blob: 9bfdf7fdfcdab2f5081894e5ce9b07377e2e8bab [file] [log] [blame]
Guido van Rossumb700df92000-03-31 14:59:30 +00001/*
2 * Secret Labs' Regular Expression Engine
Guido van Rossumb700df92000-03-31 14:59:30 +00003 *
Fredrik Lundh80946112000-06-29 18:03:25 +00004 * regular expression matching engine
Guido van Rossumb700df92000-03-31 14:59:30 +00005 *
Fredrik Lundhb0f05bd2001-07-02 16:42:49 +00006 * Copyright (c) 1997-2001 by Secret Labs AB. All rights reserved.
Guido van Rossumb700df92000-03-31 14:59:30 +00007 *
8 * See the _sre.c file for information on usage and redistribution.
9 */
10
11#ifndef SRE_INCLUDED
12#define SRE_INCLUDED
13
14#include "sre_constants.h"
15
Fredrik Lundh8f455852001-06-27 18:59:43 +000016/* size of a code word (must be unsigned short or larger, and
Antoine Pitroub83ea142012-11-20 22:30:42 +010017 large enough to hold a UCS4 character) */
Martin v. Löwis7d9c6c72004-05-07 07:18:13 +000018#define SRE_CODE Py_UCS4
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000019
Guido van Rossumb700df92000-03-31 14:59:30 +000020typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000021 PyObject_VAR_HEAD
Neal Norwitza6d80fa2006-06-12 03:05:40 +000022 Py_ssize_t groups; /* must be first! */
Guido van Rossumb700df92000-03-31 14:59:30 +000023 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000024 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000025 /* compatibility */
26 PyObject* pattern; /* pattern source (or None) */
27 int flags; /* flags used when compiling pattern source */
Raymond Hettinger027bb632004-05-31 03:09:25 +000028 PyObject *weakreflist; /* List of weak references */
Fredrik Lundh6f013982000-07-03 18:44:21 +000029 /* pattern code */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000030 Py_ssize_t codesize;
Fredrik Lundh6f013982000-07-03 18:44:21 +000031 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000032} PatternObject;
33
Fredrik Lundh6f013982000-07-03 18:44:21 +000034#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000035
36typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000037 PyObject_VAR_HEAD
Fredrik Lundhb0f05bd2001-07-02 16:42:49 +000038 PyObject* string; /* link to the target string (must be first) */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000039 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000040 PatternObject* pattern; /* link to the regex (pattern) object */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000041 Py_ssize_t pos, endpos; /* current target slice */
42 Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
43 Py_ssize_t groups; /* number of groups (start/end marks) */
44 Py_ssize_t mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000045} MatchObject;
46
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000047typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
48
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000049/* FIXME: <fl> shouldn't be a constant, really... */
50#define SRE_MARK_SIZE 200
51
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000052typedef struct SRE_REPEAT_T {
Neal Norwitza6d80fa2006-06-12 03:05:40 +000053 Py_ssize_t count;
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000054 SRE_CODE* pattern; /* points to REPEAT operator arguments */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000055 void* last_ptr; /* helper to check for infinite loops */
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000056 struct SRE_REPEAT_T *prev; /* points to previous repeat context */
57} SRE_REPEAT;
58
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000059typedef struct {
60 /* string pointers */
61 void* ptr; /* current position (also end of current slice) */
62 void* beginning; /* start of original string */
63 void* start; /* start of current slice */
64 void* end; /* end of original string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000065 /* attributes for the match object */
66 PyObject* string;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000067 Py_ssize_t pos, endpos;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000068 /* character size */
69 int charsize;
70 /* registers */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000071 Py_ssize_t lastindex;
72 Py_ssize_t lastmark;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000073 void* mark[SRE_MARK_SIZE];
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000074 /* dynamically allocated stuff */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000075 char* data_stack;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000076 size_t data_stack_size;
77 size_t data_stack_base;
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000078 /* current repeat context */
79 SRE_REPEAT *repeat;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000080 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000081 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000082} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000083
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000084typedef struct {
85 PyObject_HEAD
86 PyObject* pattern;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000087 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000088} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000089
90#endif