tags/pcre-1.09/internal.h

/*************************************************
*      Perl-Compatible Regular Expressions       *
*************************************************/


#define PCRE_VERSION       "1.01 19-Nov-1997"


/* This is a library of functions to support regular expressions whose syntax
and semantics are as close as possible to those of the Perl 5 language. See
the file Tech.Notes for some information on the internals.

Written by: Philip Hazel <ph10@cam.ac.uk>

           Copyright (c) 1997 University of Cambridge

-----------------------------------------------------------------------------
Permission is granted to anyone to use this software for any purpose on any
computer system, and to redistribute it freely, subject to the following
restrictions:

1. This software 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.

2. The origin of this software must not be misrepresented, either by
   explicit claim or by omission.

3. Altered versions must be plainly marked as such, and must not be
   misrepresented as being the original software.
-----------------------------------------------------------------------------
*/

/* This header contains definitions that are shared between the different
modules, but which are not relevant to the outside. */

/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
define a macro for memmove() if USE_BCOPY is defined. */

#ifdef USE_BCOPY
#define memmove(a, b, c) bcopy(b, a, c)
#endif

/* Standard C headers plus the external interface definition */

#include <ctype.h>
#include <limits.h>
#include <setjmp.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "pcre.h"

/* Private options flags start at the most significant end of the two bytes.
The public options defined in pcre.h start at the least significant end. Make
sure they don't overlap! */

#define PCRE_FIRSTSET           0x8000  /* first_char is set */
#define PCRE_STARTLINE          0x4000  /* start after \n for multiline */
#define PCRE_COMPILED_CASELESS  0x2000  /* like it says */

/* Options for the "extra" block produced by pcre_study(). */

#define PCRE_STUDY_CASELESS 0x01     /* study was caseless */
#define PCRE_STUDY_MAPPED   0x02     /* a map of starting chars exists */

/* Masks for identifying the public options: all permitted at compile time,
only some permitted at run or study time. */

#define PUBLIC_OPTIONS \
  (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
   PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA)

#define PUBLIC_EXEC_OPTIONS \
  (PCRE_CASELESS|PCRE_ANCHORED|PCRE_MULTILINE|PCRE_NOTBOL|PCRE_NOTEOL| \
   PCRE_DOTALL|PCRE_DOLLAR_ENDONLY)

#define PUBLIC_STUDY_OPTIONS (PCRE_CASELESS)

/* Magic number to provide a small check against being handed junk. */

#define MAGIC_NUMBER  0x50435245   /* 'PCRE' */

/* Miscellaneous definitions */

typedef int BOOL;

#define FALSE   0
#define TRUE    1

/* These are escaped items that aren't just an encoding of a particular data
value such as \n. They must have non-zero values, as check_escape() returns
their negation. Also, they must appear in the same order as in the opcode
definitions below, up to ESC_Z. The final one must be ESC_REF as subsequent
values are used for \1, \2, \3, etc. There is a test in the code for an escape
greater than ESC_b and less than ESC_X to detect the types that may be
repeated. If any new escapes are put in-between that don't consume a character,
that code will have to change. */

enum { ESC_A = 1, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W, ESC_w,

                    /* These are not Perl escapes, so can't appear in the */
       ESC_X,       /* simple table-lookup because they must be conditional */
                    /* on PCRE_EXTRA. */
       ESC_Z,
       ESC_REF };

/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
OP_EOL must correspond in order to the list of escapes immediately above. */

enum {
  OP_END,            /* End of pattern */

  /* Values corresponding to backslashed metacharacters */

  OP_SOD,            /* Start of data: \A */
  OP_NOT_WORD_BOUNDARY,  /* \B */
  OP_WORD_BOUNDARY,      /* \b */
  OP_NOT_DIGIT,          /* \D */
  OP_DIGIT,              /* \d */
  OP_NOT_WHITESPACE,     /* \S */
  OP_WHITESPACE,         /* \s */
  OP_NOT_WORDCHAR,       /* \W */
  OP_WORDCHAR,           /* \w */
  OP_CUT,            /* The analogue of Prolog's "cut" operation (extension) */
  OP_EOD,            /* End of data: or \Z. This must always be the last
                        of the backslashed meta values. */

  OP_CIRC,           /* Start of line - varies with multiline switch */
  OP_DOLL,           /* End of line - varies with multiline switch */
  OP_ANY,            /* Match any character */
  OP_CHARS,          /* Match string of characters */
  OP_NOT,            /* Match anything but the following char */

