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
 155                 //s.set_paper_size('iso_a4');
 156
 157                 p.set_default_page_setup(psetup);
 158
 159                 var s = p.get_print_settings();
 160                 print(s);
 161                 //p.set_print_settings(s);
 162                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
 163                 //Seed.quit();
 164             });
 165
 166             return;
 167             tab.get_toolbar().set_progress(0);
 168
 169             _t.add_inject(true);
 170
 171             _t.add_nsinject(true);
 172               print(typeof(Seed.argv[2]));
 173             if (Seed.argv[2] == null) {
 174                 print("NEED ID!");
 175                 Seed.quit();
 176
 177             }
 178             _t.nsdownloadNext(Seed.argv[2]  * 1)
 179
 180
 181             if (browsePage) {
 182                 print("onload: calling gather links");
 183                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 184                     'BrowserMirror.gatherlinks();'
 185
 186                 );
 187
 188             }
 189
 190
 191         };
 192
 193         var load_committed = function (web_view, web_frame)
 194         {
 195             update_url(web_view, web_frame);
 196         };
 197
 198         var clicked_link = function (web_view, web_frame, request,
 199                                      action, decision, window)
 200         {
 201             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 202                action.get_button() == 2)
 203             {
 204                 browser.new_tab(request.get_uri(), null);
 205                 return true;
 206             }
 207
 208             return false;
 209         };
 210
 211         // Public
 212
 213
 214         this.browse = function (url)
 215         {
 216             if(url.search("://") < 0)
 217                 url = "http://" + url;
 218
 219             this.open(url);
 220         };
 221
 222         this.set_tab = function (new_tab)
 223         {
 224             tab = new_tab;
 225         };
 226
 227         this.get_tab = function ()
 228         {
 229             return tab;
 230         };
 231
 232
 233         this.downloadqueue = function()
 234         {
 235             var filesList = File.list(downloaddir);
 236             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 237             if (!maxQueue || maxQueue < filesList.length) {
 238                 maxQueue  = filesList.length;
 239             }
 240
 241             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 242             if (filesList == null || filesList.length == 0  ){
 243
 244                 return false;
 245             }
 246             var url = decodeURIComponent(filesList[0]);
 247             if (!this.checkdomain(url)) {
 248                 print("SKIP (external domain) : " + url);
 249                 File.remove(downloaddir + '/' + filesList[0]);
 250                 return this.downloadqueue();
 251
 252             }
 253             this.downloadhead( url);
 254             return true;
 255
 256         };
 257
 258
 259         this.queuerun = function()
 260         {
 261             if (this.downloadqueue()) return true;
 262             return this.parsequeue();
 263         };
 264
 265         this.parsequeue = function(){
 266             var filesList = File.list(parsedir);
 267             if (filesList == null || filesList.length == 0  ){
 268                 return false;
 269             }
 270
 271             browsePage = decodeURIComponent(filesList[0]);
 272             print("parsing page:" + browsePage );
 273             this.browse(browsePage);
 274
 275
 276
 277
 278
 279             return true;
 280         };
 281
 282
 283         this.downloadpage = function(link){
 284             print("calling download page: " + link);
 285             _t.add_inject(); // just in case..
 286                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 287             this.execute_script(
 288                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 289             );
 290
 291
 292         };
 293
 294         this.downloadhead = function(link){
 295             print("calling download head: " + link);
 296             _t.add_inject(); // just in case..
