git.roojs.org Git - app.wkmirror/blob - BrowserView.js

   1 Gtk = imports.gi.Gtk;
   2 GLib = imports.gi.GLib;
   3 WebKit = imports.gi.WebKit;
   4
   5 TabbedBrowser = imports.TabbedBrowser;
   6 File = imports.File.File;
   7
   8 base64 = imports.base64.base64;
   9
  10 BrowserView = new GType({
  11     parent: WebKit.WebView.type,
  12     name: "BrowserView",
  13     init: function ()
  14     {
  15         // Private
  16
  17
  18         var _t = this;
  19
  20         var tab;
  21         var browsePage = false;
  22         var maxQueue = 0;
  23         var injected = {};
  24
  25         var storedir = '/home/alan/wkqueue';
  26         if (!File.exists(storedir)) {
  27             File.mkdir(storedir);
  28         }
  29         var parsedir = storedir + '/parse_queue';
  30         var downloaddir = storedir + '/download_queue';
  31         var donedir = storedir + '/downloaded_queue';
  32         if (!File.exists(parsedir)) {
  33             File.mkdir(parsedir);
  34         }
  35         if (!File.exists(downloaddir)) {
  36             File.mkdir(downloaddir);
  37         }
  38         if (!File.exists(donedir)) {
  39             File.mkdir(donedir);
  40         }
  41
  42         var update_title = function (web_view, web_frame, title)
  43         {
  44             if(title.length > 25)
  45                 title = title.slice(0,25) + "...";
  46
  47             tab.get_tab_label().label = title;
  48         };
  49
  50         var update_url = function (web_view, web_frame)
  51         {
  52             var toolbar = tab.get_toolbar();
  53
  54             toolbar.set_url(web_frame.get_uri());
  55             toolbar.set_can_go_back(web_view.can_go_back());
  56             toolbar.set_can_go_forward(web_view.can_go_forward());
  57         };
  58
  59         var update_progress = function (bar, progress)
  60         {
  61             tab.get_toolbar().set_progress(progress / 100);
  62         };
  63
  64         var create_new_tab = function (web_view, web_frame, new_web_view)
  65         {
  66             new_web_view = new BrowserView();
  67             new_web_view.signal.web_view_ready.connect(show_new_tab);
  68             return new_web_view;
  69         };
  70
  71         var show_new_tab = function (new_web_view)
  72         {
  73             TabbedBrowser.browser.new_tab("", new_web_view);
  74
  75             return false;
  76         };
  77
  78         var hover_link = function (web_view, link, url)
  79         {
  80             tab.get_statusbar().set_status(url);
  81         };
  82
  83
  84
  85
  86         this.add_inject = function(force)
  87         {
  88
  89             if (force || (typeof(injected[this.uri]) == 'undefined' )) {
  90                 injected[this.uri] = 0;
  91             }
  92             if (injected[this.uri] > 2) {
  93                 return;
  94             }
  95             injected[this.uri]++;
  96             var fn = __script_path__ + "/inject.js";
  97             if (File.exists(fn)) {
  98                 print("Adding inject");
  99                 var newjs = File.read(__script_path__ + "/inject.js");
 100                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 101                     newjs
 102
 103                 );
 104             }
 105
 106         }
 107          this.add_nsinject = function(force)
 108         {
 109
 110             //if (force || (typeof(injected[this.uri]) == 'undefined' )) {
 111             //    injected[this.uri] = 0;
 112             //}
 113             //if (injected[this.uri] > 2) {
 114             //    return;
 115             //}
 116             //injected[this.uri]++;
 117             var fn = __script_path__ + "/nsinject.js";
 118             //if (File.exists(fn)) {
 119                 print("Adding inject");
 120                 var newjs = File.read(__script_path__ + "/nsinject.js");
 121                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 122                     newjs
 123
 124                 );
 125             //}
 126
 127         }
 128
 129
 130
 131
 132
 133         var load_finished = function ()
 134         {
 135             print("load finished");
 136
 137             GLib.timeout_add(GLib.PRIORITY_LOW, 2000, function() {
 138
 139 //                _t.open("http://sg.lifestyleasia.com/videos/video-british-polo-day-singapore-2012/");
 140                 var mf = _t.get_main_frame();
 141
 142                  var ar = Gtk.PaperSize.get_paper_sizes();
