text-regex.c

   1 #include <stdlib.h>
   2 #include <string.h>
   3
   4 #include "text-regex.h"
   5
   6 struct Regex {
   7         regex_t regex;
   8 };
   9
  10 Regex *text_regex_new(void) {
  11         Regex *r = calloc(1, sizeof(Regex));
  12         if (!r)
  13                 return NULL;
  14         regcomp(&r->regex, "\0\0", 0); /* this should not match anything */
  15         return r;
  16 }
  17
  18 int text_regex_compile(Regex *regex, const char *string, int cflags) {
  19         int r = regcomp(&regex->regex, string, cflags);
  20         if (r)
  21                 regcomp(&regex->regex, "\0\0", 0);
  22         return r;
  23 }
  24
  25 size_t text_regex_nsub(Regex *r) {
  26         if (!r)
  27                 return 0;
  28         return r->regex.re_nsub;
  29 }
  30
  31 void text_regex_free(Regex *r) {
  32         if (!r)
  33                 return;
  34         regfree(&r->regex);
  35         free(r);
  36 }
  37
  38 int text_regex_match(Regex *r, const char *data, int eflags) {
  39         return regexec(&r->regex, data, 0, NULL, eflags);
  40 }
  41
  42 int text_search_range_forward(Text *txt, size_t pos, size_t len, Regex *r, size_t nmatch, RegexMatch pmatch[], int eflags) {
  43         char *buf = text_bytes_alloc0(txt, pos, len);
  44         if (!buf)
  45                 return REG_NOMATCH;
  46         char *cur = buf, *end = buf + len;
  47         int ret = REG_NOMATCH;
  48         regmatch_t match[MAX_REGEX_SUB];
  49         for (size_t junk = len; len > 0; len -= junk, pos += junk) {
  50                 ret = regexec(&r->regex, cur, nmatch, match, eflags);
  51                 if (!ret) {
  52                         for (size_t i = 0; i < nmatch; i++) {
  53                                 pmatch[i].start = match[i].rm_so == -1 ? EPOS : pos + match[i].rm_so;
  54                                 pmatch[i].end = match[i].rm_eo == -1 ? EPOS : pos + match[i].rm_eo;
  55                         }
  56                         break;
  57                 }
  58                 char *next = memchr(cur, 0, len);
  59                 if (!next)
  60                         break;
  61                 while (!*next && next != end)
  62                         next++;
  63                 junk = next - cur;
  64                 cur = next;
  65         }
  66         free(buf);
  67         return ret;
  68 }
  69
  70 int text_search_range_backward(Text *txt, size_t pos, size_t len, Regex *r, size_t nmatch, RegexMatch pmatch[], int eflags) {
  71         char *buf = text_bytes_alloc0(txt, pos, len);
  72         if (!buf)
  73                 return REG_NOMATCH;
  74         char *cur = buf, *end = buf + len;
  75         int ret = REG_NOMATCH;
  76         regmatch_t match[MAX_REGEX_SUB];
  77         for (size_t junk = len; len > 0; len -= junk, pos += junk) {
  78                 char *next;
  79                 if (!regexec(&r->regex, cur, nmatch, match, eflags)) {
  80                         ret = 0;
  81                         for (size_t i = 0; i < nmatch; i++) {
  82                                 pmatch[i].start = match[i].rm_so == -1 ? EPOS : pos + match[i].rm_so;
  83                                 pmatch[i].end = match[i].rm_eo == -1 ? EPOS : pos + match[i].rm_eo;
  84                         }
  85
  86                         if (match[0].rm_so == 0 && match[0].rm_eo == 0) {
  87                                 /* empty match at the beginning of cur, advance to next line */
  88                                 next = strchr(cur, '\n');
  89                                 if (!next)
  90                                         break;
  91                                 next++;
  92                         } else {
  93                                 next = cur + match[0].rm_eo;
  94                         }
  95                 } else {
  96                         next = memchr(cur, 0, len);
  97                         if (!next)
  98                                 break;
  99                         while (!*next && next != end)
 100                                 next++;
 101                 }
 102                 junk = next - cur;
 103                 cur = next;
 104                 if (cur[-1] == '\n')
 105                         eflags &= ~REG_NOTBOL;
 106                 else
 107                         eflags |= REG_NOTBOL;
 108         }
 109         free(buf);
 110         return ret;
 111 }