Only compile regex's once

src/common.c

@@ -1417,45 +1417,6 @@ set_boolean_if_valid (bool *out, const char *s)
 	return true;
 }
 
-static void
-regerror_to_str (int code, const regex_t *preg, struct str *out)
-{
-	size_t required = regerror (code, preg, NULL, 0);
-	str_ensure_space (out, required);
-	out->len += regerror (code, preg,
-		out->str + out->len, out->alloc - out->len) - 1;
-}
-
-static size_t regex_error_domain_tag;
-#define REGEX_ERROR (error_resolve_domain (&regex_error_domain_tag))
-
-enum
-{
-	REGEX_ERROR_COMPILATION_FAILED
-};
-
-static bool
-regex_match (const char *regex, const char *s, struct error **e)
-{
-	regex_t re;
-	int err = regcomp (&re, regex, REG_EXTENDED | REG_NOSUB);
-	if (err)
-	{
-		struct str desc;
-
-		str_init (&desc);
-		regerror_to_str (err, &re, &desc);
-		error_set (e, REGEX_ERROR, REGEX_ERROR_COMPILATION_FAILED,
-			"failed to compile regular expression: %s", desc.str);
-		str_free (&desc);
-		return false;
-	}
-
-	bool result = regexec (&re, s, 0, NULL, 0) != REG_NOMATCH;
-	regfree (&re);
-	return result;
-}
-
 static bool
 read_line (FILE *fp, struct str *s)
 {
@@ -1512,6 +1473,73 @@ xssl_get_error (SSL *ssl, int result, const char **error_info)
 	}
 }
 
+// --- Regular expressions -----------------------------------------------------
+
+static size_t regex_error_domain_tag;
+#define REGEX_ERROR (error_resolve_domain (&regex_error_domain_tag))
+
+enum
+{
+	REGEX_ERROR_COMPILATION_FAILED
+};
+
+static regex_t *
+regex_compile (const char *regex, int flags, struct error **e)
+{
+	regex_t *re = xmalloc (sizeof *re);
+	int err = regcomp (re, regex, flags);
+	if (!err)
+		return re;
+
+	struct str desc;
+	str_init (&desc);
+
+	size_t required = regerror (err, re, NULL, 0);
+	str_ensure_space (&desc, required);
+	desc.len += regerror (err, re,
+		desc.str + desc.len, desc.alloc - desc.len) - 1;
+
+	free (re);
+	error_set (e, REGEX_ERROR, REGEX_ERROR_COMPILATION_FAILED,
+		"%s: %s", "failed to compile regular expression", desc.str);
+	str_free (&desc);
+	return NULL;
+}
+
+static void
+regex_free (void *regex)
+{
+	regfree (regex);
+	free (regex);
+}
+
+// The cost of hashing a string is likely to be significantly smaller than that
+// of compiling the whole regular expression anew, so here is a simple cache.
+// Adding basic support for subgroups is easy: check `re_nsub' and output into
+// a `struct str_vector' (if all we want is the substrings).
+
+static void
+regex_cache_init (struct str_map *cache)
+{
+	str_map_init (cache);
+	cache->free = regex_free;
+}
+
+static bool
+regex_cache_match (struct str_map *cache, const char *regex, int flags,
+	const char *s, struct error **e)
+{
+	regex_t *re = str_map_find (cache, regex);
+	if (!re)
+	{
+		re = regex_compile (regex, flags, e);
+		if (!re)
+			return false;
+		str_map_set (cache, regex, re);
+	}
+	return regexec (re, s, 0, NULL, 0) != REG_NOMATCH;
+}
+
 // --- IRC utilities -----------------------------------------------------------
 
 struct irc_message