 143                 var psetup = new Gtk.PageSetup ();
 144                 for(var i = 0; i < ar.length; i++) {
 145                     //print(ar[i].get_name());
 146                     if (ar[i].get_name() =='iso_a2') {
 147                         psetup.set_paper_size(ar[i]);
 148                     }
 149                 }
 150
 151
 152
 153                 var p = new Gtk.PrintOperation({ export_filename : '/home/chris/test_pdf/test2.pdf' });
 154                 p.set_default_page_setup(psetup);
 155                 mf.print_full(p, Gtk.PrintOperationAction.PREVIEW)
 156                 //Seed.quit();
 157             });
 158
 159             return;
 160             tab.get_toolbar().set_progress(0);
 161
 162             _t.add_inject(true);
 163
 164             _t.add_nsinject(true);
 165               print(typeof(Seed.argv[2]));
 166             if (Seed.argv[2] == null) {
 167                 print("NEED ID!");
 168                 Seed.quit();
 169
 170             }
 171             _t.nsdownloadNext(Seed.argv[2]  * 1)
 172
 173
 174             if (browsePage) {
 175                 print("onload: calling gather links");
 176                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 177                     'BrowserMirror.gatherlinks();'
 178
 179                 );
 180
 181             }
 182
 183
 184         };
 185
 186         var load_committed = function (web_view, web_frame)
 187         {
 188             update_url(web_view, web_frame);
 189         };
 190
 191         var clicked_link = function (web_view, web_frame, request,
 192                                      action, decision, window)
 193         {
 194             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 195                action.get_button() == 2)
 196             {
 197                 browser.new_tab(request.get_uri(), null);
 198                 return true;
 199             }
 200
 201             return false;
 202         };
 203
 204         // Public
 205
 206
 207         this.browse = function (url)
 208         {
 209             if(url.search("://") < 0)
 210                 url = "http://" + url;
 211
 212             this.open(url);
 213         };
 214
 215         this.set_tab = function (new_tab)
 216         {
 217             tab = new_tab;
 218         };
 219
 220         this.get_tab = function ()
 221         {
 222             return tab;
 223         };
 224
 225
 226         this.downloadqueue = function()
 227         {
 228             var filesList = File.list(downloaddir);
 229             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 230             if (!maxQueue || maxQueue < filesList.length) {
 231                 maxQueue  = filesList.length;
 232             }
 233
 234             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 235             if (filesList == null || filesList.length == 0  ){
 236
 237                 return false;
 238             }
 239             var url = decodeURIComponent(filesList[0]);
 240             if (!this.checkdomain(url)) {
 241                 print("SKIP (external domain) : " + url);
 242                 File.remove(downloaddir + '/' + filesList[0]);
 243                 return this.downloadqueue();
 244
 245             }
 246             this.downloadhead( url);
 247             return true;
 248
 249         };
 250
 251
 252         this.queuerun = function()
 253         {
 254             if (this.downloadqueue()) return true;
 255             return this.parsequeue();
 256         };
 257
 258         this.parsequeue = function(){
 259             var filesList = File.list(parsedir);
 260             if (filesList == null || filesList.length == 0  ){
 261                 return false;
 262             }
 263
 264             browsePage = decodeURIComponent(filesList[0]);
 265             print("parsing page:" + browsePage );
 266             this.browse(browsePage);
 267
 268
 269
 270
 271
 272             return true;
 273         };
 274
 275
 276         this.downloadpage = function(link){
 277             print("calling download page: " + link);
 278             _t.add_inject(); // just in case..
 279                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 280             this.execute_script(
 281                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 282             );
 283
 284
 285         };
 286
 287         this.downloadhead = function(link){
 288             print("calling download head: " + link);
 289             _t.add_inject(); // just in case..
