blob: 735c02d094dbd7635bb3295617faf125fed2dd16 [file] [log] [blame]
Guido van Rossum51b3aa31997-10-06 14:43:11 +00001/*************************************************
2* Perl-Compatible Regular Expressions *
3*************************************************/
4
5
Guido van Rossum50700601997-12-08 17:15:20 +00006#define PCRE_VERSION "1.01 19-Nov-1997"
Guido van Rossum51b3aa31997-10-06 14:43:11 +00007
8
9/* This is a library of functions to support regular expressions whose syntax
10and semantics are as close as possible to those of the Perl 5 language. See
11the file Tech.Notes for some information on the internals.
12
13Written by: Philip Hazel <ph10@cam.ac.uk>
14
15 Copyright (c) 1997 University of Cambridge
16
17-----------------------------------------------------------------------------
18Permission is granted to anyone to use this software for any purpose on any
19computer system, and to redistribute it freely, subject to the following
20restrictions:
21
221. This software is distributed in the hope that it will be useful,
23 but WITHOUT ANY WARRANTY; without even the implied warranty of
24 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
25
262. The origin of this software must not be misrepresented, either by
27 explicit claim or by omission.
28
293. Altered versions must be plainly marked as such, and must not be
30 misrepresented as being the original software.
31-----------------------------------------------------------------------------
32*/
33
34/* This header contains definitions that are shared between the different
35modules, but which are not relevant to the outside. */
36
Guido van Rossum50700601997-12-08 17:15:20 +000037
38/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
39define a macro for memmove() if USE_BCOPY is defined. */
40
41#ifdef USE_BCOPY
42#define memmove(a, b, c) bcopy(b, a, c)
43#endif
44
Guido van Rossum51b3aa31997-10-06 14:43:11 +000045/* Standard C headers plus the external interface definition */
46
47#include <ctype.h>
48#include <limits.h>
Guido van Rossum50700601997-12-08 17:15:20 +000049#include <setjmp.h>
50#include <stddef.h>
Guido van Rossum51b3aa31997-10-06 14:43:11 +000051#include <stdio.h>
52#include <stdlib.h>
53#include <string.h>
54#include "pcre.h"
55
Guido van Rossum50700601997-12-08 17:15:20 +000056/* Private options flags start at the most significant end of the two bytes.
57The public options defined in pcre.h start at the least significant end. Make
58sure they don't overlap! */
Guido van Rossum51b3aa31997-10-06 14:43:11 +000059
Guido van Rossum50700601997-12-08 17:15:20 +000060#define PCRE_FIRSTSET 0x8000 /* first_char is set */
61#define PCRE_STARTLINE 0x4000 /* start after \n for multiline */
62#define PCRE_COMPILED_CASELESS 0x2000 /* like it says */
Guido van Rossum51b3aa31997-10-06 14:43:11 +000063
64/* Options for the "extra" block produced by pcre_study(). */
65
66#define PCRE_STUDY_CASELESS 0x01 /* study was caseless */
Guido van Rossum50700601997-12-08 17:15:20 +000067#define PCRE_STUDY_MAPPED 0x02 /* a map of starting chars exists */
Guido van Rossum51b3aa31997-10-06 14:43:11 +000068
69/* Masks for identifying the public options: all permitted at compile time,
70only some permitted at run or study time. */
71
72#ifdef FOR_PYTHON
73#define PUBLIC_OPTIONS \
Guido van Rossum50700601997-12-08 17:15:20 +000074 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
75 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_LOCALE)
Guido van Rossum51b3aa31997-10-06 14:43:11 +000076#else
77#define PUBLIC_OPTIONS \
Guido van Rossum50700601997-12-08 17:15:20 +000078 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
79 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA)
Guido van Rossum51b3aa31997-10-06 14:43:11 +000080#endif
Guido van Rossum50700601997-12-08 17:15:20 +000081#define PUBLIC_EXEC_OPTIONS \
82 (PCRE_CASELESS|PCRE_ANCHORED|PCRE_MULTILINE|PCRE_NOTBOL|PCRE_NOTEOL| \
83 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY)
84
Guido van Rossum51b3aa31997-10-06 14:43:11 +000085#define PUBLIC_STUDY_OPTIONS (PCRE_CASELESS)
86
87/* Magic number to provide a small check against being handed junk. */
88
89#define MAGIC_NUMBER 0x50435245 /* 'PCRE' */
90
91/* Miscellaneous definitions */
92
93typedef int BOOL;
94
95#define FALSE 0
96#define TRUE 1
97
Guido van Rossum51b3aa31997-10-06 14:43:11 +000098/* These are escaped items that aren't just an encoding of a particular data
99value such as \n. They must have non-zero values, as check_escape() returns
100their negation. Also, they must appear in the same order as in the opcode
101definitions below, up to ESC_Z. The final one must be ESC_REF as subsequent
102values are used for \1, \2, \3, etc. There is a test in the code for an escape
