char othercase(int ch); + */ +static char /* if no counterpart, return ch */ +othercase(ch) +int ch; +{ + assert(isalpha(ch)); + if (isupper(ch)) + return(tolower(ch)); + else if (islower(ch)) + return(toupper(ch)); + else /* peculiar, but could happen */ + return(ch); +} + +/* + - bothcases - emit a dualcase version of a two-case character + == static void bothcases(register struct parse *p, int ch); + * + * Boy, is this implementation ever a kludge... + */ +static void +bothcases(p, ch) +register struct parse *p; +int ch; +{ + register char *oldnext = p->next; + register char *oldend = p->end; + char bracket[3]; + + assert(othercase(ch) != ch); /* p_bracket() would recurse */ + p->next = bracket; + p->end = bracket+2; + bracket[0] = ch; + bracket[1] = ']'; + bracket[2] = '\0'; + p_bracket(p); + assert(p->next == bracket+2); + p->next = oldnext; + p->end = oldend; +} + +/* + - ordinary - emit an ordinary character + == static void ordinary(register struct parse *p, register int ch); + */ +static void +ordinary(p, ch) +register struct parse *p; +register int ch; +{ + register cat_t *cap = p->g->categories; + + if ((p->g->cflags®_ICASE) && isalpha(ch) && othercase(ch) != ch) + bothcases(p, ch); + else { + EMIT(OCHAR, (unsigned char)ch); + if (cap[ch] == 0) + cap[ch] = p->g->ncategories++; + } +} + +/* + - nonnewline - emit REG_NEWLINE version of OANY + == static void nonnewline(register struct parse *p); + * + * Boy, is this implementation ever a kludge... + */ +static void +nonnewline(p) +register struct parse *p; +{ + register char *oldnext = p->next; + register char *oldend = p->end; + char bracket[4]; + + p->next = bracket; + p->end = bracket+3; + bracket[0] = '^'; + bracket[1] = '\n'; + bracket[2] = ']'; + bracket[3] = '\0'; + p_bracket(p); + assert(p->next == bracket+3); + p->next = oldnext; + p->end = oldend; +} + +/* + - repeat - generate code for a bounded repetition, recursively if needed + == static void repeat(register struct parse *p, sopno start, int from, int to); + */ +static void +repeat(p, start, from, to) +register struct parse *p; +sopno start; /* operand from here to end of strip */ +int from; /* repeated from this number */ +int to; /* to this number of times (maybe INFINITY) */ +{ + register sopno finish = HERE(); +# define N 2 +# define INF 3 +# define REP(f, t) ((f)*8 + (t)) +# define MAP(n) (((n) <= 1) ? (n) : ((n) == INFINITY) ? INF : N) + register sopno copy; + + if (p->error != 0) /* head off possible runaway recursion */ + return; + + assert(from <= to); + + switch (REP(MAP(from), MAP(to))) { + case REP(0, 0): /* must be user doing this */ + DROP(finish-start); /* drop the operand */ + break; + case REP(0, 1): /* as x{1,1}? */ + case REP(0, N): /* as x{1,n}? */ + case REP(0, INF): /* as x{1,}? */ + /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */ + INSERT(OCH_, start); /* offset is wrong... */ + repeat(p, start+1, 1, to); + ASTERN(OOR1, start); + AHEAD(start); /* ... fix it */ + EMIT(OOR2, 0); + AHEAD(THERE()); + ASTERN(O_CH, THERETHERE()); + break; + case REP(1, 1): /* trivial case */ + /* done */ + break; + case REP(1, N): /* as x?x{1,n-1} */ + /* KLUDGE: emit y? as (y|) until subtle bug gets fixed */ + INSERT(OCH_, start); + ASTERN(OOR1, start); + AHEAD(start); + EMIT(OOR2, 0); /* offset very wrong... */ + AHEAD(THERE()); /* ...so fix it */ + ASTERN(O_CH, THERETHERE()); + copy = dupl(p, start+1, finish+1); + assert(copy == finish+4); + repeat(p, copy, 1, to-1); + break; + case REP(1, INF): /* as x+ */ + INSERT(OPLUS_, start); + ASTERN(O_PLUS, start); + break; + case REP(N, N): /* as xx{m-1,n-1} */ + copy = dupl(p, start, finish); + repeat(p, copy, from-1, to-1); + break; + case REP(N, INF): /* as xx{n-1,INF} */ + copy = dupl(p, start, finish); + repeat(p, copy, from-1, to); + break; + default: /* "can't happen" */ + SETERROR(REG_ASSERT); /* just in case */ + break; + } +} + +/* + - seterr - set an error condition + == static int seterr(register struct parse *p, int e); + */ +static int /* useless but makes type checking happy */ +seterr(p, e) +register struct parse *p; +int e; +{ + if (p->error == 0) /* keep earliest error condition */ + p->error = e; + p->next = nuls; /* try to bring things to a halt */ + p->end = nuls; + return(0); /* make the return value well-defined */ +} + +/* + - allocset - allocate a set of characters for [] + == static cset *allocset(register struct parse *p); + */ +static cset * +allocset(p) +register struct parse *p; +{ + register int no = p->g->ncsets++; + register size_t nc; + register size_t nbytes; + register cset *cs; + register size_t css = (size_t)p->g->csetsize; + register int i; + + if (no >= p->ncsalloc) { /* need another column of space */ + p->ncsalloc += CHAR_BIT; + nc = p->ncsalloc; + assert(nc % CHAR_BIT == 0); + nbytes = nc / CHAR_BIT * css; + if (p->g->sets == NULL) + p->g->sets = (cset *)malloc(nc * sizeof(cset)); + else + p->g->sets = (cset *)realloc((char *)p->g->sets, + nc * sizeof(cset)); + if (p->g->setbits == NULL) + p->g->setbits = (uch *)malloc(nbytes); + else { + p->g->setbits = (uch *)realloc((char *)p->g->setbits, + nbytes); + /* xxx this isn't right if setbits is now NULL */ + for (i = 0; i < no; i++) + p->g->sets[i].ptr = p->g->setbits + css*(i/CHAR_BIT); + } + if (p->g->sets != NULL && p->g->setbits != NULL) + (void) memset((char *)p->g->setbits + (nbytes - css), + 0, css); + else { + no = 0; + SETERROR(REG_ESPACE); + /* caller's responsibility not to do set ops */ + } + } + + assert(p->g->sets != NULL); /* xxx */ + cs = &p->g->sets[no]; + cs->ptr = p->g->setbits + css*((no)/CHAR_BIT); + cs->mask = 1 << ((no) % CHAR_BIT); + cs->hash = 0; + cs->smultis = 0; + cs->multis = NULL; + + return(cs); +} + +/* + - freeset - free a now-unused set + == static void freeset(register struct parse *p, register cset *cs); + */ +static void +freeset(p, cs) +register struct parse *p; +register cset *cs; +{ + register int i; + register cset *top = &p->g->sets[p->g->ncsets]; + register size_t css = (size_t)p->g->csetsize; + + for (i = 0; i < css; i++) + CHsub(cs, i); + if (cs == top-1) /* recover only the easy case */ + p->g->ncsets--; +} + +/* + - freezeset - final processing on a set of characters + == static int freezeset(register struct parse *p, register cset *cs); + * + * The main task here is merging identical sets. This is usually a waste + * of time (although the hash code minimizes the overhead), but can win + * big if REG_ICASE is being used. REG_ICASE, by the way, is why the hash + * is done using addition rather than xor -- all ASCII [aA] sets xor to + * the same value! + */ +static int /* set number */ +freezeset(p, cs) +register struct parse *p; +register cset *cs; +{ + register uch h = cs->hash; + register int i; + register cset *top = &p->g->sets[p->g->ncsets]; + register cset *cs2; + register size_t css = (size_t)p->g->csetsize; + + /* look for an earlier one which is the same */ + for (cs2 = &p->g->sets[0]; cs2 < top; cs2++) + if (cs2->hash == h && cs2 != cs) { + /* maybe */ + for (i = 0; i < css; i++) + if (!!CHIN(cs2, i) != !!CHIN(cs, i)) + break; /* no */ + if (i == css) + break; /* yes */ + } + + if (cs2 < top) { /* found one */ + freeset(p, cs); + cs = cs2; + } + + return((int)(cs - p->g->sets)); +} + +/* + - firstch - return first character in a set (which must have at least one) + == static int firstch(register struct parse *p, register cset *cs); + */ +static int /* character; there is no "none" value */ +firstch(p, cs) +register struct parse *p; +register cset *cs; +{ + register int i; + register size_t css = (size_t)p->g->csetsize; + + for (i = 0; i < css; i++) + if (CHIN(cs, i)) + return((char)i); + assert(never); + return(0); /* arbitrary */ +} + +/* + - nch - number of characters in a set + == static int nch(register struct parse *p, register cset *cs); + */ +static int +nch(p, cs) +register struct parse *p; +register cset *cs; +{ + register int i; + register size_t css = (size_t)p->g->csetsize; + register int n = 0; + + for (i = 0; i < css; i++) + if (CHIN(cs, i)) + n++; + return(n); +} + +/* + - mcadd - add a collating element to a cset + == static void mcadd(register struct parse *p, register cset *cs, \ + == register char *cp); + */ +static void +mcadd(p, cs, cp) +register struct parse *p; +register cset *cs; +register char *cp; +{ + register size_t oldend = cs->smultis; + + cs->smultis += strlen(cp) + 1; + if (cs->multis == NULL) + cs->multis = malloc(cs->smultis); + else + cs->multis = realloc(cs->multis, cs->smultis); + if (cs->multis == NULL) { + SETERROR(REG_ESPACE); + return; + } + + (void) strcpy(cs->multis + oldend - 1, cp); + cs->multis[cs->smultis - 1] = '\0'; +} + +/* + - mcsub - subtract a collating element from a cset + == static void mcsub(register cset *cs, register char *cp); + */ +static void +mcsub(cs, cp) +register cset *cs; +register char *cp; +{ + register char *fp = mcfind(cs, cp); + register size_t len = strlen(fp); + + assert(fp != NULL); + (void) memmove(fp, fp + len + 1, + cs->smultis - (fp + len + 1 - cs->multis)); + cs->smultis -= len; + + if (cs->smultis == 0) { + free(cs->multis); + cs->multis = NULL; + return; + } + + cs->multis = realloc(cs->multis, cs->smultis); + assert(cs->multis != NULL); +} + +/* + - mcin - is a collating element in a cset? + == static int mcin(register cset *cs, register char *cp); + */ +static int +mcin(cs, cp) +register cset *cs; +register char *cp; +{ + return(mcfind(cs, cp) != NULL); +} + +/* + - mcfind - find a collating element in a cset + == static char *mcfind(register cset *cs, register char *cp); + */ +static char * +mcfind(cs, cp) +register cset *cs; +register char *cp; +{ + register char *p; + + if (cs->multis == NULL) + return(NULL); + for (p = cs->multis; *p != '\0'; p += strlen(p) + 1) + if (strcmp(cp, p) == 0) + return(p); + return(NULL); +} + +/* + - mcinvert - invert the list of collating elements in a cset + == static void mcinvert(register struct parse *p, register cset *cs); + * + * This would have to know the set of possibilities. Implementation + * is deferred. + */ +static void +mcinvert(p, cs) +register struct parse *p; +register cset *cs; +{ + assert(cs->multis == NULL); /* xxx */ +} + +/* + - mccase - add case counterparts of the list of collating elements in a cset + == static void mccase(register struct parse *p, register cset *cs); + * + * This would have to know the set of possibilities. Implementation + * is deferred. + */ +static void +mccase(p, cs) +register struct parse *p; +register cset *cs; +{ + assert(cs->multis == NULL); /* xxx */ +} + +/* + - isinsets - is this character in any sets? + == static int isinsets(register struct re_guts *g, int c); + */ +static int /* predicate */ +isinsets(g, c) +register struct re_guts *g; +int c; +{ + register uch *col; + register int i; + register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT; + register unsigned uc = (unsigned char)c; + + for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize) + if (col[uc] != 0) + return(1); + return(0); +} + +/* + - samesets - are these two characters in exactly the same sets? + == static int samesets(register struct re_guts *g, int c1, int c2); + */ +static int /* predicate */ +samesets(g, c1, c2) +register struct re_guts *g; +int c1; +int c2; +{ + register uch *col; + register int i; + register int ncols = (g->ncsets+(CHAR_BIT-1)) / CHAR_BIT; + register unsigned uc1 = (unsigned char)c1; + register unsigned uc2 = (unsigned char)c2; + + for (i = 0, col = g->setbits; i < ncols; i++, col += g->csetsize) + if (col[uc1] != col[uc2]) + return(0); + return(1); +} + +/* + - categorize - sort out character categories + == static void categorize(struct parse *p, register struct re_guts *g); + */ +static void +categorize(p, g) +struct parse *p; +register struct re_guts *g; +{ + register cat_t *cats = g->categories; + register int c; + register int c2; + register cat_t cat; + + /* avoid making error situations worse */ + if (p->error != 0) + return; + + for (c = CHAR_MIN; c <= CHAR_MAX; c++) + if (cats[c] == 0 && isinsets(g, c)) { + cat = g->ncategories++; + cats[c] = cat; + for (c2 = c+1; c2 <= CHAR_MAX; c2++) + if (cats[c2] == 0 && samesets(g, c, c2)) + cats[c2] = cat; + } +} + +/* + - dupl - emit a duplicate of a bunch of sops + == static sopno dupl(register struct parse *p, sopno start, sopno finish); + */ +static sopno /* start of duplicate */ +dupl(p, start, finish) +register struct parse *p; +sopno start; /* from here */ +sopno finish; /* to this less one */ +{ + register sopno ret = HERE(); + register sopno len = finish - start; + + assert(finish >= start); + if (len == 0) + return(ret); + enlarge(p, p->ssize + len); /* this many unexpected additions */ + assert(p->ssize >= p->slen + len); + (void) memcpy((char *)(p->strip + p->slen), + (char *)(p->strip + start), (size_t)len*sizeof(sop)); + p->slen += len; + return(ret); +} + +/* + - doemit - emit a strip operator + == static void doemit(register struct parse *p, sop op, size_t opnd); + * + * It might seem better to implement this as a macro with a function as + * hard-case backup, but it's just too big and messy unless there are + * some changes to the data structures. Maybe later. + */ +static void +doemit(p, op, opnd) +register struct parse *p; +sop op; +size_t opnd; +{ + /* avoid making error situations worse */ + if (p->error != 0) + return; + + /* deal with oversize operands ("can't happen", more or less) */ + assert(opnd < 1<slen >= p->ssize) + enlarge(p, (p->ssize+1) / 2 * 3); /* +50% */ + assert(p->slen < p->ssize); + + /* finally, it's all reduced to the easy case */ + p->strip[p->slen++] = SOP(op, opnd); +} + +/* + - doinsert - insert a sop into the strip + == static void doinsert(register struct parse *p, sop op, size_t opnd, sopno pos); + */ +static void +doinsert(p, op, opnd, pos) +register struct parse *p; +sop op; +size_t opnd; +sopno pos; +{ + register sopno sn; + register sop s; + register int i; + + /* avoid making error situations worse */ + if (p->error != 0) + return; + + sn = HERE(); + EMIT(op, opnd); /* do checks, ensure space */ + assert(HERE() == sn+1); + s = p->strip[sn]; + + /* adjust paren pointers */ + assert(pos > 0); + for (i = 1; i < NPAREN; i++) { + if (p->pbegin[i] >= pos) { + p->pbegin[i]++; + } + if (p->pend[i] >= pos) { + p->pend[i]++; + } + } + + memmove((char *)&p->strip[pos+1], (char *)&p->strip[pos], + (HERE()-pos-1)*sizeof(sop)); + p->strip[pos] = s; +} + +/* + - dofwd - complete a forward reference + == static void dofwd(register struct parse *p, sopno pos, sop value); + */ +static void +dofwd(p, pos, value) +register struct parse *p; +register sopno pos; +sop value; +{ + /* avoid making error situations worse */ + if (p->error != 0) + return; + + assert(value < 1<strip[pos] = OP(p->strip[pos]) | value; +} + +/* + - enlarge - enlarge the strip + == static void enlarge(register struct parse *p, sopno size); + */ +static void +enlarge(p, size) +register struct parse *p; +register sopno size; +{ + register sop *sp; + + if (p->ssize >= size) + return; + + sp = (sop *)realloc(p->strip, size*sizeof(sop)); + if (sp == NULL) { + SETERROR(REG_ESPACE); + return; + } + p->strip = sp; + p->ssize = size; +} + +/* + - stripsnug - compact the strip + == static void stripsnug(register struct parse *p, register struct re_guts *g); + */ +static void +stripsnug(p, g) +register struct parse *p; +register struct re_guts *g; +{ + g->nstates = p->slen; + g->strip = (sop *)realloc((char *)p->strip, p->slen * sizeof(sop)); + if (g->strip == NULL) { + SETERROR(REG_ESPACE); + g->strip = p->strip; + } +} + +/* + - findmust - fill in must and mlen with longest mandatory literal string + == static void findmust(register struct parse *p, register struct re_guts *g); + * + * This algorithm could do fancy things like