sync
[app.webkitpdf] / BrowserView.js
1 Gtk = imports.gi.Gtk;
2 GLib = imports.gi.GLib;
3 WebKit = imports.gi.WebKit;
4 Soup = imports.gi.Soup;
5
6 TabbedBrowser = imports.TabbedBrowser;
7 BrowserSettings = imports.BrowserSettings;
8 BrowserTab = imports.BrowserTab;
9
10 File = imports.File.File;
11
12 base64 = imports.base64.base64;
13
14 BrowserView = new GType({
15     parent: WebKit.WebView.type,
16     name: "BrowserView",
17     init: function ()
18     {
19         // Private
20         
21         var sess  = WebKit.get_default_session();
22         sess.proxy_uri = new Soup.URI.c_new('http://localhost:3128');       
23         var _t = this;
24         
25         var tab = BrowserTab;
26         var browsePage = false;
27         var maxQueue = 0;
28         var injected = {};
29         
30         var current_url = '';
31  
32         var update_title = function (web_view, web_frame, title)
33         {
34             
35             print("calling update title");
36             if(title.length > 25)
37                 title = title.slice(0,25) + "...";
38
39             tab.get_tab_label().label = title;
40         };
41         
42         
43         
44            
45         
46         var traversedom = function(doc, fn, sub ) {
47             //print("TRAVERSE DOM?");
48             sub = sub || 0;
49             
50             var ret = '<HTML>';
51             
52             var cb =  function(s) {
53                 //print("CB:" + s);
54                 ret+=s;
55             };
56             cb.fn = fn;
57             cb.sub = sub +1;
58             print(doc);
59             print(doc.head);
60             print (doc.body);
61             
62              //Roo.select('body > div',true).each(function(el) {
63             traverseDOMTree(cb, doc.head, 1);
64             traverseDOMTree(cb, doc.body, 1);
65            //print(ret);   
66             File.write(cb.fn.replace(/\.html$/, '')+ (sub ? ('.' + sub  ) : '') + '.html' , ret +'</HTML>');
67         
68         };
69          
70         
71         
72         var traverseDOMTree = function(cb, currentElement, depth) {
73             if (currentElement) {
74                 
75                 //if (currentElement.class_name.match(/roo-dynamic/)) {
76                 //    return;
77                 //}
78                 //print(currentElement.node_name);
79                 var j;
80                 var nodeName = currentElement.node_name;
81                 var tagName = currentElement.tag_name;
82                 
83                 if  (nodeName == '#text') {
84                     cb(currentElement.node_value);
85                     return;
86                 
87                 }
88                 if(nodeName == 'BR'){
89                     cb("<BR/>");
90                     return;
91                 }
92                
93                 if (nodeName == 'SCRIPT') {
94                      
95                     return;
96                 }
97                 //if (nodeName == 'STYLE') {  return; }
98                 var i = 0;
99               // Prints the node tagName, such as <A>, <IMG>, etc
100                 var outNodeName = nodeName;
101                 if (nodeName == 'IFRAME') {
102                     //outNodeName  = 'DIV';
103                 }
104                 if (tagName) {
105                     
106                     
107                     
108                     
109                     var attr = [];
110                     
111                     for(i = 0; i < currentElement.attributes.length;i++) {
112                         var aname = currentElement.attributes.item(i).name;
113                         //if (aname =='class' || aname == 'style') {
114                         //    continue;
115                         //}
116                         if (aname =='src' && tagName == 'IFRAME') {
117                             continue;
118                         }
119                         
120                         attr.push(aname + '="' + currentElement.attributes.item(i).value + '"' );
121                     }
122                     if (tagName == 'IFRAME') {
123                         attr.push('src="' + File.basename(cb.fn).replace(/\.html$/, '')+ '.' + cb.sub + '.html"' );
124                     }
125                     
126                     //if (nodeName == 'IFRAME') {
127                     //    attr.push('src="' + currentElement.attributes.item(i).value + embeded_file '"' );
128                     //}
129                     
130                     //var style = elClassToStyle(currentElement);
131                     //if (style.length) {
132                     //    attr.push('style="' + style + '"' );
133                     //}
134                     
135                     
136                     //if (depth > 1000 && (tagName == 'BODY' || tagName == 'HEAD' )) {
137                      //   cb("<DIV"+ ( attr.length ? (' ' + attr.join(' ') ) : '') + ">");
138                     //} else {
139                         cb("<"+outNodeName + ( attr.length ? (' ' + attr.join(' ') ) : '') + ">");
140                         
141                     //}
142                 } 
143                 else {
