From cb34e117a397a6d27734cf3564001bfec97ec0ea Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@rambler-co.ru>
Date: Tue, 22 Feb 2011 17:36:34 +0300
Subject: [PATCH] * Add options section into config to organize options * Add
 one_shot option for avoiding inserting several results from one rule Fix
 parsing of headers with \t symbol [1]

Reported by: Victor Ustugov [1]
---
 rspamd.xml.sample | 33 +++++++++++++----
 src/cfg_file.h    |  1 +
 src/cfg_xml.c     | 93 +++++++++++++++++++++++++++++++++++++++++++++++
 src/cfg_xml.h     |  1 +
 src/filter.c      |  2 +-
 src/message.c     |  1 +
 6 files changed, 123 insertions(+), 8 deletions(-)

diff --git a/rspamd.xml.sample b/rspamd.xml.sample
index 1048da2ca..72fd663f0 100644
--- a/rspamd.xml.sample
+++ b/rspamd.xml.sample
@@ -1,13 +1,32 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <rspamd>
-<!-- Main section -->
-<tempdir>/tmp</tempdir>
-<pidfile>/var/run/rspamd/rspamd.pid</pidfile>
-<filters>regexp,surbl,chartable,fuzzy_check,spf</filters>
-<statfile_pool_size>262144000</statfile_pool_size>
-<raw_mode>yes</raw_mode>
+<!-- Global section -->
 <lua src="@ETC_PREFIX@/rspamd/lua/rspamd.lua" />
-<!-- End of main section -->
+
+<!-- Options -->
+<options>
+  <!-- Temporary directory -->
+  <tempdir>/tmp</tempdir>
+  <!-- Path to pid file -->
+  <pidfile>/var/run/rspamd/rspamd.pid</pidfile>
+  <!-- Turned on C filters -->
+  <filters>regexp,surbl,chartable,fuzzy_check,spf</filters>
+  <!-- Maximum size of statistics mapped in memory -->
+  <statfile_pool_size>250M</statfile_pool_size>
+  <!-- Raw mode is non-utf mode. In utf mode all messages are converted to utf8 (if possible) -->
+  <raw_mode>yes</raw_mode>
+  <!-- Check text attachements as ordinary text parts -->
+  <check_attachements>no</check_attachements>
+  <!-- If a rule has been met several times do not add additional score -->
+  <one_shot>no</one_shot>
+  <!-- DNS requests global timeout -->
+  <dns_timeout>1s</dns_timeout>
+  <!-- DNS retransmits count -->
+  <dns_retransmits>5</dns_retransmits>
+  <!-- File for saving settings of symbols cache -->
+  <cache_file>/var/run/rspamd/symbols.cache</cache_file>
+</options>
+<!-- End of options section -->
 
 <!-- Logging section -->
 <logging>
diff --git a/src/cfg_file.h b/src/cfg_file.h
index 09eea3976..3a9a7d12e 100644
--- a/src/cfg_file.h
+++ b/src/cfg_file.h
@@ -253,6 +253,7 @@ struct config_file {
 	gboolean no_fork;								/**< if 1 do not call daemon()							*/
 	gboolean config_test;							/**< if TRUE do only config file test					*/
 	gboolean raw_mode;								/**< work in raw mode instead of utf one				*/
+	gboolean one_shot_mode;							/**< rules add only one symbol							*/
 	gboolean check_text_attachements;				/**< check text attachements as text					*/
 	gboolean convert_config;						/**< convert config to XML format						*/
 
diff --git a/src/cfg_xml.c b/src/cfg_xml.c
index 885af3a57..06cdab069 100644
--- a/src/cfg_xml.c
+++ b/src/cfg_xml.c
@@ -61,6 +61,7 @@
 
