added url filtering

Makefile

@@ -22,11 +22,18 @@ options:
 surf: $(OBJ)
 	$(CC) $(SURFLDFLAGS) $(LDFLAGS) -o $@ $(OBJ) $(LIBS)
 
-$(OBJ) $(WOBJ): config.h config.mk
+$(OBJ) $(WOBJ): config.h config.mk filters_compiled
 
 config.h:
 	cp config.def.h $@
 
+filters_compiled: filters
+	sed -e '/^$$/d' -e 's|\\|\\\\|g' -e 's|$$|",|' -e 's|^|"|' < filters > $@
+
+filters:
+	@echo creating $@ from filters.def
+	@cp filters.def $@
+
 $(OBJ): $(SRC)
 	$(CC) $(SURFCFLAGS) $(CFLAGS) -c $(SRC)
 
@@ -41,7 +48,7 @@ clean:
 	rm -f $(WLIB) $(WOBJ)
 
 distclean: clean
-	rm -f config.h surf-$(VERSION).tar.gz
+	rm -f config.h surf-$(VERSION).tar.gz filters_compiled
 
 dist: distclean
 	mkdir -p surf-$(VERSION)

config.def.h

@@ -19,6 +19,14 @@ static SearchEngine searchengines[] = {
 	{ "-", "https://searx.ox2.fr/search?q=%s" },
 };
 
+/* Regular expressions to match URLs that should not be loaded */
+char *filter_patterns[] = {
+	#include "filters_compiled"
+};
+/* Define this for verbose filtering */
+// #define FILTER_VERBOSE
+
+
 /* Webkit default features */
 /* Highest priority value will be used.
  * Default parameters are priority 0

filters

@@ -0,0 +1,2 @@
+^eviladvertisments\.com$
+/favicon\.ico$

filters_compiled

@@ -0,0 +1,2 @@
+"^eviladvertisments\\.com$",
+"/favicon\\.ico$",

surf-0.6-navhist.diff

@@ -0,0 +1,138 @@
+diff --git a/config.def.h b/config.def.h
+index a221c86..9840736 100644
+--- a/config.def.h
++++ b/config.def.h
+@@ -32,6 +32,16 @@ static Bool hidebackground  = FALSE;
+ 	} \
+ }
+ 
++#define SELNAV { \
++	.v = (char *[]){ "/bin/sh", "-c", \
++		"prop=\"`xprop -id $0 _SURF_HIST" \
++		" | sed -e 's/^.[^\"]*\"//' -e 's/\"$//' -e 's/\\\\\\n/\\n/g'" \
++		" | dmenu -i -l 10`\"" \
++		" && xprop -id $0 -f _SURF_NAV 8s -set _SURF_NAV \"$prop\"", \
++		winid, NULL \
++	} \
++}
++
+ /* DOWNLOAD(URI, referer) */
+ #define DOWNLOAD(d, r) { \
+ 	.v = (char *[]){ "/bin/sh", "-c", \
+@@ -67,6 +77,7 @@ static Key keys[] = {
+ 
+     { MODKEY,               GDK_l,      navigate,   { .i = +1 } },
+     { MODKEY,               GDK_h,      navigate,   { .i = -1 } },
++    { MODKEY|GDK_SHIFT_MASK,GDK_h,      selhist,    SELNAV },
+ 
+     { MODKEY,               GDK_j,           scroll_v,   { .i = +1 } },
+     { MODKEY,               GDK_k,           scroll_v,   { .i = -1 } },
+diff --git a/surf.c b/surf.c
+index cebd469..8b6d751 100644
+--- a/surf.c
++++ b/surf.c
+@@ -32,7 +32,7 @@ char *argv0;
+ #define COOKIEJAR_TYPE          (cookiejar_get_type ())
+ #define COOKIEJAR(obj)          (G_TYPE_CHECK_INSTANCE_CAST ((obj), COOKIEJAR_TYPE, CookieJar))
+ 
+-enum { AtomFind, AtomGo, AtomUri, AtomLast };
++enum { AtomFind, AtomGo, AtomUri, AtomHist, AtomNav, AtomLast };
+ 
+ typedef union Arg Arg;
+ union Arg {
+@@ -137,6 +137,8 @@ static void loadstatuschange(WebKitWebView *view, GParamSpec *pspec,
+ 		Client *c);
+ static void loaduri(Client *c, const Arg *arg);
+ static void navigate(Client *c, const Arg *arg);
++static void selhist(Client *c, const Arg *arg);
++static void navhist(Client *c, const Arg *arg);
+ static Client *newclient(void);
+ static void newwindow(Client *c, const Arg *arg, gboolean noembed);
+ static void pasteuri(GtkClipboard *clipboard, const char *text, gpointer d);
+@@ -649,6 +651,59 @@ navigate(Client *c, const Arg *arg) {
+ 	webkit_web_view_go_back_or_forward(c->view, steps);
+ }
+ 
++static void
++selhist(Client *c, const Arg *arg) {
++	WebKitWebBackForwardList *lst;
++	WebKitWebHistoryItem *cur;
++	gint i;
++	gchar *out;
++	gchar *tmp;
++	gchar *line;
++
++	out = g_strdup("");
++
++	if(!(lst = webkit_web_view_get_back_forward_list(c->view)))
++		return;
++
++	for(i = webkit_web_back_forward_list_get_back_length(lst); i > 0; i--) {
++		if(!(cur = webkit_web_back_forward_list_get_nth_item(lst, -i)))
++			break;
++		line = g_strdup_printf("%d: %s\n", -i,
++		                       webkit_web_history_item_get_original_uri(cur));
++		tmp = g_strconcat(out, line, NULL);
++		g_free(out);
++		out = tmp;
++	}
++
++	if((cur = webkit_web_back_forward_list_get_nth_item(lst, 0))) {
++		line = g_strdup_printf("%d: %s", 0,
++		                       webkit_web_history_item_get_original_uri(cur));
++		tmp = g_strconcat(out, line, NULL);
++		g_free(out);
++		out = tmp;
++	}
++
++	for(i = 1; i <= webkit_web_back_forward_list_get_forward_length(lst); i++) {
++		if(!(cur = webkit_web_back_forward_list_get_nth_item(lst, i)))
++			break;
++		line = g_strdup_printf("\n%d: %s", i,
++		                       webkit_web_history_item_get_original_uri(cur));
++		tmp = g_strconcat(out, line, NULL);
++		g_free(out);
++		out = tmp;
++	}
++
++	setatom(c, AtomHist, out);
++	g_free(out);
++	spawn(c, arg);
++}
++
++static void
++navhist(Client *c, const Arg *arg) {
++	Arg a = { .i = atoi(arg->v) };
++	navigate(c, &a);
++}
++
+ static Client *
+ newclient(void) {
+ 	Client *c;
+@@ -805,6 +860,7 @@ newclient(void) {
+ 
+ 	setatom(c, AtomFind, "");
+ 	setatom(c, AtomUri, "about:blank");
++	setatom(c, AtomHist, "");
+ 	if(hidebackground)
+ 		webkit_web_view_set_transparent(c->view, TRUE);
+ 
+@@ -923,6 +979,9 @@ processx(GdkXEvent *e, GdkEvent *event, gpointer d) {
+ 				arg.v = getatom(c, AtomGo);
+ 				loaduri(c, &arg);
+ 				return GDK_FILTER_REMOVE;
++			} else if(ev->atom == atoms[AtomNav]) {
++				arg.v = getatom(c, AtomNav);
++				navhist(c, &arg);
+ 			}
+ 		}
+ 	}
+@@ -1004,6 +1063,8 @@ setup(void) {
+ 	atoms[AtomFind] = XInternAtom(dpy, "_SURF_FIND", False);
+ 	atoms[AtomGo] = XInternAtom(dpy, "_SURF_GO", False);
+ 	atoms[AtomUri] = XInternAtom(dpy, "_SURF_URI", False);
++	atoms[AtomHist] = XInternAtom(dpy, "_SURF_HIST", False);
++	atoms[AtomNav] = XInternAtom(dpy, "_SURF_NAV", False);
+ 
+ 	/* dirs and files */
+ 	cookiefile = buildpath(cookiefile);