 290                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 291             this.execute_script(
 292                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 293             );
 294
 295         };
 296         var nsqueue = false;
 297         this.nsdownloadNext = function()
 298         {
 299             /*if (nsqueue === false) {
 300                 nsqueue = [
 301                            //282,273,285,395,271,272,278,394,402,279,432,280,284,281,433,437,404,444,151,152,1,283,287,288,418,407,398,449,147,150,131,149,443,148,411,405,415,417,416,406,408,399,410,409,412,400,155,133,124,157,118,183,450,434,301,435,130,286,162,292,431,161,290,146,145,158,159,120,137,128,139,163,293,160,289,291,141,140,138,142,144,110,401,156,154,153,170,167,295,166,294,277,425,164,165,168,296,169,297,113,171,119,180,300,176,175,172,302,299,
 302                            //298,421,187,307,428,275,269,304,181,182,174,173,184,303,112,179,114,305,177,178,186,306,403,420,111,109,309,188,308,446,445,310,311,53,189,52,312,419,270,190,191,192,193,54,194,108,268,121,195,206,396,205,207,324,220,340,212,331,219,339,222,342,436,209,327,424,208,326,210,328,330,325,211,329,218,337,338,224,223,217,336,214,333,213,332,216,335,221,341,215,334,197,314,
 303                             //196,313,198,316,201,320,317,204,323,203,322,202,321,200,319,199,318,346,58,267,393,265,391,438,264,390,261,387,262,388,260,386,266,392,413,414,263,389,347,225,343,227,345,226,344,422,228,348,237,358,231,352,229,349,351,233,354,238,359,360,239,361,230,350,232,353,234,355,439,236,357,240,362,235,356,254,380,257,383,259,385,255,381,256,448,382,258,384,374,252,375,376,251,373,378,429,430,379,253,244,366,247,
 304                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 305                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 306                         ];
 307             }
 308
 309             if (!nsqueue.length) {
 310                 print("DONE");
 311                 return;
 312             }
 313             var pg = nsqueue.shift() ;
 314
 315
 316             */
 317
 318             //print(typeof(Seed.argv[2]));
 319             if (!Seed.argv[2]) {
 320                 print("NEED ID!");
 321                 Seed.quit();
 322
 323             }
 324             var pg = Seed.argv[2]  * 1;
 325             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 326                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 327                 Seed.quit();
 328                 return;
 329             }
 330
 331             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 332
 333             print("downloadnext : " + pg);
 334             this.execute_script(
 335                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 336
 337             );
 338
 339
 340
 341         }
 342
 343         // Implementation
 344         //this.set_scroll_adjustments(null, null);
 345
 346         this.signal.title_changed.connect(update_title);
 347         this.signal.load_committed.connect(load_committed);
 348         this.signal.load_finished.connect(load_finished);
 349         this.signal.load_progress_changed.connect(update_progress);
 350
 351         // For some reason, this segfaults seed in the instance init closure handler
 352         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 353         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 354
 355         this.signal.hovering_over_link.connect(hover_link);
 356
 357         this.signal.create_web_view.connect(create_new_tab);
 358
 359
 360
 361         print("ADDing console message sig handler");
 362
 363
 364
 365
 366         this.signal.console_message.connect(function(wv, msg, line, sid) {
 367             // print('BrowserView.js got ' + msg);
 368             var methodname;
 369             var ret;
 370             try {
 371                 ret = JSON.parse(msg);
 372
 373             } catch(e) {
 374                 print("GOT INVALID message:" + msg)
 375                 return true;
 376
 377
 378             }
 379             print("got method : " + ret.method);
 380
 381             if (ret.method == 'exit') {
 382                 Seed.quit();
 383             }
 384
 385
 386             if (ret.method == 'nsdownloadpage'){
 387                 try {
 388                     var mt = ret.contentType.split(';').shift();
 389                 } catch( e) {
 390                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 391                     mt='';
 392                 }
 393
 394                 switch(mt) {
 395
 396                     case '':
 397                         print("CONSOLE GOT BLANK - call download next?");
 398                           Seed.quit();
 399                          //_t.nsdownloadNext();
 400                        // _t.moveToDone(ret.requesturl);
 401                         break;
 402
 403                     default:
 404                         var info = false;
 405                         var info_f = _t.dupeCheck(ret.requesturl);
 406                         if (info_f) {
 407                             info  = JSON.parse(File.read(info_f));
 408                         }
 409                          var target  = storedir+'/output/' + ret.filename
 410                          // flag it as done..
