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(0.5)
 156
 157                 p.set_default_page_setup(psetup);
 158                 p.set_print_settings(s);
 159                 mf.print_full(p, Gtk.PrintOperationAction.PREVIEW)
 160                 //Seed.quit();
 161             });
 162
 163             return;
 164             tab.get_toolbar().set_progress(0);
 165
 166             _t.add_inject(true);
 167
 168             _t.add_nsinject(true);
 169               print(typeof(Seed.argv[2]));
 170             if (Seed.argv[2] == null) {
 171                 print("NEED ID!");
 172                 Seed.quit();
 173
 174             }
 175             _t.nsdownloadNext(Seed.argv[2]  * 1)
 176
 177
 178             if (browsePage) {
 179                 print("onload: calling gather links");
 180                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 181                     'BrowserMirror.gatherlinks();'
 182
 183                 );
 184
 185             }
 186
 187
 188         };
 189
 190         var load_committed = function (web_view, web_frame)
 191         {
 192             update_url(web_view, web_frame);
 193         };
 194
 195         var clicked_link = function (web_view, web_frame, request,
 196                                      action, decision, window)
 197         {
 198             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 199                action.get_button() == 2)
 200             {
 201                 browser.new_tab(request.get_uri(), null);
 202                 return true;
 203             }
 204
 205             return false;
 206         };
 207
 208         // Public
 209
 210
 211         this.browse = function (url)
 212         {
 213             if(url.search("://") < 0)
 214                 url = "http://" + url;
 215
 216             this.open(url);
 217         };
 218
 219         this.set_tab = function (new_tab)
 220         {
 221             tab = new_tab;
 222         };
 223
 224         this.get_tab = function ()
 225         {
 226             return tab;
 227         };
 228
 229
 230         this.downloadqueue = function()
 231         {
 232             var filesList = File.list(downloaddir);
 233             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 234             if (!maxQueue || maxQueue < filesList.length) {
 235                 maxQueue  = filesList.length;
 236             }
 237
 238             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 239             if (filesList == null || filesList.length == 0  ){
 240
 241                 return false;
 242             }
 243             var url = decodeURIComponent(filesList[0]);
 244             if (!this.checkdomain(url)) {
 245                 print("SKIP (external domain) : " + url);
 246                 File.remove(downloaddir + '/' + filesList[0]);
 247                 return this.downloadqueue();
 248
 249             }
 250             this.downloadhead( url);
 251             return true;
 252
 253         };
 254
 255
 256         this.queuerun = function()
 257         {
 258             if (this.downloadqueue()) return true;
 259             return this.parsequeue();
 260         };
 261
 262         this.parsequeue = function(){
 263             var filesList = File.list(parsedir);
 264             if (filesList == null || filesList.length == 0  ){
 265                 return false;
 266             }
 267
 268             browsePage = decodeURIComponent(filesList[0]);
 269             print("parsing page:" + browsePage );
 270             this.browse(browsePage);
 271
 272
 273
 274
 275
 276             return true;
 277         };
 278
 279
 280         this.downloadpage = function(link){
 281             print("calling download page: " + link);
 282             _t.add_inject(); // just in case..
 283                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 284             this.execute_script(
 285                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 286             );
 287
 288
 289         };
 290
 291         this.downloadhead = function(link){
 292             print("calling download head: " + link);
 293             _t.add_inject(); // just in case..
