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