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