Reintroducing simple filter rules

--HG--
branch : experimental

9 files changed, 130 insertions, 20 deletions

diff --git a/doc/dwb.1 b/doc/dwb.1
index bdae1ba4..c09c5f0e 100644
--- a/doc/dwb.1
+++ b/doc/dwb.1
@@ -550,12 +550,14 @@ If a string value is set to
 the default value will be used. The settings in detail are: 
 .TP
 .BR adblocker
-Block advertisements using a filterlist. The filter can be specified in
-.IR $XDG_CONFIG_HOME/dwb/adblock .
-Regular expressions can be used by putting a '@' at the beginning of the line. 
-default value:
+Block advertisements using a filterlist, see also 
+.IR adblocker-filterlist
 .IR false .
 .TP
+.BR adblocker-filterlist
+A path to a adblock plus compatible filterlist for the adblocker.
+.IR NULL .
+.TP
 .BR auto-load-images  
 Load images automatically. Possible values: true/false, 
 default value:
diff --git a/src/adblock.c b/src/adblock.c
index 69f9d19a..03d967d0 100644
--- a/src/adblock.c
+++ b/src/adblock.c
@@ -97,6 +97,8 @@ typedef struct _AdblockElementHider {
 
 
 /* Static variables {{{*/
+static GPtrArray *_simple_rules;
+static GPtrArray *_simple_exceptions;
 static GPtrArray *_rules;
 static GPtrArray *_exceptions;
 static GHashTable *_hider_rules;
@@ -167,9 +169,18 @@ adblock_do_match(AdblockRule *rule, const char *uri) {
   return false;
 }/*}}}*/
 
-/* adblock_match(GPtrArray *, SoupURI *, const char *base_domain, * AdblockAttribute, gboolean thirdparty) {{{*/
+/* adblock_match(GPtrArray *, SoupURI *, const char *base_domain, * AdblockAttribute, gboolean thirdparty) 
+ * Params: 
+ * array      - the filtes array
+ * uri        - the uri to check
+ * uri_host   - the hostname of the request
+ * uri_base   - the domainname of the request
+ * host       - the hostname of the page
+ * domain     - the domainname of the page
+ * thirdparty - the domainname of the page
+ * {{{*/
 gboolean                
-adblock_match(GPtrArray *array, const char *uri, const char *uri_host, const char *uri_base, const char *host, const char *base_domain, AdblockAttribute attributes, gboolean thirdparty) {
+adblock_match(GPtrArray *array, const char *uri, const char *uri_host, const char *uri_base, const char *host, const char *domain, AdblockAttribute attributes, gboolean thirdparty) {
   gboolean match = false;
   const char *base_start = strstr(uri, uri_base);
   const char *uri_start = strstr(uri, uri_host);
@@ -199,7 +210,7 @@ adblock_match(GPtrArray *array, const char *uri, const char *uri_host, const cha
       if (rule->attributes & ADBLOCK_CLEAR_UPPER && ! (rule->attributes & attributes) ) 
         continue;
     }
-    if (rule->domains && !domain_match(rule->domains, host, base_domain)) {
+    if (rule->domains && !domain_match(rule->domains, host, domain)) {
       continue;
     }
     if    ( (rule->options & AO_THIRDPARTY && !thirdparty) 
@@ -291,6 +302,47 @@ adblock_frame_created_cb(WebKitWebView *wv, WebKitWebFrame *frame, GList *gl) {
   g_signal_connect(frame, "load-committed", G_CALLBACK(adblock_frame_load_committed_cb), gl);
 }/*}}}*/
 
