Modules/sre.h

   1 /*
   2  * Secret Labs' Regular Expression Engine
   3  *
   4  * regular expression matching engine
   5  *
   6  * Copyright (c) 1997-2001 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
  16 /* 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
  19 #define SRE_CODE unsigned long
  20 #else
  21 #define SRE_CODE unsigned short
  22 #endif
  23
  24 typedef struct {
  25     PyObject_VAR_HEAD
  26     int groups; /* must be first! */
  27     PyObject* groupindex;
  28     PyObject* indexgroup;
  29     /* compatibility */
  30     PyObject* pattern; /* pattern source (or None) */
  31     int flags; /* flags used when compiling pattern source */
  32     /* pattern code */
  33     int codesize;
  34     SRE_CODE code[1];
  35 } PatternObject;
  36
  37 #define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
  38
  39 typedef struct {
  40     PyObject_VAR_HEAD
  41     PyObject* string; /* link to the target string (must be first) */
  42     PyObject* regs; /* cached list of matching spans */
  43     PatternObject* pattern; /* link to the regex (pattern) object */
  44     int pos, endpos; /* current target slice */
  45     int lastindex; /* last index marker seen by the engine (-1 if none) */
  46     int groups; /* number of groups (start/end marks) */
  47     int mark[1];
  48 } MatchObject;
  49
  50 typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
  51
  52 /* FIXME: <fl> shouldn't be a constant, really... */
  53 #define SRE_MARK_SIZE 200
  54
  55 typedef struct SRE_REPEAT_T {
  56     int count;
  57     SRE_CODE* pattern; /* points to REPEAT operator arguments */
  58     struct SRE_REPEAT_T *prev; /* points to previous repeat context */
  59 } SRE_REPEAT;
  60
  61 typedef struct {
  62     /* string pointers */
  63     void* ptr; /* current position (also end of current slice) */
  64     void* beginning; /* start of original string */
  65     void* start; /* start of current slice */
  66     void* end; /* end of original string */
  67     /* attributes for the match object */
  68     PyObject* string;
  69     int pos, endpos;
  70     /* character size */
  71     int charsize;
  72     /* registers */
  73     int lastindex;
  74     int lastmark;
  75     void* mark[SRE_MARK_SIZE];
  76     /* dynamically allocated stuff */
  77     void** mark_stack;
  78     int mark_stack_size;
  79     int mark_stack_base;
  80     SRE_REPEAT *repeat; /* current repeat context */
  81     /* hooks */
  82     SRE_TOLOWER_HOOK lower;
  83 } SRE_STATE;
  84
  85 typedef struct {
  86     PyObject_HEAD
  87     PyObject* pattern;
  88     SRE_STATE state;
  89 } ScannerObject;
  90
  91 #endif