  OP_STAR,           /* The maximizing and minimizing versions of */
  OP_MINSTAR,        /* all these opcodes must come in pairs, with */
  OP_PLUS,           /* the minimizing one second. */
  OP_MINPLUS,        /* This first set applies to single characters */
  OP_QUERY,
  OP_MINQUERY,
  OP_UPTO,           /* From 0 to n matches */
  OP_MINUPTO,
  OP_EXACT,          /* Exactly n matches */

  OP_NOTSTAR,        /* The maximizing and minimizing versions of */
  OP_NOTMINSTAR,     /* all these opcodes must come in pairs, with */
  OP_NOTPLUS,        /* the minimizing one second. */
  OP_NOTMINPLUS,     /* This first set applies to "not" single characters */
  OP_NOTQUERY,
  OP_NOTMINQUERY,
  OP_NOTUPTO,        /* From 0 to n matches */
  OP_NOTMINUPTO,
  OP_NOTEXACT,       /* Exactly n matches */

  OP_TYPESTAR,       /* The maximizing and minimizing versions of */
  OP_TYPEMINSTAR,    /* all these opcodes must come in pairs, with */
  OP_TYPEPLUS,       /* the minimizing one second. These codes must */
  OP_TYPEMINPLUS,    /* be in exactly the same order as those above. */
  OP_TYPEQUERY,      /* This set applies to character types such as \d */
  OP_TYPEMINQUERY,
  OP_TYPEUPTO,       /* From 0 to n matches */
  OP_TYPEMINUPTO,
  OP_TYPEEXACT,      /* Exactly n matches */

  OP_CRSTAR,         /* The maximizing and minimizing versions of */
  OP_CRMINSTAR,      /* all these opcodes must come in pairs, with */
  OP_CRPLUS,         /* the minimizing one second. These codes must */
  OP_CRMINPLUS,      /* be in exactly the same order as those above. */
  OP_CRQUERY,        /* These are for character classes and back refs */
  OP_CRMINQUERY,
  OP_CRRANGE,        /* These are different to the three seta above. */
  OP_CRMINRANGE,

  OP_CLASS,          /* Match a character class */
  OP_REF,            /* Match a back reference */

  OP_ALT,            /* Start of alternation */
  OP_KET,            /* End of group that doesn't have an unbounded repeat */
  OP_KETRMAX,        /* These two must remain together and in this */
  OP_KETRMIN,        /* order. They are for groups the repeat for ever. */

  OP_ASSERT,
  OP_ASSERT_NOT,
  OP_ONCE,           /* Once matched, don't back up into the subpattern */

  OP_BRAZERO,        /* These two must remain together and in this */
  OP_BRAMINZERO,     /* order. */

  OP_BRA             /* This and greater values are used for brackets that
                        extract substrings. */
};

/* The highest extraction number. This is limited by the number of opcodes
left after OP_BRA, i.e. 255 - OP_BRA. We actually set it somewhat lower. */

#define EXTRACT_MAX  99

/* The texts of compile-time error messages are defined as macros here so that
they can be accessed by the POSIX wrapper and converted into error codes.  Yes,
I could have used error codes in the first place, but didn't feel like changing
just to accommodate the POSIX wrapper. */

#define ERR1  "\\ at end of pattern"
#define ERR2  "\\c at end of pattern"
#define ERR3  "unrecognized character follows \\"
#define ERR4  "numbers out of order in {} quantifier"
#define ERR5  "number too big in {} quantifier"
#define ERR6  "missing terminating ] for character class"
#define ERR7  "invalid escape sequence in character class"
#define ERR8  "range out of order in character class"
#define ERR9  "nothing to repeat"
#define ERR10 "operand of unlimited repeat could match the empty string"
#define ERR11 "internal error: unexpected repeat"
#define ERR12 "unrecognized character after (?"
#define ERR13 "too many capturing parenthesized sub-patterns"
#define ERR14 "missing )"
#define ERR15 "back reference to non-existent subpattern"
#define ERR16 "erroffset passed as NULL"
#define ERR17 "unknown option bit(s) set"
#define ERR18 "missing ) after comment"
#define ERR19 "too many sets of parentheses"
#define ERR20 "regular expression too large"
#define ERR21 "failed to get memory"
#define ERR22 "unmatched brackets"
#define ERR23 "internal error: code overflow"

