first commit

1 files changed, 160 insertions, 0 deletions

diff --git a/surf-webext-dom.c b/surf-webext-dom.c
new file mode 100644
index 0000000..31e1308
--- /dev/null
+++ b/surf-webext-dom.c
@@ -0,0 +1,160 @@
+//
+// !!! THIS IS A BUGGY CONCEPT/PROTOTYPE. USE AT YOUR OWN RISK !!!
+//
+// The following simple Webkit2GTK+ extension was hacked together
+// in the sleepless night between 09th and 10th March 2018 by:
+//
+// Michal Idziorek <m.i@gmx.at>.
+//
+// It regjsters a document_loaded_callback, where the DOM is manipulated,
+// i.e: a <span> with a number added to every link encountered.
+//
+// The urls are extracted from the DOM and shared with the browsers 
+// process over a system V memory segment. 
+//
+// As an example, this can be used to enable keyboardless navigation 
+// in surf by simply typing in the number of the link. 
+//
+// TODO: Support multiple clients! See source for more TODOS!
+//       THE biggest limitation by now is the lazily hardcoded:
+//       key_t my_ftok = ftok("~/surf-webext-dom-shared-mem",'a');
+//
+// Last Update: 10th March 2018
+// 
+// Last Tested: with surf commit 81f0452bc6c2a110239fa303ad1e72f11c33dc94
+//              from htps://git.suckless.org/surf
+// 
+
+#include <webkit2/webkit-web-extension.h>
+
+#include <sys/ipc.h>
+#include <sys/shm.h>
+#include <errno.h>
+#include <string.h>
+
+bool str_prefix(const char *str, const char *pre);
+
+void
+document_loaded_callback (WebKitWebPage *web_page,
+                          GPtrArray     *elements,
+                          GVariant      *user_data)
+{
+
+    // TODO: g_free where appropriate !!
+    // TODO: allow focus on input boxes/forms 
+    // TODO: Put this somehwere else? (but I dislike #defines)
+    // share this info with surf itself! maybe surf can supply this along
+    // the ftok path via the user_data GVariant.
+    int MAX_LINKS=1024;
+    int MAX_LINK_LENGTH=1024;
+    int DEBUG=3;
+
+    // get html doc
+    WebKitDOMHTMLDocument *doc	   = webkit_web_page_get_dom_document(web_page);
+    if(!WEBKIT_DOM_IS_HTML_DOCUMENT(doc))return;
+
+    // get url
+    char *url = webkit_dom_document_get_url (doc); 
+    if(DEBUG>1) g_print("document_loaded (callback): %s\n", url);
+
+    // tokenize url
+    char *url_proto=strtok(url,"/");
+    char *url_host=strtok(NULL,"/");
+    if(DEBUG>2) g_print("proto: %s\n", url_proto);
+    if(DEBUG>2) g_print("host: %s\n", url_host);
+
+    // get html body
+    WebKitDOMHTMLElement *body  = webkit_dom_document_get_body(doc);
+
+    // get links (alt)
+    /* altenative way, what is the difference (TODO)?
+    WebKitDOMNodeList *links = webkit_dom_document_query_selector_all(doc,"a",NULL);
+    gulong links_count		   = webkit_dom_node_list_get_length (links);
+    g_print("extracted %i links\n",links_count);
+    */
+
+    // get links 
+    WebKitDOMHTMLCollection *links = webkit_dom_document_get_links(doc);
+    gulong links_count		   = webkit_dom_html_collection_get_length (links);
+    if(DEBUG>1) g_print("extracted %i links\n",links_count);
+
+    // attach to shared memory;
+    key_t my_ftok = ftok("~/surf-webext-dom-shared-mem",'a');
+    
+    // TODO: should we need the size again here?
+    int mem_seg=shmget(my_ftok,MAX_LINKS*MAX_LINK_LENGTH,IPC_CREAT|0660);
+    if(mem_seg==-1)
+    {
+	g_print("shmget failed: %s\n",strerror(errno));
+    }
+
+    char *shared_buf=shmat(mem_seg,NULL,0);
+    if(shared_buf==(void*)-1)
+    {
+	g_print("shmat failed: %s\n",strerror(errno));
+    }
+    
+    // iterate over links
+    for(gulong i=0;i<links_count;i++)
+    {
+	char buf[MAX_LINK_LENGTH]; // TODO: dynamic!
+
+	WebKitDOMHTMLLinkElement * node = webkit_dom_html_collection_item (links,i);
+
+	// add hint for keyboard navigation
+	char *txt = webkit_dom_html_element_get_inner_html (node); 
+
+	// TODO: style via stylesheet
+	snprintf(buf,MAX_LINK_LENGTH, 
+		"<span style='padding-right:5px;border-bottom:1px solid black;'>\
+		<span style='font-size:13px;font-weight:bold;font-face:Arial;\
+		      border:1px solid black;display:inline;color:white;\
+		      background-color:#c0474c;padding:0 5px;margin:0;\
+		      margin-right:5px;'>%i</span>%s</span>", i,txt);
+
+	webkit_dom_html_element_set_inner_html (node,buf,NULL);
+
+	// alternatively add hints at the end of the body
+	// webkit_dom_element_insert_adjacent_html (body,"beforeend",buf,NULL);
+
+	// TODO: this does not work for some sad reason. 
+	// sadly, likely due to my misunderstanding.
+	// node seems not to be a LinkElement at all :(
+	// txt = webkit_dom_html_link_element_get_href (node); 
+	// so we have to use this:
+	txt = webkit_dom_element_get_attribute (node,"href");
+	if(str_prefix(txt,"//"))     snprintf(buf,MAX_LINK_LENGTH,"%s%s",url_proto,txt);
+	else if(str_prefix(txt,"/")) snprintf(buf,MAX_LINK_LENGTH,"%s//%s%s",url_proto,url_host,txt);
+	else                         snprintf(buf,MAX_LINK_LENGTH,"%s",txt);
+	
+	if(i<MAX_LINKS) snprintf(&(shared_buf[MAX_LINK_LENGTH*i]),MAX_LINK_LENGTH,"%s\0",buf);
+        if(DEBUG>10) g_print("extracted link : %s\n",buf);
+    }
+}
+
+static void
+page_created_callback (WebKitWebExtension *extension,
+                       WebKitWebPage      *web_page,
+                       GVariant           *user_data)
+{
+    g_signal_connect (web_page, "document-loaded", 
+                      G_CALLBACK (document_loaded_callback), 
+                      user_data);
+}
+
+G_MODULE_EXPORT void
+webkit_web_extension_initialize_with_user_data (WebKitWebExtension *extension,
+					        GVariant           *user_data)
+{
+    g_signal_connect (extension, "page-created", 
+                      G_CALLBACK (page_created_callback),
+                      user_data );
+}
+
+//https://stackoverflow.com/questions/4770985/how-to-check-if-a-string-starts-with-another-string-in-c
+bool str_prefix(const char *str, const char *pre)
+{
+    size_t lenpre = strlen(pre),
+           lenstr = strlen(str);
+    return lenstr < lenpre ? false : strncmp(pre, str, lenpre) == 0;
+}