[epiphany-extensions] Catched up with latest version of Midori's adblock implementation.
- From: Mario Sanchez Prada <msanchez src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [epiphany-extensions] Catched up with latest version of Midori's adblock implementation.
- Date: Tue, 6 Dec 2011 01:28:04 +0000 (UTC)
commit 67bdb045d72a7b198e4719356fd6fe759ddd276a
Author: Mario Sanchez Prada <msanchez igalia com>
Date: Mon Dec 5 19:06:31 2011 +0100
Catched up with latest version of Midori's adblock implementation.
Current code in Epiphany's adblock extension contained some bugs that
have been already addressed in Midori's adblock extension, which was
also more feature complete (it was able to block more things).
So, it makes sense to backport that code to here.
extensions/adblock/uri-tester.c | 365 ++++++++++++++++++++++++---------------
1 files changed, 226 insertions(+), 139 deletions(-)
---
diff --git a/extensions/adblock/uri-tester.c b/extensions/adblock/uri-tester.c
index 84049ec..6b6b5f0 100644
--- a/extensions/adblock/uri-tester.c
+++ b/extensions/adblock/uri-tester.c
@@ -48,6 +48,9 @@ struct _UriTesterPrivate
GHashTable* keys;
GHashTable* optslist;
GHashTable* urlcache;
+
+ GString* blockcss;
+ GString* blockcssprivate;
};
enum
@@ -63,13 +66,13 @@ G_DEFINE_DYNAMIC_TYPE (UriTester, uri_tester, G_TYPE_OBJECT);
static void uri_tester_class_init (UriTesterClass *klass);
static void uri_tester_init (UriTester *dialog);
-static char *
-uri_tester_fixup_regexp (char* src);
+static GString*
+uri_tester_fixup_regexp (const gchar* prefix, char* src);
static gboolean
uri_tester_parse_file_at_uri (UriTester* tester, const char* fileuri);
-static char *
+static char*
uri_tester_ensure_data_dir ()
{
char* folder = NULL;
@@ -285,20 +288,28 @@ uri_tester_save_filters (UriTester *tester)
g_free (filepath);
}
-static inline gboolean
-uri_tester_check_filter_options (GRegex* regex,
- const char* opts,
- const char* req_uri,
- const char* page_uri)
+static inline gint
+uri_tester_check_rule (UriTester *tester,
+ GRegex* regex,
+ const gchar* patt,
+ const gchar* req_uri,
+ const gchar* page_uri)
{
- if (g_regex_match_simple (",third-party", opts,
- G_REGEX_CASELESS, G_REGEX_MATCH_NOTEMPTY))
+ gchar* opts;
+
+ if (!g_regex_match_full (regex, req_uri, -1, 0, 0, NULL, NULL))
+ return FALSE;
+
+ opts = g_hash_table_lookup (tester->priv->optslist, patt);
+ if (opts && g_regex_match_simple (",third-party", opts,
+ G_REGEX_CASELESS, G_REGEX_MATCH_NOTEMPTY))
{
if (page_uri && g_regex_match_full (regex, page_uri, -1, 0, 0, NULL, NULL))
- return TRUE;
+ return FALSE;
}
-
- return FALSE;
+ /* TODO: Domain opt check */
+ LOG ("blocked by pattern regexp=%s -- %s", g_regex_get_pattern (regex), req_uri);
+ return TRUE;
}
static inline gboolean
@@ -308,22 +319,12 @@ uri_tester_is_matched_by_pattern (UriTester *tester,
{
GHashTableIter iter;
gpointer patt, regex;
- char* opts;
g_hash_table_iter_init (&iter, tester->priv->pattern);
while (g_hash_table_iter_next (&iter, &patt, ®ex))
{
- if (g_regex_match_full (regex, req_uri, -1, 0, 0, NULL, NULL))
