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             var scr_a =  " var r = document.getElementsByTagName('link');" +
 281                 "for (var i=0;i < r.length;i++) { " +
 282                     "var a = r[i]; " +
 283                     "if (  a.getAttribute('media') == '')  { continue; } " +
 284                     "if (  a.getAttribute('media') == 'screen')  { "+
 285                         "a.removeAttribute('media'); continue;" +
 286                     "} " +
 287                     //"a.parentNode.removeChild(a);" +
 288                 "}";
 289             var scr_b =  "var a = document.body.querySelectorAll('*'); "+
 290                 "var dnodes=[]; "+
 291                 "for(var i =i;i<a.length;i++) { "+
 292                 "   var cs = window.getComputedStyle(a[i],null);"+
 293                 "   if ('none' == cs.display) {"+
 294                 "    dnodes.push(a[i]); "+
 295                 "  }"+
 296                 "}"+
 297                 "console.log(dnodes.length);" +
 298                 "dnodes.forEach(function(n) { "+
 299                 "    try {  n.parentNode.removeChild(n);   } catch(e) { } "+
 300                 "});";
 301
 302
 303             //TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 304             //    scr_a
 305             //);
 306             //print(scr_b);
 307             //TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 308             //    scr_b
 309             //);
 310
 311             if (current_url.match(/\.coconuts\.co\//) ) {
 312                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 313                     File.read( __script_path__ + "/domains/coconuts.co.js")
 314                 );
 315
 316             }
 317             if (current_url.match(/\.inquirer\.net\//) ) {
 318                 print("add inquirer.net code");
 319                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 320                     File.read( __script_path__ + "/domains/inquirer.net.js")
 321                 );
 322
 323             }
 324
 325
 326             // motorme - requires a scroll to make content appear..
 327             if (current_url.match(/\.motorme\.my\//) ) {
 328                 TabbedBrowser.browser.current_tab().get_web_view().execute_script("window.scrollTo(0,500);");
 329
 330             }
 331
 332             if (current_url.match(/\.efytimes\.com\//) ||  current_url.match(/\/efytimes\.com\//) ) {
 333                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 334                     File.read( __script_path__ + "/domains/efytimes.com.js")
 335                 );
 336             }
 337
 338
 339             print("Delay:" + BrowserSettings.delay);
 340
 341             GLib.timeout_add(GLib.PRIORITY_DEFAULT, BrowserSettings.delay *1, function() {
 342
 343                 try {
 344
 345                     if (current_url.match(/theindependent\.sg\//) ) {
 346                         TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 347                             File.read( __script_path__ + "/domains/theindependent.sg.js")
 348                         );
 349
 350                     }
 351                     print("FETCHING");
 352                     //print(web_frame);
 353                     if (web_frame &&  BrowserSettings.export_filename_html  ) {
 354                         var html = traversedom(web_frame.get_dom_document(), BrowserSettings.export_filename_html);
 355                         //File.write(BrowserSettings.export_filename_html, html);
 356
 357                     }
 358
 359                     if (!BrowserSettings.export_filename) {
 360                         if (!BrowserSettings.export_filename_html) {
 361                              print("no export filename / and html filename");
 362                             return;
 363                         }
 364
 365                         Seed.quit();
 366                     }
 367                     var mf = _t.get_main_frame();
 368
 369                     print("Creating paper");
 370                     var ar = Gtk.PaperSize.get_paper_sizes();
 371                     var psetup = new Gtk.PageSetup();
 372                     for(var i = 0; i < ar.length; i++) {
 373                         if (ar[i].get_name() =='iso_a2') {
 374                             psetup.set_paper_size(ar[i]);
 375                         }
 376                     }
 377                     print("doing print operation");
 378                     var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
 379                     print("doing print operation - set page");
 380
 381                     p.set_default_page_setup(psetup);
 382
 383                     print("doing print operation - print_full");
 384                     mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
 385                     print("made image - exiting");
 386
 387                     Seed.quit();
 388                     return true;
 389                 } catch(e) {
 390                     print("error occured");
 391                     print(JSON.stringify(e));
 392                 }
 393             });
 394         }
 395
 396
 397         var weibo_added = false;
 398
 399         var is_weibo = false;
 400
 401
 402         var load_finished = function (webkit, web_frame, wb)
 403         {
 404
 405
 406
 407             if (!is_weibo) {
 408                 do_print(web_frame);
 409                 return;
 410             }
 411
 412
 413              print("load finished");
 414              if (weibo_added) {
 415
 416                  if (after_login === true) {
 417                      print("do print");
 418                      do_print();
 419                      return;
 420
 421                  }
 422                  if (after_login === false) {
 423                      return; // not here?
