blob: 1c4bf687283f38b005ebacd39807deb3435f6c33 [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) */
17#define SRE_CODE unsigned short
18
Guido van Rossumb700df92000-03-31 14:59:30 +000019typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000020 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000021 int groups;
22 PyObject* groupindex;
Fredrik Lundhc2301732000-07-02 22:25:39 +000023 PyObject* indexgroup;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000024 /* compatibility */
25 PyObject* pattern; /* pattern source (or None) */
26 int flags; /* flags used when compiling pattern source */
Fredrik Lundh6f013982000-07-03 18:44:21 +000027 /* pattern code */
28 SRE_CODE code[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000029} PatternObject;
30
Fredrik Lundh6f013982000-07-03 18:44:21 +000031#define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
Guido van Rossumb700df92000-03-31 14:59:30 +000032
33typedef struct {
Fredrik Lundh6f013982000-07-03 18:44:21 +000034 PyObject_VAR_HEAD
Guido van Rossumb700df92000-03-31 14:59:30 +000035 PyObject* string; /* link to the target string */
36 PatternObject* pattern; /* link to the regex (pattern) object */
Fredrik Lundh6f013982000-07-03 18:44:21 +000037 int pos, endpos; /* current target slice */
38 int lastindex; /* last index marker seen by the engine (-1 if none) */
Guido van Rossumb700df92000-03-31 14:59:30 +000039 int groups; /* number of groups (start/end marks) */
Fredrik Lundh6f013982000-07-03 18:44:21 +000040 int mark[1];
Guido van Rossumb700df92000-03-31 14:59:30 +000041} MatchObject;
42
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000043typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
44
45typedef struct {
46 /* stack elements */
47 SRE_CODE* pattern;
48 void* ptr;
Fredrik Lundh72b82ba2000-07-03 21:31:48 +000049 int mark;
50 void* mark0;
51 void* mark1;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000052} SRE_STACK;
53
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000054/* FIXME: <fl> shouldn't be a constant, really... */
55#define SRE_MARK_SIZE 200
56
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000057typedef struct {
58 /* string pointers */
59 void* ptr; /* current position (also end of current slice) */
60 void* beginning; /* start of original string */
61 void* start; /* start of current slice */
62 void* end; /* end of original string */
63 /* character size */
64 int charsize;
65 /* registers */
Fredrik Lundh6f013982000-07-03 18:44:21 +000066 int lastindex;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000067 int lastmark;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000068 void* mark[SRE_MARK_SIZE];
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000069 /* backtracking stack */
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000070 SRE_STACK* stack;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000071 int stacksize;
72 int stackbase;
Fredrik Lundh102f3ad2000-06-29 08:55:54 +000073 /* hooks */
Fredrik Lundhb389df32000-06-29 12:48:37 +000074 SRE_TOLOWER_HOOK lower;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000075} SRE_STATE;
Guido van Rossumb700df92000-03-31 14:59:30 +000076
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000077typedef struct {
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000078 /* scanner (internal helper object) */
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000079 PyObject_HEAD
80 PyObject* pattern;
81 PyObject* string;
82 SRE_STATE state;
Fredrik Lundhbe2211e2000-06-29 16:57:40 +000083} ScannerObject;
Jeremy Hyltonb1aa1952000-06-01 17:39:12 +000084
85#endif