 enum xml_config_section {
 	XML_SECTION_MAIN,
+	XML_SECTION_OPTIONS,
     XML_SECTION_LOGGING, 
 	XML_SECTION_WORKER,
 	XML_SECTION_METRIC,
@@ -187,6 +188,83 @@ static struct xml_parser_rule grammar[] = {
 		},
 		NULL_DEF_ATTR
 	},
+	{ XML_SECTION_OPTIONS, {
+			{
+				"statfile_pool_size",
+				xml_handle_size,
+				G_STRUCT_OFFSET (struct config_file, max_statfile_size),
+				NULL
+			},
+			{
+				"user_settings",
+				handle_user_settings,
+				0,
+				NULL
+			},
+			{
+				"domain_settings",
+				handle_domain_settings,
+				0,
+				NULL
+			},
+			{
+				"cache_file",
+				xml_handle_string,
+				G_STRUCT_OFFSET (struct config_file, cache_filename),
+				NULL
+			},
+			{
+				"dns_timeout",
+				xml_handle_seconds,
+				G_STRUCT_OFFSET (struct config_file, dns_timeout),
+				NULL
+			},
+			{
+				"dns_retransmits",
+				xml_handle_uint32,
+				G_STRUCT_OFFSET (struct config_file, dns_retransmits),
+				NULL
+			},
+			{
+				"raw_mode",
+				xml_handle_boolean,
+				G_STRUCT_OFFSET (struct config_file, raw_mode),
+				NULL
+			},
+			{
+				"one_shot",
+				xml_handle_boolean,
+				G_STRUCT_OFFSET (struct config_file, one_shot_mode),
+				NULL
+			},
+			{
+				"check_attachements",
+				xml_handle_boolean,
+				G_STRUCT_OFFSET (struct config_file, check_text_attachements),
+				NULL
+			},
+			{
+				"tempdir",
+				xml_handle_string,
+				G_STRUCT_OFFSET (struct config_file, temp_dir),
+				NULL
+			},
+			{
+				"pidfile",
+				xml_handle_string,
+				G_STRUCT_OFFSET (struct config_file, pid_file),
+				NULL
+			},
+			{
+				"filters",
+				xml_handle_string,
+				G_STRUCT_OFFSET (struct config_file, filters_str),
+				NULL
+			},
+			NULL_ATTR
+		},
+		NULL_DEF_ATTR
+	},
 	{ XML_SECTION_LOGGING, {
 			{
 				"type",
@@ -465,6 +543,8 @@ xml_state_to_string (struct rspamd_xml_userdata *ud)
 			return "read param";
 		case XML_READ_MODULE:
 			return "read module section";
+		case XML_READ_OPTIONS:
+			return "read options section";
 		case XML_READ_MODULES:
 			return "read modules section";
 		case XML_READ_CLASSIFIER:
@@ -1441,6 +1521,9 @@ rspamd_xml_start_element (GMarkupParseContext *context, const gchar *element_nam
 			else if (g_ascii_strcasecmp (element_name, "modules") == 0) {
 				ud->state = XML_READ_MODULES;	
 			}
+			else if (g_ascii_strcasecmp (element_name, "options") == 0) {
+				ud->state = XML_READ_OPTIONS;
+			}
 			else if (g_ascii_strcasecmp (element_name, "logging") == 0) {
 				ud->state = XML_READ_LOGGING;	
 			}
@@ -1519,6 +1602,7 @@ rspamd_xml_start_element (GMarkupParseContext *context, const gchar *element_nam
 		case XML_READ_WORKER:
 		case XML_READ_LOGGING:
 		case XML_READ_VIEW:
+		case XML_READ_OPTIONS:
 			rspamd_strlcpy (ud->section_name, element_name, sizeof (ud->section_name));
 			/* Save attributes */
 			ud->cur_attrs = process_attrs (ud->cfg, attribute_names, attribute_values);
@@ -1618,6 +1702,9 @@ rspamd_xml_end_element (GMarkupParseContext	*context, const gchar *element_name,
 		case XML_READ_MODULES:
 			CHECK_TAG ("modules", FALSE);
 			break;
+		case XML_READ_OPTIONS:
+			CHECK_TAG ("options", FALSE);
+			break;
 		case XML_READ_METRIC:
 			CHECK_TAG ("metric", FALSE);
 			if (res) {
@@ -1703,6 +1790,12 @@ rspamd_xml_text (GMarkupParseContext *context, const gchar *text, gsize text_len
 				ud->state = XML_ERROR;
 			}
 			break;
+		case XML_READ_OPTIONS:
+			if (!call_param_handler (ud, ud->section_name, val, cfg, XML_SECTION_OPTIONS)) {
+				*error = g_error_new (xml_error_quark (), XML_EXTRA_ELEMENT, "cannot parse tag '%s' data: %s", ud->section_name, val);
+				ud->state = XML_ERROR;
+			}
+			break;
 		case XML_READ_CLASSIFIER:
 			if (!call_param_handler (ud, ud->section_name, val, ud->section_pointer, XML_SECTION_CLASSIFIER)) {
 				*error = g_error_new (xml_error_quark (), XML_EXTRA_ELEMENT, "cannot parse tag '%s' data: %s", ud->section_name, val);
diff --git a/src/cfg_xml.h b/src/cfg_xml.h
index 2abdf650b..cd6f2eff4 100644
--- a/src/cfg_xml.h
+++ b/src/cfg_xml.h
@@ -23,6 +23,7 @@ enum xml_read_state {
 	XML_READ_WORKER,
 	XML_READ_VIEW,
 	XML_READ_LOGGING,
+	XML_READ_OPTIONS,
 	XML_READ_VALUE,
 	XML_SKIP_ELEMENTS,
 	XML_ERROR,
diff --git a/src/filter.c b/src/filter.c
index 4b8585a7c..d41c85ff0 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -171,7 +171,7 @@ insert_result_common (struct worker_task *task, const gchar *symbol, double flag
 void
 insert_result (struct worker_task *task, const gchar *symbol, double flag, GList * opts)
 {
-	insert_result_common (task, symbol, flag, opts, FALSE);
+	insert_result_common (task, symbol, flag, opts, task->cfg->one_shot_mode);
 }
 
 /* Insert result as a single option */
diff --git a/src/message.c b/src/message.c
index bf2896009..38f9a663e 100644
--- a/src/message.c
+++ b/src/message.c
@@ -513,6 +513,7 @@ process_raw_headers (struct worker_task *task)
 			/* We got header's name, so skip any \t or spaces */
 			if (*p == '\t') {
 				new->tab_separated = TRUE;
+				p ++;
 			}
 			else if (*p == ' '){
 				p ++;
-- 
2.39.5