 424                  }
 425                  print("adding timeout?");
 426                  //return;
 427                 GLib.timeout_add(GLib.PRIORITY_LOW, 2000, function() {
 428                     print("Redirecting after login?" + after_login);
 429
 430                     _t.browse(after_login);
 431                     after_login  = true;
 432                 });
 433                 return;
 434              }
 435
 436             weibo_added = true;
 437             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 438                 File.read( __script_path__ + "/weibo.js")
 439             );
 440             print("run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");");
 441
 442             TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 443                     "run_weibo(" + JSON.stringify( BrowserSettings.username ) + ", " + JSON.stringify(BrowserSettings.passwd) + ");"
 444             );
 445             return;
 446
 447             //    return;
 448             print(after_login);
 449             if (after_login !== false) {
 450
 451 //                return;
 452                 // wait a bit then load the real page..
 453                 GLib.timeout_add(GLib.PRIORITY_LOW, 500, function() {
 454                     print("Redirecting after login?");
 455                     _t.browse(after_login);
 456                 });
 457
 458                 return;
 459             }
 460
 461             if (load_finished_called) {
 462                 return ;
 463
 464             }
 465             load_finished_called = true;
 466
 467
 468
 469 //            if(document.location.host=='weibo.com') {
 470 //                // clear login dialog from weibo.
 471 //                //check and hidden the login dialog and overlay .....
 472 //                var els = document.querySelectorAll('div[node-type]');
 473 //                var bbc = [];
 474 //                for (i = 0; i < els.length; i++) {
 475 //                    if (els[i].hasAttribute('node-type')) {
 476 //                        if(els[i].getAttribute('node-type') == 'outer'){
 477 //                            bbc.push(els[i]);
 478 //                        }
 479 //                    }
 480 //                }
 481 //                for (i = 0; i < bbc.length; i++) {// hide all the outer.....
 482 //                    bbc[i].style.display = 'none';
 483 //                }
 484 //
 485 //                //try to login
 486 //                if(document.location.pathname == '/login.php'){
 487 //
 488 //                    // fills in the user name and password
 489 //                    var els = document.querySelectorAll('input[node-type]');
 490 //
 491 //                    for (i = 0; i < els.length; i++) {
 492 //                        if (els[i].hasAttribute('node-type')) {
 493 //                            if(els[i].getAttribute('node-type') == 'username'){
 494 //                                els[i].value = BrowserSettings.username;
 495 //                            }
 496 //                            if(els[i].getAttribute('node-type') == 'password'){
 497 //                                els[i].value = BrowserSettings.passwd;
 498 //                            }
 499 //                        }
 500 //                    }
 501 //
 502 //
 503 //                    // simulating click event to login ....
 504 //                    var elss = document.querySelectorAll('a[node-type]');
 505 //
 506 //                    for (i = 0; i < elss.length; i++) {
 507 //                        if (elss[i].hasAttribute('node-type')) {
 508 //                            if(elss[i].getAttribute('node-type') == 'submitBtn'){
 509 //                                elss[i].click();
 510 //                            }
 511 //                        }
 512 //                    }
 513 //
 514 //
 515 //                }
 516 //
 517 //            }
 518
 519
 520             // clear login dialog from weibo.
 521 //            TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 522 //                " if (document.location.host=='weibo.com') { " +
 523 //                " try {  " +
 524 //                " var a = document.getElementsByClassName('W_layer')[0]; " +
 525 //                " a.parentNode.removeChild(a.previousSibling); " +
 526 //                " a.parentNode.removeChild(a); " +
 527 //                " } catch(e) { }  } " );
 528
 529
 530             return;
 531             GLib.timeout_add(GLib.PRIORITY_LOW, BrowserSettings.delay, function() {
 532
 533                 var mf = _t.get_main_frame();
 534
 535                 var ar = Gtk.PaperSize.get_paper_sizes();
 536                 var psetup = new Gtk.PageSetup();
 537                 for(var i = 0; i < ar.length; i++) {
 538                     if (ar[i].get_name() =='iso_a2') {
 539                         psetup.set_paper_size(ar[i]);
 540                     }
 541                 }
 542
 543                 var p = new Gtk.PrintOperation({ export_filename : BrowserSettings.export_filename });
 544                 p.set_default_page_setup(psetup);
 545                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT);
 546                 print("made image - exiting");
 547
 548                 Seed.quit();
 549                 return true;
 550             });
 551
 552             return;
 553         };
 554
 555
 556
 557
 558         var load_committed = function (web_view, web_frame)
 559         {
 560             print("load commited");
 561
 562             update_url(web_view, web_frame);
 563
 564             // call load finished after 20seconds??
 565             // so even if it never complets we try and print the thing.
