* Add functions to operate with content-type header and describe them in documentation

diff --git a/README.utf8.txt b/README.utf8.txt
index eed4010a20f771cd50e07f560c190f9c8aa934c7..3ea18664ad969ea33ac9fb10208df366e15a1e1d 100644 (file)
--- a/README.utf8.txt
+++ b/README.utf8.txt
@@ -139,3 +139,11 @@ SOME_SYMBOL = "${to_blah} & !(${from_blah} | ${subject_blah})"
     между частями письма. Функция работает с сообщениями, содержащими 2 текстовые части (text/plain и text/html) и
        возвращает true тогда, когда эти части различаются более чем на n процентов. Если аргумент не указан, то
        по умолчанию ищется различие в 100% (полностью разные части).
+  * content_type_compare_param - сравнивает параметр content-type заголовка с регулярным выражением или строкой:
+     content_type_compare_param(Charset, /windows-\d+/)
+        content_type_compare_param(Charset, ascii)
+  * content_type_has_param - проверяет, есть ли в заголовке content-type определенный параметр
+  * content_type_is_subtype - сравнивает подтип content-type с регулярным выражением или строкой
+  * content_type_is_type - сравнивает тип content-type с регулярным выражением или строкой
+     content_type_is_type(text)
+     content_type_is_subtype(/?.html/)

diff --git a/src/expressions.c b/src/expressions.c
index a6b6ec9a12c7622d4f15c8d7da900d9212f0ec14..00fec06659673fb20713d2577a22c2a53191247e 100644 (file)
--- a/src/expressions.c
+++ b/src/expressions.c
@@ -34,7 +34,12 @@ typedef gboolean (*rspamd_internal_func_t)(struct worker_task *, GList *args);
 
 gboolean rspamd_compare_encoding (struct worker_task *task, GList *args);
 gboolean rspamd_header_exists (struct worker_task *task, GList *args);
+gboolean rspamd_content_type_compare_param (struct worker_task *task, GList *args);
+gboolean rspamd_content_type_has_param (struct worker_task *task, GList *args);
+gboolean rspamd_content_type_is_subtype (struct worker_task *task, GList *args);
+gboolean rspamd_content_type_is_type (struct worker_task *task, GList *args);
 gboolean rspamd_parts_distance (struct worker_task *task, GList *args);
+
 /*
  * List of internal functions of rspamd
  * Sorted by name to use bsearch
@@ -45,6 +50,10 @@ static struct _fl {
 } rspamd_functions_list[] = {
        { "compare_encoding", rspamd_compare_encoding },
        { "compare_parts_distance", rspamd_parts_distance },
+       { "content_type_compare_param", rspamd_content_type_compare_param },
+       { "content_type_has_param", rspamd_content_type_has_param },
+       { "content_type_is_subtype", rspamd_content_type_is_subtype },
+       { "content_type_is_type", rspamd_content_type_is_type },
        { "header_exists", rspamd_header_exists },
 };
 
@@ -57,6 +66,29 @@ fl_cmp (const void *s1, const void *s2)
        return strcmp (fl1->name, fl2->name);
 }
 
+/* Cache for regular expressions that are used in functions */
+static GHashTable *re_cache = NULL;
+
+static inline void *
+re_cache_check (const char *line)
+{
+       if (re_cache == NULL) {
+               re_cache = g_hash_table_new (g_str_hash, g_str_equal);
+               return NULL;
+       }
+       return g_hash_table_lookup (re_cache, line);
+}
+
+static inline void
+re_cache_add (char *line, void *pointer)
+{
+       if (re_cache == NULL) {
+               re_cache = g_hash_table_new (g_str_hash, g_str_equal);
+       }
+       
+       g_hash_table_insert (re_cache, line, pointer);
+}
+
 /*
  * Functions for parsing expressions
  */
@@ -652,6 +684,166 @@ rspamd_parts_distance (struct worker_task *task, GList *args)
        return FALSE;
 }
 