144                   cb("[unknown tag]");
145                 }
146                 
147                  if (nodeName == 'IFRAME') {
148                     
149                    
150                     traversedom(currentElement.content_document,  cb.fn,  cb.sub )
151                     cb.sub++;
152                     cb("</IFRAME>");
153                     return;
154                 }
155                 
156                 
157                 // Traverse the tree
158                 i = 0;
159                 var currentElementChild = currentElement.child_nodes.item(i);
160                 var allText = true;
161                 while (currentElementChild) {
162                     // Formatting code (indent the tree so it looks nice on the screen)
163                     
164                     if  (currentElementChild.node_name == '#text') {
165                        // if (currentElementChild.node_value.length) {
166                             cb(currentElementChild.node_value);
167                        // }
168                         
169                         i++;
170                         currentElementChild=currentElement.child_nodes.item(i);
171                         continue;
172                     }   
173                     allText = false;
174                     //cb("\n");
175                     //for (j = 0; j < depth; j++) {
176                       // &#166 is just a vertical line
177                     //  cb("  ");
178                     //}//               
179                     
180                         
181                     // Recursively traverse the tree structure of the child node
182                     traverseDOMTree(cb, currentElementChild, depth+1);
183                     i++;
184                     currentElementChild=currentElement.child_nodes.item(i);
185                 }
186                 if (!allText) {
187                     // The remaining code is mostly for formatting the tree
188                     //cb("\n");
189                     //for (j = 0; j < depth - 1; j++) {
190                       //cb("  ");
191                     //}     
192                 }
193                 if (tagName) {
194                    // if (depth > 1000 && (tagName == 'BODY' || tagName == 'HEAD' )) {
195                    //     cb("</DIV>");
196                    // } else {
197                         cb("</"+outNodeName+">");
198                    // }
199                     
200                 }
201             }
202         };
203         
204
205         var update_url = function (web_view, web_frame)
206         {
207             var toolbar = tab.get_toolbar();
208
209             toolbar.set_url(web_frame.get_uri());
210             toolbar.set_can_go_back(web_view.can_go_back());
211             toolbar.set_can_go_forward(web_view.can_go_forward());
212             
213             
214             
215             
216         };
217
218     
219       
220
221         var create_new_tab = function (web_view, web_frame, new_web_view)
222         {
223             new_web_view = new BrowserView();
224             new_web_view.signal.web_view_ready.connect(show_new_tab);
225             return new_web_view;
226         };
227
228         var show_new_tab = function (new_web_view)
229         {
230             TabbedBrowser.browser.new_tab("", new_web_view);
231
232             return false;
233         };
234
235         var hover_link = function (web_view, link, url)
236         {
237             tab.get_statusbar().set_status(url);
238         };
239          
240         /*
241         
242         this.add_inject = function(force)
243         {
244             
245             if (force || (typeof(injected[this.uri]) == 'undefined' )) {
246                 injected[this.uri] = 0;
247             }
248             if (injected[this.uri] > 2) {
249                 return;
250             }
251             injected[this.uri]++;
252             var fn = __script_path__ + "/inject.js";
253             if (File.exists(fn)) {
254 //                print("Adding inject");
255                 var newjs = File.read(__script_path__ + "/inject.js");
256                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
257                     newjs
258                     
259                 );
260             }
261             
262         }
263         */
264         var after_login = false;
265         
266         
267         var load_finished_called = false;
268         
269         
270         var do_print = function(web_frame) {
271             
272              
273             
274             var scr_a =  " var r = document.getElementsByTagName('link');" +
275                 "for (var i=0;i < r.length;i++) { " +
276                     "var a = r[i]; " +
277                     "if (  a.getAttribute('media') == '')  { continue; } " +
278                     "if (  a.getAttribute('media') == 'screen')  { "+
279                         "a.removeAttribute('media'); continue;" +
280                     "} " +
281                     //"a.parentNode.removeChild(a);" +
282                 "}";
283             var scr_b =  "var a = document.body.querySelectorAll('*'); "+