 566             GLib.timeout_add(GLib.PRIORITY_LOW, 40000, function() {
 567                 print("load commited - 3000 ms?");
 568                 load_finished();
 569             });
 570
 571         };
 572
 573         var clicked_link = function (web_view, web_frame, request,
 574                                      action, decision, window)
 575         {
 576             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 577                action.get_button() == 2)
 578             {
 579                 browser.new_tab(request.get_uri(), null);
 580                 return true;
 581             }
 582
 583             return false;
 584         };
 585
 586         // Public
 587
 588
 589         this.browse = function (url)
 590         {
 591             if(url.search("://") < 0)
 592                 url = "http://" + url;
 593
 594             current_url = url;
 595
 596             print("BROWSE: " + url);
 597             if (url.match(/\/weibo\.com\//) && after_login === false) {
 598                 is_weibo = true;
 599                 after_login = url;
 600                 url = "http://weibo.com/login.php";
 601             }
 602             else  if (url.match(/\/weibo\.com\//) && after_login !== false) {
 603                 after_login = false;
 604             }
 605             print("BROWSE -really: " + url);
 606             this.open(url);
 607         };
 608
 609         this.set_tab = function (new_tab)
 610         {
 611             tab = new_tab;
 612         };
 613
 614         this.get_tab = function ()
 615         {
 616             return tab;
 617         };
 618
 619
 620         // Implementation
 621         //this.set_scroll_adjustments(null, null);
 622
 623         //this.signal.title_changed.connect(update_title);
 624         //this.signal.load_committed.connect(load_committed);
 625         this.signal.load_finished.connect(load_finished);
 626
 627
 628         // For some reason, this segfaults seed in the instance init closure handler
 629         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 630         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 631
 632         //this.signal.hovering_over_link.connect(hover_link);
 633
 634         this.signal.create_web_view.connect(create_new_tab);
 635
 636
 637
 638 //        print("ADDing console message sig handler");
 639
 640
 641
 642
 643         this.toFilename = function(url)
 644         {
 645             url = url.replace(/^http[s]*:\/\//, '');
 646             var p = url.split('/');
 647             p.unshift(storedir+'/output');
 648             for (var i =1 ;i < p.length; i++) {
 649                 p[i] = encodeURIComponent(p[i]);
 650
 651             }
 652             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 653             ret = p.join('/');
 654             var dir = File.dirname(ret);
 655             File.mkdirall(dir);
 656             return ret;
 657
 658         }
 659         this.checkdomain = function(comp)
 660         {
 661             var b = parseUri(this.uri);
 662             var d = parseUri(comp);
 663             return (d.host == b.host && d.protocol == b.protocol);
 664
 665
 666         }
 667
 668         this.dupeCheck = function(url)
 669         {
 670
 671            // order - return highest up the queue first..
 672             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 673                 return downloaddir +'/' + encodeURIComponent(url);
 674             }
 675              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 676                 return parsedir +'/' + encodeURIComponent(url);
 677             }
 678             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 679                 return donedir +'/' + encodeURIComponent(url);
 680             }
 681             return  false;
 682
 683
 684         }
 685         this.moveToParse = function(url)
 686         {
 687             var old = this.dupeCheck(url);
 688             var target =parsedir +'/' + encodeURIComponent(url);
 689             if (old == target) {
 690                 return;
 691             }
 692             File.write(target, old ? File.read(old) : '');
 693             if (old) {
 694                 File.remove(old);
 695             }
 696
 697         }
 698
 699         this.moveToDownload= function(url)
 700         {
 701             var old = this.dupeCheck(url);
 702             var target =downloaddir +'/' + encodeURIComponent(url);
 703             if (old == target) {
 704                 return;
 705             }
 706             File.write(target, old ? File.read(old) : '');
 707             if (old) {
 708                 File.remove(old);
 709             }
 710
 711         }
 712         this.moveToDone= function(url)
 713         {
 714             var old = this.dupeCheck(url);
 715             var target = donedir +'/' + encodeURIComponent(url);
 716             if (old == target) {
 717                 return;
 718             }
 719             File.write(target, old ? File.read(old) : '');
 720             if (old) {
 721                 File.remove(old);
 722             }
 723
 724         }
 725
 726     }
 727 });
 728
 729 function parseUri (str) {
 730         var     o   = parseUri.options,
 731                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 732                 uri = {},
 733                 i   = 14;
 734
 735         while (i--) uri[o.key[i]] = m[i] || "";
 736
 737         uri[o.q.name] = {};
 738         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 739                 if ($1) uri[o.q.name][$1] = $2;
 740         });
 741
 742         return uri;
 743 };
 744
 745 parseUri.options = {
 746         strictMode: false,
 747         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 748         q:   {
 749                 name:   "queryKey",
 750                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 751         },
 752         parser: {
 753                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 754                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 755         }
 756 };