From: Peter Hofmann <scm@uninformativ.de>
Date: Sun, 30 Nov 2014 14:50:51 +0000 (+0100)
Subject: Re-implement adblock as a web extension
X-Git-Tag: v1.0.0~207
X-Git-Url: https://git.armaanb.net/?p=chorizo.git;a=commitdiff_plain;h=dae6061a900777aff7c4bfbddf2cef191743cfce

Re-implement adblock as a web extension
---

diff --git a/Makefile b/Makefile
index 5dad98b..6d1c811 100644
--- a/Makefile
+++ b/Makefile
@@ -3,6 +3,9 @@ __NAME__ = lariza
 __NAME_UPPERCASE__ = `echo $(__NAME__) | sed 's/.*/\U&/'`
 __NAME_CAPITALIZED__ = `echo $(__NAME__) | sed 's/^./\U&\E/'`
 
+.PHONY: all
+all: $(__NAME__) we_adblock.so
+
 $(__NAME__): browser.c
 	$(CC) $(CFLAGS) $(LDFLAGS) \
 		-D__NAME__=\"$(__NAME__)\" \
@@ -11,5 +14,13 @@ $(__NAME__): browser.c
 		-o $@ $< \
 		`pkg-config --cflags --libs gtk+-3.0 glib-2.0 webkit2gtk-4.0`
 
+we_adblock.so: we_adblock.c
+	$(CC) $(CFLAGS) $(LDFLAGS) \
+		-D__NAME__=\"$(__NAME__)\" \
+		-D__NAME_UPPERCASE__=\"$(__NAME_UPPERCASE__)\" \
+		-D__NAME_CAPITALIZED__=\"$(__NAME_CAPITALIZED__)\" \
+		-shared -o $@ -fPIC $< \
+		`pkg-config --cflags --libs glib-2.0 webkit2gtk-4.0`
+
 clean:
-	rm -f $(__NAME__)
+	rm -f $(__NAME__) we_adblock.so
diff --git a/README b/README
index e105f89..b706069 100644
--- a/README
+++ b/README
@@ -17,6 +17,8 @@ Features:
     - Global content zoom
     - Cooperative instances using FIFOs
     - Support for Flash and Java
+    - Bundled web extensions:
+        - Adblock
 
 
 ==============================================
@@ -30,9 +32,6 @@ http://blogs.igalia.com/carlosgc/2014/08/01/webkitgtk-2-5-1-good-bye-webkit1/
 
 There's a number of issues on this branch:
 
-    - No adblock support. As stated in commit f1174ff, this has to be
-      implemented as a "WebKit2 web extension". Maybe do this in a
-      project of its own?
     - "View source" mode is not yet ported. This feature has been
       removed from WebKit2, so we have to invent some way to re-create
       this...
@@ -302,6 +301,35 @@ won't be touched. Instead, the new file name will have a suffix such as
 ".1", ".2", ".3" and so on.
 
 
+======================
+Bundled web extensions
+======================
+
+On startup, WebKit checks ~/.local/share/lariza/web_extensions for any
+.so files. See this blog post for further information on these
+extensions:
+
+http://blogs.igalia.com/carlosgc/2013/09/10/webkit2gtk-web-process-extensions/
+
+lariza comes with the following extensions:
+
+    we_adblock.so
+
+        Generic adblock. Reads patterns from the following file:
+
+            ~/.config/lariza/adblock.black
+
+        Each line can contain a regular expression. These expressions
+        match case-insensitive and partially, i.e. ".*foo.*" is the same
+        as ".*FOO.*" and you can use anchors like "^https?://...".
+
+        Lines starting with "#" are ignored.
+
+Those bundled web extensions are automatically compiled when you run
+make. To use them, though, make sure to copy them to the directory
+mentioned above.
+
+
 ====================
 WebKit local storage
 ====================
@@ -334,3 +362,8 @@ API references:
     - http://webkitgtk.org/reference/webkit2gtk/stable/index.html
     - https://developer.gnome.org/gtk3/stable/index.html
     - https://developer.gnome.org/glib/stable/index.html
+
+Regular expressions supported by GRegex, you can use these in your
+adblock patterns:
+
+    - https://developer.gnome.org/glib/stable/glib-regex-syntax.html
diff --git a/we_adblock.c b/we_adblock.c
new file mode 100644
index 0000000..be4e7b7
--- /dev/null
+++ b/we_adblock.c
@@ -0,0 +1,81 @@
+#include <stdio.h>
+
+#include <glib.h>
+#include <webkit2/webkit-web-extension.h>
+
+
+static GSList *adblock_patterns = NULL;
+
+
+static void
+adblock_load(void)
+{
+	GRegex *re = NULL;
+	GError *err = NULL;
+	GIOChannel *channel = NULL;
+	gchar *path = NULL, *buf = NULL;
+
+	path = g_build_filename(g_get_user_config_dir(), __NAME__, "adblock.black",
+	                        NULL);
+	channel = g_io_channel_new_file(path, "r", &err);
+	if (channel != NULL)
+	{
+		while (g_io_channel_read_line(channel, &buf, NULL, NULL, NULL)
+		       == G_IO_STATUS_NORMAL)
+		{
+			g_strstrip(buf);
+			if (buf[0] != '#')
+			{
+				re = g_regex_new(buf,
+				                 G_REGEX_CASELESS | G_REGEX_OPTIMIZE,
+				                 G_REGEX_MATCH_PARTIAL, &err);
+				if (err != NULL)
+				{
+					fprintf(stderr, __NAME__": Could not compile regex: %s\n", buf);
+					g_error_free(err);
+					err = NULL;
+				}
+				else
+					adblock_patterns = g_slist_append(adblock_patterns, re);
+			}
+			g_free(buf);
+		}
+		g_io_channel_shutdown(channel, FALSE, NULL);
+	}
+	g_free(path);
+}
+
+static gboolean
+web_page_send_request(WebKitWebPage *web_page, WebKitURIRequest *request,
+                      WebKitURIResponse *redirected_response, gpointer user_data)
+{
+	GSList *it = adblock_patterns;
+	const gchar *uri;
+
+	uri = webkit_uri_request_get_uri(request);
+
+	while (it)
+	{
+		if (g_regex_match((GRegex *)(it->data), uri, 0, NULL))
+			return TRUE;
+		it = g_slist_next(it);
+	}
+
+	return FALSE;
+}
+
+static void
+web_page_created_callback(WebKitWebExtension *extension, WebKitWebPage *web_page,
+                          gpointer user_data)
+{
+	g_signal_connect_object(web_page, "send-request",
+	                        G_CALLBACK(web_page_send_request), NULL, 0);
+}
+
+G_MODULE_EXPORT void
+webkit_web_extension_initialize(WebKitWebExtension *extension)
+{
+	adblock_load();
+	g_signal_connect(extension, "page-created",
+	                 G_CALLBACK(web_page_created_callback), NULL);
+}