 411
 412
 413
 414
 415
 416                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 417                         print("GOT array sized: " + ret.data.length);
 418
 419                         File.writeBinaryArray(target,ret.data);
 420                         print("Wrote to file: " + target);
 421                         // get next..
 422                         return true;
 423                         //Seed.quit();
 424                         //_t.nsdownloadNext();
 425
 426
 427
 428                         break;
 429                 }
 430                 return true; //???
 431             }
 432
 433             if (ret.method == 'gatherlinks'){
 434                 // flag the page as parsed.
 435
 436                 if (browsePage) {
 437                     _t.moveToDone(browsePage);
 438
 439                 }
 440                 var sourcePage = browsePage;
 441                 browsePage = false;
 442                 print(typeof(ret.data));
 443                 if (typeof(ret.data) != 'object' ) {
 444                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 445                     ret.data= [];
 446
 447                 }
 448                 ret.data.forEach(function(ln) {
 449                     if (!ln.href.match(/^http[s]*:\/\//)) {
 450                         print("SKIP link: " + ln.href);
 451                         return;
 452                     }
 453                     ln.href= ln.href.replace(/#.*$/,'');
 454
 455                     if (!_t.checkdomain(ln.href) ) {
 456                         print("SKIP link (external domain): " + ln.href);
 457                         return;
 458                     }
 459
 460                     // this is just for our purposes..
 461
 462                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 463                         print("SKIP link (ingore unnamed pages): " + ln.href);
 464                         return;
 465                     }
 466
 467
 468
 469                     var fn = encodeURIComponent(ln.href);
 470
 471                     var dupe  = _t.dupeCheck(ln.href);
 472                     if (dupe) {
 473
 474                         if (dupe == downloaddir + '/'  + fn) {
 475                             var info = JSON.parse(File.read(dupe));
 476                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 477                                 print("SKIP link (in queue already): " + ln.href);
 478                                 return;
 479                             }
 480                             print("found a longer link for url")
 481                         } else {
 482                             print("SKIP link (in another queue): " + ln.href);
 483                             return;
 484                         }
 485                     }
 486
 487                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 488                         label : ln.label,
 489                         fromURL : sourcePage
 490                     })); // write an empyt file indicating it needs downloading..
 491                 });
 492                 var filesList = File.list(downloaddir);
 493
 494                 maxQueue =  filesList.length;
 495             }
 496
 497             if (ret.method == 'downloadpage'){
 498                 // got the results from download page:
 499                 // requesturl
 500                 // remove from downloadqueue.
 501
 502
 503                 //
 504                 try {
 505                     var mt = ret.contentType.split(';').shift();
 506                 } catch( e) {
 507                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 508                     mt='';
 509                 }
 510
 511                 switch(mt) {
 512
 513                     case '':
 514                         _t.moveToDone(ret.requesturl);
 515                         break;
 516
 517                     default:
 518                         var info = false;
 519                         var info_f = _t.dupeCheck(ret.requesturl);
 520                         if (info_f) {
 521                             info  = JSON.parse(File.read(info_f));
 522                         }
 523
 524                         _t.moveToDone(ret.requesturl);
 525                         // flag it as done..
 526
 527                         // write it...
 528                         var target  = _t.toFilename(ret.requesturl);
 529                         if (info && info.fromURL) {
 530                             var bn = decodeURIComponent(File.basename(target));
 531                             target = _t.toFilename(info.fromURL+'/'+ bn);
 532                         }
 533
 534
 535
 536
 537                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 538                         print("GOT array sized: " + ret.data.length);
 539
 540                         File.writeBinaryArray(target,ret.data);
 541                         print("Wrote to file: " + target);
 542                         break;
 543                 }
 544
 545                 // if it's HTML then add it to parse queue
 546                 // otehrwise save it.. and run the queue again.