+static void 
+adblock_resource_request_cb(WebKitWebView *wv, WebKitWebFrame *frame,
+    WebKitWebResource *resource, WebKitNetworkRequest *request,
+    WebKitNetworkResponse *response, GList *gl) {
+  if (request == NULL) 
+    return;
+  AdblockAttribute attribute = webkit_web_view_get_main_frame(wv) == frame ? AA_DOCUMENT : AA_SUBDOCUMENT;
+
+  const char *uri = webkit_network_request_get_uri(request);
+  if (uri == NULL)
+    return;
+  SoupMessage *msg = webkit_network_request_get_message(request);
+  if (msg == NULL)
+    return;
+  SoupURI *suri = soup_message_get_uri(msg);
+  const char *host = soup_uri_get_host(suri);
+  if (host == NULL)
+    return;
+  const char *domain = domain_get_base_for_host(host);
+  if (domain == NULL)
+    return;
+
+  SoupURI *sfirst_party = soup_message_get_first_party(msg);
+  if (sfirst_party == NULL)
+    return;
+  const char *firsthost = soup_uri_get_host(sfirst_party);
+  if (firsthost == NULL)
+    return;
+  const char *firstdomain = domain_get_base_for_host(firsthost);
+  if (firstdomain == NULL)
+    return;
+  gboolean thirdparty = strcmp(domain, firstdomain);
+
+  if (!adblock_match(_simple_exceptions, uri, host, domain, firsthost, firstdomain, attribute, thirdparty)) {
+    if (adblock_match(_simple_rules, uri, host, domain, firsthost, firstdomain, attribute, thirdparty)) {
+      webkit_network_request_set_uri(request, "about:blank");
+    }
+  }
+
+ 
+}
 /* adblock_load_status_cb(WebKitWebView *, GParamSpec *, GList *) {{{*/
 static void
 adblock_load_status_cb(WebKitWebView *wv, GParamSpec *p, GList *gl) {
@@ -310,10 +362,16 @@ adblock_load_status_cb(WebKitWebView *wv, GParamSpec *p, GList *gl) {
     AdblockElementHider *hider;
     WebKitWebDataSource *datasource = webkit_web_frame_get_data_source(frame);
     WebKitNetworkRequest *request = webkit_web_data_source_get_request(datasource);
+
     SoupMessage *msg = webkit_network_request_get_message(request);
+    g_return_if_fail(msg != NULL);
+
     SoupURI *suri = soup_message_get_uri(msg);
+    g_return_if_fail(suri != NULL);
+
     const char *host = soup_uri_get_host(suri);
     const char *base_domain = domain_get_base_for_host(host);
+    g_return_if_fail(host != NULL);
     g_return_if_fail(base_domain != NULL);
 
     GString *css_rule = NULL;
@@ -389,6 +447,11 @@ adblock_load_status_cb(WebKitWebView *wv, GParamSpec *p, GList *gl) {
 
 
 /* START AND END {{{*/
+
+gboolean
+adblock_running() {
+  return _init && GET_BOOL("adblocker");
+}
 /* adblock_disconnect(GList *) {{{*/
 void 
 adblock_disconnect(GList *gl) {
@@ -401,10 +464,13 @@ adblock_disconnect(GList *gl) {
     g_signal_handler_disconnect(WEBVIEW(gl), (VIEW(gl)->status->signals[SIG_AD_FRAME_CREATED]));
     v->status->signals[SIG_AD_FRAME_CREATED] = 0;
   }
+  if (v->status->signals[SIG_AD_RESOURCE_REQUEST] > 0) {
+    g_signal_handler_disconnect(WEBVIEW(gl), (VIEW(gl)->status->signals[SIG_AD_RESOURCE_REQUEST]));
+    v->status->signals[SIG_AD_RESOURCE_REQUEST] = 0;
+  }
 }/*}}}*/
 
 
-
 /* adblock_connect() {{{*/
 void 
 adblock_connect(GList *gl) {
@@ -414,6 +480,9 @@ adblock_connect(GList *gl) {
     VIEW(gl)->status->signals[SIG_AD_LOAD_STATUS] = g_signal_connect(WEBVIEW(gl), "notify::load-status", G_CALLBACK(adblock_load_status_cb), gl);
     VIEW(gl)->status->signals[SIG_AD_FRAME_CREATED] = g_signal_connect(WEBVIEW(gl), "frame-created", G_CALLBACK(adblock_frame_created_cb), gl);
   }
+  if (_simple_rules->len > 0) {
+    VIEW(gl)->status->signals[SIG_AD_RESOURCE_REQUEST] = g_signal_connect(WEBVIEW(gl), "resource-request-starting", G_CALLBACK(adblock_resource_request_cb), gl);
+  }
 }/*}}}*/
 