Guido van Rossum50700601997-12-08 17:15:20 +0000103greater than ESC_b and less than ESC_X to detect the types that may be
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000104repeated. If any new escapes are put in-between that don't consume a character,
105that code will have to change. */
106
107enum { ESC_A = 1, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W, ESC_w,
Guido van Rossum50700601997-12-08 17:15:20 +0000108
109 /* These are not Perl escapes, so can't appear in the */
110 ESC_X, /* simple table-lookup because they must be conditional */
111 /* on PCRE_EXTRA. */
112 ESC_Z,
113 ESC_REF };
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000114
115/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
116that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
117OP_EOL must correspond in order to the list of escapes immediately above. */
118
119enum {
120 OP_END, /* End of pattern */
121
122 /* Values corresponding to backslashed metacharacters */
123
124 OP_SOD, /* Start of data: \A */
Guido van Rossum50700601997-12-08 17:15:20 +0000125 OP_NOT_WORD_BOUNDARY, /* \B */
126 OP_WORD_BOUNDARY, /* \b */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000127 OP_NOT_DIGIT, /* \D */
128 OP_DIGIT, /* \d */
129 OP_NOT_WHITESPACE, /* \S */
130 OP_WHITESPACE, /* \s */
131 OP_NOT_WORDCHAR, /* \W */
132 OP_WORDCHAR, /* \w */
Guido van Rossum50700601997-12-08 17:15:20 +0000133 OP_CUT, /* The analogue of Prolog's "cut" operation (extension) */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000134 OP_EOD, /* End of data: or \Z. This must always be the last
135 of the backslashed meta values. */
136
Guido van Rossum50700601997-12-08 17:15:20 +0000137 OP_NOT_WORD_BOUNDARY_L, /* localized \B */
138 OP_WORD_BOUNDARY_L, /* localized \b */
139 OP_NOT_WORDCHAR_L, /* localized \W */
140 OP_WORDCHAR_L, /* localized \w */
141
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000142 OP_CIRC, /* Start of line - varies with multiline switch */
143 OP_DOLL, /* End of line - varies with multiline switch */
144 OP_ANY, /* Match any character */
145 OP_CHARS, /* Match string of characters */
Guido van Rossum50700601997-12-08 17:15:20 +0000146 OP_NOT, /* Match anything but the following char */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000147
148 OP_STAR, /* The maximizing and minimizing versions of */
149 OP_MINSTAR, /* all these opcodes must come in pairs, with */
150 OP_PLUS, /* the minimizing one second. */
151 OP_MINPLUS, /* This first set applies to single characters */
152 OP_QUERY,
153 OP_MINQUERY,
Guido van Rossum50700601997-12-08 17:15:20 +0000154 OP_UPTO, /* From 0 to n matches */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000155 OP_MINUPTO,
Guido van Rossum50700601997-12-08 17:15:20 +0000156 OP_EXACT, /* Exactly n matches */
157
158 OP_NOTSTAR, /* The maximizing and minimizing versions of */
159 OP_NOTMINSTAR, /* all these opcodes must come in pairs, with */
160 OP_NOTPLUS, /* the minimizing one second. */
161 OP_NOTMINPLUS, /* This first set applies to "not" single characters */
162 OP_NOTQUERY,
163 OP_NOTMINQUERY,
164 OP_NOTUPTO, /* From 0 to n matches */
165 OP_NOTMINUPTO,
166 OP_NOTEXACT, /* Exactly n matches */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000167
168 OP_TYPESTAR, /* The maximizing and minimizing versions of */
169 OP_TYPEMINSTAR, /* all these opcodes must come in pairs, with */
170 OP_TYPEPLUS, /* the minimizing one second. These codes must */
171 OP_TYPEMINPLUS, /* be in exactly the same order as those above. */
172 OP_TYPEQUERY, /* This set applies to character types such as \d */
173 OP_TYPEMINQUERY,
Guido van Rossum50700601997-12-08 17:15:20 +0000174 OP_TYPEUPTO, /* From 0 to n matches */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000175 OP_TYPEMINUPTO,
Guido van Rossum50700601997-12-08 17:15:20 +0000176 OP_TYPEEXACT, /* Exactly n matches */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000177
178 OP_CRSTAR, /* The maximizing and minimizing versions of */
179 OP_CRMINSTAR, /* all these opcodes must come in pairs, with */
180 OP_CRPLUS, /* the minimizing one second. These codes must */
181 OP_CRMINPLUS, /* be in exactly the same order as those above. */
182 OP_CRQUERY, /* These are for character classes and back refs */
183 OP_CRMINQUERY,
Guido van Rossum50700601997-12-08 17:15:20 +0000184 OP_CRRANGE, /* These are different to the three seta above. */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000185 OP_CRMINRANGE,
186
187 OP_CLASS, /* Match a character class */
Guido van Rossum50700601997-12-08 17:15:20 +0000188 OP_CLASS_L, /* Match a character class */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000189 OP_REF, /* Match a back reference */
190