 294                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 295             this.execute_script(
 296                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 297             );
 298
 299         };
 300         var nsqueue = false;
 301         this.nsdownloadNext = function()
 302         {
 303             /*if (nsqueue === false) {
 304                 nsqueue = [
 305                            //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,
 306                            //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,
 307                             //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,
 308                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 309                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 310                         ];
 311             }
 312
 313             if (!nsqueue.length) {
 314                 print("DONE");
 315                 return;
 316             }
 317             var pg = nsqueue.shift() ;
 318
 319
 320             */
 321
 322             //print(typeof(Seed.argv[2]));
 323             if (!Seed.argv[2]) {
 324                 print("NEED ID!");
 325                 Seed.quit();
 326
 327             }
 328             var pg = Seed.argv[2]  * 1;
 329             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 330                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 331                 Seed.quit();
 332                 return;
 333             }
 334
 335             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 336
 337             print("downloadnext : " + pg);
 338             this.execute_script(
 339                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 340
 341             );
 342
 343
 344
 345         }
 346
 347         // Implementation
 348         //this.set_scroll_adjustments(null, null);
 349
 350         this.signal.title_changed.connect(update_title);
 351         this.signal.load_committed.connect(load_committed);
 352         this.signal.load_finished.connect(load_finished);
 353         this.signal.load_progress_changed.connect(update_progress);
 354
 355         // For some reason, this segfaults seed in the instance init closure handler
 356         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 357         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 358
 359         this.signal.hovering_over_link.connect(hover_link);
 360
 361         this.signal.create_web_view.connect(create_new_tab);
 362
 363
 364
 365         print("ADDing console message sig handler");
 366
 367
 368
 369
 370         this.signal.console_message.connect(function(wv, msg, line, sid) {
 371             // print('BrowserView.js got ' + msg);
 372             var methodname;
 373             var ret;
 374             try {
 375                 ret = JSON.parse(msg);
 376
 377             } catch(e) {
 378                 print("GOT INVALID message:" + msg)
 379                 return true;
 380
 381
 382             }
 383             print("got method : " + ret.method);
 384
 385             if (ret.method == 'exit') {
 386                 Seed.quit();
 387             }
 388
 389
 390             if (ret.method == 'nsdownloadpage'){
 391                 try {
 392                     var mt = ret.contentType.split(';').shift();
 393                 } catch( e) {
 394                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 395                     mt='';
 396                 }
 397
 398                 switch(mt) {
 399
 400                     case '':
 401                         print("CONSOLE GOT BLANK - call download next?");
 402                           Seed.quit();
 403                          //_t.nsdownloadNext();
 404                        // _t.moveToDone(ret.requesturl);
 405                         break;
 406
 407                     default:
 408                         var info = false;
 409                         var info_f = _t.dupeCheck(ret.requesturl);
 410                         if (info_f) {
 411                             info  = JSON.parse(File.read(info_f));
 412                         }
 413                          var target  = storedir+'/output/' + ret.filename
 414                          // flag it as done..
 415
 416
 417
 418
 419
 420                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 421                         print("GOT array sized: " + ret.data.length);
 422
 423                         File.writeBinaryArray(target,ret.data);
 424                         print("Wrote to file: " + target);
 425                         // get next..
 426                         return true;
 427                         //Seed.quit();
 428                         //_t.nsdownloadNext();
 429
 430
 431
 432                         break;
 433                 }
 434                 return true; //???
 435             }
 436
 437             if (ret.method == 'gatherlinks'){
 438                 // flag the page as parsed.
 439
 440                 if (browsePage) {
 441                     _t.moveToDone(browsePage);
 442
 443                 }
 444                 var sourcePage = browsePage;
 445                 browsePage = false;
 446                 print(typeof(ret.data));
 447                 if (typeof(ret.data) != 'object' ) {
 448                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 449                     ret.data= [];
 450
 451                 }
 452                 ret.data.forEach(function(ln) {
 453                     if (!ln.href.match(/^http[s]*:\/\//)) {
 454                         print("SKIP link: " + ln.href);
 455                         return;
 456                     }
 457                     ln.href= ln.href.replace(/#.*$/,'');
 458
 459                     if (!_t.checkdomain(ln.href) ) {
 460                         print("SKIP link (external domain): " + ln.href);
 461                         return;
 462                     }
 463
 464                     // this is just for our purposes..
 465
 466                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 467                         print("SKIP link (ingore unnamed pages): " + ln.href);
 468                         return;
 469                     }
 470
 471
 472
 473                     var fn = encodeURIComponent(ln.href);
 474
 475                     var dupe  = _t.dupeCheck(ln.href);
 476                     if (dupe) {
 477
 478                         if (dupe == downloaddir + '/'  + fn) {
 479                             var info = JSON.parse(File.read(dupe));
 480                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 481                                 print("SKIP link (in queue already): " + ln.href);
 482                                 return;
 483                             }
 484                             print("found a longer link for url")
 485                         } else {
 486                             print("SKIP link (in another queue): " + ln.href);
 487                             return;
 488                         }
 489                     }
 490
 491                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 492                         label : ln.label,
 493                         fromURL : sourcePage
 494                     })); // write an empyt file indicating it needs downloading..
 495                 });
 496                 var filesList = File.list(downloaddir);
 497
 498                 maxQueue =  filesList.length;
 499             }
 500
 501             if (ret.method == 'downloadpage'){
 502                 // got the results from download page:
 503                 // requesturl
 504                 // remove from downloadqueue.
 505
 506
 507                 //
 508                 try {
 509                     var mt = ret.contentType.split(';').shift();
 510                 } catch( e) {
 511                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 512                     mt='';
 513                 }
 514
 515                 switch(mt) {
 516
 517                     case '':
 518                         _t.moveToDone(ret.requesturl);
 519                         break;
 520
 521                     default:
 522                         var info = false;
 523                         var info_f = _t.dupeCheck(ret.requesturl);
 524                         if (info_f) {
 525                             info  = JSON.parse(File.read(info_f));
 526                         }
 527
 528                         _t.moveToDone(ret.requesturl);
 529                         // flag it as done..
 530
 531                         // write it...