 /* adblock_warn_ignored(const char *message, const char *rule){{{*/
@@ -438,7 +507,6 @@ adblock_rule_parse(char *pattern) {
   if (pattern[0] == '!' || pattern[0] == '[') {
     return STATUS_IGNORE;
   }
-  //AdblockRule *rule = adblock_rule_new();
   char *tmp = NULL;
   char *tmp_a = NULL, *tmp_b = NULL, *tmp_c = NULL;
   /* Element hiding rules */
@@ -662,10 +730,18 @@ adblock_rule_parse(char *pattern) {
     adrule->pattern = rule;
     adrule->options = option;
     adrule->domains = domains;
-    if (exception) 
-      g_ptr_array_add(_exceptions, adrule);
-    else 
-      g_ptr_array_add(_rules, adrule);
+    if (!(attributes & ~(AA_DOCUMENT | AA_FRAME))) {
+      if (exception) 
+        g_ptr_array_add(_simple_exceptions, adrule);
+      else 
+        g_ptr_array_add(_simple_rules, adrule);
+    }
+    else {
+      if (exception) 
+        g_ptr_array_add(_exceptions, adrule);
+      else 
+        g_ptr_array_add(_rules, adrule);
+    }
   }
 error_out:
   FREE(tmp_a);
@@ -682,6 +758,10 @@ adblock_end() {
     g_string_free(_css_exceptions, true);
   if (_rules != NULL) 
     g_ptr_array_free(_rules, true);
+  if (_simple_rules != NULL) 
+    g_ptr_array_free(_simple_rules, true);
+  if (_simple_exceptions != NULL) 
+    g_ptr_array_free(_simple_exceptions, true);
   if(_exceptions != NULL) 
     g_ptr_array_free(_exceptions, true);
   if (_hider_rules != NULL) 
@@ -709,6 +789,8 @@ adblock_init() {
   }
   _rules              = g_ptr_array_new_with_free_func((GDestroyNotify)adblock_rule_free);
   _exceptions         = g_ptr_array_new_with_free_func((GDestroyNotify)adblock_rule_free);
+  _simple_rules       = g_ptr_array_new_with_free_func((GDestroyNotify)adblock_rule_free);
+  _simple_exceptions  = g_ptr_array_new_with_free_func((GDestroyNotify)adblock_rule_free);
   _hider_rules        = g_hash_table_new_full((GHashFunc)g_str_hash, (GEqualFunc)g_str_equal, (GDestroyNotify)g_free, NULL);
   _css_exceptions     = g_string_new(NULL);
   _css_rules          = g_string_new(NULL);
diff --git a/src/adblock.h b/src/adblock.h
index 0015372d..2b138836 100644
--- a/src/adblock.h
+++ b/src/adblock.h
@@ -23,9 +23,8 @@
 #include "dwb.h"
 
 gboolean adblock_init();
+gboolean adblock_running();
 void adblock_end();
-void adblock_resource_request_cb(WebKitWebView *, WebKitWebFrame *, WebKitWebResource *, 
-    WebKitNetworkRequest  *, WebKitNetworkResponse *, GList *gl);
 void adblock_connect(GList *gl);
 void adblock_disconnect(GList *gl);
 void adblock_set_user_stylesheet(const char *file);
diff --git a/src/commands.c b/src/commands.c
index 336ac532..e170f1f4 100644
--- a/src/commands.c
+++ b/src/commands.c
@@ -18,6 +18,9 @@
 
 #include "commands.h"
 #include "local.h"
+#ifdef DWB_ADBLOCKER
+#include "adblock.h"
+#endif
 
 static int inline dwb_floor(double x) { 
   return x >= 0 ? (int) x : (int) x - 1;
@@ -695,6 +698,20 @@ commands_toggle_scripts(KeyMap *km, Arg *arg) {
   return STATUS_OK;
 }/*}}}*/
 
+#ifdef DWB_ADBLOCKER
+/* commands_toggle_adblocker {{{ */
+DwbStatus 
+commands_toggle_adblocker(KeyMap *km, Arg *arg) {
+  gboolean running = adblock_running();
+  WebSettings *s = g_hash_table_lookup(dwb.settings, "adblocker");
+  s->arg.b = !running;
+  dwb_set_adblock(NULL, s);
+
+  dwb_set_normal_message(dwb.state.fview, true, "Adblocker %s", running ? "disabled" : "enabled");
+  return STATUS_OK;
+}/*}}}*/
+#endif
+
 /* commands_new_window_or_view{{{*/
 DwbStatus 
 commands_new_window_or_view(KeyMap *km, Arg *arg) {
diff --git a/src/commands.h b/src/commands.h
index 3400a3db..b8e3a4a9 100644
--- a/src/commands.h
+++ b/src/commands.h
@@ -93,5 +93,8 @@ DwbStatus commands_open_editor(KeyMap *, Arg *);
 DwbStatus commands_insert_mode(KeyMap *, Arg *);
 DwbStatus commands_command_mode(KeyMap *, Arg *);
 DwbStatus commands_only(KeyMap *, Arg *);
+#ifdef DWB_ADBLOCKER
+DwbStatus commands_toggle_adblocker(KeyMap *, Arg *);
+#endif
 