+gboolean 
+rspamd_content_type_compare_param (struct worker_task *task, GList *args)
+{
+       char *param_name, *param_pattern;
+       const char *param_data;
+       struct rspamd_regexp *re;
+       
+       if (args == NULL) {
+               msg_warn ("rspamd_content_type_compare_param: no parameters to function");
+               return FALSE;
+       }
+       param_name = args->data;
+       args = g_list_next (args);
+       if (args == NULL) {
+               msg_warn ("rspamd_content_type_compare_param: too few params to function");
+               return FALSE;
+       }
+       param_pattern = args->data;
+
+       if ((param_data = g_mime_content_type_get_parameter (g_mime_object_get_content_type (GMIME_OBJECT (task->message)), param_name)) == NULL) {
+               return FALSE;
+       }
+       
+       if (*param_pattern == '/') {
+               /* This is regexp, so compile and create g_regexp object */
+               if ((re = re_cache_check (param_pattern)) == NULL) {
+                       re = parse_regexp (task->task_pool, param_pattern);
+                       if (re == NULL) {
+                               msg_warn ("rspamd_content_type_compare_param: cannot compile regexp for function");
+                               return FALSE;
+                       }
+                       re_cache_add (param_pattern, re);
+               }
+               if (g_regex_match (re->regexp, param_data, 0, NULL) == TRUE) {
+                       return TRUE;
+               }
+       }
+       else {
+               /* Just do strcasecmp */
+               if (g_ascii_strcasecmp (param_data, param_pattern) == 0) {
+                       return TRUE;
+               }
+       }
+
+       return FALSE;
+}      
+
+gboolean 
+rspamd_content_type_has_param (struct worker_task *task, GList *args)
+{
+       char *param_name;
+       const char *param_data;
+       
+       if (args == NULL) {
+               msg_warn ("rspamd_content_type_compare_param: no parameters to function");
+               return FALSE;
+       }
+       param_name = args->data;
+       if ((param_data = g_mime_content_type_get_parameter (g_mime_object_get_content_type (GMIME_OBJECT (task->message)), param_name)) == NULL) {
+               return FALSE;
+       }
+       
+       return TRUE;
+}
+
+/* In gmime24 this function is opaque, so define it here to avoid errors when compiling with gmime24 */
+typedef struct {
+       char *type;
+       char *subtype;
+       
+       GMimeParam *params;
+       GHashTable *param_hash;
+} localContentType;
+
+gboolean 
+rspamd_content_type_is_subtype (struct worker_task *task, GList *args)
+{
+       char *param_pattern;
+       struct rspamd_regexp *re;
+       localContentType *ct;
+       
+       if (args == NULL) {
+               msg_warn ("rspamd_content_type_compare_param: no parameters to function");
+               return FALSE;
+       }
+
+       param_pattern = args->data;
+       ct = (localContentType *)g_mime_object_get_content_type (GMIME_OBJECT (task->message));
+
+       if (ct == NULL) {
+               return FALSE;
+       }
+       
+       if (*param_pattern == '/') {
+               /* This is regexp, so compile and create g_regexp object */
+               if ((re = re_cache_check (param_pattern)) == NULL) {
+                       re = parse_regexp (task->task_pool, param_pattern);
+                       if (re == NULL) {
+                               msg_warn ("rspamd_content_type_compare_param: cannot compile regexp for function");
+                               return FALSE;
+                       }
+                       re_cache_add (param_pattern, re);
+               }
+               if (g_regex_match (re->regexp, ct->subtype, 0, NULL) == TRUE) {
+                       return TRUE;
+               }
+       }
+       else {
+               /* Just do strcasecmp */
+               if (g_ascii_strcasecmp (ct->subtype, param_pattern) == 0) {
+                       return TRUE;
+               }
+       }
+
+       return FALSE;
+}
+
+gboolean 
+rspamd_content_type_is_type (struct worker_task *task, GList *args)
+{
+       char *param_pattern;
+       struct rspamd_regexp *re;
+       localContentType *ct;
+       
+       if (args == NULL) {
+               msg_warn ("rspamd_content_type_compare_param: no parameters to function");
+               return FALSE;
+       }
+
+       param_pattern = args->data;
+       ct = (localContentType *)g_mime_object_get_content_type (GMIME_OBJECT (task->message));
+
+       if (ct == NULL) {
+               return FALSE;
+       }
+       
+       if (*param_pattern == '/') {
+               /* This is regexp, so compile and create g_regexp object */
+               if ((re = re_cache_check (param_pattern)) == NULL) {
+                       re = parse_regexp (task->task_pool, param_pattern);
+                       if (re == NULL) {
+                               msg_warn ("rspamd_content_type_compare_param: cannot compile regexp for function");
+                               return FALSE;
+                       }
+                       re_cache_add (param_pattern, re);
+               }
+               if (g_regex_match (re->regexp, ct->type, 0, NULL) == TRUE) {
+                       return TRUE;
+               }
+       }
+       else {
+               /* Just do strcasecmp */
+               if (g_ascii_strcasecmp (ct->type, param_pattern) == 0) {
+                       return TRUE;
+               }
+       }
+
+       return FALSE;
+}
+
 /*
  * vi:ts=4
  */