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 Py_UCS4
  20 #else
  21 #define SRE_CODE unsigned short
  22 #endif
  23
  24 typedef struct {
  25     PyObject_VAR_HEAD
  26     Py_ssize_t 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     PyObject *weakreflist; /* List of weak references */
  33         int charsize; /* pattern charsize (or -1) */
  34     /* pattern code */
  35     Py_ssize_t codesize;
  36     SRE_CODE code[1];
  37 } PatternObject;
  38
  39 #define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
  40
  41 typedef struct {
  42     PyObject_VAR_HEAD
  43     PyObject* string; /* link to the target string (must be first) */
  44     PyObject* regs; /* cached list of matching spans */
  45     PatternObject* pattern; /* link to the regex (pattern) object */
  46     Py_ssize_t pos, endpos; /* current target slice */
  47     Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
  48     Py_ssize_t groups; /* number of groups (start/end marks) */
  49     Py_ssize_t mark[1];
  50 } MatchObject;
  51
  52 typedef unsigned int (*SRE_TOLOWER_HOOK)(unsigned int ch);
  53
  54 /* FIXME: <fl> shouldn't be a constant, really... */
  55 #define SRE_MARK_SIZE 200
  56
  57 typedef struct SRE_REPEAT_T {
  58     Py_ssize_t count;
  59     SRE_CODE* pattern; /* points to REPEAT operator arguments */
  60     void* last_ptr; /* helper to check for infinite loops */
  61     struct SRE_REPEAT_T *prev; /* points to previous repeat context */
  62 } SRE_REPEAT;
  63
  64 typedef struct {
  65     /* string pointers */
  66     void* ptr; /* current position (also end of current slice) */
  67     void* beginning; /* start of original string */
  68     void* start; /* start of current slice */
  69     void* end; /* end of original string */
  70     /* attributes for the match object */
  71     PyObject* string;
  72     Py_ssize_t pos, endpos;
  73     /* character size */
  74     int charsize;
  75     /* registers */
  76     Py_ssize_t lastindex;
  77     Py_ssize_t lastmark;
  78     void* mark[SRE_MARK_SIZE];
  79     /* dynamically allocated stuff */
  80     char* data_stack;
  81     size_t data_stack_size;
  82     size_t data_stack_base;
  83     /* current repeat context */
  84     SRE_REPEAT *repeat;
  85     /* hooks */
  86     SRE_TOLOWER_HOOK lower;
  87 } SRE_STATE;
  88
  89 typedef struct {
  90     PyObject_HEAD
  91     PyObject* pattern;
  92     SRE_STATE state;
  93 } ScannerObject;
  94
  95 #endif