 #endif
diff --git a/src/config.h b/src/config.h
index 8d10c6d6..197bcdcc 100644
--- a/src/config.h
+++ b/src/config.h
@@ -139,6 +139,7 @@ static KeyValue KEYS[] = {
   { "toggle_plugins_host",      {   "ph",           0 }, }, 
   { "toggle_plugins_uri_tmp",   {   "ptu",           0 }, }, 
   { "toggle_plugins_host_tmp",  {   "pth",           0 }, }, 
+  { "toggle_adblocker",         {   "a",           GDK_CONTROL_MASK }, }, 
   { "new_view",                 {   "V",            0 }, }, 
   { "new_window",               {   "W",            0 }, }, 
   { "save",                     {   "sf",            0 }, }, 
@@ -432,6 +433,10 @@ static FunctionMap FMAP [] = {
     (Func) commands_toggle_plugin_blocker, NULL,      POST_SM,    { .n = ALLOW_HOST | ALLOW_TMP } }, 
   { { "toggle_plugins_uri_tmp", "Toggle block content for current url for this session" },   1, 
     (Func) commands_toggle_plugin_blocker, NULL,       POST_SM,    { .n = ALLOW_URI | ALLOW_TMP } }, 
+#ifdef DWB_ADBLOCKER
+  { { "toggle_adblocker", "Toggle adblocker" },   1, 
+    (Func) commands_toggle_adblocker, NULL,       POST_SM,    { 0 } }, 
+#endif
   { { "toggle_hidden_files",   "Toggle hidden files in directory listing" },  1, 
     (Func) commands_toggle_hidden_files, NULL,                  ALWAYS_SM,    { 0 } }, 
   { { "print",                 "Print current page" },                         1, 
diff --git a/src/domain.c b/src/domain.c
index 125ee6fd..072501fb 100644
--- a/src/domain.c
+++ b/src/domain.c
@@ -76,7 +76,8 @@ domain_match(char **domains, const char *host, const char *base_domain) {
 
 const char *
 domain_get_base_for_host(const char *host) {
-  g_return_val_if_fail(host != NULL, NULL);
+  if (host == NULL)
+    return NULL;
   g_return_val_if_fail(_tld_table != NULL, NULL);
 
   const char *cur_domain = host;
diff --git a/src/dwb.c b/src/dwb.c
index 2c3866f7..a6442702 100644
--- a/src/dwb.c
+++ b/src/dwb.c
@@ -45,9 +45,6 @@ static void dwb_set_startpage(GList *, WebSettings *);
 static void dwb_set_message_delay(GList *, WebSettings *);
 static void dwb_set_history_length(GList *, WebSettings *);
 static void dwb_set_plugin_blocker(GList *, WebSettings *);
-#ifdef DWB_ADBLOCKER
-static void dwb_set_adblock(GList *, WebSettings *);
-#endif
 static void dwb_set_hide_tabbar(GList *, WebSettings *);
 static void dwb_set_sync_interval(GList *, WebSettings *);
 static void dwb_set_private_browsing(GList *, WebSettings *);
@@ -123,7 +120,7 @@ dwb_set_plugin_blocker(GList *gl, WebSettings *s) {
 
 #ifdef DWB_ADBLOCKER
 /* dwb_set_adblock {{{*/
-static void
+void
 dwb_set_adblock(GList *gl, WebSettings *s) {
   if (s->arg.b) {
     for (GList *l = dwb.state.views; l; l=l->next) 
diff --git a/src/dwb.h b/src/dwb.h
index ab2928d4..78e1a293 100644
--- a/src/dwb.h
+++ b/src/dwb.h
@@ -374,6 +374,7 @@ enum Signal {
 #ifdef DWB_ADBLOCKER
   SIG_AD_LOAD_STATUS,
   SIG_AD_FRAME_CREATED,
+  SIG_AD_RESOURCE_REQUEST,
 #endif
 
   SIG_PLUGINS_LOAD,
@@ -796,5 +797,8 @@ void dwb_set_open_mode(Open);
 
 DwbStatus dwb_set_clipboard(const char *text, GdkAtom atom);
 DwbStatus dwb_open_in_editor(void);
+#ifdef DWB_ADBLOCKER
+void dwb_set_adblock(GList *, WebSettings *);
+#endif
 
 #endif