- {
- opts = g_hash_table_lookup (tester->priv->optslist, patt);
- if (opts && uri_tester_check_filter_options (regex, opts, req_uri, page_uri) == TRUE)
- return FALSE;
- else
- {
- LOG ("blocked by pattern regexp=%s -- %s", g_regex_get_pattern (regex), req_uri);
- return TRUE;
- }
- }
+ if (uri_tester_check_rule(tester, regex, patt, req_uri, page_uri))
+ return TRUE;
}
return FALSE;
}
@@ -335,48 +336,39 @@ uri_tester_is_matched_by_key (UriTester *tester,
const char* page_uri)
{
UriTesterPrivate* priv = NULL;
- char* uri;
- int len;
+ gchar* uri;
+ gint len;
int pos = 0;
GList* regex_bl = NULL;
+ GString* guri;
+ gboolean ret = FALSE;
+ gchar sig[SIGNATURE_SIZE + 1];
priv = tester->priv;
- uri = uri_tester_fixup_regexp ((char*)req_uri);
- len = strlen (uri);
+ memset (&sig[0], 0, sizeof (sig));
+ /* Signatures are made on pattern, so we need to convert url to a pattern as well */
+ guri = uri_tester_fixup_regexp ("", (gchar*)req_uri);
+ uri = guri->str;
+ len = guri->len;
+
for (pos = len - SIGNATURE_SIZE; pos >= 0; pos--)
{
- char* sig = g_strndup (uri + pos, SIGNATURE_SIZE);
- GRegex* regex = g_hash_table_lookup (priv->keys, sig);
- char* opts = NULL;
-
- if (regex && !g_list_find (regex_bl, regex))
- {
- if (g_regex_match_full (regex, req_uri, -1, 0, 0, NULL, NULL))
- {
- opts = g_hash_table_lookup (tester->priv->optslist, sig);
- g_free (sig);
- if (opts && uri_tester_check_filter_options (regex, opts, req_uri, page_uri))
- {
- g_free (uri);
- g_list_free (regex_bl);
- return FALSE;
- }
- else
- {
- LOG ("blocked by regexp=%s -- %s", g_regex_get_pattern (regex), uri);
- g_free (uri);
- g_list_free (regex_bl);
- return TRUE;
- }
- }
- regex_bl = g_list_prepend (regex_bl, regex);
- }
- g_free (sig);
+ GRegex* regex;
+ strncpy (sig, uri + pos, SIGNATURE_SIZE);
+ regex = g_hash_table_lookup (priv->keys, sig);
+
+ /* Dont check if regex is already blacklisted */
+ if (!regex || g_list_find (regex_bl, regex))
+ continue;
+ ret = uri_tester_check_rule (tester, regex, sig, req_uri, page_uri);
+ if (ret)
+ break;
+ regex_bl = g_list_prepend (regex_bl, regex);
}
- g_free (uri);
+ g_string_free (guri, TRUE);
g_list_free (regex_bl);
- return FALSE;
+ return ret;
}
static gboolean
@@ -386,8 +378,6 @@ uri_tester_is_matched (UriTester *tester,
const char* page_uri)
{
UriTesterPrivate* priv = NULL;
- gboolean foundbykey;
- gboolean foundbypattern;
char* value;
priv = tester->priv;
@@ -397,31 +387,35 @@ uri_tester_is_matched (UriTester *tester,
return (value[0] != '0') ? TRUE : FALSE;
/* Look for a match either by key or by pattern. */
- foundbykey = uri_tester_is_matched_by_key (tester, opts, req_uri, page_uri);
- foundbypattern = uri_tester_is_matched_by_pattern (tester, req_uri, page_uri);
+ if (uri_tester_is_matched_by_key (tester, opts, req_uri, page_uri))
+ {
+ g_hash_table_insert (priv->urlcache, g_strdup (req_uri), g_strdup("1"));
+ return TRUE;
+ }
- if (foundbykey || foundbypattern)
+ /* Matching by pattern is pretty expensive, so do it if needed only. */
+ if (uri_tester_is_matched_by_pattern (tester, req_uri, page_uri))
