blob: d4af05c045e2065d78d3fba38d77a9e5d2e65792 [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
17 large enough to hold a Py_UNICODE character) */
18#ifdef Py_UNICODE_WIDE
Martin v. Löwis7d9c6c72004-05-07 07:18:13 +000019#define SRE_CODE Py_UCS4
Fredrik Lundh1294ad02001-06-26 17:17:07 +000020#else
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000021#define SRE_CODE unsigned short
Fredrik Lundh1294ad02001-06-26 17:17:07 +000022#endif
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000023
Guido van Rossumb700df92000-03-31 14:59:30 +000024typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000025 PyObject_VAR_HEAD
Neal Norwitza6d80fa2006-06-12 03:05:40 +000026 Py_ssize_t groups; /* must be first! */
Guido van Rossumb700df92000-03-31 14:59:30 +000027 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000028 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000029 /* compatibility */
30 PyObject* pattern; /* pattern source (or None) */
31 int flags; /* flags used when compiling pattern source */
Raymond Hettinger027bb632004-05-31 03:09:25 +000032 PyObject *weakreflist; /* List of weak references */
Fredrik Lundh6f013982000-07-03 18:44:21 +000033 /* pattern code */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000034 Py_ssize_t codesize;
Fredrik Lundh6f013982000-07-03 18:44:21 +000035 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000036} PatternObject;
37
Fredrik Lundh6f013982000-07-03 18:44:21 +000038#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000039
40typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000041 PyObject_VAR_HEAD
Fredrik Lundhb0f05bd2001-07-02 16:42:49 +000042 PyObject* string; /* link to the target string (must be first) */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000043 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000044 PatternObject* pattern; /* link to the regex (pattern) object */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000045 Py_ssize_t pos, endpos; /* current target slice */
46 Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
47 Py_ssize_t groups; /* number of groups (start/end marks) */
48 Py_ssize_t mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000049} MatchObject;
50
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000051typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
52
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000053/* FIXME: <fl> shouldn't be a constant, really... */
54#define SRE_MARK_SIZE 200
55
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000056typedef struct SRE_REPEAT_T {
Neal Norwitza6d80fa2006-06-12 03:05:40 +000057 Py_ssize_t count;
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000058 SRE_CODE* pattern; /* points to REPEAT operator arguments */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000059 void* last_ptr; /* helper to check for infinite loops */
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000060 struct SRE_REPEAT_T *prev; /* points to previous repeat context */
61} SRE_REPEAT;
62
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000063typedef struct {
64 /* string pointers */
65 void* ptr; /* current position (also end of current slice) */
66 void* beginning; /* start of original string */
67 void* start; /* start of current slice */
68 void* end; /* end of original string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000069 /* attributes for the match object */
70 PyObject* string;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000071 Py_ssize_t pos, endpos;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000072 /* character size */
73 int charsize;
74 /* registers */
Neal Norwitza6d80fa2006-06-12 03:05:40 +000075 Py_ssize_t lastindex;
76 Py_ssize_t lastmark;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000077 void* mark[SRE_MARK_SIZE];
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000078 /* dynamically allocated stuff */
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000079 char* data_stack;
Neal Norwitza6d80fa2006-06-12 03:05:40 +000080 size_t data_stack_size;
81 size_t data_stack_base;
Gustavo Niemeyerad3fc442003-10-17 22:13:16 +000082 /* current repeat context */
83 SRE_REPEAT *repeat;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000084 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000085 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000086} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000087
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000088typedef struct {
89 PyObject_HEAD
90 PyObject* pattern;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000091 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000092} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000093
94#endif