git.roojs.org Git - app.webkitpdf/blob - 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         var done_print = false;
 270         var do_print = function(web_frame) {
 271
 272
 273             if (done_print) {
 274                 print("do_print called -- while already doing this...");
 275                 return;
 276             }
 277             done_print = true;
 278             print("do_print called");
 279
 280
 281             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 282                     File.read( __script_path__ + "/domains/remove_print_css.js")
 283             );
 284
 285             var url_p = current_url.split('/');
 286             var dom = url_p[2];
 287             var dom_p = url_p[2].split('.');
 288             var chk = dom_p.pop();
 289             chk = dom_p.pop() + '.' + chk;
 290
 291
 292
 293             print("CHECK: " + chk +"\n");
 294             if (File.exists( __script_path__ + "/domains/" + chk + '.js')) {
 295                 print("INCLUDING: " + __script_path__ + "/domains/"+ chk + ".js\n");
 296                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 297                     File.read( __script_path__ + "/domains/"+ chk + ".js")
 298                 );
 299             }
 300
 301
 302             // motorme - requires a scroll to make content appear..
 303             if (current_url.match(/\.motorme\.my\//) ) {
 304                 TabbedBrowser.browser.current_tab().get_web_view().execute_script("window.scrollTo(0,500);");
 305
 306             }
 307
 308             if (current_url.match('/forbes.com')) {
 309                 v
 310             }
 311
 312
 313             print("Delay:" + BrowserSettings.delay);
 314
 315
 316             var realprint = function() {
 317
 318                 try {
 319
 320                     if (current_url.match(/theindependent\.sg\//) ) {
 321                         TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 322                             File.read( __script_path__ + "/domains/theindependent.sg.js")
 323                         );
 324
 325                     }
 326                     print("FETCHING");
 327                     //print(web_frame);
 328                     if (web_frame &&  BrowserSettings.export_filename_html  ) {
 329                         var html = traversedom(web_frame.get_dom_document(), BrowserSettings.export_filename_html);
 330                         //File.write(BrowserSettings.export_filename_html, html);
 331
 332                     }
 333
 334                     if (!BrowserSettings.export_filename) {
 335                         if (!BrowserSettings.export_filename_html) {
 336                              print("no export filename / and html filename");
 337                             return;
 338                         }
 339
 340                         Seed.quit();
 341                     }
 342                     var mf = _t.get_main_frame();
 343
 344                     print("Creating paper");
 345                     var ar = Gtk.PaperSize.get_paper_sizes();
 346                     var psetup = new Gtk.PageSetup();
 347                     for(var i = 0; i < ar.length; i++) {
 348                         if (ar[i].get_name() =='iso_a2') {
 349                             psetup.set_paper_size(ar[i]);
 350                         }
 351                     }
 352                     print("doing print operation");
 353                     var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
 354                     print("doing print operation - set page");
 355
 356                     p.set_default_page_setup(psetup);
 357
 358                     print("doing print operation - print_full");
 359                     mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
 360                     print("made image - exiting");
 361
 362                     Seed.quit();
 363                     return true;
 364                 } catch(e) {
 365                     print("error occured");
 366                     print(JSON.stringify(e));
 367                 }
 368             };
 369
 370
 371             GLib.timeout_add(GLib.PRIORITY_DEFAULT, BrowserSettings.delay *1, realprint);
 372         }
 373
 374
 375         var weibo_added = false;
 376
 377         var is_weibo = false;
 378
 379
 380         var load_finished = function (webkit, web_frame, wb)
 381         {
 382
 383
 384
 385             if (!is_weibo) {
 386                 do_print(web_frame);
 387                 return;
 388             }
 389
 390
 391              print("load finished");
 392              if (weibo_added) {
 393
 394                  if (after_login === true) {
 395                      print("do print");
 396                      do_print();
 397                      return;
 398
 399                  }
 400                  if (after_login === false) {
 401                      return; // not here?
 402                  }
 403                  print("adding timeout?");
 404                  //return;
 405                 GLib.timeout_add(GLib.PRIORITY_LOW, 2000, function() {
 406                     print("Redirecting after login?" + after_login);
 407
 408                     _t.browse(after_login);
 409                     after_login  = true;
 410                 });
 411                 return;
 412              }
 413
 414             weibo_added = true;
 415             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 416                 File.read( __script_path__ + "/weibo.js")
 417             );
 418             print("run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");");
 419
 420             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 421                     "run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");"
 422             );
 423             return;
 424
 425             //    return;
 426             print(after_login);
 427             if (after_login !== false) {
 428
 429 //                return;
 430                 // wait a bit then load the real page..