 547
 548
 549                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 550             }
 551
 552
 553             if (ret.method == 'downloadhead'){
 554                 // got the results from download page:
 555                 // requesturl
 556                 // remove from downloadqueue.
 557
 558                 //
 559                 try {
 560                     var mt = ret.contentType.split(';').shift();
 561                 } catch( e) {
 562                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 563                     mt='';
 564                 }
 565
 566                 switch(mt) {
 567                     case 'text/html':
 568                         // add to parse QUEUE..
 569                         print("moving to parse queue");
 570                         _t.moveToParse(ret.requesturl);
 571                         break;
 572                     // stuf we do not care about..
 573                     case 'application/atom+xml':
 574                     case '':
 575                         print("moving to done queue");
 576                          _t.moveToDone(ret.requesturl);
 577                         break;
 578
 579                     default:
 580                         print("calling download file");
 581                         _t.downloadpage( ret.requesturl );
 582                         // keep it on the queue..
 583                         // do not run the queue..
 584                         return true;
 585
 586                 }
 587
 588                 // if it's HTML then add it to parse queue
 589                 // otehrwise save it.. and run the queue again.
 590
 591
 592                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 593             }
 594             _t.queuerun();
 595
 596             return true;
 597         });
 598
 599
 600         this.toFilename = function(url)
 601         {
 602             url = url.replace(/^http[s]*:\/\//, '');
 603             var p = url.split('/');
 604             p.unshift(storedir+'/output');
 605             for (var i =1 ;i < p.length; i++) {
 606                 p[i] = encodeURIComponent(p[i]);
 607
 608             }
 609             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 610             ret = p.join('/');
 611             var dir = File.dirname(ret);
 612             File.mkdirall(dir);
 613             return ret;
 614
 615         }
 616         this.checkdomain = function(comp)
 617         {
 618             var b = parseUri(this.uri);
 619             var d = parseUri(comp);
 620             return (d.host == b.host && d.protocol == b.protocol);
 621
 622
 623         }
 624
 625         this.dupeCheck = function(url)
 626         {
 627
 628            // order - return highest up the queue first..
 629             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 630                 return downloaddir +'/' + encodeURIComponent(url);
 631             }
 632              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 633                 return parsedir +'/' + encodeURIComponent(url);
 634             }
 635             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 636                 return donedir +'/' + encodeURIComponent(url);
 637             }
 638             return  false;
 639
 640
 641         }
 642         this.moveToParse = function(url)
 643         {
 644             var old = this.dupeCheck(url);
 645             var target =parsedir +'/' + encodeURIComponent(url);
 646             if (old == target) {
 647                 return;
 648             }
 649             File.write(target, old ? File.read(old) : '');
 650             if (old) {
 651                 File.remove(old);
 652             }
 653
 654         }
 655
 656         this.moveToDownload= function(url)
 657         {
 658             var old = this.dupeCheck(url);
 659             var target =downloaddir +'/' + encodeURIComponent(url);
 660             if (old == target) {
 661                 return;
 662             }
 663             File.write(target, old ? File.read(old) : '');
 664             if (old) {
 665                 File.remove(old);
 666             }
 667
 668         }
 669         this.moveToDone= function(url)
 670         {
 671             var old = this.dupeCheck(url);
 672             var target = donedir +'/' + encodeURIComponent(url);
 673             if (old == target) {
 674                 return;
 675             }
 676             File.write(target, old ? File.read(old) : '');
 677             if (old) {
 678                 File.remove(old);
 679             }
 680
 681         }
 682
 683     }
 684 });
 685
 686 function parseUri (str) {
 687         var     o   = parseUri.options,
 688                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 689                 uri = {},
 690                 i   = 14;
 691
 692         while (i--) uri[o.key[i]] = m[i] || "";
 693
 694         uri[o.q.name] = {};
 695         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 696                 if ($1) uri[o.q.name][$1] = $2;
 697         });
 698
 699         return uri;
 700 };
 701
 702 parseUri.options = {
 703         strictMode: false,
 704         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 705         q:   {
 706                 name:   "queryKey",
 707                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 708         },
 709         parser: {
 710                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 711                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 712         }
 713 };