From 24ff611b744d26d7f95f9dd2f46292d774b0ab16 Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Sat, 26 Oct 2013 22:08:05 +0100
Subject: [PATCH] Allow xml->ucl config conversion.

Now it is possible to read UCL configs directly and convert legacy XML
to UCL by --convert-config option.

Default config is now rspamd.conf, not rspamd.xml.
---
 src/cfg_file.h  |  2 +-
 src/cfg_utils.c | 72 ++++++++++++++++++++++++++++++++++++++++---------
 src/main.c      |  4 ++-
 src/main.h      |  2 +-
 4 files changed, 65 insertions(+), 15 deletions(-)

diff --git a/src/cfg_file.h b/src/cfg_file.h
index 875ce3373..659d6981e 100644
--- a/src/cfg_file.h
+++ b/src/cfg_file.h
@@ -532,7 +532,7 @@ gboolean parse_normalizer (struct config_file *cfg, struct statfile *st, const g
 /*
  * Read XML configuration file
  */
-gboolean read_xml_config (struct config_file *cfg, const gchar *filename);
+gboolean read_rspamd_config (struct config_file *cfg, const gchar *filename, const gchar *convert_to);
 
 /*
  * Register symbols of classifiers inside metrics
diff --git a/src/cfg_utils.c b/src/cfg_utils.c
index d41f364ba..0e698cdcd 100644
--- a/src/cfg_utils.c
+++ b/src/cfg_utils.c
@@ -955,18 +955,31 @@ static GMarkupParser xml_parser = {
 	.error = rspamd_xml_error,
 };
 
+static const char*
+get_filename_extension (const char *filename)
+{
+	const char *dot_pos = strrchr (filename, '.');
+
+	if (dot_pos != NULL) {
+		return (dot_pos + 1);
+	}
+
+	return NULL;
+}
+
 gboolean
-read_xml_config (struct config_file *cfg, const gchar *filename)
+read_rspamd_config (struct config_file *cfg, const gchar *filename, const gchar *convert_to)
 {
 	struct stat                     st;
 	gint                            fd;
 	gchar                          *data, *rcl;
+	const gchar                    *ext;
 	GMarkupParseContext            *ctx;
 	GError                         *err = NULL;
 	struct rspamd_rcl_section     *top;
-	gboolean res;
-
+	gboolean res, is_xml = FALSE;
 	struct rspamd_xml_userdata ud;
+	struct ucl_parser *parser;
 
 	if (stat (filename, &st) == -1) {
 		msg_err ("cannot stat %s: %s", filename, strerror (errno));
@@ -985,23 +998,58 @@ read_xml_config (struct config_file *cfg, const gchar *filename)
 	}
 	close (fd);
 	
-	/* Prepare xml parser */
-	memset (&ud, 0, sizeof (ud));
-	ud.cfg = cfg;
-	ud.state = 0;
-	ctx = g_markup_parse_context_new (&xml_parser, G_MARKUP_TREAT_CDATA_AS_TEXT, &ud, NULL);
-	res = g_markup_parse_context_parse (ctx, data, st.st_size, &err);
+	ext = get_filename_extension (filename);
+	if (ext != NULL && strcmp (ext, "xml") == 0) {
+		is_xml = TRUE;
+	}
 
-	munmap (data, st.st_size);
+	if (is_xml) {
+		/* Prepare xml parser */
+		memset (&ud, 0, sizeof (ud));
+		ud.cfg = cfg;
+		ud.state = 0;
+		ctx = g_markup_parse_context_new (&xml_parser, G_MARKUP_TREAT_CDATA_AS_TEXT, &ud, NULL);
+		res = g_markup_parse_context_parse (ctx, data, st.st_size, &err);
+
+		munmap (data, st.st_size);
+
+		top = rspamd_rcl_config_init ();
+
+		err = NULL;
+	}
+	else {
+		parser = ucl_parser_new (UCL_PARSER_KEY_LOWERCASE);
+		if (!ucl_parser_add_chunk (parser, data, st.st_size)) {
+			msg_err ("ucl parser error: %s", ucl_parser_get_error (parser));
+			munmap (data, st.st_size);
+			return FALSE;
+		}
+		munmap (data, st.st_size);
+		cfg->rcl_obj = ucl_parser_get_object (parser);
+	}
 
-	top = rspamd_rcl_config_init ();
 
-	err = NULL;
 	if (!res || !rspamd_read_rcl_config (top, cfg, cfg->rcl_obj, &err)) {
 		msg_err ("rcl parse error: %s", err->message);
 		return FALSE;
 	}
 
+	if (is_xml && convert_to != NULL) {
+		/* Convert XML config to UCL */
+		rcl = ucl_object_emit (cfg->rcl_obj, UCL_EMIT_CONFIG);
+		if (rcl != NULL) {
+			fd = open (convert_to, O_CREAT|O_TRUNC|O_WRONLY, 00644);
+			if (fd == -1) {
+				msg_err ("cannot open %s: %s", convert_to, strerror (errno));
+			}
+			else if (write (fd, rcl, strlen (rcl)) == -1) {
+				msg_err ("cannot write rcl %s: %s", convert_to, strerror (errno));
+			}
+			close (fd);
+			free (rcl);
+		}
+	}
+
 	return TRUE;
 }
 
diff --git a/src/main.c b/src/main.c
index 8f37f5cc9..6c4f0c318 100644
--- a/src/main.c
+++ b/src/main.c
@@ -75,6 +75,7 @@ static gboolean                 dump_vars = FALSE;
 static gboolean                 dump_cache = FALSE;
 static gboolean                 is_debug = FALSE;
 static gboolean                 is_insecure = FALSE;
+static gchar                   *convert_config = FALSE;
 
 /* List of workers that are pending to start */
 static GList                   *workers_pending = NULL;
@@ -103,6 +104,7 @@ static GOptionEntry entries[] =
   { "test-lua", 0, 0, G_OPTION_ARG_FILENAME_ARRAY, &lua_tests, "Specify lua file(s) to test", NULL },
   { "sign-config", 0, 0, G_OPTION_ARG_FILENAME_ARRAY, &sign_configs, "Specify config file(s) to sign", NULL },
   { "private-key", 0, 0, G_OPTION_ARG_FILENAME, &privkey, "Specify private key to sign", NULL },
+  { "convert-config", 0, 0, G_OPTION_ARG_FILENAME, &convert_config, "Convert cnfiguration to UCL", NULL},
   { NULL, 0, 0, G_OPTION_ARG_NONE, NULL, NULL, NULL }
 };
 
@@ -754,7 +756,7 @@ load_rspamd_config (struct config_file *cfg, gboolean init_modules)
 	struct filter                  *filt;
 	struct module_ctx              *cur_module = NULL;
 
-	if (! read_xml_config (cfg, cfg->cfg_name)) {
+	if (! read_rspamd_config (cfg, cfg->cfg_name, convert_config)) {
 		return FALSE;
 	}
 
diff --git a/src/main.h b/src/main.h
index e0545408c..f169b4f5d 100644
--- a/src/main.h
+++ b/src/main.h
@@ -22,7 +22,7 @@
 #include "roll_history.h"
 
 /* Default values */
-#define FIXED_CONFIG_FILE ETC_PREFIX "/rspamd.xml"
+#define FIXED_CONFIG_FILE ETC_PREFIX "/rspamd.conf"
 /* Time in seconds to exit for old worker */
 #define SOFT_SHUTDOWN_TIME 10
 /* Default metric name */
-- 
2.39.5