analyzing the operands of | + * for common subsequences. Someday. This code is simple and finds most + * of the interesting cases. + * + * Note that must and mlen got initialized during setup. + */ +static void +findmust(p, g) +struct parse *p; +register struct re_guts *g; +{ + register sop *scan; + sop *start; + register sop *newstart; + register sopno newlen; + register sop s; + register char *cp; + register sopno i; + + /* avoid making error situations worse */ + if (p->error != 0) + return; + + /* find the longest OCHAR sequence in strip */ + newlen = 0; + scan = g->strip + 1; + do { + s = *scan++; + switch (OP(s)) { + case OCHAR: /* sequence member */ + if (newlen == 0) /* new sequence */ + newstart = scan - 1; + newlen++; + break; + case OPLUS_: /* things that don't break one */ + case OLPAREN: + case ORPAREN: + break; + case OQUEST_: /* things that must be skipped */ + case OCH_: + scan--; + do { + scan += OPND(s); + s = *scan; + /* assert() interferes w debug printouts */ + if (OP(s) != O_QUEST && OP(s) != O_CH && + OP(s) != OOR2) { + g->iflags |= BAD; + return; + } + } while (OP(s) != O_QUEST && OP(s) != O_CH); + /* fallthrough */ + default: /* things that break a sequence */ + if (newlen > g->mlen) { /* ends one */ + start = newstart; + g->mlen = newlen; + } + newlen = 0; + break; + } + } while (OP(s) != OEND); + + if (g->mlen == 0) /* there isn't one */ + return; + + /* turn it into a character string */ + g->must = malloc((size_t)g->mlen + 1); + if (g->must == NULL) { /* argh; just forget it */ + g->mlen = 0; + return; + } + cp = g->must; + scan = start; + for (i = g->mlen; i > 0; i--) { + while (OP(s = *scan++) != OCHAR) + continue; + assert(cp < g->must + g->mlen); + *cp++ = (char)OPND(s); + } + assert(cp == g->must + g->mlen); + *cp++ = '\0'; /* just on general principles */ +} + +/* + - pluscount - count + nesting + == static sopno pluscount(register struct parse *p, register struct re_guts *g); + */ +static sopno /* nesting depth */ +pluscount(p, g) +struct parse *p; +register struct re_guts *g; +{ + register sop *scan; + register sop s; + register sopno plusnest = 0; + register sopno maxnest = 0; + + if (p->error != 0) + return(0); /* there may not be an OEND */ + + scan = g->strip + 1; + do { + s = *scan++; + switch (OP(s)) { + case OPLUS_: + plusnest++; + break; + case O_PLUS: + if (plusnest > maxnest) + maxnest = plusnest; + plusnest--; + break; + } + } while (OP(s) != OEND); + if (plusnest != 0) + g->iflags |= BAD; + return(maxnest); +} diff --git a/win32/regex/regcomp.ih b/win32/regex/regcomp.ih new file mode 100644 index 0000000..0776e71 --- /dev/null +++ b/win32/regex/regcomp.ih @@ -0,0 +1,51 @@ +/* ========= begin header generated by ./mkh ========= */ +#ifdef __cplusplus +extern "C" { +#endif + +/* === regcomp.c === */ +static void p_ere(register struct parse *p, int stop); +static void p_ere_exp(register struct parse *p); +static void p_str(register struct parse *p); +static void p_bre(register struct parse *p, register int end1, register int end2); +static int p_simp_re(register struct parse *p, int starordinary); +static int p_count(register struct parse *p); +static void p_bracket(register struct parse *p); +static void p_b_term(register struct parse *p, register cset *cs); +static void p_b_cclass(register struct parse *p, register cset *cs); +static void p_b_eclass(register struct parse *p, register cset *cs); +static char p_b_symbol(register struct parse *p); +static char p_b_coll_elem(register struct parse *p, int endc); +static char othercase(int ch); +static void bothcases(register struct parse *p, int ch); +static void ordinary(register struct parse *p, register int ch); +static void nonnewline(register struct parse *p); +static void repeat(register struct parse *p, sopno start, int from, int to); +static int seterr(register struct parse *p, int e); +static cset *allocset(register struct parse *p); +static void freeset(register struct parse *p, register cset *cs); +static int freezeset(register struct parse *p, register cset *cs); +static int firstch(register struct parse *p, register cset *cs); +static int nch(register struct parse *p, register cset *cs); +static void mcadd(register struct parse *p, register cset *cs, register char *cp); +static void mcsub(register cset *cs, register char *cp); +static int mcin(register cset *cs, register char *cp); +static char *mcfind(register cset *cs, register char *cp); +static void mcinvert(register struct parse *p, register cset *cs); +static void mccase(register struct parse *p, register cset *cs); +static int isinsets(register struct re_guts *g, int c); +static int samesets(register struct re_guts *g, int c1, int c2); +static void categorize(struct parse *p, register struct re_guts *g); +static sopno dupl(register struct parse *p, sopno start, sopno finish); +static void doemit(register struct parse *p, sop op, size_t opnd); +static void doinsert(register struct parse *p, sop op, size_t opnd, sopno pos); +static void dofwd(register struct parse *p, sopno pos, sop value); +static void enlarge(register struct parse *p, sopno size); +static void stripsnug(register struct parse *p, register struct re_guts *g); +static void findmust(register struct parse *p, register struct re_guts *g); +static sopno pluscount(register struct parse *p, register struct re_guts *g); + +#ifdef __cplusplus +} +#endif +/* ========= end header generated by ./mkh ========= */ diff --git a/win32/regex/regerror.c b/win32/regex/regerror.c new file mode 100644 index 0000000..e53dafc --- /dev/null +++ b/win32/regex/regerror.c @@ -0,0 +1,126 @@ +#include +#include +#include +#include +#include +#include +#include + +#include "utils.h" +#include "regerror.ih" + +/* + = #define REG_OKAY 0 + = #define REG_NOMATCH 1 + = #define REG_BADPAT 2 + = #define REG_ECOLLATE 3 + = #define REG_ECTYPE 4 + = #define REG_EESCAPE 5 + = #define REG_ESUBREG 6 + = #define REG_EBRACK 7 + = #define REG_EPAREN 8 + = #define REG_EBRACE 9 + = #define REG_BADBR 10 + = #define REG_ERANGE 11 + = #define REG_ESPACE 12 + = #define REG_BADRPT 13 + = #define REG_EMPTY 14 + = #define REG_ASSERT 15 + = #define REG_INVARG 16 + = #define REG_ATOI 255 // convert name to number (!) + = #define REG_ITOA 0400 // convert number to name (!) + */ +static struct rerr { + int code; + char *name; + char *explain; +} rerrs[] = { + REG_OKAY, "REG_OKAY", "no errors detected", + REG_NOMATCH, "REG_NOMATCH", "regexec() failed to match", + REG_BADPAT, "REG_BADPAT", "invalid regular expression", + REG_ECOLLATE, "REG_ECOLLATE", "invalid collating element", + REG_ECTYPE, "REG_ECTYPE", "invalid character class", + REG_EESCAPE, "REG_EESCAPE", "trailing backslash (\\)", + REG_ESUBREG, "REG_ESUBREG", "invalid backreference number", + REG_EBRACK, "REG_EBRACK", "brackets ([ ]) not balanced", + REG_EPAREN, "REG_EPAREN", "parentheses not balanced", + REG_EBRACE, "REG_EBRACE", "braces not balanced", + REG_BADBR, "REG_BADBR", "invalid repetition count(s)", + REG_ERANGE, "REG_ERANGE", "invalid character range", + REG_ESPACE, "REG_ESPACE", "out of memory", + REG_BADRPT, "REG_BADRPT", "repetition-operator operand invalid", + REG_EMPTY, "REG_EMPTY", "empty (sub)expression", + REG_ASSERT, "REG_ASSERT", "\"can't happen\" -- you found a bug", + REG_INVARG, "REG_INVARG", "invalid argument to regex routine", + -1, "", "*** unknown regexp error code ***", +}; + +/* + - regerror - the interface to error numbers + = extern size_t regerror(int, const regex_t *, char *, size_t); + */ +/* ARGSUSED */ +size_t +regerror(errcode, preg, errbuf, errbuf_size) +int errcode; +const regex_t *preg; +char *errbuf; +size_t errbuf_size; +{ + register struct rerr *r; + register size_t len; + register int target = errcode &~ REG_ITOA; + register char *s; + char convbuf[50]; + + if (errcode == REG_ATOI) + s = regatoi(preg, convbuf); + else { + for (r = rerrs; r->code >= 0; r++) + if (r->code == target) + break; + + if (errcode®_ITOA) { + if (r->code >= 0) + (void) strcpy(convbuf, r->name); + else + sprintf(convbuf, "REG_0x%x", target); + assert(strlen(convbuf) < sizeof(convbuf)); + s = convbuf; + } else + s = r->explain; + } + + len = strlen(s) + 1; + if (errbuf_size > 0) { + if (errbuf_size > len) + (void) strcpy(errbuf, s); + else { + (void) strncpy(errbuf, s, errbuf_size-1); + errbuf[errbuf_size-1] = '\0'; + } + } + + return(len); +} + +/* + - regatoi - internal routine to implement REG_ATOI + == static char *regatoi(const regex_t *preg, char *localbuf); + */ +static char * +regatoi(preg, localbuf) +const regex_t *preg; +char *localbuf; +{ + register struct rerr *r; + + for (r = rerrs; r->code >= 0; r++) + if (strcmp(r->name, preg->re_endp) == 0) + break; + if (r->code < 0) + return("0"); + + sprintf(localbuf, "%d", r->code); + return(localbuf); +} diff --git a/win32/regex/regerror.ih b/win32/regex/regerror.ih new file mode 100644 index 0000000..2cb668c --- /dev/null +++ b/win32/regex/regerror.ih @@ -0,0 +1,12 @@ +/* ========= begin header generated by ./mkh ========= */ +#ifdef __cplusplus +extern "C" { +#endif + +/* === regerror.c === */ +static char *regatoi(const regex_t *preg, char *localbuf); + +#ifdef __cplusplus +} +#endif +/* ========= end header generated by ./mkh ========= */ diff --git a/win32/regex/regex2.h b/win32/regex/regex2.h new file mode 100644 index 0000000..58fd8d8 --- /dev/null +++ b/win32/regex/regex2.h @@ -0,0 +1,134 @@ +/* + * First, the stuff that ends up in the outside-world include file + = typedef off_t regoff_t; + = typedef struct { + = int re_magic; + = size_t re_nsub; // number of parenthesized subexpressions + = const char *re_endp; // end pointer for REG_PEND + = struct re_guts *re_g; // none of your business :-) + = } regex_t; + = typedef struct { + = regoff_t rm_so; // start of match + = regoff_t rm_eo; // end of match + = } regmatch_t; + */ +/* + * internals of regex_t + */ +#define MAGIC1 ((('r'^0200)<<8) | 'e') + +/* + * The internal representation is a *strip*, a sequence of + * operators ending with an endmarker. (Some terminology etc. is a + * historical relic of earlier versions which used multiple strips.) + * Certain oddities in the representation are there to permit running + * the machinery backwards; in particular, any deviation from sequential + * flow must be marked at both its source and its destination. Some + * fine points: + * + * - OPLUS_ and O_PLUS are *inside* the loop they create. + * - OQUEST_ and O_QUEST are *outside* the bypass they create. + * - OCH_ and O_CH are *outside* the multi-way branch they