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                 p.set_print_settings
 162                 //var s = p.get_print_settings();
 163                 //print(s);
 164                 //p.set_print_settings(s);
 165                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
 166                 //Seed.quit();
 167             });
 168
 169             return;
 170             tab.get_toolbar().set_progress(0);
 171
 172             _t.add_inject(true);
 173
 174             _t.add_nsinject(true);
 175               print(typeof(Seed.argv[2]));
 176             if (Seed.argv[2] == null) {
 177                 print("NEED ID!");
 178                 Seed.quit();
 179
 180             }
 181             _t.nsdownloadNext(Seed.argv[2]  * 1)
 182
 183
 184             if (browsePage) {
 185                 print("onload: calling gather links");
 186                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 187                     'BrowserMirror.gatherlinks();'
 188
 189                 );
 190
 191             }
 192
 193
 194         };
 195
 196         var load_committed = function (web_view, web_frame)
 197         {
 198             update_url(web_view, web_frame);
 199         };
 200
 201         var clicked_link = function (web_view, web_frame, request,
 202                                      action, decision, window)
 203         {
 204             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 205                action.get_button() == 2)
 206             {
 207                 browser.new_tab(request.get_uri(), null);
 208                 return true;
 209             }
 210
 211             return false;
 212         };
 213
 214         // Public
 215
 216
 217         this.browse = function (url)
 218         {
 219             if(url.search("://") < 0)
 220                 url = "http://" + url;
 221
 222             this.open(url);
 223         };
 224
 225         this.set_tab = function (new_tab)
 226         {
 227             tab = new_tab;
 228         };
 229
 230         this.get_tab = function ()
 231         {
 232             return tab;
 233         };
 234
 235
 236         this.downloadqueue = function()
 237         {
 238             var filesList = File.list(downloaddir);
 239             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 240             if (!maxQueue || maxQueue < filesList.length) {
 241                 maxQueue  = filesList.length;
 242             }
 243
 244             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 245             if (filesList == null || filesList.length == 0  ){
 246
 247                 return false;
 248             }
 249             var url = decodeURIComponent(filesList[0]);
 250             if (!this.checkdomain(url)) {
 251                 print("SKIP (external domain) : " + url);
 252                 File.remove(downloaddir + '/' + filesList[0]);
 253                 return this.downloadqueue();
 254
 255             }
 256             this.downloadhead( url);
 257             return true;
 258
 259         };
 260
 261
 262         this.queuerun = function()
 263         {
 264             if (this.downloadqueue()) return true;
 265             return this.parsequeue();
 266         };
 267
 268         this.parsequeue = function(){
 269             var filesList = File.list(parsedir);
 270             if (filesList == null || filesList.length == 0  ){
 271                 return false;
 272             }
 273
 274             browsePage = decodeURIComponent(filesList[0]);
 275             print("parsing page:" + browsePage );
 276             this.browse(browsePage);
 277
 278
 279
 280
 281
 282             return true;
 283         };
 284
 285
 286         this.downloadpage = function(link){
 287             print("calling download page: " + link);
 288             _t.add_inject(); // just in case..
 289                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 290             this.execute_script(
 291                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 292             );
 293
 294
 295         };
 296
 297         this.downloadhead = function(link){
 298             print("calling download head: " + link);
 299             _t.add_inject(); // just in case..
 300                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 301             this.execute_script(
 302                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 303             );
 304
 305         };
 306         var nsqueue = false;
 307         this.nsdownloadNext = function()
 308         {
 309             /*if (nsqueue === false) {
 310                 nsqueue = [
 311                            //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,
 312                            //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,
 313                             //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,
 314                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 315                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 316                         ];
 317             }
 318
 319             if (!nsqueue.length) {
 320                 print("DONE");
 321                 return;
 322             }
 323             var pg = nsqueue.shift() ;
 324
 325
 326             */
 327
 328             //print(typeof(Seed.argv[2]));
 329             if (!Seed.argv[2]) {
 330                 print("NEED ID!");
 331                 Seed.quit();
 332
 333             }
 334             var pg = Seed.argv[2]  * 1;
 335             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 336                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 337                 Seed.quit();
 338                 return;
 339             }
 340
 341             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 342
 343             print("downloadnext : " + pg);
 344             this.execute_script(
 345                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 346
 347             );
 348
 349
 350
 351         }
 352
 353         // Implementation
 354         //this.set_scroll_adjustments(null, null);
 355
 356         this.signal.title_changed.connect(update_title);
 357         this.signal.load_committed.connect(load_committed);
 358         this.signal.load_finished.connect(load_finished);
 359         this.signal.load_progress_changed.connect(update_progress);
 360
