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, 1000, 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                 var s = new Gtk.PrintSettings();
 155
 156
 157                 //s.set_paper_size('iso_a4');
 158 //                psetup.set_scale(0.5);
 159                 p.set_default_page_setup(psetup);
 160
 161                 s.set_scale(0.5);
 162
 163
 164
 165                 p.set_print_settings(s);
 166                 //var s = p.get_print_settings();
 167                 //print(s);
 168                 //p.set_print_settings(s);
 169                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
 170                 //Seed.quit();
 171             });
 172
 173             return;
 174             tab.get_toolbar().set_progress(0);
 175
 176             _t.add_inject(true);
 177
 178             _t.add_nsinject(true);
 179               print(typeof(Seed.argv[2]));
 180             if (Seed.argv[2] == null) {
 181                 print("NEED ID!");
 182                 Seed.quit();
 183
 184             }
 185             _t.nsdownloadNext(Seed.argv[2]  * 1)
 186
 187
 188             if (browsePage) {
 189                 print("onload: calling gather links");
 190                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 191                     'BrowserMirror.gatherlinks();'
 192
 193                 );
 194
 195             }
 196
 197
 198         };
 199
 200         var load_committed = function (web_view, web_frame)
 201         {
 202             update_url(web_view, web_frame);
 203         };
 204
 205         var clicked_link = function (web_view, web_frame, request,
 206                                      action, decision, window)
 207         {
 208             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 209                action.get_button() == 2)
 210             {
 211                 browser.new_tab(request.get_uri(), null);
 212                 return true;
 213             }
 214
 215             return false;
 216         };
 217
 218         // Public
 219
 220
 221         this.browse = function (url)
 222         {
 223             if(url.search("://") < 0)
 224                 url = "http://" + url;
 225
 226             this.open(url);
 227         };
 228
 229         this.set_tab = function (new_tab)
 230         {
 231             tab = new_tab;
 232         };
 233
 234         this.get_tab = function ()
 235         {
 236             return tab;
 237         };
 238
 239
 240         this.downloadqueue = function()
 241         {
 242             var filesList = File.list(downloaddir);
 243             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 244             if (!maxQueue || maxQueue < filesList.length) {
 245                 maxQueue  = filesList.length;
 246             }
 247
 248             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 249             if (filesList == null || filesList.length == 0  ){
 250
 251                 return false;
 252             }
 253             var url = decodeURIComponent(filesList[0]);
 254             if (!this.checkdomain(url)) {
 255                 print("SKIP (external domain) : " + url);
 256                 File.remove(downloaddir + '/' + filesList[0]);
 257                 return this.downloadqueue();
 258
 259             }
 260             this.downloadhead( url);
 261             return true;
 262
 263         };
 264
 265
 266         this.queuerun = function()
 267         {
 268             if (this.downloadqueue()) return true;
 269             return this.parsequeue();
 270         };
 271
 272         this.parsequeue = function(){
 273             var filesList = File.list(parsedir);
 274             if (filesList == null || filesList.length == 0  ){
 275                 return false;
 276             }
 277
 278             browsePage = decodeURIComponent(filesList[0]);
 279             print("parsing page:" + browsePage );
 280             this.browse(browsePage);
 281
 282
 283
 284
 285
 286             return true;
 287         };
 288
 289
 290         this.downloadpage = function(link){
 291             print("calling download page: " + link);
 292             _t.add_inject(); // just in case..
 293                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 294             this.execute_script(
 295                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 296             );
 297
 298
 299         };
 300
 301         this.downloadhead = function(link){
 302             print("calling download head: " + link);
 303             _t.add_inject(); // just in case..