create, while + * OOR1 and OOR2 are respectively the end and the beginning of one of + * the branches. Note that there is an implicit OOR2 following OCH_ + * and an implicit OOR1 preceding O_CH. + * + * In state representations, an operator's bit is on to signify a state + * immediately *preceding* "execution" of that operator. + */ +typedef long sop; /* strip operator */ +typedef long sopno; +#define OPRMASK 0x7c000000 +#define OPDMASK 0x03ffffff +#define OPSHIFT (26) +#define OP(n) ((n)&OPRMASK) +#define OPND(n) ((n)&OPDMASK) +#define SOP(op, opnd) ((op)|(opnd)) +/* operators meaning operand */ +/* (back, fwd are offsets) */ +#define OEND (1< uch [csetsize] */ + uch mask; /* bit within array */ + uch hash; /* hash code */ + size_t smultis; + char *multis; /* -> char[smulti] ab\0cd\0ef\0\0 */ +} cset; +/* note that CHadd and CHsub are unsafe, and CHIN doesn't yield 0/1 */ +#define CHadd(cs, c) ((cs)->ptr[(uch)(c)] |= (cs)->mask, (cs)->hash += (c)) +#define CHsub(cs, c) ((cs)->ptr[(uch)(c)] &= ~(cs)->mask, (cs)->hash -= (c)) +#define CHIN(cs, c) ((cs)->ptr[(uch)(c)] & (cs)->mask) +#define MCadd(p, cs, cp) mcadd(p, cs, cp) /* regcomp() internal fns */ +#define MCsub(p, cs, cp) mcsub(p, cs, cp) +#define MCin(p, cs, cp) mcin(p, cs, cp) + +/* stuff for character categories */ +typedef unsigned char cat_t; + +/* + * main compiled-expression structure + */ +struct re_guts { + int magic; +# define MAGIC2 ((('R'^0200)<<8)|'E') + sop *strip; /* malloced area for strip */ + int csetsize; /* number of bits in a cset vector */ + int ncsets; /* number of csets in use */ + cset *sets; /* -> cset [ncsets] */ + uch *setbits; /* -> uch[csetsize][ncsets/CHAR_BIT] */ + int cflags; /* copy of regcomp() cflags argument */ + sopno nstates; /* = number of sops */ + sopno firststate; /* the initial OEND (normally 0) */ + sopno laststate; /* the final OEND */ + int iflags; /* internal flags */ +# define USEBOL 01 /* used ^ */ +# define USEEOL 02 /* used $ */ +# define BAD 04 /* something wrong */ + int nbol; /* number of ^ used */ + int neol; /* number of $ used */ + int ncategories; /* how many character categories */ + cat_t *categories; /* ->catspace[-CHAR_MIN] */ + char *must; /* match must contain this string */ + int mlen; /* length of must */ + size_t nsub; /* copy of re_nsub */ + int backrefs; /* does it use back references? */ + sopno nplus; /* how deep does it nest +s? */ + /* catspace must be last */ + cat_t catspace[1]; /* actually [NC] */ +}; + +/* misc utilities */ +#define OUT (CHAR_MAX+1) /* a non-character value */ +#define ISWORD(c) (isalnum(c) || (c) == '_') diff --git a/win32/regex/regexec.c b/win32/regex/regexec.c new file mode 100644 index 0000000..dcb11b2 --- /dev/null +++ b/win32/regex/regexec.c @@ -0,0 +1,138 @@ +/* + * the outer shell of regexec() + * + * This file includes engine.c *twice*, after muchos fiddling with the + * macros that code uses. This lets the same code operate on two different + * representations for state sets. + */ +#include +#include +#include +#include +#include +#include +#include + +#include "utils.h" +#include "regex2.h" + +static int nope = 0; /* for use in asserts; shuts lint up */ + +/* macros for manipulating states, small version */ +#define states unsigned +#define states1 unsigned /* for later use in regexec() decision */ +#define CLEAR(v) ((v) = 0) +#define SET0(v, n) ((v) &= ~((unsigned)1 << (n))) +#define SET1(v, n) ((v) |= (unsigned)1 << (n)) +#define ISSET(v, n) ((v) & ((unsigned)1 << (n))) +#define ASSIGN(d, s) ((d) = (s)) +#define EQ(a, b) ((a) == (b)) +#define STATEVARS int dummy /* dummy version */ +#define STATESETUP(m, n) /* nothing */ +#define STATETEARDOWN(m) /* nothing */ +#define SETUP(v) ((v) = 0) +#define onestate unsigned +#define INIT(o, n) ((o) = (unsigned)1 << (n)) +#define INC(o) ((o) <<= 1) +#define ISSTATEIN(v, o) ((v) & (o)) +/* some abbreviations; note that some of these know variable names! */ +/* do "if I'm here, I can also be there" etc without branches */ +#define FWD(dst, src, n) ((dst) |= ((unsigned)(src)&(here)) << (n)) +#define BACK(dst, src, n) ((dst) |= ((unsigned)(src)&(here)) >> (n)) +#define ISSETBACK(v, n) ((v) & ((unsigned)here >> (n))) +/* function names */ +#define SNAMES /* engine.c looks after details */ + +#include "engine.c" + +/* now undo things */ +#undef states +#undef CLEAR +#undef SET0 +#undef SET1 +#undef ISSET +#undef ASSIGN +#undef EQ +#undef STATEVARS +#undef STATESETUP +#undef STATETEARDOWN +#undef SETUP +#undef onestate +#undef INIT +#undef INC +#undef ISSTATEIN +#undef FWD +#undef BACK +#undef ISSETBACK +#undef SNAMES + +/* macros for manipulating states, large version */ +#define states char * +#define CLEAR(v) memset(v, 0, m->g->nstates) +#define SET0(v, n) ((v)[n] = 0) +#define SET1(v, n) ((v)[n] = 1) +#define ISSET(v, n) ((v)[n]) +#define ASSIGN(d, s) memcpy(d, s, m->g->nstates) +#define EQ(a, b) (memcmp(a, b, m->g->nstates) == 0) +#define STATEVARS int vn; char *space +#define STATESETUP(m, nv) { (m)->space = malloc((nv)*(m)->g->nstates); \ + if ((m)->space == NULL) return(REG_ESPACE); \ + (m)->vn = 0; } +#define STATETEARDOWN(m) { free((m)->space); } +#define SETUP(v) ((v) = &m->space[m->vn++ * m->g->nstates]) +#define onestate int +#define INIT(o, n) ((o) = (n)) +#define