blob: a62b9171d99c3e96146822b0ca59f3623967a992 [file] [log] [blame]
Guido van Rossumb700df92000-03-31 14:59:30 +00001/*
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +00002 *
Guido van Rossumb700df92000-03-31 14:59:30 +00003 * Secret Labs' Regular Expression Engine
Guido van Rossumb700df92000-03-31 14:59:30 +00004 *
Fredrik Lundh80946112000-06-29 18:03:25 +00005 * regular expression matching engine
Guido van Rossumb700df92000-03-31 14:59:30 +00006 *
7 * Copyright (c) 1997-2000 by Secret Labs AB. All rights reserved.
8 *
9 * See the _sre.c file for information on usage and redistribution.
10 */
11
12#ifndef SRE_INCLUDED
13#define SRE_INCLUDED
14
15#include "sre_constants.h"
16
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000017/* size of a code word (must be unsigned short or larger) */
18#define SRE_CODE unsigned short
19
Guido van Rossumb700df92000-03-31 14:59:30 +000020typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000021 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000022 int groups;
23 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 */
Fredrik Lundh6f013982000-07-03 18:44:21 +000028 /* pattern code */
29 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000030} PatternObject;
31
Fredrik Lundh6f013982000-07-03 18:44:21 +000032#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000033
34typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000035 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000036 PyObject* string; /* link to the target string */
Fredrik Lundh8a3ebf82000-07-23 21:46:17 +000037 PyObject* regs; /* cached list of matching spans */
Guido van Rossumb700df92000-03-31 14:59:30 +000038 PatternObject* pattern; /* link to the regex (pattern) object */
Fredrik Lundh6f013982000-07-03 18:44:21 +000039 int pos, endpos; /* current target slice */
40 int lastindex; /* last index marker seen by the engine (-1 if none) */
Guido van Rossumb700df92000-03-31 14:59:30 +000041 int groups; /* number of groups (start/end marks) */
Fredrik Lundh6f013982000-07-03 18:44:21 +000042 int mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000043} MatchObject;
44
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000045typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
46
47typedef struct {
48 /* stack elements */
49 SRE_CODE* pattern;
50 void* ptr;
Fredrik Lundh72b82ba2000-07-03 21:31:48 +000051 int mark;
52 void* mark0;
53 void* mark1;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000054} SRE_STACK;
55
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000056/* FIXME: <fl> shouldn't be a constant, really... */
57#define SRE_MARK_SIZE 200
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];
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000074 /* backtracking stack */
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000075 SRE_STACK* stack;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000076 int stacksize;
77 int stackbase;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000078 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000079 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000080} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000081
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000082typedef struct {
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000083 /* scanner (internal helper object) */
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000084 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