 304                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 305             this.execute_script(
 306                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 307             );
 308
 309         };
 310         var nsqueue = false;
 311         this.nsdownloadNext = function()
 312         {
 313             /*if (nsqueue === false) {
 314                 nsqueue = [
 315                            //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,
 316                            //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,
 317                             //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,
 318                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 319                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 320                         ];
 321             }
 322
 323             if (!nsqueue.length) {
 324                 print("DONE");
 325                 return;
 326             }
 327             var pg = nsqueue.shift() ;
 328
 329
 330             */
 331
 332             //print(typeof(Seed.argv[2]));
 333             if (!Seed.argv[2]) {
 334                 print("NEED ID!");
 335                 Seed.quit();
 336
 337             }
 338             var pg = Seed.argv[2]  * 1;
 339             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 340                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 341                 Seed.quit();
 342                 return;
 343             }
 344
 345             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 346
 347             print("downloadnext : " + pg);
 348             this.execute_script(
 349                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 350
 351             );
 352
 353
 354
 355         }
 356
 357         // Implementation
 358         //this.set_scroll_adjustments(null, null);
 359
 360         this.signal.title_changed.connect(update_title);
 361         this.signal.load_committed.connect(load_committed);
 362         this.signal.load_finished.connect(load_finished);
 363         this.signal.load_progress_changed.connect(update_progress);
 364
 365         // For some reason, this segfaults seed in the instance init closure handler
 366         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 367         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 368
 369         this.signal.hovering_over_link.connect(hover_link);
 370
 371         this.signal.create_web_view.connect(create_new_tab);
 372
 373
 374
 375         print("ADDing console message sig handler");
 376
 377
 378
 379
 380         this.signal.console_message.connect(function(wv, msg, line, sid) {
 381             // print('BrowserView.js got ' + msg);
 382             var methodname;
 383             var ret;
 384             try {
 385                 ret = JSON.parse(msg);
 386
 387             } catch(e) {
 388                 print("GOT INVALID message:" + msg)
 389                 return true;
 390
 391
 392             }
 393             print("got method : " + ret.method);
 394
 395             if (ret.method == 'exit') {
 396                 Seed.quit();
 397             }
 398
 399
 400             if (ret.method == 'nsdownloadpage'){
 401                 try {
 402                     var mt = ret.contentType.split(';').shift();
 403                 } catch( e) {
 404                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 405                     mt='';
 406                 }
 407
 408                 switch(mt) {
 409
 410                     case '':
 411                         print("CONSOLE GOT BLANK - call download next?");
 412                           Seed.quit();
 413                          //_t.nsdownloadNext();
 414                        // _t.moveToDone(ret.requesturl);
 415                         break;
 416
 417                     default:
 418                         var info = false;
 419                         var info_f = _t.dupeCheck(ret.requesturl);
 420                         if (info_f) {
 421                             info  = JSON.parse(File.read(info_f));
 422                         }
 423                          var target  = storedir+'/output/' + ret.filename
 424                          // flag it as done..
 425
 426
 427
 428
 429
 430                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 431                         print("GOT array sized: " + ret.data.length);
 432
 433                         File.writeBinaryArray(target,ret.data);
 434                         print("Wrote to file: " + target);
 435                         // get next..
 436                         return true;
 437                         //Seed.quit();
 438                         //_t.nsdownloadNext();
 439
 440
 441
 442                         break;
 443                 }
 444                 return true; //???
 445             }
 446
 447             if (ret.method == 'gatherlinks'){
 448                 // flag the page as parsed.
 449
 450                 if (browsePage) {
 451                     _t.moveToDone(browsePage);
 452
 453                 }
 454                 var sourcePage = browsePage;
 455                 browsePage = false;
 456                 print(typeof(ret.data));
 457                 if (typeof(ret.data) != 'object' ) {
 458                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 459                     ret.data= [];
 460
 461                 }
 462                 ret.data.forEach(function(ln) {
 463                     if (!ln.href.match(/^http[s]*:\/\//)) {
 464                         print("SKIP link: " + ln.href);
 465                         return;
 466                     }
 467                     ln.href= ln.href.replace(/#.*$/,'');
 468
 469                     if (!_t.checkdomain(ln.href) ) {
 470                         print("SKIP link (external domain): " + ln.href);
 471                         return;
 472                     }
 473
 474                     // this is just for our purposes..
 475
 476                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 477                         print("SKIP link (ingore unnamed pages): " + ln.href);
 478                         return;
 479                     }
 480
 481
 482
 483                     var fn = encodeURIComponent(ln.href);
 484
 485                     var dupe  = _t.dupeCheck(ln.href);
 486                     if (dupe) {
 487
 488                         if (dupe == downloaddir + '/'  + fn) {
 489                             var info = JSON.parse(File.read(dupe));
 490                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 491                                 print("SKIP link (in queue already): " + ln.href);
 492                                 return;
 493                             }
 494                             print("found a longer link for url")
 495                         } else {
 496                             print("SKIP link (in another queue): " + ln.href);
 497                             return;
 498                         }
 499                     }
 500
 501                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 502                         label : ln.label,
 503                         fromURL : sourcePage
 504                     })); // write an empyt file indicating it needs downloading..
 505                 });
 506                 var filesList = File.list(downloaddir);
 507
 508                 maxQueue =  filesList.length;
 509             }
 510
 511             if (ret.method == 'downloadpage'){
 512                 // got the results from download page:
 513                 // requesturl
 514                 // remove from downloadqueue.
 515
 516
 517                 //
 518                 try {
 519                     var mt = ret.contentType.split(';').shift();
 520                 } catch( e) {
 521                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 522                     mt='';
 523                 }
 524
 525                 switch(mt) {
 526
 527                     case '':
 528                         _t.moveToDone(ret.requesturl);
 529                         break;
 530
 531                     default:
 532                         var info = false;
 533                         var info_f = _t.dupeCheck(ret.requesturl);
 534                         if (info_f) {
 535                             info  = JSON.parse(File.read(info_f));
 536                         }
 537
 538                         _t.moveToDone(ret.requesturl);
 539                         // flag it as done..