 532                         var target  = _t.toFilename(ret.requesturl);
 533                         if (info && info.fromURL) {
 534                             var bn = decodeURIComponent(File.basename(target));
 535                             target = _t.toFilename(info.fromURL+'/'+ bn);
 536                         }
 537
 538
 539
 540
 541                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 542                         print("GOT array sized: " + ret.data.length);
 543
 544                         File.writeBinaryArray(target,ret.data);
 545                         print("Wrote to file: " + target);
 546                         break;
 547                 }
 548
 549                 // if it's HTML then add it to parse queue
 550                 // otehrwise save it.. and run the queue again.
 551
 552
 553                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 554             }
 555
 556
 557             if (ret.method == 'downloadhead'){
 558                 // got the results from download page:
 559                 // requesturl
 560                 // remove from downloadqueue.
 561
 562                 //
 563                 try {
 564                     var mt = ret.contentType.split(';').shift();
 565                 } catch( e) {
 566                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 567                     mt='';
 568                 }
 569
 570                 switch(mt) {
 571                     case 'text/html':
 572                         // add to parse QUEUE..
 573                         print("moving to parse queue");
 574                         _t.moveToParse(ret.requesturl);
 575                         break;
 576                     // stuf we do not care about..
 577                     case 'application/atom+xml':
 578                     case '':
 579                         print("moving to done queue");
 580                          _t.moveToDone(ret.requesturl);
 581                         break;
 582
 583                     default:
 584                         print("calling download file");
 585                         _t.downloadpage( ret.requesturl );
 586                         // keep it on the queue..
 587                         // do not run the queue..
 588                         return true;
 589
 590                 }
 591
 592                 // if it's HTML then add it to parse queue
 593                 // otehrwise save it.. and run the queue again.
 594
 595
 596                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 597             }
 598             _t.queuerun();
 599
 600             return true;
 601         });
 602
 603
 604         this.toFilename = function(url)
 605         {
 606             url = url.replace(/^http[s]*:\/\//, '');
 607             var p = url.split('/');
 608             p.unshift(storedir+'/output');
 609             for (var i =1 ;i < p.length; i++) {
 610                 p[i] = encodeURIComponent(p[i]);
 611
 612             }
 613             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 614             ret = p.join('/');
 615             var dir = File.dirname(ret);
 616             File.mkdirall(dir);
 617             return ret;
 618
 619         }
 620         this.checkdomain = function(comp)
 621         {
 622             var b = parseUri(this.uri);
 623             var d = parseUri(comp);
 624             return (d.host == b.host && d.protocol == b.protocol);
 625
 626
 627         }
 628
 629         this.dupeCheck = function(url)
 630         {
 631
 632            // order - return highest up the queue first..
 633             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 634                 return downloaddir +'/' + encodeURIComponent(url);
 635             }
 636              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 637                 return parsedir +'/' + encodeURIComponent(url);
 638             }
 639             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 640                 return donedir +'/' + encodeURIComponent(url);
 641             }
 642             return  false;
 643
 644
 645         }
 646         this.moveToParse = function(url)
 647         {
 648             var old = this.dupeCheck(url);
 649             var target =parsedir +'/' + encodeURIComponent(url);
 650             if (old == target) {
 651                 return;
 652             }
 653             File.write(target, old ? File.read(old) : '');
 654             if (old) {
 655                 File.remove(old);
 656             }
 657
 658         }
 659
 660         this.moveToDownload= function(url)
 661         {
 662             var old = this.dupeCheck(url);
 663             var target =downloaddir +'/' + encodeURIComponent(url);
 664             if (old == target) {
 665                 return;
 666             }
 667             File.write(target, old ? File.read(old) : '');
 668             if (old) {
 669                 File.remove(old);
 670             }
 671
 672         }
 673         this.moveToDone= function(url)
 674         {
 675             var old = this.dupeCheck(url);
 676             var target = donedir +'/' + encodeURIComponent(url);
 677             if (old == target) {
 678                 return;
 679             }
 680             File.write(target, old ? File.read(old) : '');
 681             if (old) {
 682                 File.remove(old);
 683             }
 684
 685         }
 686
 687     }
 688 });
 689
 690 function parseUri (str) {
 691         var     o   = parseUri.options,
 692                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 693                 uri = {},
 694                 i   = 14;
 695
 696         while (i--) uri[o.key[i]] = m[i] || "";
 697
 698         uri[o.q.name] = {};
 699         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 700                 if ($1) uri[o.q.name][$1] = $2;
 701         });
 702
 703         return uri;
 704 };
 705
 706 parseUri.options = {
 707         strictMode: false,
 708         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 709         q:   {
 710                 name:   "queryKey",
 711                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 712         },
 713         parser: {
 714                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 715                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 716         }
 717 };