191 OP_ALT, /* Start of alternation */
192 OP_KET, /* End of group that doesn't have an unbounded repeat */
193 OP_KETRMAX, /* These two must remain together and in this */
194 OP_KETRMIN, /* order. They are for groups the repeat for ever. */
195
196 OP_ASSERT,
197 OP_ASSERT_NOT,
Guido van Rossum50700601997-12-08 17:15:20 +0000198 OP_ONCE, /* Once matched, don't back up into the subpattern */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000199
200 OP_BRAZERO, /* These two must remain together and in this */
201 OP_BRAMINZERO, /* order. */
202
203 OP_BRA /* This and greater values are used for brackets that
204 extract substrings. */
205};
206
207/* The highest extraction number. This is limited by the number of opcodes
208left after OP_BRA, i.e. 255 - OP_BRA. We actually set it somewhat lower. */
209
210#define EXTRACT_MAX 99
211
Guido van Rossum50700601997-12-08 17:15:20 +0000212/* The texts of compile-time error messages are defined as macros here so that
213they can be accessed by the POSIX wrapper and converted into error codes. Yes,
214I could have used error codes in the first place, but didn't feel like changing
215just to accommodate the POSIX wrapper. */
216
217#define ERR1 "\\ at end of pattern"
218#define ERR2 "\\c at end of pattern"
219#define ERR3 "unrecognized character follows \\"
220#define ERR4 "numbers out of order in {} quantifier"
221#define ERR5 "number too big in {} quantifier"
222#define ERR6 "missing terminating ] for character class"
223#define ERR7 "invalid escape sequence in character class"
224#define ERR8 "range out of order in character class"
225#define ERR9 "nothing to repeat"
226#define ERR10 "operand of unlimited repeat could match the empty string"
227#define ERR11 "internal error: unexpected repeat"
228#define ERR12 "unrecognized character after (?"
229#define ERR13 "too many capturing parenthesized sub-patterns"
230#define ERR14 "missing )"
231#define ERR15 "back reference to non-existent subpattern"
232#define ERR16 "erroffset passed as NULL"
233#define ERR17 "unknown option bit(s) set"
234#define ERR18 "missing ) after comment"
235#define ERR19 "too many sets of parentheses"
236#define ERR20 "regular expression too large"
237#define ERR21 "failed to get memory"
238#define ERR22 "unmatched brackets"
239#define ERR23 "internal error: code overflow"
240
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000241/* All character handling must be done as unsigned characters. Otherwise there
242are problems with top-bit-set characters and functions such as isspace().
243However, we leave the interface to the outside world as char *, because that
244should make things easier for callers. We define a short type for unsigned char
245to save lots of typing. I tried "uchar", but it causes problems on Digital
246Unix, where it is defined in sys/types, so use "uschar" instead. */
247
248typedef unsigned char uschar;
249
250/* The real format of the start of the pcre block; the actual code vector
251runs on as long as necessary after the end. */
252
253typedef struct real_pcre {
254 unsigned int magic_number;
Guido van Rossum50700601997-12-08 17:15:20 +0000255 unsigned short int options;
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000256 unsigned char top_bracket;
Guido van Rossum50700601997-12-08 17:15:20 +0000257 unsigned char top_backref;
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000258 unsigned char first_char;
259 unsigned char code[1];
260} real_pcre;
261
262/* The real format of the extra block returned by pcre_study(). */
263
264typedef struct real_pcre_extra {
265 unsigned char options;
266 unsigned char start_bits[32];
267} real_pcre_extra;
268
Guido van Rossum50700601997-12-08 17:15:20 +0000269/* Global tables from chartables.c */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000270
271extern uschar pcre_lcc[];
Guido van Rossum50700601997-12-08 17:15:20 +0000272extern uschar pcre_fcc[];
273extern uschar pcre_cbits[];
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000274extern uschar pcre_ctypes[];
275
276/* Bit definitions for entries in pcre_ctypes[]. */
277
278#define ctype_space 0x01
Guido van Rossum50700601997-12-08 17:15:20 +0000279#define ctype_letter 0x02
280#define ctype_digit 0x04
281#define ctype_xdigit 0x08
282#define ctype_word 0x10 /* alphameric or '_' */
283#define ctype_odigit 0x20 /* octal digit */
Guido van Rossum51b3aa31997-10-06 14:43:11 +0000284#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
285
Guido van Rossum50700601997-12-08 17:15:20 +0000286/* Offsets for the bitmap tables */
287
288#define cbit_digit 0
289#define cbit_letter 32
290#define cbit_word 64
291#define cbit_space 96
292#define cbit_length 128 /* Length of the cbits table */
293
294/* End of internal.h */