blob: 22a2d5dd96c4ac9dbf5bc8750ae79af1d3b36f76 [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) */
Benjamin Peterson5ca88d22013-01-01 23:04:16 -060018#ifdef Py_USING_UNICODE
Serhiy Storchakae18e05c2013-02-16 16:47:15 +020019# define SRE_CODE Py_UCS4
20# if SIZEOF_SIZE_T > 4
21# define SRE_MAXREPEAT (~(SRE_CODE)0)
22# else
Serhiy Storchaka3ade66c2013-08-03 19:26:33 +030023# define SRE_MAXREPEAT ((SRE_CODE)PY_SSIZE_T_MAX)
Serhiy Storchakae18e05c2013-02-16 16:47:15 +020024# endif
Benjamin Peterson5ca88d22013-01-01 23:04:16 -060025#else
Serhiy Storchakacb1a0282013-05-21 22:53:33 +030026# define SRE_CODE unsigned int
27# if SIZEOF_SIZE_T > SIZEOF_INT
Serhiy Storchakae18e05c2013-02-16 16:47:15 +020028# define SRE_MAXREPEAT (~(SRE_CODE)0)
29# else
Serhiy Storchaka3ade66c2013-08-03 19:26:33 +030030# define SRE_MAXREPEAT ((SRE_CODE)PY_SSIZE_T_MAX)
Serhiy Storchakae18e05c2013-02-16 16:47:15 +020031# endif
Benjamin Peterson5ca88d22013-01-01 23:04:16 -060032#endif
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000033
Guido van Rossumb700df92000-03-31 14:59:30 +000034typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000035 PyObject_VAR_HEAD
Neal Norwitza6d80fa2006-06-12 03:05:40 +000036 Py_ssize_t groups; /* must be first! */
Guido van Rossumb700df92000-03-31 14:59:30 +000037 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000038 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000039 /* compatibility */
40 PyObject* pattern; /* pattern source (or None) */
41 int flags; /* flags used when compiling pattern source */
Raymond Hettinger027bb632004-05-31 03:09:25 +000042 PyObject *weakreflist; /* List of weak references */
Fredrik Lundh6f013982000-07-03 18:44:21 +000043 /* pattern code */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000044 Py_ssize_t codesize;
Fredrik Lundh6f013982000-07-03 18:44:21 +000045 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000046} PatternObject;
47
Fredrik Lundh6f013982000-07-03 18:44:21 +000048#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000049
50typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000051 PyObject_VAR_HEAD
Fredrik Lundhb0f05bd2001-07-02 16:42:49 +000052 PyObject* string; /* link to the target string (must be first) */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000053 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000054 PatternObject* pattern; /* link to the regex (pattern) object */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000055 Py_ssize_t pos, endpos; /* current target slice */
56 Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
57 Py_ssize_t groups; /* number of groups (start/end marks) */
58 Py_ssize_t mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000059} MatchObject;
60
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000061typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
62
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000063/* FIXME: <fl> shouldn't be a constant, really... */
64#define SRE_MARK_SIZE 200
65
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000066typedef struct SRE_REPEAT_T {
Neal Norwitza6d80fa2006-06-12 03:05:40 +000067 Py_ssize_t count;
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000068 SRE_CODE* pattern; /* points to REPEAT operator arguments */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000069 void* last_ptr; /* helper to check for infinite loops */
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000070 struct SRE_REPEAT_T *prev; /* points to previous repeat context */
71} SRE_REPEAT;
72
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000073typedef struct {
74 /* string pointers */
75 void* ptr; /* current position (also end of current slice) */
76 void* beginning; /* start of original string */
77 void* start; /* start of current slice */
78 void* end; /* end of original string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000079 /* attributes for the match object */
80 PyObject* string;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000081 Py_ssize_t pos, endpos;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000082 /* character size */
83 int charsize;
84 /* registers */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000085 Py_ssize_t lastindex;
86 Py_ssize_t lastmark;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000087 void* mark[SRE_MARK_SIZE];
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000088 /* dynamically allocated stuff */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000089 char* data_stack;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000090 size_t data_stack_size;
91 size_t data_stack_base;
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000092 /* current repeat context */
93 SRE_REPEAT *repeat;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000094 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000095 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000096} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000097
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000098typedef struct {
99 PyObject_HEAD
100 PyObject* pattern;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000101 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +0000102} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +0000103
104#endif