 431                 GLib.timeout_add(GLib.PRIORITY_LOW, 500, function() {
 432                     print("Redirecting after login?");
 433                     _t.browse(after_login);
 434                 });
 435
 436                 return;
 437             }
 438
 439             if (load_finished_called) {
 440                 return ;
 441
 442             }
 443             load_finished_called = true;
 444
 445
 446
 447 //            if(document.location.host=='weibo.com') {
 448 //                // clear login dialog from weibo.
 449 //                //check and hidden the login dialog and overlay .....
 450 //                var els = document.querySelectorAll('div[node-type]');
 451 //                var bbc = [];
 452 //                for (i = 0; i < els.length; i++) {
 453 //                    if (els[i].hasAttribute('node-type')) {
 454 //                        if(els[i].getAttribute('node-type') == 'outer'){
 455 //                            bbc.push(els[i]);
 456 //                        }
 457 //                    }
 458 //                }
 459 //                for (i = 0; i < bbc.length; i++) {// hide all the outer.....
 460 //                    bbc[i].style.display = 'none';
 461 //                }
 462 //
 463 //                //try to login
 464 //                if(document.location.pathname == '/login.php'){
 465 //
 466 //                    // fills in the user name and password
 467 //                    var els = document.querySelectorAll('input[node-type]');
 468 //
 469 //                    for (i = 0; i < els.length; i++) {
 470 //                        if (els[i].hasAttribute('node-type')) {
 471 //                            if(els[i].getAttribute('node-type') == 'username'){
 472 //                                els[i].value = BrowserSettings.username;
 473 //                            }
 474 //                            if(els[i].getAttribute('node-type') == 'password'){
 475 //                                els[i].value = BrowserSettings.passwd;
 476 //                            }
 477 //                        }
 478 //                    }
 479 //
 480 //
 481 //                    // simulating click event to login ....
 482 //                    var elss = document.querySelectorAll('a[node-type]');
 483 //
 484 //                    for (i = 0; i < elss.length; i++) {
 485 //                        if (elss[i].hasAttribute('node-type')) {
 486 //                            if(elss[i].getAttribute('node-type') == 'submitBtn'){
 487 //                                elss[i].click();
 488 //                            }
 489 //                        }
 490 //                    }
 491 //
 492 //
 493 //                }
 494 //
 495 //            }
 496
 497
 498             // clear login dialog from weibo.
 499 //            TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 500 //                " if (document.location.host=='weibo.com') { " +
 501 //                " try {  " +
 502 //                " var a = document.getElementsByClassName('W_layer')[0]; " +
 503 //                " a.parentNode.removeChild(a.previousSibling); " +
 504 //                " a.parentNode.removeChild(a); " +
 505 //                " } catch(e) { }  } " );
 506
 507
 508             return;
 509             GLib.timeout_add(GLib.PRIORITY_LOW, BrowserSettings.delay, function() {
 510
 511                 var mf = _t.get_main_frame();
 512
 513                 var ar = Gtk.PaperSize.get_paper_sizes();
 514                 var psetup = new Gtk.PageSetup();
 515                 for(var i = 0; i < ar.length; i++) {
 516                     if (ar[i].get_name() =='iso_a2') {
 517                         psetup.set_paper_size(ar[i]);
 518                     }
 519                 }
 520
 521                 var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
 522                 p.set_default_page_setup(psetup);
 523                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
 524                 print("made image - exiting");
 525
 526                 Seed.quit();
 527                 return true;
 528             });
 529
 530             return;
 531         };
 532
 533
 534
 535
 536         var load_committed = function (web_view, web_frame)
 537         {
 538             print("load commited");
 539
 540             update_url(web_view, web_frame);
 541
 542             // call load finished after 20seconds??
 543             // so even if it never complets we try and print the thing.
