From: cebka@mailsupport.rambler.ru <cebka@mailsupport.rambler.ru>
Date: Tue, 14 Oct 2008 01:03:54 +0000 (+0400)
Subject: * Add utility function for parsing regexp from given string (type flags and pcre... 
X-Git-Tag: 0.2.7~362
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=fd84d363d87414274dc24e77f301ab5c966ede0b;p=rspamd.git

* Add utility function for parsing regexp from given string (type flags and pcre flags)
---

diff --git a/cfg_file.h b/cfg_file.h
index 5b4e2d84b..21c8b26fa 100644
--- a/cfg_file.h
+++ b/cfg_file.h
@@ -50,6 +50,20 @@
 
 enum { VAL_UNDEF=0, VAL_TRUE, VAL_FALSE };
 
+enum rspamd_regexp_type {
+	REGEXP_NONE = 0,
+	REGEXP_HEADER,
+	REGEXP_MIME,
+	REGEXP_MESSAGE,
+	REGEXP_URL,
+};
+
+struct rspamd_regexp {
+	enum rspamd_regexp_type type;
+	char *regexp_text;
+	GRegex *regexp;
+};
+
 struct memcached_server {
 	struct upstream up;
 	struct in_addr addr;
@@ -117,6 +131,7 @@ unsigned int parse_seconds (const char *t);
 char parse_flag (const char *str);
 char* substitute_variable (struct config_file *cfg, char *str, u_char recursive);
 void post_load_config (struct config_file *cfg);
+struct rspamd_regexp* parse_regexp (memory_pool_t *pool, char *line);
 
 int yylex (void);
 int yyparse (void);
diff --git a/cfg_utils.c b/cfg_utils.c
index 6f290d2d8..3f46b0dd8 100644
--- a/cfg_utils.c
+++ b/cfg_utils.c
@@ -399,6 +399,106 @@ post_load_config (struct config_file *cfg)
 	parse_filters_str (cfg, cfg->url_filters_str, SCRIPT_URL);
 }
 
+/*
+ * Rspamd regexp utility functions
+ */
+struct rspamd_regexp*
+parse_regexp (memory_pool_t *pool, char *line)
+{
+	char *begin, *end, *p;
+	struct rspamd_regexp *result;
+	int regexp_flags = 0;
+	enum rspamd_regexp_type type = REGEXP_NONE;
+	GError *err;
+	
+	/* Find begin */
+	while (*line != '/') {
+		line ++;
+	}
+	if (*line != '\0') {
+		begin = line + 1;
+	}
+	else {
+		return NULL;
+	}
+	/* Find end */
+	end = begin;
+	while (*end && (*end != '/' || *(end - 1) == '\\')) {
+		end ++;
+	}
+	if (end == begin || *end != '/') {
+		return NULL;
+	}
+	/* Parse flags */
+	p = end + 1;
+	while (p != NULL) {
+		switch (*p) {
+			case 'i':
+				regexp_flags |= G_REGEX_CASELESS;
+				p ++;
+				break;
+			case 'm':
+				regexp_flags |= G_REGEX_MULTILINE;
+				p ++;
+				break;
+			case 's':
+				regexp_flags |= G_REGEX_DOTALL;
+				p ++;
+				break;
+			case 'x':
+				regexp_flags |= G_REGEX_EXTENDED;
+				p ++;
+				break;
+			case 'u':
+				regexp_flags |= G_REGEX_UNGREEDY;
+				p ++;
+				break;
+			case 'o':
+				regexp_flags |= G_REGEX_OPTIMIZE;
+				p ++;
+				break;
+			/* Type flags */
+			case 'H':
+				if (type != REGEXP_NONE) {
+					type = REGEXP_HEADER;
+				}
+				p ++;
+				break;
+			case 'M':
+				if (type != REGEXP_NONE) {
+					type = REGEXP_MESSAGE;
+				}
+				p ++;
+				break;
+			case 'P':
+				if (type != REGEXP_NONE) {
+					type = REGEXP_MIME;
+				}
+				p ++;
+				break;
+			case 'U':
+				if (type != REGEXP_NONE) {
+					type = REGEXP_URL;
+				}
+				p ++;
+				break;
+			/* Stop flags parsing */
+			default:
+				p = NULL;
+				break;
+		}
+	}
+
+	result = memory_pool_alloc (pool, sizeof (struct rspamd_regexp));
+	result->type = type;
+	*end = '\0';
+	result->regexp = g_regex_new (begin, regexp_flags, 0, &err);
+	result->regexp_text = memory_pool_strdup (pool, begin);
+	*end = '/';
+
+	return result;
+}
+
 /*
  * vi:ts=4
  */