/* All character handling must be done as unsigned characters. Otherwise there
are problems with top-bit-set characters and functions such as isspace().
However, we leave the interface to the outside world as char *, because that
should make things easier for callers. We define a short type for unsigned char
to save lots of typing. I tried "uchar", but it causes problems on Digital
Unix, where it is defined in sys/types, so use "uschar" instead. */

typedef unsigned char uschar;

/* The real format of the start of the pcre block; the actual code vector
runs on as long as necessary after the end. */

typedef struct real_pcre {
  unsigned int  magic_number;
  unsigned short int options;
  unsigned char top_bracket;
  unsigned char top_backref;
  unsigned char first_char;
  unsigned char code[1];
} real_pcre;

/* The real format of the extra block returned by pcre_study(). */

typedef struct real_pcre_extra {
  unsigned char options;
  unsigned char start_bits[32];
} real_pcre_extra;

/* Global tables from chartables.c */

extern uschar pcre_lcc[];
extern uschar pcre_fcc[];
extern uschar pcre_cbits[];
extern uschar pcre_ctypes[];

/* Bit definitions for entries in pcre_ctypes[]. */

#define ctype_space   0x01
#define ctype_letter  0x02
#define ctype_digit   0x04
#define ctype_xdigit  0x08
#define ctype_word    0x10   /* alphameric or '_' */
#define ctype_meta    0x80   /* regexp meta char or zero (end pattern) */

/* Offsets for the bitmap tables */

#define cbit_digit    0
#define cbit_letter  32
#define cbit_word    64
#define cbit_space   96
#define cbit_length 128      /* Length of the cbits table */

