code / tomo

   1 // A simple text parsing primitive:
   2 //     const char *line = "foo.txt:15";
   3 //     const char *filename; int line;
   4 //     if (strparse(line, &filename, ":", &line)) { success...}
   5 // or:
   6 //     FILE *f = ...;
   7 //     if (fparse(f, &filename, ":", &line)) { success... }
   8 
   9 #include <ctype.h>
  10 #include <gc.h>
  11 #include <stdbool.h>
  12 #include <stdint.h>
  13 #include <stdio.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16 
  17 #include "simpleparse.h"
  18 #include "util.h"
  19 
  20 static bool _match_word(const char **str, const char *target) {
  21     size_t len = strlen(target);
  22     if (strncasecmp(*str, target, len) == 0 && !isalnum((*str)[len]) && (*str)[len] != '_') {
  23         *str += len;
  24         return true;
  25     }
  26     return false;
  27 }
  28 
  29 public
  30 const char *simpleparse(const char *str, int n, parse_type_e types[n], void *destinations[n]) {
  31     for (int i = 0; i < n; i++) {
  32         switch (types[i]) {
  33         case PARSE_SOME_OF: {
  34             if (destinations[i]) str += strspn(str, (char *)destinations[i]);
  35             break;
  36         }
  37         case PARSE_LITERAL: {
  38             const char *target = (const char *)destinations[i];
  39             if (target) {
  40                 if (strncmp(str, target, strlen(target)) != 0) return str;
  41                 str += strlen(target);
  42             }
  43             break;
  44         }
  45         case PARSE_STRING: {
  46             size_t len;
  47             static const char matching_pair[256] = {[(int)'('] = ')', [(int)'{'] = '}',   [(int)'['] = ']',
  48                                                     [(int)'"'] = '"', [(int)'\''] = '\'', [(int)'`'] = '`',
  49                                                     [(int)'<'] = '>'};
  50             if (i > 0 && i + 1 < n && types[i - 1] == PARSE_LITERAL && types[i + 1] == PARSE_LITERAL
  51                 && destinations[i - 1] && destinations[i + 1] && strlen((char *)destinations[i - 1]) == 1
  52                 && strlen((char *)destinations[i + 1]) == 1
  53                 && *(char *)destinations[i + 1] == matching_pair[(int)*(char *)destinations[i - 1]]) {
  54                 len = 0;
  55                 char special_characters[4] = {'\\', *(char *)destinations[i - 1], *(char *)destinations[i + 1], 0};
  56                 for (int depth = 1; depth > 0;) {
  57                     len += strcspn(str + len, special_characters);
  58                     if (str[len] == '\0') {
  59                         return str;
  60                     } else if (str[len] == '\\'
  61                                && (special_characters[1] == '"' || special_characters[1] == '\''
  62                                    || special_characters[1] == '`')) {
  63                         if (str[len + 1] == '\0') return str;
  64                         len += 2;
  65                     } else if (str[len] == special_characters[2]) { // Check for closing quotes before opening quotes
  66                         depth -= 1;
  67                         if (depth > 0) len += 1;
  68                     } else if (str[len] == special_characters[1]) {
  69                         depth += 1;
  70                         if (depth > 999999) return str;
  71                         len += 1;
  72                     }
  73                 }
  74             } else if (i + 1 < n && types[i + 1] == PARSE_LITERAL) {
  75                 const char *terminator = (const char *)destinations[i + 1];
  76                 if (terminator) {
  77                     const char *end = strstr(str, terminator);
  78                     if (!end) return str;
  79                     len = (size_t)((ptrdiff_t)end - (ptrdiff_t)str);
  80                 } else {
  81                     len = strlen(str);
  82                 }
  83             } else if (i + 1 < n && types[i + 1] == PARSE_SOME_OF) {
  84                 len = destinations[i + 1] ? strcspn(str, (char *)destinations[i + 1]) : strlen(str);
  85                 ;
  86             } else {
  87                 len = strlen(str);
  88             }
  89             if (destinations[i]) {
  90                 char *matched = GC_MALLOC_ATOMIC(len + 1);
  91                 memcpy(matched, str, len);
  92                 matched[len] = '\0';
  93                 *(const char **)destinations[i] = matched;
  94             }
  95             str += len;
  96             break;
  97         }
  98         case PARSE_DOUBLE: {
  99             char *end = NULL;
 100             double val = strtod(str, &end);
 101             if (end == str) return str;
 102             if (destinations[i]) *(double *)destinations[i] = val;
 103             str = end;
 104             break;
 105         }
 106         case PARSE_LONG: {
 107             char *end = NULL;
 108             long val = strtol(str, &end, 10);
 109             if (end == str) return str;
 110             if (destinations[i]) *(long *)destinations[i] = val;
 111             str = end;
 112             break;
 113         }
 114         case PARSE_BOOL: {
 115             if (_match_word(&str, "true") || _match_word(&str, "yes") || _match_word(&str, "on")
 116                 || _match_word(&str, "1")) {
 117                 if (destinations[i]) *(bool *)destinations[i] = true;
 118             } else if (_match_word(&str, "false") || _match_word(&str, "no") || _match_word(&str, "off")
 119                        || _match_word(&str, "0")) {
 120                 if (destinations[i]) *(bool *)destinations[i] = false;
 121             } else {
 122                 return str;
 123             }
 124             break;
 125         }
 126         default: break;
 127         }
 128     }
 129     return NULL;
 130 }