{
g_hash_table_insert (priv->urlcache, g_strdup (req_uri), g_strdup("1"));
return TRUE;
}
+
g_hash_table_insert (priv->urlcache, g_strdup (req_uri), g_strdup("0"));
return FALSE;
}
-static char *
-uri_tester_fixup_regexp (char* src)
+static GString*
+uri_tester_fixup_regexp (const gchar* prefix, char* src)
{
- char* dst;
GString* str;
- int len;
+ int len = 0;
if (!src)
return NULL;
- str = g_string_new ("");
+ str = g_string_new (prefix);
- /* Lets strip first .* */
+ /* lets strip first .* */
if (src[0] == '*')
{
(void)*src++;
@@ -434,22 +428,17 @@ uri_tester_fixup_regexp (char* src)
case '*':
g_string_append (str, ".*");
break;
- /* case '.': */
- /* g_string_append (str, "\\."); */
- /* break; */
+ /*case '.':
+ g_string_append (str, "\\.");
+ break;*/
case '?':
g_string_append (str, "\\?");
break;
case '|':
- g_string_append (str, "");
- break;
- /* FIXME: We actually need to match :[0-9]+ or '/'. Sign
- * means "here could be port number or nothing". So bla.com^
- * will match bla.com/ or bla.com:8080/ but not bla.com.au/.
- */
+ /* FIXME: We actually need to match :[0-9]+ or '/'. Sign means
+ "here could be port number or nothing". So bla.com^ will match
+ bla.com/ or bla.com:8080/ but not bla.com.au/ */
case '^':
- g_string_append (str, "");
- break;
case '+':
break;
default:
@@ -460,41 +449,47 @@ uri_tester_fixup_regexp (char* src)
}
while (*src);
- dst = g_strdup (str->str);
- g_string_free (str, TRUE);
+ len = str->len;
+ /* We dont need .* in the end of url. Thats stupid */
+ if (str->str && str->str[len-1] == '*' && str->str[len-2] == '.')
+ g_string_erase (str, len-2, 2);
- /* We dont need .* in the end of url. Thats stupid. */
- len = strlen (dst);
- if (dst && dst[len-1] == '*' && dst[len-2] == '.')
- {
- dst[len-2] = '\0';
- }
- return dst;
+ return str;
}
-static void
+static gboolean
uri_tester_compile_regexp (UriTester* tester,
- char* patt,
+ GString* gpatt,
char* opts)
{
GRegex* regex;
GError* error = NULL;
- int pos = 0;
- char *sig;
+ gchar *patt;
+ int len;
+ if (!gpatt)
+ return FALSE;
+
+ patt = gpatt->str;
+ len = gpatt->len;
+
+ /* TODO: Play with optimization flags */
regex = g_regex_new (patt, G_REGEX_OPTIMIZE,
G_REGEX_MATCH_NOTEMPTY, &error);
if (error)
{
g_warning ("%s: %s", G_STRFUNC, error->message);
g_error_free (error);
- return;
+ g_regex_unref (regex);
+ return TRUE;
}
if (!g_regex_match_simple ("^/.*[\\^\\$\\*].*/$", patt, G_REGEX_UNGREEDY, G_REGEX_MATCH_NOTEMPTY))
{
- int len = strlen (patt);
int signature_count = 0;
+ int pos = 0;
+ gchar *sig;
+
for (pos = len - SIGNATURE_SIZE; pos >= 0; pos--) {
sig = g_strndup (patt + pos, SIGNATURE_SIZE);
if (!g_regex_match_simple ("[\\*]", sig, G_REGEX_UNGREEDY, G_REGEX_MATCH_NOTEMPTY) &&
@@ -517,59 +512,141 @@ uri_tester_compile_regexp (UriTester* tester,
}
g_free (sig);
}
- if (signature_count > 1 && g_hash_table_lookup (tester->priv->pattern, opts))
- g_hash_table_steal (tester->priv->pattern, patt);
+ g_regex_unref (regex);
+
+ if (signature_count > 1 && g_hash_table_lookup (tester->priv->pattern, patt))