/* End of internal.h */
1	nigel	3	/*************************************************
2			* Perl-Compatible Regular Expressions *
3			*************************************************/
4
5
6	nigel	5	#define PCRE_VERSION "1.01 19-Nov-1997"
7	nigel	3
8
9			/* This is a library of functions to support regular expressions whose syntax
10			and semantics are as close as possible to those of the Perl 5 language. See
11			the file Tech.Notes for some information on the internals.
12
13			Written by: Philip Hazel <ph10@cam.ac.uk>
14
15			Copyright (c) 1997 University of Cambridge
16
17			-----------------------------------------------------------------------------
18			Permission is granted to anyone to use this software for any purpose on any
19			computer system, and to redistribute it freely, subject to the following
20			restrictions:
21
22			1. 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
26			2. The origin of this software must not be misrepresented, either by
27			explicit claim or by omission.
28
29			3. 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
35			modules, but which are not relevant to the outside. */
36
37			/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
38			define a macro for memmove() if USE_BCOPY is defined. */
39
40			#ifdef USE_BCOPY
41			#define memmove(a, b, c) bcopy(b, a, c)
42			#endif
43
44			/* Standard C headers plus the external interface definition */
45
46			#include <ctype.h>
47			#include <limits.h>
48			#include <setjmp.h>
49			#include <stddef.h>
50			#include <stdio.h>
51			#include <stdlib.h>
52			#include <string.h>
53			#include "pcre.h"
54
55			/* Private options flags start at the most significant end of the two bytes.
56			The public options defined in pcre.h start at the least significant end. Make
57			sure they don't overlap! */
58
59			#define PCRE_FIRSTSET 0x8000 /* first_char is set */
60			#define PCRE_STARTLINE 0x4000 /* start after \n for multiline */
61			#define PCRE_COMPILED_CASELESS 0x2000 /* like it says */
62
63			/* Options for the "extra" block produced by pcre_study(). */
64
65			#define PCRE_STUDY_CASELESS 0x01 /* study was caseless */
66			#define PCRE_STUDY_MAPPED 0x02 /* a map of starting chars exists */
67
68			/* Masks for identifying the public options: all permitted at compile time,
69			only some permitted at run or study time. */
70
71			#define PUBLIC_OPTIONS \
72			(PCRE_CASELESS\|PCRE_EXTENDED\|PCRE_ANCHORED\|PCRE_MULTILINE\| \
73			PCRE_DOTALL\|PCRE_DOLLAR_ENDONLY\|PCRE_EXTRA)
74
75			#define PUBLIC_EXEC_OPTIONS \
76			(PCRE_CASELESS\|PCRE_ANCHORED\|PCRE_MULTILINE\|PCRE_NOTBOL\|PCRE_NOTEOL\| \
77			PCRE_DOTALL\|PCRE_DOLLAR_ENDONLY)
78
79			#define PUBLIC_STUDY_OPTIONS (PCRE_CASELESS)
80
81			/* Magic number to provide a small check against being handed junk. */
82
83			#define MAGIC_NUMBER 0x50435245 /* 'PCRE' */
84
85			/* Miscellaneous definitions */
86
87			typedef int BOOL;
88
89			#define FALSE 0
90			#define TRUE 1
91
92			/* These are escaped items that aren't just an encoding of a particular data
93			value such as \n. They must have non-zero values, as check_escape() returns
94			their negation. Also, they must appear in the same order as in the opcode
95			definitions below, up to ESC_Z. The final one must be ESC_REF as subsequent
96			values are used for \1, \2, \3, etc. There is a test in the code for an escape
97			greater than ESC_b and less than ESC_X to detect the types that may be
98			repeated. If any new escapes are put in-between that don't consume a character,
99			that code will have to change. */
100
101			enum { ESC_A = 1, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W, ESC_w,
102
103			/* These are not Perl escapes, so can't appear in the */
104			ESC_X, /* simple table-lookup because they must be conditional */
105			/* on PCRE_EXTRA. */
106			ESC_Z,
107			ESC_REF };
108
109			/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
110			that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
111			OP_EOL must correspond in order to the list of escapes immediately above. */
112
113			enum {
114			OP_END, /* End of pattern */
115
116			/* Values corresponding to backslashed metacharacters */
117
118			OP_SOD, /* Start of data: \A */
119			OP_NOT_WORD_BOUNDARY, /* \B */
120			OP_WORD_BOUNDARY, /* \b */
121			OP_NOT_DIGIT, /* \D */
122			OP_DIGIT, /* \d */
123			OP_NOT_WHITESPACE, /* \S */
124			OP_WHITESPACE, /* \s */
125			OP_NOT_WORDCHAR, /* \W */
126			OP_WORDCHAR, /* \w */
127			OP_CUT, /* The analogue of Prolog's "cut" operation (extension) */
128			OP_EOD, /* End of data: or \Z. This must always be the last
129			of the backslashed meta values. */
130
131			OP_CIRC, /* Start of line - varies with multiline switch */
132			OP_DOLL, /* End of line - varies with multiline switch */
133			OP_ANY, /* Match any character */
134			OP_CHARS, /* Match string of characters */
135			OP_NOT, /* Match anything but the following char */
136
137			OP_STAR, /* The maximizing and minimizing versions of */
138			OP_MINSTAR, /* all these opcodes must come in pairs, with */
139			OP_PLUS, /* the minimizing one second. */
140			OP_MINPLUS, /* This first set applies to single characters */
141			OP_QUERY,
142			OP_MINQUERY,
143			OP_UPTO, /* From 0 to n matches */
144			OP_MINUPTO,
145			OP_EXACT, /* Exactly n matches */
146
147			OP_NOTSTAR, /* The maximizing and minimizing versions of */
148			OP_NOTMINSTAR, /* all these opcodes must come in pairs, with */
