Add support for -accept-regex and --reject-regex.

[wget] / src / utils.c

diff --git a/src/utils.c b/src/utils.c
index 4188ced7d2ab409097b3d4a60650209255732674..55a8a8d286959dc50aa4d15d8b51a1840c7ca9e7 100644 (file)
--- a/src/utils.c
+++ b/src/utils.c
@@ -73,6 +73,11 @@ as that of the covered work.  */
 #include <signal.h>
 #include <setjmp.h>
 
+#include <regex.h>
+#ifdef HAVE_LIBPCRE
+# include <pcre.h>
+#endif
+
 #ifndef HAVE_SIGSETJMP
 /* If sigsetjmp is a macro, configure won't pick it up. */
 # ifdef sigsetjmp
@@ -917,6 +922,19 @@ acceptable (const char *s)
   return true;
 }
 
+/* Determine whether an URL is acceptable to be followed, according to
+   regex patterns to accept/reject.  */
+bool
+accept_url (const char *s)
+{
+  if (opt.acceptregex && !opt.regex_match_fun (opt.acceptregex, s))
+    return false;
+  if (opt.rejectregex && opt.regex_match_fun (opt.rejectregex, s))
+    return false;
+
+  return true;
+}
+
 /* Check if D2 is a subdirectory of D1.  E.g. if D1 is `/something', subdir_p()
    will return true if and only if D2 begins with `/something/' or is exactly
    '/something'.  */
@@ -2309,6 +2327,89 @@ base64_decode (const char *base64, void *dest)
   return q - (char *) dest;
 }
 
+#ifdef HAVE_LIBPCRE
+/* Compiles the PCRE regex. */
+void *
+compile_pcre_regex (const char *str)
+{
+  const char *errbuf;
+  int erroffset;
+  pcre *regex = pcre_compile (str, 0, &errbuf, &erroffset, 0);
+  if (! regex)
+    {
+      fprintf (stderr, _("Invalid regular expression %s, %s\n"),
+               quote (str), errbuf);
+      return false;
+    }
+  return regex;
+}
+#endif
+
+/* Compiles the POSIX regex. */
+void *
+compile_posix_regex (const char *str)
+{
+  regex_t *regex = xmalloc (sizeof (regex_t));
+  int errcode = regcomp ((regex_t *) regex, str, REG_EXTENDED | REG_NOSUB);
+  if (errcode != 0)
+    {
+      int errbuf_size = regerror (errcode, (regex_t *) regex, NULL, 0);
+      char *errbuf = xmalloc (errbuf_size);
+      errbuf_size = regerror (errcode, (regex_t *) regex, errbuf, errbuf_size);
+      fprintf (stderr, _("Invalid regular expression %s, %s\n"),
+               quote (str), errbuf);
+      xfree (errbuf);
+      return NULL;
+    }
+
+  return regex;
+}
+
+#ifdef HAVE_LIBPCRE
+#define OVECCOUNT 30
+/* Matches a PCRE regex.  */
+bool
+match_pcre_regex (const void *regex, const char *str)
+{
+  int l = strlen (str);
+  int ovector[OVECCOUNT];
+
+  int rc = pcre_exec ((pcre *) regex, 0, str, l, 0, 0, ovector, OVECCOUNT);
+  if (rc == PCRE_ERROR_NOMATCH)
+    return false;
+  else if (rc < 0)
+    {
+      logprintf (LOG_VERBOSE, _("Error while matching %s: %d\n"),
+                 quote (str), rc);
+      return false;
+    }
+  else
+    return true;
+}
+#undef OVECCOUNT
+#endif
+
+/* Matches a POSIX regex.  */
+bool
+match_posix_regex (const void *regex, const char *str)
+{
+  int rc = regexec ((regex_t *) regex, str, 0, NULL, 0);
+  if (rc == REG_NOMATCH)
+    return false;
+  else if (rc == 0)
+    return true;
+  else
+    {
+      int errbuf_size = regerror (rc, opt.acceptregex, NULL, 0);
+      char *errbuf = xmalloc (errbuf_size);
+      errbuf_size = regerror (rc, opt.acceptregex, errbuf, errbuf_size);
+      logprintf (LOG_VERBOSE, _("Error while matching %s: %d\n"),
+                 quote (str), rc);
+      xfree (errbuf);
+      return false;
+    }
+}
+
 #undef IS_ASCII
 #undef NEXT_CHAR
 \f