 540
 541                         // write it...
 542                         var target  = _t.toFilename(ret.requesturl);
 543                         if (info && info.fromURL) {
 544                             var bn = decodeURIComponent(File.basename(target));
 545                             target = _t.toFilename(info.fromURL+'/'+ bn);
 546                         }
 547
 548
 549
 550
 551                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 552                         print("GOT array sized: " + ret.data.length);
 553
 554                         File.writeBinaryArray(target,ret.data);
 555                         print("Wrote to file: " + target);
 556                         break;
 557                 }
 558
 559                 // if it's HTML then add it to parse queue
 560                 // otehrwise save it.. and run the queue again.
 561
 562
 563                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 564             }
 565
 566
 567             if (ret.method == 'downloadhead'){
 568                 // got the results from download page:
 569                 // requesturl
 570                 // remove from downloadqueue.
 571
 572                 //
 573                 try {
 574                     var mt = ret.contentType.split(';').shift();
 575                 } catch( e) {
 576                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 577                     mt='';
 578                 }
 579
 580                 switch(mt) {
 581                     case 'text/html':
 582                         // add to parse QUEUE..
 583                         print("moving to parse queue");
 584                         _t.moveToParse(ret.requesturl);
 585                         break;
 586                     // stuf we do not care about..
 587                     case 'application/atom+xml':
 588                     case '':
 589                         print("moving to done queue");
 590                          _t.moveToDone(ret.requesturl);
 591                         break;
 592
 593                     default:
 594                         print("calling download file");
 595                         _t.downloadpage( ret.requesturl );
 596                         // keep it on the queue..
 597                         // do not run the queue..
 598                         return true;
 599
 600                 }
 601
 602                 // if it's HTML then add it to parse queue
 603                 // otehrwise save it.. and run the queue again.
 604
 605
 606                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 607             }
 608             _t.queuerun();
 609
 610             return true;
 611         });
 612
 613
 614         this.toFilename = function(url)
 615         {
 616             url = url.replace(/^http[s]*:\/\//, '');
 617             var p = url.split('/');
 618             p.unshift(storedir+'/output');
 619             for (var i =1 ;i < p.length; i++) {
 620                 p[i] = encodeURIComponent(p[i]);
 621
 622             }
 623             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 624             ret = p.join('/');
 625             var dir = File.dirname(ret);
 626             File.mkdirall(dir);
 627             return ret;
 628
 629         }
 630         this.checkdomain = function(comp)
 631         {
 632             var b = parseUri(this.uri);
 633             var d = parseUri(comp);
 634             return (d.host == b.host && d.protocol == b.protocol);
 635
 636
 637         }
 638
 639         this.dupeCheck = function(url)
 640         {
 641
 642            // order - return highest up the queue first..
 643             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 644                 return downloaddir +'/' + encodeURIComponent(url);
 645             }
 646              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 647                 return parsedir +'/' + encodeURIComponent(url);
 648             }
 649             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 650                 return donedir +'/' + encodeURIComponent(url);
 651             }
 652             return  false;
 653
 654
 655         }
 656         this.moveToParse = function(url)
 657         {
 658             var old = this.dupeCheck(url);
 659             var target =parsedir +'/' + 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
 670         this.moveToDownload= function(url)
 671         {
 672             var old = this.dupeCheck(url);
 673             var target =downloaddir +'/' + encodeURIComponent(url);
 674             if (old == target) {
 675                 return;
 676             }
 677             File.write(target, old ? File.read(old) : '');
 678             if (old) {
 679                 File.remove(old);
 680             }
 681
 682         }
 683         this.moveToDone= function(url)
 684         {
 685             var old = this.dupeCheck(url);
 686             var target = donedir +'/' + encodeURIComponent(url);
 687             if (old == target) {
 688                 return;
 689             }
 690             File.write(target, old ? File.read(old) : '');
 691             if (old) {
 692                 File.remove(old);
 693             }
 694
 695         }
 696
 697     }
 698 });
 699
 700 function parseUri (str) {
 701         var     o   = parseUri.options,
 702                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 703                 uri = {},
 704                 i   = 14;
 705
 706         while (i--) uri[o.key[i]] = m[i] || "";
 707
 708         uri[o.q.name] = {};
 709         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 710                 if ($1) uri[o.q.name][$1] = $2;
 711         });
 712
 713         return uri;
 714 };
 715
 716 parseUri.options = {
 717         strictMode: false,
 718         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 719         q:   {
 720                 name:   "queryKey",
 721                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 722         },
 723         parser: {
 724                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 725                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 726         }
 727 };