blob: b632165a1f516591e5c3640152c13a8282742ac4 [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 Pitrou766a16e2012-06-23 14:17:39 +020017 large enough to hold a UCS4 character) */
Martin v. Löwis7d9c6c72004-05-07 07:18:13 +000018#define SRE_CODE Py_UCS4
Serhiy Storchaka70ca0212013-02-16 16:47:47 +020019#if SIZEOF_SIZE_T > 4
20# define SRE_MAXREPEAT (~(SRE_CODE)0)
Serhiy Storchaka9baa5b22014-09-29 22:49:23 +030021# define SRE_MAXGROUPS ((~(SRE_CODE)0) / 2)
Serhiy Storchaka70ca0212013-02-16 16:47:47 +020022#else
Serhiy Storchaka1f35ae02013-08-03 19:18:38 +030023# define SRE_MAXREPEAT ((SRE_CODE)PY_SSIZE_T_MAX)
Serhiy Storchaka9baa5b22014-09-29 22:49:23 +030024# define SRE_MAXGROUPS ((SRE_CODE)PY_SSIZE_T_MAX / SIZEOF_SIZE_T / 2)
Serhiy Storchaka70ca0212013-02-16 16:47:47 +020025#endif
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000026
Guido van Rossumb700df92000-03-31 14:59:30 +000027typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000028 PyObject_VAR_HEAD
Thomas Wouters0e3f5912006-08-11 14:57:12 +000029 Py_ssize_t groups; /* must be first! */
Guido van Rossumb700df92000-03-31 14:59:30 +000030 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000031 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000032 /* compatibility */
33 PyObject* pattern; /* pattern source (or None) */
34 int flags; /* flags used when compiling pattern source */
Raymond Hettinger027bb632004-05-31 03:09:25 +000035 PyObject *weakreflist; /* List of weak references */
Serhiy Storchaka9eabac62013-10-26 10:45:48 +030036 int isbytes; /* pattern type (1 - bytes, 0 - string, -1 - None) */
Fredrik Lundh6f013982000-07-03 18:44:21 +000037 /* pattern code */
Thomas Wouters0e3f5912006-08-11 14:57:12 +000038 Py_ssize_t codesize;
Fredrik Lundh6f013982000-07-03 18:44:21 +000039 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000040} PatternObject;
41
Fredrik Lundh6f013982000-07-03 18:44:21 +000042#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000043
44typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000045 PyObject_VAR_HEAD
Fredrik Lundhb0f05bd2001-07-02 16:42:49 +000046 PyObject* string; /* link to the target string (must be first) */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000047 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000048 PatternObject* pattern; /* link to the regex (pattern) object */
Thomas Wouters0e3f5912006-08-11 14:57:12 +000049 Py_ssize_t pos, endpos; /* current target slice */
50 Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
51 Py_ssize_t groups; /* number of groups (start/end marks) */
52 Py_ssize_t mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000053} MatchObject;
54
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000055typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
56
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000057typedef struct SRE_REPEAT_T {
Thomas Wouters0e3f5912006-08-11 14:57:12 +000058 Py_ssize_t count;
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000059 SRE_CODE* pattern; /* points to REPEAT operator arguments */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000060 void* last_ptr; /* helper to check for infinite loops */
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000061 struct SRE_REPEAT_T *prev; /* points to previous repeat context */
62} SRE_REPEAT;
63
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000064typedef struct {
65 /* string pointers */
66 void* ptr; /* current position (also end of current slice) */
67 void* beginning; /* start of original string */
68 void* start; /* start of current slice */
69 void* end; /* end of original string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000070 /* attributes for the match object */
71 PyObject* string;
Thomas Wouters0e3f5912006-08-11 14:57:12 +000072 Py_ssize_t pos, endpos;
Serhiy Storchaka9eabac62013-10-26 10:45:48 +030073 int isbytes;
74 int charsize; /* character size */
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000075 /* registers */
Thomas Wouters0e3f5912006-08-11 14:57:12 +000076 Py_ssize_t lastindex;
77 Py_ssize_t lastmark;
Serhiy Storchaka9baa5b22014-09-29 22:49:23 +030078 void** mark;
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000079 /* dynamically allocated stuff */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000080 char* data_stack;
Thomas Wouters0e3f5912006-08-11 14:57:12 +000081 size_t data_stack_size;
82 size_t data_stack_base;
Benjamin Petersone48944b2012-03-07 14:50:25 -060083 Py_buffer buffer;
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000084 /* current repeat context */
85 SRE_REPEAT *repeat;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000086 /* hooks */
Serhiy Storchaka4b8f8942014-10-31 12:36:56 +020087 SRE_TOLOWER_HOOK lower, upper;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000088} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000089
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000090typedef struct {
91 PyObject_HEAD
92 PyObject* pattern;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000093 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000094} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000095
96#endif