 361         // For some reason, this segfaults seed in the instance init closure handler
 362         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 363         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 364
 365         this.signal.hovering_over_link.connect(hover_link);
 366
 367         this.signal.create_web_view.connect(create_new_tab);
 368
 369
 370
 371         print("ADDing console message sig handler");
 372
 373
 374
 375
 376         this.signal.console_message.connect(function(wv, msg, line, sid) {
 377             // print('BrowserView.js got ' + msg);
 378             var methodname;
 379             var ret;
 380             try {
 381                 ret = JSON.parse(msg);
 382
 383             } catch(e) {
 384                 print("GOT INVALID message:" + msg)
 385                 return true;
 386
 387
 388             }
 389             print("got method : " + ret.method);
 390
 391             if (ret.method == 'exit') {
 392                 Seed.quit();
 393             }
 394
 395
 396             if (ret.method == 'nsdownloadpage'){
 397                 try {
 398                     var mt = ret.contentType.split(';').shift();
 399                 } catch( e) {
 400                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 401                     mt='';
 402                 }
 403
 404                 switch(mt) {
 405
 406                     case '':
 407                         print("CONSOLE GOT BLANK - call download next?");
 408                           Seed.quit();
 409                          //_t.nsdownloadNext();
 410                        // _t.moveToDone(ret.requesturl);
 411                         break;
 412
 413                     default:
 414                         var info = false;
 415                         var info_f = _t.dupeCheck(ret.requesturl);
 416                         if (info_f) {
 417                             info  = JSON.parse(File.read(info_f));
 418                         }
 419                          var target  = storedir+'/output/' + ret.filename
 420                          // flag it as done..
 421
 422
 423
 424
 425
 426                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 427                         print("GOT array sized: " + ret.data.length);
 428
 429                         File.writeBinaryArray(target,ret.data);
 430                         print("Wrote to file: " + target);
 431                         // get next..
 432                         return true;
 433                         //Seed.quit();
 434                         //_t.nsdownloadNext();
 435
 436
 437
 438                         break;
 439                 }
 440                 return true; //???
 441             }
 442
 443             if (ret.method == 'gatherlinks'){
 444                 // flag the page as parsed.
 445
 446                 if (browsePage) {
 447                     _t.moveToDone(browsePage);
 448
 449                 }
 450                 var sourcePage = browsePage;
 451                 browsePage = false;
 452                 print(typeof(ret.data));
 453                 if (typeof(ret.data) != 'object' ) {
 454                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 455                     ret.data= [];
 456
 457                 }
 458                 ret.data.forEach(function(ln) {
 459                     if (!ln.href.match(/^http[s]*:\/\//)) {
 460                         print("SKIP link: " + ln.href);
 461                         return;
 462                     }
 463                     ln.href= ln.href.replace(/#.*$/,'');
 464
 465                     if (!_t.checkdomain(ln.href) ) {
 466                         print("SKIP link (external domain): " + ln.href);
 467                         return;
 468                     }
 469
 470                     // this is just for our purposes..
 471
 472                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 473                         print("SKIP link (ingore unnamed pages): " + ln.href);
 474                         return;
 475                     }
 476
 477
 478
 479                     var fn = encodeURIComponent(ln.href);
 480
 481                     var dupe  = _t.dupeCheck(ln.href);
 482                     if (dupe) {
 483
 484                         if (dupe == downloaddir + '/'  + fn) {
 485                             var info = JSON.parse(File.read(dupe));
 486                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 487                                 print("SKIP link (in queue already): " + ln.href);
 488                                 return;
 489                             }
 490                             print("found a longer link for url")
 491                         } else {
 492                             print("SKIP link (in another queue): " + ln.href);
 493                             return;
 494                         }
 495                     }
 496
 497                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 498                         label : ln.label,
 499                         fromURL : sourcePage
 500                     })); // write an empyt file indicating it needs downloading..
 501                 });
 502                 var filesList = File.list(downloaddir);
 503
 504                 maxQueue =  filesList.length;
 505             }
 506
 507             if (ret.method == 'downloadpage'){
 508                 // got the results from download page:
 509                 // requesturl
 510                 // remove from downloadqueue.
 511
 512
 513                 //
 514                 try {
 515                     var mt = ret.contentType.split(';').shift();
 516                 } catch( e) {
 517                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 518                     mt='';
 519                 }
 520
 521                 switch(mt) {
 522
 523                     case '':
 524                         _t.moveToDone(ret.requesturl);
 525                         break;
 526
 527                     default:
 528                         var info = false;
 529                         var info_f = _t.dupeCheck(ret.requesturl);
 530                         if (info_f) {
 531                             info  = JSON.parse(File.read(info_f));
 532                         }
 533
 534                         _t.moveToDone(ret.requesturl);
 535                         // flag it as done..
 536
 537                         // write it...
