From 8525f71340f4edaeb9cdf9f57766dd4a4a0f993a Mon Sep 17 00:00:00 2001
From: Vsevolod Stakhov <vsevolod@highsecure.ru>
Date: Wed, 17 Dec 2014 13:59:52 +0000
Subject: [PATCH] Make words from text parts.

---
 src/libmime/message.c       | 20 ++++++++++++++++++--
 src/libmime/message.h       |  3 ++-
 src/libserver/task.c        | 10 ++++++++++
 src/tokenizers/tokenizers.c |  4 ++--
 src/tokenizers/tokenizers.h |  3 ++-
 5 files changed, 34 insertions(+), 6 deletions(-)

diff --git a/src/libmime/message.c b/src/libmime/message.c
index 0e33967ca..f465b1c06 100644
--- a/src/libmime/message.c
+++ b/src/libmime/message.c
@@ -30,6 +30,8 @@
 #include "html.h"
 #include "images.h"
 #include "utlist.h"
+#include "tokenizers/tokenizers.h"
+
 #include <iconv.h>
 
 #define RECURSION_LIMIT 30
@@ -1033,6 +1035,8 @@ process_text_part (struct rspamd_task *task,
 {
 	struct mime_text_part *text_part;
 	const gchar *cd;
+	gchar *pos;
+	rspamd_fstring_t token, buf;
 
 	/* Skip attachements */
 #ifndef GMIME24
@@ -1056,7 +1060,6 @@ process_text_part (struct rspamd_task *task,
 
 	if (g_mime_content_type_is_type (type, "text",
 		"html") || g_mime_content_type_is_type (type, "text", "xhtml")) {
-		debug_task ("got urls from text/html part");
 
 		text_part =
 			rspamd_mempool_alloc0 (task->task_pool,
@@ -1097,7 +1100,6 @@ process_text_part (struct rspamd_task *task,
 		task->text_parts = g_list_prepend (task->text_parts, text_part);
 	}
 	else if (g_mime_content_type_is_type (type, "text", "*")) {
-		debug_task ("got urls from text/plain part");
 
 		text_part =
 			rspamd_mempool_alloc0 (task->task_pool,
@@ -1120,6 +1122,20 @@ process_text_part (struct rspamd_task *task,
 		rspamd_fuzzy_from_text_part (text_part, task->task_pool, task->cfg->max_diff);
 		task->text_parts = g_list_prepend (task->text_parts, text_part);
 	}
+	else {
+		return;
+	}
+
+	/* Post process part */
+	buf.begin = text_part->content->data;
+	buf.len = text_part->content->len;
+	buf.size = buf.len;
+
+	text_part->words = g_array_new (FALSE, FALSE, sizeof (rspamd_fstring_t));
+	while ((pos = rspamd_tokenizer_get_word (&buf,
+			&token, &text_part->urls_offset)) != NULL) {
+		g_array_append_val (text_part->words, token);
+	}
 }
 
 #ifdef GMIME24
diff --git a/src/libmime/message.h b/src/libmime/message.h
index ddc5939fa..8287db9b0 100644
--- a/src/libmime/message.h
+++ b/src/libmime/message.h
@@ -30,12 +30,13 @@ struct mime_text_part {
 	GByteArray *orig;
 	GByteArray *content;
 	GNode *html_nodes;
-	GList *urls_offset;                                         /**< list of offsets of urls						*/
+	GList *urls_offset;	/**< list of offsets of urls						*/
 	rspamd_fuzzy_t *fuzzy;
 	rspamd_fuzzy_t *double_fuzzy;
 	GMimeObject *parent;
 	GUnicodeScript script;
 	rspamd_fstring_t *diff_str;
+	GArray *words;
 };
 
 struct received_header {
diff --git a/src/libserver/task.c b/src/libserver/task.c
index b6a910a91..bfa9184fb 100644
--- a/src/libserver/task.c
+++ b/src/libserver/task.c
@@ -226,6 +226,7 @@ rspamd_task_free (struct rspamd_task *task, gboolean is_soft)
 {
 	GList *part;
 	struct mime_part *p;
+	struct mime_text_part *tp;
 
 	if (task) {
 		debug_task ("free pointer %p", task);
@@ -236,6 +237,15 @@ rspamd_task_free (struct rspamd_task *task, gboolean is_soft)
 			g_list_free_1 (part);
 		}
 		if (task->text_parts) {
+			part = task->text_parts;
+			while (part) {
+				tp = (struct mime_text_part *)part->data;
+				if (tp->words) {
+					g_array_free (tp->words, TRUE);
+				}
+				part = g_list_next (part);
+			}
+
 			g_list_free (task->text_parts);
 		}
 		if (task->images) {
diff --git a/src/tokenizers/tokenizers.c b/src/tokenizers/tokenizers.c
index 4789d8f62..2adc86ff9 100644
--- a/src/tokenizers/tokenizers.c
+++ b/src/tokenizers/tokenizers.c
@@ -31,7 +31,7 @@
 #include "tokenizers.h"
 
 struct tokenizer tokenizers[] = {
-	{"osb-text", osb_tokenize_text, get_next_word},
+	{"osb-text", osb_tokenize_text, rspamd_tokenizer_get_word},
 };
 
 const int primes[] = {
@@ -104,7 +104,7 @@ token_node_compare_func (gconstpointer a, gconstpointer b)
 
 /* Get next word from specified f_str_t buf */
 gchar *
-get_next_word (rspamd_fstring_t * buf, rspamd_fstring_t * token, GList **exceptions)
+rspamd_tokenizer_get_word (rspamd_fstring_t * buf, rspamd_fstring_t * token, GList **exceptions)
 {
 	gsize remain, pos;
 	guchar *p;
diff --git a/src/tokenizers/tokenizers.h b/src/tokenizers/tokenizers.h
index b10729614..efce5b307 100644
--- a/src/tokenizers/tokenizers.h
+++ b/src/tokenizers/tokenizers.h
@@ -30,7 +30,8 @@ int token_node_compare_func (gconstpointer a, gconstpointer b);
 /* Get tokenizer structure by name or return NULL if this name is not found */
 struct tokenizer * get_tokenizer (const char *name);
 /* Get next word from specified f_str_t buf */
-gchar * get_next_word (rspamd_fstring_t *buf, rspamd_fstring_t *token, GList **exceptions);
+gchar * rspamd_tokenizer_get_word (rspamd_fstring_t *buf,
+		rspamd_fstring_t *token, GList **exceptions);
 /* OSB tokenize function */
 int osb_tokenize_text (struct tokenizer *tokenizer,
 	rspamd_mempool_t *pool,
-- 
2.39.5