surf.c

@@ -70,6 +70,8 @@
 #define LENGTH(x)			   (sizeof(x) / sizeof(x[0]))
 #define CLEANMASK(mask)		 (mask & (MODKEY|GDK_SHIFT_MASK))
 
+regex_t *filter_expressions;
+
 enum { AtomFind, AtomGo, AtomUri, AtomUTF8, AtomLast };
 
 enum {
@@ -287,6 +289,10 @@ static void toggleinspector(Client *c, const Arg *a);
 static void find(Client *c, const Arg *a);
 static void insert(Client *c, const Arg *a);
 
+/* filter url */
+static bool filter_init(void);
+static bool filter_request(const gchar *uri);
+
 /* Buttons */
 static void clicknavigate(Client *c, const Arg *a, WebKitHitTestResult *h);
 static void clicknewwindow(Client *c, const Arg *a, WebKitHitTestResult *h);
@@ -744,6 +750,49 @@ die(const char *errstr, ...)
 	   exit(1);
 }
 
+static bool
+filter_init(void) {
+	bool errors = false;
+	char *errorbuf;
+
+	errorbuf = malloc(sizeof(char) * BUFSIZ);
+	filter_expressions = malloc(sizeof(regex_t) * LENGTH(filter_patterns));
+
+	for (off_t idx = 0; idx < LENGTH(filter_patterns); idx++) {
+		char *pat = filter_patterns[idx];
+		int err = regcomp(&filter_expressions[idx], pat,
+				            REG_EXTENDED | REG_ICASE | REG_NOSUB);
+		if (err != 0) {
+			/* regerror always ends messages with 0x00 */
+			(void) regerror(err, &filter_expressions[idx], errorbuf, BUFSIZ);
+			fprintf(stderr, "Failed to compile \"%s\": %s\n", pat, errorbuf);
+			errors = true;
+		}
+	}
+
+	free(errorbuf);
+	return !errors;
+}
+
+static bool
+filter_request(const gchar *uri) {
+	if (!strcmp(uri, "about:blank"))
+		return false;
+	for (off_t idx = 0; idx < LENGTH(filter_patterns); idx++) {
+		if (regexec(&filter_expressions[idx], uri, 0, NULL, 0) == REG_NOMATCH) {
+			continue;
+		}
+#ifdef FILTER_VERBOSE
+		fprintf(stderr, "filtering \"%s\"\n", uri);
+#endif
+		return true;
+	}
+#ifdef FILTER_VERBOSE
+	fprintf(stderr, "not filtering \"%s\"\n", uri);
+#endif
+	return false;
+}
+
 void
 usage(void)
 {
@@ -849,6 +898,10 @@ setup(void)
 				uriparams[i].config[j] = defconfig[j];
 		}
 	}
+
+	if (!filter_init()) {
+		die("Failed to compile one or more filter expressions\n");
+	}
 }
 
 void
@@ -2225,7 +2278,7 @@ decideresource(WebKitPolicyDecision *d, Client *c)
 		webkit_response_policy_decision_get_response(r);
 	const gchar *uri = webkit_uri_response_get_uri(res);
 
-	if (g_str_has_suffix(uri, "/favicon.ico")) {
+	if (filter_request(uri)) {
 		webkit_policy_decision_ignore(d);
 		return;
 	}