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