blob: 61a020896ab0b19119412c165953654abfbf1dca [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 *
6 * Copyright (c) 1997-2000 by Secret Labs AB. All rights reserved.
7 *
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 Lundh102f3ad2000-06-29 08:55:54 +000016/* size of a code word (must be unsigned short or larger) */
Fredrik Lundh1294ad02001-06-26 17:17:07 +000017#ifdef USE_UCS4_STORAGE
18#define SRE_CODE unsigned long
19#else
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000020#define SRE_CODE unsigned short
Fredrik Lundh1294ad02001-06-26 17:17:07 +000021#endif
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000022
Guido van Rossumb700df92000-03-31 14:59:30 +000023typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000024 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000025 int groups;
26 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000027 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000028 /* compatibility */
29 PyObject* pattern; /* pattern source (or None) */
30 int flags; /* flags used when compiling pattern source */
Fredrik Lundh6f013982000-07-03 18:44:21 +000031 /* pattern code */
32 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000033} PatternObject;
34
Fredrik Lundh6f013982000-07-03 18:44:21 +000035#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000036
37typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000038 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000039 PyObject* string; /* link to the target string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000040 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000041 PatternObject* pattern; /* link to the regex (pattern) object */
Fredrik Lundh6f013982000-07-03 18:44:21 +000042 int pos, endpos; /* current target slice */
43 int lastindex; /* last index marker seen by the engine (-1 if none) */
Guido van Rossumb700df92000-03-31 14:59:30 +000044 int groups; /* number of groups (start/end marks) */
Fredrik Lundh6f013982000-07-03 18:44:21 +000045 int mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000046} MatchObject;
47
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000048typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
49
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000050/* FIXME: <fl> shouldn't be a constant, really... */
51#define SRE_MARK_SIZE 200
52
Fredrik Lundh29c4ba92000-08-01 18:20:07 +000053typedef struct SRE_REPEAT_T {
54 int count;
55 SRE_CODE* pattern; /* points to REPEAT operator arguments */
56 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;
67 int pos, endpos;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000068 /* character size */
69 int charsize;
70 /* registers */
Fredrik Lundh6f013982000-07-03 18:44:21 +000071 int lastindex;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000072 int 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 */
75 void** mark_stack;
76 int mark_stack_size;
77 int mark_stack_base;
78 SRE_REPEAT *repeat; /* current repeat context */
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000079 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000080 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000081} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000082
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000083typedef struct {
84 PyObject_HEAD
85 PyObject* pattern;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000086 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000087} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000088
89#endif