git.oblomov.eu Git - ohcount/blob - ext/ohcount_native/compiled_state.c

   1 /*
   2  *  compiled_state.c
   3  *  Ohcount
   4  *
   5  *  Created by Jason Allen on 6/23/06.
   6  *  Copyright 2006 Ohloh. All rights reserved.
   7  *
   8  */
   9
  10 #include "common.h"
  11
  12
  13 Transition *compiled_state_get_transition(CompiledState *compiled_state, int nth_transition) {
  14         return compiled_state->transition_map[nth_transition];
  15 }
  16
  17 /*
  18  * compiled_state_initialize
  19  *
  20  * a 'compiled_state' helps the parser walk the DFA faster by precomputing the regular expression
  21  * for each state. Basically any transition __from__ a state is added to the regex term. We also
  22  * ensure the "\n" transition (to force end-of-line processing).
  23  *
  24  */
  25 void compiled_state_initialize(CompiledState *compiled_state, State *state, Transition *transitions[]) {
  26         compiled_state->state = state;
  27
  28         Transition **pp_t_cur = transitions;
  29         Transition **pp_t_map = compiled_state->transition_map;
  30         char *regex_cur = compiled_state->regex;
  31         bool seen_newline = false;
  32
  33         for (; (*pp_t_cur) != NULL; pp_t_cur++) {
  34                 Transition *t_cur = *pp_t_cur;
  35                 if (t_cur->from_state == state) {
  36
  37                         // hey! this is us! add this transition to our regex
  38                         *regex_cur++ = '(';
  39                         if (!seen_newline) {
  40                                 seen_newline = *t_cur->regex == '\n';
  41                         }
  42                         char *regex_src = t_cur->regex;
  43                         while ((*regex_cur++ = *regex_src++)) {}
  44                         regex_cur--; // dont want the nil!
  45                         *regex_cur++ = ')';
  46                         *regex_cur++ = '|';
  47
  48                         // add this transition to our map
  49                         (*pp_t_map) = t_cur;
  50                         pp_t_map++;
  51                 }
  52         }
  53
  54         // add newline (only if it's not already there...)
  55         if (!seen_newline) {
  56                 *regex_cur++ = '(';
  57                 *regex_cur++ = '\n';
  58                 *regex_cur++ = ')';
  59                 *regex_cur++ = '|'; // just to stay similar to loop pattern
  60                 (*pp_t_map) = NULL; // make sure we are pointing to a NULL transition
  61         }
  62
  63         // terminate the string
  64         *--regex_cur = 0;
  65
  66         // some sanity checks
  67 #ifndef NDEBUG
  68         int transition_count = pp_t_map - compiled_state->transition_map;
  69         if (transition_count >= MAX_TRANSITIONS) {
  70                 log("[ohcount] - ASSERT FAILED: transition_count > MAX (%d)\n", transition_count);
  71         }
  72 #endif
  73
  74         const char *error;
  75         int erroffset;
  76         compiled_state->pcre = pcre_compile(compiled_state->regex,          /* the pattern */
  77                                                                                                                                                         0,                              /* default options */
  78                                                                                                                                                         &error,                         /* for error message */
  79                                                                                                                                                         &erroffset,                     /* for error offset */
  80                                                                                                                                                         NULL);                          /* use default character tables */
  81
  82         const char *errptr;
  83         /* since we're likely going to reuse this often, go ahead and study it */
  84         /* we dont care about errors - it will return NULL which is fine */
  85         compiled_state->pcre_extra = pcre_study(compiled_state->pcre,  /* result of pcre_compile() */
  86                                                                                                                                                                         0,                     /* no options exist */
  87                                                                                                                                                                         &errptr);                 /* set to NULL or points to a message */
  88 }
  89