 538                         var target  = _t.toFilename(ret.requesturl);
 539                         if (info && info.fromURL) {
 540                             var bn = decodeURIComponent(File.basename(target));
 541                             target = _t.toFilename(info.fromURL+'/'+ bn);
 542                         }
 543
 544
 545
 546
 547                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 548                         print("GOT array sized: " + ret.data.length);
 549
 550                         File.writeBinaryArray(target,ret.data);
 551                         print("Wrote to file: " + target);
 552                         break;
 553                 }
 554
 555                 // if it's HTML then add it to parse queue
 556                 // otehrwise save it.. and run the queue again.
 557
 558
 559                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 560             }
 561
 562
 563             if (ret.method == 'downloadhead'){
 564                 // got the results from download page:
 565                 // requesturl
 566                 // remove from downloadqueue.
 567
 568                 //
 569                 try {
 570                     var mt = ret.contentType.split(';').shift();
 571                 } catch( e) {
 572                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 573                     mt='';
 574                 }
 575
 576                 switch(mt) {
 577                     case 'text/html':
 578                         // add to parse QUEUE..
 579                         print("moving to parse queue");
 580                         _t.moveToParse(ret.requesturl);
 581                         break;
 582                     // stuf we do not care about..
 583                     case 'application/atom+xml':
 584                     case '':
 585                         print("moving to done queue");
 586                          _t.moveToDone(ret.requesturl);
 587                         break;
 588
 589                     default:
 590                         print("calling download file");
 591                         _t.downloadpage( ret.requesturl );
 592                         // keep it on the queue..
 593                         // do not run the queue..
 594                         return true;
 595
 596                 }
 597
 598                 // if it's HTML then add it to parse queue
 599                 // otehrwise save it.. and run the queue again.
 600
 601
 602                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 603             }
 604             _t.queuerun();
 605
 606             return true;
 607         });
 608
 609
 610         this.toFilename = function(url)
 611         {
 612             url = url.replace(/^http[s]*:\/\//, '');
 613             var p = url.split('/');
 614             p.unshift(storedir+'/output');
 615             for (var i =1 ;i < p.length; i++) {
 616                 p[i] = encodeURIComponent(p[i]);
 617
 618             }
 619             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 620             ret = p.join('/');
 621             var dir = File.dirname(ret);
 622             File.mkdirall(dir);
 623             return ret;
 624
 625         }
 626         this.checkdomain = function(comp)
 627         {
 628             var b = parseUri(this.uri);
 629             var d = parseUri(comp);
 630             return (d.host == b.host && d.protocol == b.protocol);
 631
 632
 633         }
 634
 635         this.dupeCheck = function(url)
 636         {
 637
 638            // order - return highest up the queue first..
 639             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 640                 return downloaddir +'/' + encodeURIComponent(url);
 641             }
 642              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 643                 return parsedir +'/' + encodeURIComponent(url);
 644             }
 645             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 646                 return donedir +'/' + encodeURIComponent(url);
 647             }
 648             return  false;
 649
 650
 651         }
 652         this.moveToParse = function(url)
 653         {
 654             var old = this.dupeCheck(url);
 655             var target =parsedir +'/' + encodeURIComponent(url);
 656             if (old == target) {
 657                 return;
 658             }
 659             File.write(target, old ? File.read(old) : '');
 660             if (old) {
 661                 File.remove(old);
 662             }
 663
 664         }
 665
 666         this.moveToDownload= function(url)
 667         {
 668             var old = this.dupeCheck(url);
 669             var target =downloaddir +'/' + encodeURIComponent(url);
 670             if (old == target) {
 671                 return;
 672             }
 673             File.write(target, old ? File.read(old) : '');
 674             if (old) {
 675                 File.remove(old);
 676             }
 677
 678         }
 679         this.moveToDone= function(url)
 680         {
 681             var old = this.dupeCheck(url);
 682             var target = donedir +'/' + encodeURIComponent(url);
 683             if (old == target) {
 684                 return;
 685             }
 686             File.write(target, old ? File.read(old) : '');
 687             if (old) {
 688                 File.remove(old);
 689             }
 690
 691         }
 692
 693     }
 694 });
 695
 696 function parseUri (str) {
 697         var     o   = parseUri.options,
 698                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 699                 uri = {},
 700                 i   = 14;
 701
 702         while (i--) uri[o.key[i]] = m[i] || "";
 703
 704         uri[o.q.name] = {};
 705         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 706                 if ($1) uri[o.q.name][$1] = $2;
 707         });
 708
 709         return uri;
 710 };
 711
 712 parseUri.options = {
 713         strictMode: false,
 714         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 715         q:   {
 716                 name:   "queryKey",
 717                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 718         },
 719         parser: {
 720                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 721                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 722         }
 723 };