 297                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 298             this.execute_script(
 299                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 300             );
 301
 302         };
 303         var nsqueue = false;
 304         this.nsdownloadNext = function()
 305         {
 306             /*if (nsqueue === false) {
 307                 nsqueue = [
 308                            //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,
 309                            //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,
 310                             //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,
 311                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 312                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 313                         ];
 314             }
 315
 316             if (!nsqueue.length) {
 317                 print("DONE");
 318                 return;
 319             }
 320             var pg = nsqueue.shift() ;
 321
 322
 323             */
 324
 325             //print(typeof(Seed.argv[2]));
 326             if (!Seed.argv[2]) {
 327                 print("NEED ID!");
 328                 Seed.quit();
 329
 330             }
 331             var pg = Seed.argv[2]  * 1;
 332             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 333                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 334                 Seed.quit();
 335                 return;
 336             }
 337
 338             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 339
 340             print("downloadnext : " + pg);
 341             this.execute_script(
 342                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 343
 344             );
 345
 346
 347
 348         }
 349
 350         // Implementation
 351         //this.set_scroll_adjustments(null, null);
 352
 353         this.signal.title_changed.connect(update_title);
 354         this.signal.load_committed.connect(load_committed);
 355         this.signal.load_finished.connect(load_finished);
 356         this.signal.load_progress_changed.connect(update_progress);
 357
 358         // For some reason, this segfaults seed in the instance init closure handler
 359         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 360         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 361
 362         this.signal.hovering_over_link.connect(hover_link);
 363
 364         this.signal.create_web_view.connect(create_new_tab);
 365
 366
 367
 368         print("ADDing console message sig handler");
 369
 370
 371
 372
 373         this.signal.console_message.connect(function(wv, msg, line, sid) {
 374             // print('BrowserView.js got ' + msg);
 375             var methodname;
 376             var ret;
 377             try {
 378                 ret = JSON.parse(msg);
 379
 380             } catch(e) {
 381                 print("GOT INVALID message:" + msg)
 382                 return true;
 383
 384
 385             }
 386             print("got method : " + ret.method);
 387
 388             if (ret.method == 'exit') {
 389                 Seed.quit();
 390             }
 391
 392
 393             if (ret.method == 'nsdownloadpage'){
 394                 try {
 395                     var mt = ret.contentType.split(';').shift();
 396                 } catch( e) {
 397                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 398                     mt='';
 399                 }
 400
 401                 switch(mt) {
 402
 403                     case '':
 404                         print("CONSOLE GOT BLANK - call download next?");
 405                           Seed.quit();
 406                          //_t.nsdownloadNext();
 407                        // _t.moveToDone(ret.requesturl);
 408                         break;
 409
 410                     default:
 411                         var info = false;
 412                         var info_f = _t.dupeCheck(ret.requesturl);
 413                         if (info_f) {
 414                             info  = JSON.parse(File.read(info_f));
 415                         }
 416                          var target  = storedir+'/output/' + ret.filename
 417                          // flag it as done..
 418
 419
 420
 421
 422
 423                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 424                         print("GOT array sized: " + ret.data.length);
 425
 426                         File.writeBinaryArray(target,ret.data);
 427                         print("Wrote to file: " + target);
 428                         // get next..
 429                         return true;
 430                         //Seed.quit();
 431                         //_t.nsdownloadNext();
 432
 433
 434
 435                         break;
 436                 }
 437                 return true; //???
 438             }
 439
 440             if (ret.method == 'gatherlinks'){
 441                 // flag the page as parsed.
 442
 443                 if (browsePage) {
 444                     _t.moveToDone(browsePage);
 445
 446                 }
 447                 var sourcePage = browsePage;
 448                 browsePage = false;
 449                 print(typeof(ret.data));
 450                 if (typeof(ret.data) != 'object' ) {
 451                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 452                     ret.data= [];
 453
 454                 }
 455                 ret.data.forEach(function(ln) {
 456                     if (!ln.href.match(/^http[s]*:\/\//)) {
 457                         print("SKIP link: " + ln.href);
 458                         return;
 459                     }
 460                     ln.href= ln.href.replace(/#.*$/,'');
 461
 462                     if (!_t.checkdomain(ln.href) ) {
 463                         print("SKIP link (external domain): " + ln.href);
 464                         return;
 465                     }
 466
 467                     // this is just for our purposes..
 468
 469                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 470                         print("SKIP link (ingore unnamed pages): " + ln.href);
 471                         return;
 472                     }
 473
 474
 475
 476                     var fn = encodeURIComponent(ln.href);
 477
 478                     var dupe  = _t.dupeCheck(ln.href);
 479                     if (dupe) {
 480
 481                         if (dupe == downloaddir + '/'  + fn) {
 482                             var info = JSON.parse(File.read(dupe));
 483                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 484                                 print("SKIP link (in queue already): " + ln.href);
 485                                 return;
 486                             }
 487                             print("found a longer link for url")
 488                         } else {
 489                             print("SKIP link (in another queue): " + ln.href);
 490                             return;
 491                         }
 492                     }
 493
 494                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 495                         label : ln.label,
 496                         fromURL : sourcePage
 497                     })); // write an empyt file indicating it needs downloading..
 498                 });
 499                 var filesList = File.list(downloaddir);
 500
 501                 maxQueue =  filesList.length;
 502             }
 503
 504             if (ret.method == 'downloadpage'){
 505                 // got the results from download page:
 506                 // requesturl
 507                 // remove from downloadqueue.
 508
 509
 510                 //
 511                 try {
 512                     var mt = ret.contentType.split(';').shift();
 513                 } catch( e) {
 514                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 515                     mt='';
 516                 }
 517
 518                 switch(mt) {
 519
 520                     case '':
 521                         _t.moveToDone(ret.requesturl);
 522                         break;
 523
 524                     default:
 525                         var info = false;
 526                         var info_f = _t.dupeCheck(ret.requesturl);
 527                         if (info_f) {
 528                             info  = JSON.parse(File.read(info_f));
 529                         }
 530
 531                         _t.moveToDone(ret.requesturl);
 532                         // flag it as done..
 533
 534                         // write it...