284                 "var dnodes=[]; "+
285                 "for(var i =i;i<a.length;i++) { "+
286                 "   var cs = window.getComputedStyle(a[i],null);"+
287                 "   if ('none' == cs.display) {"+
288                 "    dnodes.push(a[i]); "+
289                 "  }"+ 
290                 "}"+
291                 "console.log(dnodes.length);" +
292                 "dnodes.forEach(function(n) { "+
293                 "    try {  n.parentNode.removeChild(n);   } catch(e) { } "+
294                 "});";
295             
296             
297             //TabbedBrowser.browser.current_tab().get_web_view().execute_script(
298             //    scr_a
299             //);
300             //print(scr_b);
301             //TabbedBrowser.browser.current_tab().get_web_view().execute_script(
302             //    scr_b
303             //);
304             
305             if (current_url.match(/\.coconuts\.co\//) ) {
306                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
307                     File.read( __script_path__ + "/domains/coconuts.co.js")
308                 );
309                               
310             }
311             // motorme - requires a scroll to make content appear..
312             if (current_url.match(/\.motorme\.my\//) ) {
313                 TabbedBrowser.browser.current_tab().get_web_view().execute_script("window.scrollTo(0,500);"); 
314                               
315             }
316             
317             
318            
319             print("Delay:" + BrowserSettings.delay);
320          
321             GLib.timeout_add(GLib.PRIORITY_LOW, BrowserSettings.delay *1, function() {
322                  
323                 print("FETCHING");
324                 //print(web_frame);
325                 if (web_frame &&  BrowserSettings.export_filename_html  ) {
326                     var html = traversedom(web_frame.get_dom_document(), BrowserSettings.export_filename_html);
327                     //File.write(BrowserSettings.export_filename_html, html);
328                     
329                 } 
330                  
331                 if (!BrowserSettings.export_filename) {
332                     if (!BrowserSettings.export_filename_html) {
333                         return;
334                     }
335                     
336                     Seed.quit();
337                 }
338                 var mf = _t.get_main_frame();
339                  
340                 
341                 var ar = Gtk.PaperSize.get_paper_sizes();
342                 var psetup = new Gtk.PageSetup();
343                 for(var i = 0; i < ar.length; i++) {
344                     if (ar[i].get_name() =='iso_a2') {
345                         psetup.set_paper_size(ar[i]);
346                     }
347                 }
348                 
349                 var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
350                 p.set_default_page_setup(psetup);
351                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
352                 print("made image - exiting");
353                 
354                 Seed.quit();
355                 return true;
356             });
357         }
358         
359         
360         var weibo_added = false;
361         
362         var is_weibo = false;
363         
364         
365         var load_finished = function (webkit, web_frame, wb)
366         {
367             
368            
369             
370             if (!is_weibo) {
371                 do_print(web_frame);
372                 return;
373             }
374              
375              
376              print("load finished");
377              if (weibo_added) {
378                  
379                  if (after_login === true) {
380                      print("do print");
381                      do_print();
382                      return;
383                      
384                  }
385                  if (after_login === false) {
386                      return; // not here?
387                  }
388                  print("adding timeout?");
389                  //return;
390                 GLib.timeout_add(GLib.PRIORITY_LOW, 2000, function() {
391                     print("Redirecting after login?" + after_login);
392                     
393                     _t.browse(after_login);
394                     after_login  = true;
395                 });
396                 return;
397              }
398             
399             weibo_added = true;
400             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
401                 File.read( __script_path__ + "/weibo.js") 
402             );
403             print("run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");");
404     
405             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
406                     "run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");"
407             );
408             return;
409     
410             //    return;
411             print(after_login);
412             if (after_login !== false) {
413                 
414 //                return;
415                 // wait a bit then load the real page..