+ {
+ g_hash_table_steal (tester->priv->pattern, patt);
+ return TRUE;
+ }
+
+ return FALSE;
}
else
{
LOG ("patt: %s%s", patt, "");
- /* Pattern is a regexp chars. */
- g_hash_table_insert (tester->priv->pattern, g_strdup (patt), g_regex_ref (regex));
+ /* Pattern is a regexp chars */
+ g_hash_table_insert (tester->priv->pattern, g_strdup (patt), regex);
g_hash_table_insert (tester->priv->optslist, g_strdup (patt), g_strdup (opts));
+ return FALSE;
}
-
- g_regex_unref (regex);
}
static char*
uri_tester_add_url_pattern (UriTester* tester,
- char* format,
+ char* prefix,
char* type,
char* line)
{
- char** data;
- char* patt;
- char* fixed_patt;
- char* format_patt;
- char* opts;
+ gchar** data;
+ gchar* patt;
+ GString* format_patt;
+ gchar* opts;
+ gboolean should_free;
+
+ data = g_strsplit (line, "$", -1);
+ if (!data || !data[0])
+ {
+ g_strfreev (data);
+ return NULL;
+ }
- data = g_strsplit (line, "$", -1);
- if (data && data[0] && data[1] && data[2])
+ if (data[1] && data[2])
{
- patt = g_strdup_printf ("%s%s", data[0], data[1]);
- opts = g_strdup_printf ("type=%s,regexp=%s,%s", type, patt, data[2]);
+ patt = g_strconcat (data[0], data[1], NULL);
+ opts = g_strconcat (type, ",", data[2], NULL);
}
- else if (data && data[0] && data[1])
+ else if (data[1])
{
- patt = g_strdup (data[0]);
- opts = g_strdup_printf ("type=%s,regexp=%s,%s", type, patt, data[1]);
+ patt = data[0];
+ opts = g_strconcat (type, ",", data[1], NULL);
}
- else
+ else
{
- patt = g_strdup (data[0]);
- opts = g_strdup_printf ("type=%s,regexp=%s", type, patt);
+ patt = data[0];
+ opts = type;
}
- fixed_patt = uri_tester_fixup_regexp (patt);
- format_patt = g_strdup_printf (format, fixed_patt);
+ if (g_regex_match_simple ("subdocument", opts,
+ G_REGEX_CASELESS, G_REGEX_MATCH_NOTEMPTY))
+ {
+ if (data[1] && data[2])
+ g_free (patt);
+ if (data[1])
+ g_free (opts);
+ g_strfreev (data);
+ return NULL;
+ }
- LOG ("got: %s opts %s", format_patt, opts);
- uri_tester_compile_regexp (tester, format_patt, opts);
+ format_patt = uri_tester_fixup_regexp (prefix, patt);
+
+ LOG ("got: %s opts %s", format_patt->str, opts);
+ should_free = uri_tester_compile_regexp (tester, format_patt, opts);
+
+ if (data[1] && data[2])
+ g_free (patt);
+ if (data[1])
+ g_free (opts);
+ g_strfreev (data);
+
+ return g_string_free (format_patt, should_free);
+}
+
+static inline void
+uri_tester_frame_add (UriTester* tester, gchar* line)
+{
+ const gchar* separator = " , ";
+
+ (void)*line++;
+ (void)*line++;
+ if (strchr (line, '\'')
+ || (strchr (line, ':')
+ && !g_regex_match_simple (".*\\[.*:.*\\].*", line,
+ G_REGEX_CASELESS, G_REGEX_MATCH_NOTEMPTY)))
+ {
+ return;
+ }
+ g_string_append (tester->priv->blockcss, separator);
+ g_string_append (tester->priv->blockcss, line);
+}
+static inline void
+uri_tester_frame_add_private (UriTester* tester,
+ const gchar* line,
+ const gchar* sep)
+{
+ gchar** data;
+ data = g_strsplit (line, sep, 2);
+
+ if (!(data[1] && *data[1])
+ || strchr (data[1], '\'')
+ || (strchr (data[1], ':')
