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 UCS4 character) */
      18  #define SRE_CODE Py_UCS4
      19  #if SIZEOF_SIZE_T > 4
      20  # define SRE_MAXREPEAT (~(SRE_CODE)0)
      21  # define SRE_MAXGROUPS ((SRE_CODE)INT32_MAX / 2)
      22  #else
      23  # define SRE_MAXREPEAT ((SRE_CODE)PY_SSIZE_T_MAX)
      24  # define SRE_MAXGROUPS ((SRE_CODE)PY_SSIZE_T_MAX / SIZEOF_VOID_P / 2)
      25  #endif
      26  
      27  typedef struct {
      28      PyObject_VAR_HEAD
      29      Py_ssize_t groups; /* must be first! */
      30      PyObject* groupindex; /* dict */
      31      PyObject* indexgroup; /* tuple */
      32      /* compatibility */
      33      PyObject* pattern; /* pattern source (or None) */
      34      int flags; /* flags used when compiling pattern source */
      35      PyObject *weakreflist; /* List of weak references */
      36      int isbytes; /* pattern type (1 - bytes, 0 - string, -1 - None) */
      37      /* pattern code */
      38      Py_ssize_t codesize;
      39      SRE_CODE code[1];
      40  } PatternObject;
      41  
      42  #define PatternObject_GetCode(o) (((PatternObject*)(o))->code)
      43  
      44  typedef struct {
      45      PyObject_VAR_HEAD
      46      PyObject* string; /* link to the target string (must be first) */
      47      PyObject* regs; /* cached list of matching spans */
      48      PatternObject* pattern; /* link to the regex (pattern) object */
      49      Py_ssize_t pos, endpos; /* current target slice */
      50      Py_ssize_t lastindex; /* last index marker seen by the engine (-1 if none) */
      51      Py_ssize_t groups; /* number of groups (start/end marks) */
      52      Py_ssize_t mark[1];
      53  } MatchObject;
      54  
      55  typedef struct {
      56      PyObject_VAR_HEAD
      57      Py_ssize_t chunks;  /* the number of group references and non-NULL literals
      58                           * self->chunks <= 2*Py_SIZE(self) + 1 */
      59      PyObject *literal;
      60      struct {
      61          Py_ssize_t index;
      62          PyObject *literal;  /* NULL if empty */
      63      } items[0];
      64  } TemplateObject;
      65  
      66  typedef struct SRE_REPEAT_T {
      67      Py_ssize_t count;
      68      const SRE_CODE* pattern; /* points to REPEAT operator arguments */
      69      const void* last_ptr; /* helper to check for infinite loops */
      70      struct SRE_REPEAT_T *prev; /* points to previous repeat context */
      71  } SRE_REPEAT;
      72  
      73  typedef struct {
      74      /* string pointers */
      75      const void* ptr; /* current position (also end of current slice) */
      76      const void* beginning; /* start of original string */
      77      const void* start; /* start of current slice */
      78      const void* end; /* end of original string */
      79      /* attributes for the match object */
      80      PyObject* string;
      81      Py_buffer buffer;
      82      Py_ssize_t pos, endpos;
      83      int isbytes;
      84      int charsize; /* character size */
      85      int match_all;
      86      int must_advance;
      87      /* marks */
      88      int lastmark;
      89      int lastindex;
      90      const void** mark;
      91      /* dynamically allocated stuff */
      92      char* data_stack;
      93      size_t data_stack_size;
      94      size_t data_stack_base;
      95      /* current repeat context */
      96      SRE_REPEAT *repeat;
      97  } SRE_STATE;
      98  
      99  typedef struct {
     100      PyObject_HEAD
     101      PyObject* pattern;
     102      SRE_STATE state;
     103      int executing;
     104  } ScannerObject;
     105  
     106  #endif