416                 GLib.timeout_add(GLib.PRIORITY_LOW, 500, function() {
417                     print("Redirecting after login?");
418                     _t.browse(after_login);
419                 });
420                 
421                 return;
422             }
423          
424             if (load_finished_called) {
425                 return ;
426             
427             }
428             load_finished_called = true;
429             
430           
431             
432 //            if(document.location.host=='weibo.com') {
433 //                // clear login dialog from weibo.
434 //                //check and hidden the login dialog and overlay .....
435 //                var els = document.querySelectorAll('div[node-type]');
436 //                var bbc = [];
437 //                for (i = 0; i < els.length; i++) {
438 //                    if (els[i].hasAttribute('node-type')) {
439 //                        if(els[i].getAttribute('node-type') == 'outer'){
440 //                            bbc.push(els[i]);
441 //                        }
442 //                    }
443 //                }
444 //                for (i = 0; i < bbc.length; i++) {// hide all the outer.....
445 //                    bbc[i].style.display = 'none';
446 //                }
447 //                
448 //                //try to login 
449 //                if(document.location.pathname == '/login.php'){
450 //                    
451 //                    // fills in the user name and password 
452 //                    var els = document.querySelectorAll('input[node-type]');
453 //                    
454 //                    for (i = 0; i < els.length; i++) {
455 //                        if (els[i].hasAttribute('node-type')) {
456 //                            if(els[i].getAttribute('node-type') == 'username'){
457 //                                els[i].value = BrowserSettings.username;
458 //                            }
459 //                            if(els[i].getAttribute('node-type') == 'password'){
460 //                                els[i].value = BrowserSettings.passwd;
461 //                            }
462 //                        }
463 //                    }
464 //                    
465 //                    
466 //                    // simulating click event to login ....
467 //                    var elss = document.querySelectorAll('a[node-type]');
468 //                    
469 //                    for (i = 0; i < elss.length; i++) {
470 //                        if (elss[i].hasAttribute('node-type')) {
471 //                            if(elss[i].getAttribute('node-type') == 'submitBtn'){
472 //                                elss[i].click();
473 //                            }
474 //                        }
475 //                    }
476 //                    
477 //                    
478 //                }
479 //                
480 //            }
481             
482             
483             // clear login dialog from weibo.
484 //            TabbedBrowser.browser.current_tab().get_web_view().execute_script(
485 //                " if (document.location.host=='weibo.com') { " +
486 //                " try {  " + 
487 //                " var a = document.getElementsByClassName('W_layer')[0]; " + 
488 //                " a.parentNode.removeChild(a.previousSibling); " + 
489 //                " a.parentNode.removeChild(a); " + 
490 //                " } catch(e) { }  } " );
491             
492             
493             return;
494             GLib.timeout_add(GLib.PRIORITY_LOW, BrowserSettings.delay, function() {
495                  
496                 var mf = _t.get_main_frame();
497                 
498                 var ar = Gtk.PaperSize.get_paper_sizes();
499                 var psetup = new Gtk.PageSetup();
500                 for(var i = 0; i < ar.length; i++) {
501                     if (ar[i].get_name() =='iso_a2') {
502                         psetup.set_paper_size(ar[i]);
503                     }
504                 }
505                 
506                 var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
507                 p.set_default_page_setup(psetup);
508                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
509                 print("made image - exiting");
510                 
511                 Seed.quit();
512                 return true;
513             });
514             
515             return;
516         };
517  
518  
519  
520         
521         var load_committed = function (web_view, web_frame)
522         {
523             print("load commited");
524             
525             update_url(web_view, web_frame);
526              
527             // call load finished after 20seconds??
528             // so even if it never complets we try and print the thing.