+ && !g_regex_match_simple (".*\\[.*:.*\\].*", data[1],
+ G_REGEX_CASELESS, G_REGEX_MATCH_NOTEMPTY)))
+ {
+ g_strfreev (data);
+ return;
+ }
+
+ if (strchr (data[0], ','))
+ {
+ gchar** domains;
+ gint i;
+
+ domains = g_strsplit (data[0], ",", -1);
+ for (i = 0; domains[i]; i++)
+ {
+ g_string_append_printf (tester->priv->blockcssprivate, ";sites['%s']+=',%s'",
+ g_strstrip (domains[i]), data[1]);
+ }
+ g_strfreev (domains);
+ }
+ else
+ {
+ g_string_append_printf (tester->priv->blockcssprivate, ";sites['%s']+=',%s'",
+ data[0], data[1]);
+ }
g_strfreev (data);
- g_free (patt);
- g_free (fixed_patt);
- return format_patt;
}
static char*
@@ -578,52 +655,56 @@ uri_tester_parse_line (UriTester* tester, char* line)
if (!line)
return NULL;
g_strchomp (line);
-
- /* Ignore comments and new lines. */
+ /* Ignore comments and new lines */
if (line[0] == '!')
return NULL;
-
- /* FIXME: No support for whitelisting. */
+ /* FIXME: No support for whitelisting */
if (line[0] == '@' && line[1] == '@')
return NULL;
-
- /* FIXME: No support for [include] and [exclude] tags. */
+ /* FIXME: No support for [include] and [exclude] tags */
if (line[0] == '[')
return NULL;
- /* Got CSS block hider. */
+ /* Skip garbage */
+ if (line[0] == ' ' || !line[0])
+ return NULL;
+
+ /* Got CSS block hider */
if (line[0] == '#' && line[1] == '#' )
{
+ uri_tester_frame_add (tester, line);
return NULL;
}
- /* Got CSS block hider. Workaround. */
+ /* Got CSS block hider. Workaround */
if (line[0] == '#')
return NULL;
- /* Got per domain CSS hider rule. */
+ /* Got per domain CSS hider rule */
if (strstr (line, "##"))
{
+ uri_tester_frame_add_private (tester, line, "##");
return NULL;
}
- /* Got per domain CSS hider rule. Workaround. */
+ /* Got per domain CSS hider rule. Workaround */
if (strchr (line, '#'))
{
+ uri_tester_frame_add_private (tester, line, "#");
return NULL;
}
- /* Got URL blocker rule. */
+ /* Got URL blocker rule */
if (line[0] == '|' && line[1] == '|' )
{
(void)*line++;
(void)*line++;
- return uri_tester_add_url_pattern (tester, "%s", "fulluri", line);
+ return uri_tester_add_url_pattern (tester, "", "fulluri", line);
}
if (line[0] == '|')
{
(void)*line++;
- return uri_tester_add_url_pattern (tester, "^%s", "fulluri", line);
+ return uri_tester_add_url_pattern (tester, "^", "fulluri", line);
}
- return uri_tester_add_url_pattern (tester, "%s", "uri", line);
+ return uri_tester_add_url_pattern (tester, "", "uri", line);
}
static gboolean
@@ -672,6 +753,9 @@ uri_tester_init (UriTester *tester)
(GDestroyNotify)g_free,
(GDestroyNotify)g_free);
+ priv->blockcss = g_string_new ("z-non-exist");
+ priv->blockcssprivate = g_string_new ("");
+
uri_tester_load_filters (tester);
uri_tester_load_patterns (tester);
}
@@ -710,6 +794,9 @@ uri_tester_finalize (GObject *object)
g_hash_table_destroy (priv->optslist);
g_hash_table_destroy (priv->urlcache);
+ g_string_free (priv->blockcss, TRUE);
+ g_string_free (priv->blockcssprivate, TRUE);
+
G_OBJECT_CLASS (uri_tester_parent_class)->finalize (object);
}
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]