 544             GLib.timeout_add(GLib.PRIORITY_LOW, 40000, function() {
 545                 print("load commited - 3000 ms?");
 546                 load_finished();
 547             });
 548
 549         };
 550
 551         var clicked_link = function (web_view, web_frame, request,
 552                                      action, decision, window)
 553         {
 554             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 555                action.get_button() == 2)
 556             {
 557                 browser.new_tab(request.get_uri(), null);
 558                 return true;
 559             }
 560
 561             return false;
 562         };
 563
 564         // Public
 565
 566
 567         this.browse = function (url)
 568         {
 569             if(url.search("://") < 0)
 570                 url = "http://" + url;
 571
 572             current_url = url;
 573
 574             print("BROWSE: " + url);
 575             if (url.match(/\/weibo\.com\//) && after_login === false) {
 576                 is_weibo = true;
 577                 after_login = url;
 578                 url = "http://weibo.com/login.php";
 579             }
 580             else  if (url.match(/\/weibo\.com\//) && after_login !== false) {
 581                 after_login = false;
 582             }
 583             print("BROWSE -really: " + url);
 584             this.open(url);
 585         };
 586
 587         this.set_tab = function (new_tab)
 588         {
 589             tab = new_tab;
 590         };
 591
 592         this.get_tab = function ()
 593         {
 594             return tab;
 595         };
 596
 597
 598         // Implementation
 599         //this.set_scroll_adjustments(null, null);
 600
 601         //this.signal.title_changed.connect(update_title);
 602         //this.signal.load_committed.connect(load_committed);
 603         this.signal.load_finished.connect(load_finished);
 604
 605
 606         // For some reason, this segfaults seed in the instance init closure handler
 607         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 608         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 609
 610         //this.signal.hovering_over_link.connect(hover_link);
 611
 612         this.signal.create_web_view.connect(create_new_tab);
 613
 614
 615
 616 //        print("ADDing console message sig handler");
 617
 618
 619
 620
 621         this.toFilename = function(url)
 622         {
 623             url = url.replace(/^http[s]*:\/\//, '');
 624             var p = url.split('/');
 625             p.unshift(storedir+'/output');
 626             for (var i =1 ;i < p.length; i++) {
 627                 p[i] = encodeURIComponent(p[i]);
 628
 629             }
 630             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 631             ret = p.join('/');
 632             var dir = File.dirname(ret);
 633             File.mkdirall(dir);
 634             return ret;
 635
 636         }
 637         this.checkdomain = function(comp)
 638         {
 639             var b = parseUri(this.uri);
 640             var d = parseUri(comp);
 641             return (d.host == b.host && d.protocol == b.protocol);
 642
 643
 644         }
 645
 646         this.dupeCheck = function(url)
 647         {
 648
 649            // order - return highest up the queue first..
 650             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 651                 return downloaddir +'/' + encodeURIComponent(url);
 652             }
 653              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 654                 return parsedir +'/' + encodeURIComponent(url);
 655             }
 656             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 657                 return donedir +'/' + encodeURIComponent(url);
 658             }
 659             return  false;
 660
 661
 662         }
 663         this.moveToParse = function(url)
 664         {
 665             var old = this.dupeCheck(url);
 666             var target =parsedir +'/' + encodeURIComponent(url);
 667             if (old == target) {
 668                 return;
 669             }
 670             File.write(target, old ? File.read(old) : '');
 671             if (old) {
 672                 File.remove(old);
 673             }
 674
 675         }
 676
 677         this.moveToDownload= function(url)
 678         {
 679             var old = this.dupeCheck(url);
 680             var target =downloaddir +'/' + encodeURIComponent(url);
 681             if (old == target) {
 682                 return;
 683             }
 684             File.write(target, old ? File.read(old) : '');
 685             if (old) {
 686                 File.remove(old);
 687             }
 688
 689         }
 690         this.moveToDone= function(url)
 691         {
 692             var old = this.dupeCheck(url);
 693             var target = donedir +'/' + encodeURIComponent(url);
 694             if (old == target) {
 695                 return;
 696             }
 697             File.write(target, old ? File.read(old) : '');
 698             if (old) {
 699                 File.remove(old);
 700             }
 701
 702         }
 703
 704     }
 705 });
 706
 707 function parseUri (str) {
 708         var     o   = parseUri.options,
 709                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 710                 uri = {},
 711                 i   = 14;
 712
 713         while (i--) uri[o.key[i]] = m[i] || "";
 714
 715         uri[o.q.name] = {};
 716         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 717                 if ($1) uri[o.q.name][$1] = $2;
 718         });
 719
 720         return uri;
 721 };
 722
 723 parseUri.options = {
 724         strictMode: false,
 725         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 726         q:   {
 727                 name:   "queryKey",
 728                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 729         },
 730         parser: {
 731                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 732                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 733         }
 734 };