 535                         var target  = _t.toFilename(ret.requesturl);
 536                         if (info && info.fromURL) {
 537                             var bn = decodeURIComponent(File.basename(target));
 538                             target = _t.toFilename(info.fromURL+'/'+ bn);
 539                         }
 540
 541
 542
 543
 544                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 545                         print("GOT array sized: " + ret.data.length);
 546
 547                         File.writeBinaryArray(target,ret.data);
 548                         print("Wrote to file: " + target);
 549                         break;
 550                 }
 551
 552                 // if it's HTML then add it to parse queue
 553                 // otehrwise save it.. and run the queue again.
 554
 555
 556                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 557             }
 558
 559
 560             if (ret.method == 'downloadhead'){
 561                 // got the results from download page:
 562                 // requesturl
 563                 // remove from downloadqueue.
 564
 565                 //
 566                 try {
 567                     var mt = ret.contentType.split(';').shift();
 568                 } catch( e) {
 569                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 570                     mt='';
 571                 }
 572
 573                 switch(mt) {
 574                     case 'text/html':
 575                         // add to parse QUEUE..
 576                         print("moving to parse queue");
 577                         _t.moveToParse(ret.requesturl);
 578                         break;
 579                     // stuf we do not care about..
 580                     case 'application/atom+xml':
 581                     case '':
 582                         print("moving to done queue");
 583                          _t.moveToDone(ret.requesturl);
 584                         break;
 585
 586                     default:
 587                         print("calling download file");
 588                         _t.downloadpage( ret.requesturl );
 589                         // keep it on the queue..
 590                         // do not run the queue..
 591                         return true;
 592
 593                 }
 594
 595                 // if it's HTML then add it to parse queue
 596                 // otehrwise save it.. and run the queue again.
 597
 598
 599                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 600             }
 601             _t.queuerun();
 602
 603             return true;
 604         });
 605
 606
 607         this.toFilename = function(url)
 608         {
 609             url = url.replace(/^http[s]*:\/\//, '');
 610             var p = url.split('/');
 611             p.unshift(storedir+'/output');
 612             for (var i =1 ;i < p.length; i++) {
 613                 p[i] = encodeURIComponent(p[i]);
 614
 615             }
 616             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 617             ret = p.join('/');
 618             var dir = File.dirname(ret);
 619             File.mkdirall(dir);
 620             return ret;
 621
 622         }
 623         this.checkdomain = function(comp)
 624         {
 625             var b = parseUri(this.uri);
 626             var d = parseUri(comp);
 627             return (d.host == b.host && d.protocol == b.protocol);
 628
 629
 630         }
 631
 632         this.dupeCheck = function(url)
 633         {
 634
 635            // order - return highest up the queue first..
 636             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 637                 return downloaddir +'/' + encodeURIComponent(url);
 638             }
 639              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 640                 return parsedir +'/' + encodeURIComponent(url);
 641             }
 642             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 643                 return donedir +'/' + encodeURIComponent(url);
 644             }
 645             return  false;
 646
 647
 648         }
 649         this.moveToParse = function(url)
 650         {
 651             var old = this.dupeCheck(url);
 652             var target =parsedir +'/' + encodeURIComponent(url);
 653             if (old == target) {
 654                 return;
 655             }
 656             File.write(target, old ? File.read(old) : '');
 657             if (old) {
 658                 File.remove(old);
 659             }
 660
 661         }
 662
 663         this.moveToDownload= function(url)
 664         {
 665             var old = this.dupeCheck(url);
 666             var target =downloaddir +'/' + encodeURIComponent(url);
 667             if (old == target) {
 668                 return;
 669             }
 670             File.write(target, old ? File.read(old) : '');
 671             if (old) {
 672                 File.remove(old);
 673             }
 674
 675         }
 676         this.moveToDone= function(url)
 677         {
 678             var old = this.dupeCheck(url);
 679             var target = donedir +'/' + encodeURIComponent(url);
 680             if (old == target) {
 681                 return;
 682             }
 683             File.write(target, old ? File.read(old) : '');
 684             if (old) {
 685                 File.remove(old);
 686             }
 687
 688         }
 689
 690     }
 691 });
 692
 693 function parseUri (str) {
 694         var     o   = parseUri.options,
 695                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 696                 uri = {},
 697                 i   = 14;
 698
 699         while (i--) uri[o.key[i]] = m[i] || "";
 700
 701         uri[o.q.name] = {};
 702         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 703                 if ($1) uri[o.q.name][$1] = $2;
 704         });
 705
 706         return uri;
 707 };
 708
 709 parseUri.options = {
 710         strictMode: false,
 711         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 712         q:   {
 713                 name:   "queryKey",
 714                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 715         },
 716         parser: {
 717                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 718                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 719         }
 720 };