529             GLib.timeout_add(GLib.PRIORITY_LOW, 40000, function() {
530                 print("load commited - 3000 ms?");
531                 load_finished();
532             });
533             
534         };
535
536         var clicked_link = function (web_view, web_frame, request,
537                                      action, decision, window)
538         {
539             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
540                action.get_button() == 2)
541             {
542                 browser.new_tab(request.get_uri(), null);
543                 return true;
544             }
545
546             return false;
547         };
548
549         // Public
550         
551         
552         this.browse = function (url)
553         {
554             if(url.search("://") < 0)
555                 url = "http://" + url;
556             
557             current_url = url;
558             
559             print("BROWSE: " + url);
560             if (url.match(/\/weibo\.com\//) && after_login === false) {
561                 is_weibo = true;
562                 after_login = url;
563                 url = "http://weibo.com/login.php";
564             }
565             else  if (url.match(/\/weibo\.com\//) && after_login !== false) {
566                 after_login = false;
567             }
568             print("BROWSE -really: " + url);
569             this.open(url);
570         };
571
572         this.set_tab = function (new_tab)
573         {
574             tab = new_tab;
575         };
576
577         this.get_tab = function ()
578         {
579             return tab;
580         };
581
582     
583         // Implementation
584         //this.set_scroll_adjustments(null, null);
585         
586         //this.signal.title_changed.connect(update_title);
587         //this.signal.load_committed.connect(load_committed);
588         this.signal.load_finished.connect(load_finished);
589        
590         
591         // For some reason, this segfaults seed in the instance init closure handler
592         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
593         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
594
595         //this.signal.hovering_over_link.connect(hover_link);
596
597         this.signal.create_web_view.connect(create_new_tab);
598         
599         
600          
601 //        print("ADDing console message sig handler");
602         
603         
604          
605         
606         this.toFilename = function(url)
607         {
608             url = url.replace(/^http[s]*:\/\//, '');
609             var p = url.split('/');
610             p.unshift(storedir+'/output');
611             for (var i =1 ;i < p.length; i++) {
612                 p[i] = encodeURIComponent(p[i]);
613             
614             }
615             p[p.length-1] = decodeURIComponent(p[p.length-1]);
616             ret = p.join('/');
617             var dir = File.dirname(ret);
618             File.mkdirall(dir);
619             return ret;
620             
621         }
622         this.checkdomain = function(comp)
623         {
624             var b = parseUri(this.uri);
625             var d = parseUri(comp);
626             return (d.host == b.host && d.protocol == b.protocol);
627             
628             
629         }
630         
631         this.dupeCheck = function(url)
632         {
633             
634            // order - return highest up the queue first..
635             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
636                 return downloaddir +'/' + encodeURIComponent(url);
637             }
638              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
639                 return parsedir +'/' + encodeURIComponent(url);
640             }
641             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
642                 return donedir +'/' + encodeURIComponent(url);
643             }
644             return  false;
645             
646             
647         }
648         this.moveToParse = function(url)
649         {
650             var old = this.dupeCheck(url);
651             var target =parsedir +'/' + encodeURIComponent(url);
652             if (old == target) {
653                 return;
654             }
655             File.write(target, old ? File.read(old) : '');
656             if (old) {
657                 File.remove(old);
658             }
659             
660         }
661         
662         this.moveToDownload= function(url)
663         {
664             var old = this.dupeCheck(url);
665             var target =downloaddir +'/' + encodeURIComponent(url);
666             if (old == target) {
667                 return;
668             }
669             File.write(target, old ? File.read(old) : '');
670             if (old) {
671                 File.remove(old);
672             }
673             
674         }
675         this.moveToDone= function(url)
676         {
677             var old = this.dupeCheck(url);
678             var target = donedir +'/' + encodeURIComponent(url);
679             if (old == target) {
680                 return;
681             }
682             File.write(target, old ? File.read(old) : '');
683             if (old) {
684                 File.remove(old);
685             }
686             
687         }
688         
689     }
690 });
691
692 function parseUri (str) {
693         var     o   = parseUri.options,
694                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
695                 uri = {},
696                 i   = 14;
697
698         while (i--) uri[o.key[i]] = m[i] || "";
699
700         uri[o.q.name] = {};
701         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
702                 if ($1) uri[o.q.name][$1] = $2;
703         });
704
705         return uri;
706 };
707
708 parseUri.options = {
709         strictMode: false,
710         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
711         q:   {
712                 name:   "queryKey",
713                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
714         },
715         parser: {
716                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
717                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
718         }
719 };