149			OP_NOTPLUS, /* the minimizing one second. */
150			OP_NOTMINPLUS, /* This first set applies to "not" single characters */
151			OP_NOTQUERY,
152			OP_NOTMINQUERY,
153			OP_NOTUPTO, /* From 0 to n matches */
154			OP_NOTMINUPTO,
155			OP_NOTEXACT, /* Exactly n matches */
156
157			OP_TYPESTAR, /* The maximizing and minimizing versions of */
158			OP_TYPEMINSTAR, /* all these opcodes must come in pairs, with */
159			OP_TYPEPLUS, /* the minimizing one second. These codes must */
160			OP_TYPEMINPLUS, /* be in exactly the same order as those above. */
161			OP_TYPEQUERY, /* This set applies to character types such as \d */
162			OP_TYPEMINQUERY,
163			OP_TYPEUPTO, /* From 0 to n matches */
164			OP_TYPEMINUPTO,
165			OP_TYPEEXACT, /* Exactly n matches */
166
167			OP_CRSTAR, /* The maximizing and minimizing versions of */
168			OP_CRMINSTAR, /* all these opcodes must come in pairs, with */
169			OP_CRPLUS, /* the minimizing one second. These codes must */
170			OP_CRMINPLUS, /* be in exactly the same order as those above. */
171			OP_CRQUERY, /* These are for character classes and back refs */
172			OP_CRMINQUERY,
173			OP_CRRANGE, /* These are different to the three seta above. */
174			OP_CRMINRANGE,
175
176			OP_CLASS, /* Match a character class */
177			OP_REF, /* Match a back reference */
178
179			OP_ALT, /* Start of alternation */
180			OP_KET, /* End of group that doesn't have an unbounded repeat */
181			OP_KETRMAX, /* These two must remain together and in this */
182			OP_KETRMIN, /* order. They are for groups the repeat for ever. */
183
184			OP_ASSERT,
185			OP_ASSERT_NOT,
186			OP_ONCE, /* Once matched, don't back up into the subpattern */
187
188			OP_BRAZERO, /* These two must remain together and in this */
189			OP_BRAMINZERO, /* order. */
190
191			OP_BRA /* This and greater values are used for brackets that
192			extract substrings. */
193			};
194
195			/* The highest extraction number. This is limited by the number of opcodes
196			left after OP_BRA, i.e. 255 - OP_BRA. We actually set it somewhat lower. */
197
198			#define EXTRACT_MAX 99
199
200			/* The texts of compile-time error messages are defined as macros here so that
201			they can be accessed by the POSIX wrapper and converted into error codes. Yes,
202			I could have used error codes in the first place, but didn't feel like changing
203			just to accommodate the POSIX wrapper. */
204
205			#define ERR1 "\\ at end of pattern"
206			#define ERR2 "\\c at end of pattern"
207			#define ERR3 "unrecognized character follows \\"
208			#define ERR4 "numbers out of order in {} quantifier"
209			#define ERR5 "number too big in {} quantifier"
210			#define ERR6 "missing terminating ] for character class"
211			#define ERR7 "invalid escape sequence in character class"
212			#define ERR8 "range out of order in character class"
213			#define ERR9 "nothing to repeat"
214			#define ERR10 "operand of unlimited repeat could match the empty string"
215			#define ERR11 "internal error: unexpected repeat"
216			#define ERR12 "unrecognized character after (?"
217			#define ERR13 "too many capturing parenthesized sub-patterns"
218			#define ERR14 "missing )"
219			#define ERR15 "back reference to non-existent subpattern"
220			#define ERR16 "erroffset passed as NULL"
221			#define ERR17 "unknown option bit(s) set"
222			#define ERR18 "missing ) after comment"
223			#define ERR19 "too many sets of parentheses"
224			#define ERR20 "regular expression too large"
225			#define ERR21 "failed to get memory"
226			#define ERR22 "unmatched brackets"
227			#define ERR23 "internal error: code overflow"
228
229			/* All character handling must be done as unsigned characters. Otherwise there
230			are problems with top-bit-set characters and functions such as isspace().
231			However, we leave the interface to the outside world as char *, because that
232			should make things easier for callers. We define a short type for unsigned char
233			to save lots of typing. I tried "uchar", but it causes problems on Digital
234			Unix, where it is defined in sys/types, so use "uschar" instead. */
235
236			typedef unsigned char uschar;
237
238			/* The real format of the start of the pcre block; the actual code vector
239			runs on as long as necessary after the end. */
240
241			typedef struct real_pcre {
242			unsigned int magic_number;
243			unsigned short int options;
244			unsigned char top_bracket;
245			unsigned char top_backref;
246			unsigned char first_char;
247			unsigned char code[1];
248			} real_pcre;
249
250			/* The real format of the extra block returned by pcre_study(). */
251
252			typedef struct real_pcre_extra {
253			unsigned char options;
254			unsigned char start_bits[32];
255			} real_pcre_extra;
256
257			/* Global tables from chartables.c */
258
259			extern uschar pcre_lcc[];
260			extern uschar pcre_fcc[];
261			extern uschar pcre_cbits[];
262			extern uschar pcre_ctypes[];
263
264			/* Bit definitions for entries in pcre_ctypes[]. */
265
266			#define ctype_space 0x01
267			#define ctype_letter 0x02
268			#define ctype_digit 0x04
269			#define ctype_xdigit 0x08
270			#define ctype_word 0x10 /* alphameric or '_' */
271			#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
272
273			/* Offsets for the bitmap tables */
274
275			#define cbit_digit 0
276			#define cbit_letter 32
277			#define cbit_word 64
278			#define cbit_space 96
279			#define cbit_length 128 /* Length of the cbits table */
280
281			/* End of internal.h */