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