INC(o) ((o)++) +#define ISSTATEIN(v, o) ((v)[o]) +/* some abbreviations; note that some of these know variable names! */ +/* do "if I'm here, I can also be there" etc without branches */ +#define FWD(dst, src, n) ((dst)[here+(n)] |= (src)[here]) +#define BACK(dst, src, n) ((dst)[here-(n)] |= (src)[here]) +#define ISSETBACK(v, n) ((v)[here - (n)]) +/* function names */ +#define LNAMES /* flag */ + +#include "engine.c" + +/* + - regexec - interface for matching + = extern int regexec(const regex_t *, const char *, size_t, \ + = regmatch_t [], int); + = #define REG_NOTBOL 00001 + = #define REG_NOTEOL 00002 + = #define REG_STARTEND 00004 + = #define REG_TRACE 00400 // tracing of execution + = #define REG_LARGE 01000 // force large representation + = #define REG_BACKR 02000 // force use of backref code + * + * We put this here so we can exploit knowledge of the state representation + * when choosing which matcher to call. Also, by this point the matchers + * have been prototyped. + */ +int /* 0 success, REG_NOMATCH failure */ +regexec(preg, string, nmatch, pmatch, eflags) +const regex_t *preg; +const char *string; +size_t nmatch; +regmatch_t pmatch[]; +int eflags; +{ + register struct re_guts *g = preg->re_g; +#ifdef REDEBUG +# define GOODFLAGS(f) (f) +#else +# define GOODFLAGS(f) ((f)&(REG_NOTBOL|REG_NOTEOL|REG_STARTEND)) +#endif + + if (preg->re_magic != MAGIC1 || g->magic != MAGIC2) + return(REG_BADPAT); + assert(!(g->iflags&BAD)); + if (g->iflags&BAD) /* backstop for no-debug case */ + return(REG_BADPAT); + eflags = GOODFLAGS(eflags); + + if (g->nstates <= CHAR_BIT*sizeof(states1) && !(eflags®_LARGE)) + return(smatcher(g, (char *)string, nmatch, pmatch, eflags)); + else + return(lmatcher(g, (char *)string, nmatch, pmatch, eflags)); +} diff --git a/win32/regex/regfree.c b/win32/regex/regfree.c new file mode 100644 index 0000000..9a6acf1 --- /dev/null +++ b/win32/regex/regfree.c @@ -0,0 +1,37 @@ +#include +#include +#include +#include + +#include "utils.h" +#include "regex2.h" + +/* + - regfree - free everything + = extern void regfree(regex_t *); + */ +void +regfree(preg) +regex_t *preg; +{ + register struct re_guts *g; + + if (preg->re_magic != MAGIC1) /* oops */ + return; /* nice to complain, but hard */ + + g = preg->re_g; + if (g == NULL || g->magic != MAGIC2) /* oops again */ + return; + preg->re_magic = 0; /* mark it invalid */ + g->magic = 0; /* mark it invalid */ + + if (g->strip != NULL) + free((char *)g->strip); + if (g->sets != NULL) + free((char *)g->sets); + if (g->setbits != NULL) + free((char *)g->setbits); + if (g->must != NULL) + free(g->must); + free((char *)g); +} diff --git a/win32/regex/split.c b/win32/regex/split.c new file mode 100644 index 0000000..188bdb7 --- /dev/null +++ b/win32/regex/split.c @@ -0,0 +1,316 @@ +#include +#include + +/* + - split - divide a string into fields, like awk split() + = int split(char *string, char *fields[], int nfields, char *sep); + */ +int /* number of fields, including overflow */ +split(string, fields, nfields, sep) +char *string; +char *fields[]; /* list is not NULL-terminated */ +int nfields; /* number of entries available in fields[] */ +char *sep; /* "" white, "c" single char, "ab" [ab]+ */ +{ + register char *p = string; + register char c; /* latest character */ + register char sepc = sep[0]; + register char sepc2; + register int fn; + register char **fp = fields; + register char *sepp; + register int trimtrail; + + /* white space */ + if (sepc == '\0') { + while ((c = *p++) == ' ' || c == '\t') + continue; + p--; + trimtrail = 1; + sep = " \t"; /* note, code below knows this is 2 long */ + sepc = ' '; + } else + trimtrail = 0; + sepc2 = sep[1]; /* now we can safely pick this up */ + + /* catch empties */ + if (*p == '\0') + return(0); + + /* single separator */ + if (sepc2 == '\0') { + fn = nfields; + for (;;) { + *fp++ = p; + fn--; + if (fn == 0) + break; + while ((c = *p++) != sepc) + if (c == '\0') + return(nfields - fn); + *(p-1) = '\0'; + } + /* we have overflowed the fields vector -- just count them */ + fn = nfields; + for (;;) { + while ((c = *p++) != sepc) + if (c == '\0') + return(fn); + fn++; + } + /* not reached */ + } + + /* two separators */ + if (sep[2] == '\0') { + fn = nfields; + for (;;) { + *fp++ = p; + fn--; + while ((c = *p++) != sepc && c != sepc2) + if (c == '\0') { + if (trimtrail && **(fp-1) == '\0') + fn++; + return(nfields - fn); + } + if (fn == 0) + break; + *(p-1) = '\0'; + while ((c = *p++) == sepc || c == sepc2) + continue; + p--; + } + /* we have overflowed the fields vector -- just count them */ + fn = nfields; + while (c != '\0') { + while ((c = *p++) == sepc || c == sepc2) + continue; + p--; + fn++; + while ((c = *p++) != '\0' && c != sepc && c != sepc2) + continue; + } + /* might have to trim trailing white space */ + if (trimtrail) { + p--; + while ((c = *--p) == sepc || c == sepc2) + continue; + p++; + if (*p != '\0') { + if (fn == nfields+1) + *p = '\0'; + fn--; + } + } + return(fn); + } + + /* n separators */ + fn = 0; + for (;;) { + if (fn < nfields) + *fp++ = p; + fn++; + for (;;) { + c = *p++; + if (c == '\0') + return(fn); + sepp = sep; + while ((sepc = *sepp++) != '\0' && sepc != c) + continue; + if (sepc != '\0') /* it was a separator */ + break; + } + if (fn < nfields) + *(p-1) = '\0'; + for (;;) { + c = *p++; + sepp = sep; + while ((sepc = *sepp++) != '\0' && sepc != c) + continue; + if (sepc == '\0') /* it wasn't a separator */ + break; + } + p--; + } + + /* not reached */ +} + +#ifdef TEST_SPLIT + + +/* + * test program + * pgm runs regression + * pgm sep splits stdin lines by sep + * pgm str sep splits str by sep + * pgm str sep n splits str by sep n times + */ +int +main(argc, argv) +int argc; +char *argv[]; +{ + char buf[512]; + register int n; +# define MNF 10 + char *fields[MNF]; + + if (argc > 4) + for (n = atoi(argv[3]); n > 0; n--) { + (void) strcpy(buf, argv[1]); + } + else if (argc > 3) + for (n = atoi(argv[3]); n > 0; n--) { + (void) strcpy(buf, argv[1]); + (void) split(buf, fields, MNF, argv[2]); + } + else if (argc > 2) + dosplit(argv[1], argv[2]); + else if (argc > 1) + while (fgets(buf, sizeof(buf), stdin) != NULL) { + buf[strlen(buf)-1] = '\0'; /* stomp newline */ + dosplit(buf, argv[1]); + } + else + regress(); + + exit(0); +} + +dosplit(string, seps) +char *string; +char *seps; +{ +# define NF 5 + char *fields[NF]; + register int nf; + + nf = split(string, fields, NF, seps); + print(nf, NF, fields); +} + +print(nf, nfp, fields) +int nf; +int nfp; +char *fields[]; +{ + register int fn; + register int bound; + + bound = (nf > nfp) ? nfp : nf; + printf("%d:\t", nf); + for (fn = 0; fn < bound; fn++) + printf("\"%s\"%s", fields[fn], (fn+1 < nf) ? ", " : "\n"); +} + +#define RNF 5 /* some table entries know this */ +struct { + char *str; + char *seps; + int nf; + char *fi[RNF]; +} tests[] = { + "", " ", 0, { "" }, + " ", " ", 2, { "", "" }, + "x", " ", 1, { "x" }, + "xy", " ", 1, { "xy" }, + "x y", " ", 2, { "x", "y" }, + "abc def g ", " ", 5, { "abc", "def", "", "g", "" }, + " a bcd", " ", 4, { "", "", "a", "bcd" }, + "a b c d e f", " ", 6, { "a", "b", "c", "d", "e f" }, + " a b c d ", " ", 6, { "", "a", "b", "c", "d " }, + + "", " _", 0, { "" }, + " ", " _", 2, { "", "" }, + "x", " _", 1, { "x" }, + "x y", " _", 2, { "x", "y" }, + "ab _ cd", " _", 2, { "ab", "cd" }, + " a_b c ", " _", 5, { "", "a", "b", "c", "" }, + "a b c_d e f", " _", 6, { "a", "b", "c", "d", "e f" }, + " a b c d ", " _", 6, { "", "a", "b", "c", "d " }, + + "", " _~", 0, { "" }, + " ", " _~", 2, { "", "" }, + "x", " _~", 1, { "x" }, + "x y", " _~", 2, { "x", "y" }, + "ab _~ cd", " _~", 2, { "ab", "cd" }, + " a_b c~", " _~", 5, { "", "a", "b", "c", "" }, + "a b_c d~e f", " _~", 6, { "a", "b", "c", "d", "e f" }, + "~a b c d ", " _~", 6, { "", "a", "b", "c", "d " }, + + "", " _~-", 0, { "" }, + " ", " _~-", 2, { "", "" }, + "x", " _~-", 1, { "x" }, + "x y", " _~-", 2, { "x", "y" }, + "ab _~- cd", " _~-", 2, { "ab", "cd" }, + " a_b c~", " _~-", 5, { "", "a", "b", "c", "" }, + "a b_c-d~e f", " _~-", 6, { "a", "b", "c", "d", "e f" }, + "~a-b c d ", " _~-", 6, { "", "a", "b", "c", "d " }, + + "", " ", 0, { "" }, + " ", " ", 2, { "", "" }, + "x", " ", 1, { "x" }, + "xy", " ", 1, { "xy" }, + "x y", " ", 2, { "x", "y" }, + "abc def g ", " ", 4, { "abc", "def", "g", "" }, + " a bcd", " ", 3, { "", "a", "bcd" }, + "a b c d e f", " ", 6, { "a", "b", "c", "d", "e f" }, + " a b c d ", " ", 6, { "", "a", "b", "c", "d " }, + + "", "", 0, { "" }, + " ", "", 0, { "" }, + "x", "", 1, { "x" }, + "xy", "", 1, { "xy" }, + "x y", "", 2, { "x", "y" }, + "abc def g ", "", 3, { "abc", "def", "g" }, + "\t a bcd", "", 2, { "a", "bcd" }, + " a \tb\t c ", "", 3, { "a", "b", "c" }, + "a b c d e ", "", 5, { "a", "b", "c", "d", "e" }, + "a b\tc d e f", "", 6, { "a", "b", "c", "d", "e f" }, + " a b c d e f ", "", 6, { "a", "b", "c", "d", "e f " }, + + NULL, NULL, 0, { NULL }, +}; + +regress() +{ + char buf[512]; + register int n; + char *fields[RNF+1]; + register int nf; + register int i; + register int printit; + register char *f; + + for (n = 0; tests[n].str != NULL; n++) { + (void) strcpy(buf, tests[n].str); + fields[RNF] = NULL; + nf = split(buf, fields, RNF, tests[n].seps); + printit = 0; + if (nf != tests[n].nf) { + printf("split `%s' by `%s' gave %d fields, not %d\n", + tests[n].str, tests[n].seps, nf, tests[n].nf); + printit = 1; + } else if (fields[RNF] != NULL) { + printf("split() went beyond array end\n"); + printit = 1; + } else { + for (i = 0; i < nf && i < RNF; i++) { + f = fields[i]; + if (f == NULL) + f = "(NULL)"; + if (strcmp(f, tests[n].fi[i]) != 0) { + printf("split `%s' by `%s', field %d is `%s', not `%s'\n", + tests[n].str, tests[n].seps, + i, fields[i], tests[n].fi[i]); + printit = 1; + } + } + } + if (printit) + print(nf, RNF, fields); + } +} +#endif diff --git a/win32/regex/utils.h b/win32/regex/utils.h new file mode 100644 index 0000000..1a997ac --- /dev/null +++ b/win32/regex/utils.h @@ -0,0 +1,22 @@ +/* utility definitions */ +#ifdef _POSIX2_RE_DUP_MAX +#define DUPMAX _POSIX2_RE_DUP_MAX +#else +#define DUPMAX 255 +#endif +#define INFINITY (DUPMAX + 1) +#define NC (CHAR_MAX - CHAR_MIN + 1) +typedef unsigned char uch; + +/* switch off assertions (if not already off) if no REDEBUG */ +#ifndef REDEBUG +#ifndef NDEBUG +#define NDEBUG /* no assertions please */ +#endif +#endif +#include + +/* for old systems with bcopy() but no memmove() */ +#ifdef USEBCOPY +#define memmove(d, s, c) bcopy(s, d, c) +#endif -- cgit v1.2.3