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, 5000, function() {
 138                 _t.set_zoom_level(0.5);
 139 //                _t.open("http://sg.lifestyleasia.com/videos/video-british-polo-day-singapore-2012/");
 140                 var mf = _t.get_main_frame();
 141                 var p = new Gtk.PrintOperation({ export_filename : '/home/chris/test_pdf/test2.pdf' });
 142
 143                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
 144                 Seed.quit();
 145             });
 146
 147             return;
 148             tab.get_toolbar().set_progress(0);
 149
 150             _t.add_inject(true);
 151
 152             _t.add_nsinject(true);
 153               print(typeof(Seed.argv[2]));
 154             if (Seed.argv[2] == null) {
 155                 print("NEED ID!");
 156                 Seed.quit();
 157
 158             }
 159             _t.nsdownloadNext(Seed.argv[2]  * 1)
 160
 161
 162             if (browsePage) {
 163                 print("onload: calling gather links");
 164                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 165                     'BrowserMirror.gatherlinks();'
 166
 167                 );
 168
 169             }
 170
 171
 172         };
 173
 174         var load_committed = function (web_view, web_frame)
 175         {
 176             update_url(web_view, web_frame);
 177         };
 178
 179         var clicked_link = function (web_view, web_frame, request,
 180                                      action, decision, window)
 181         {
 182             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 183                action.get_button() == 2)
 184             {
 185                 browser.new_tab(request.get_uri(), null);
 186                 return true;
 187             }
 188
 189             return false;
 190         };
 191
 192         // Public
 193
 194
 195         this.browse = function (url)
 196         {
 197             if(url.search("://") < 0)
 198                 url = "http://" + url;
 199
 200             this.open(url);
 201         };
 202
 203         this.set_tab = function (new_tab)
 204         {
 205             tab = new_tab;
 206         };
 207
 208         this.get_tab = function ()
 209         {
 210             return tab;
 211         };
 212
 213
 214         this.downloadqueue = function()
 215         {
 216             var filesList = File.list(downloaddir);
 217             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 218             if (!maxQueue || maxQueue < filesList.length) {
 219                 maxQueue  = filesList.length;
 220             }
 221
 222             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 223             if (filesList == null || filesList.length == 0  ){
 224
 225                 return false;
 226             }
 227             var url = decodeURIComponent(filesList[0]);
 228             if (!this.checkdomain(url)) {
 229                 print("SKIP (external domain) : " + url);
 230                 File.remove(downloaddir + '/' + filesList[0]);
 231                 return this.downloadqueue();
 232
 233             }
 234             this.downloadhead( url);
 235             return true;
 236
 237         };
 238
 239
 240         this.queuerun = function()
 241         {
 242             if (this.downloadqueue()) return true;
 243             return this.parsequeue();
 244         };
 245
 246         this.parsequeue = function(){
 247             var filesList = File.list(parsedir);
 248             if (filesList == null || filesList.length == 0  ){
 249                 return false;
 250             }
 251
 252             browsePage = decodeURIComponent(filesList[0]);
 253             print("parsing page:" + browsePage );
 254             this.browse(browsePage);
 255
 256
 257
 258
 259
 260             return true;
 261         };
 262
 263
 264         this.downloadpage = function(link){
 265             print("calling download page: " + link);
 266             _t.add_inject(); // just in case..
 267                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 268             this.execute_script(
 269                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 270             );
 271
 272
 273         };
 274
 275         this.downloadhead = function(link){
 276             print("calling download head: " + link);
 277             _t.add_inject(); // just in case..
 278                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 279             this.execute_script(
 280                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 281             );
 282
 283         };
 284         var nsqueue = false;
 285         this.nsdownloadNext = function()
 286         {
 287             /*if (nsqueue === false) {
 288                 nsqueue = [
 289                            //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,
 290                            //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,
 291                             //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,
 292                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 293                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 294                         ];
 295             }
 296
 297             if (!nsqueue.length) {
 298                 print("DONE");
 299                 return;
 300             }
 301             var pg = nsqueue.shift() ;
 302
 303
 304             */
 305
 306             //print(typeof(Seed.argv[2]));
 307             if (!Seed.argv[2]) {
 308                 print("NEED ID!");
 309                 Seed.quit();
 310
 311             }
 312             var pg = Seed.argv[2]  * 1;
 313             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 314                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 315                 Seed.quit();
 316                 return;
 317             }
 318
 319             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 320
 321             print("downloadnext : " + pg);
 322             this.execute_script(
 323                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 324
 325             );
 326
 327
 328
 329         }
 330
 331         // Implementation
 332         //this.set_scroll_adjustments(null, null);
 333
 334         this.signal.title_changed.connect(update_title);
 335         this.signal.load_committed.connect(load_committed);
 336         this.signal.load_finished.connect(load_finished);
 337         this.signal.load_progress_changed.connect(update_progress);
 338
 339         // For some reason, this segfaults seed in the instance init closure handler
 340         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 341         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 342
 343         this.signal.hovering_over_link.connect(hover_link);
 344
 345         this.signal.create_web_view.connect(create_new_tab);
 346
 347
 348         print("ADDing console message sig handler");
 349
 350
 351
 352
 353         this.signal.console_message.connect(function(wv, msg, line, sid) {
 354             // print('BrowserView.js got ' + msg);
 355             var methodname;
 356             var ret;
 357             try {
 358                 ret = JSON.parse(msg);
 359
 360             } catch(e) {
 361                 print("GOT INVALID message:" + msg)
 362                 return true;
 363
 364
 365             }
 366             print("got method : " + ret.method);
 367
 368             if (ret.method == 'exit') {
 369                 Seed.quit();
 370             }
 371
 372
 373             if (ret.method == 'nsdownloadpage'){
 374                 try {
 375                     var mt = ret.contentType.split(';').shift();
 376                 } catch( e) {
 377                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 378                     mt='';
 379                 }
 380
 381                 switch(mt) {
 382
 383                     case '':
 384                         print("CONSOLE GOT BLANK - call download next?");
 385                           Seed.quit();
 386                          //_t.nsdownloadNext();
 387                        // _t.moveToDone(ret.requesturl);
 388                         break;
 389
 390                     default:
 391                         var info = false;
 392                         var info_f = _t.dupeCheck(ret.requesturl);
 393                         if (info_f) {
 394                             info  = JSON.parse(File.read(info_f));
 395                         }
 396                          var target  = storedir+'/output/' + ret.filename
 397                          // flag it as done..
 398
 399
 400
 401
 402
 403                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 404                         print("GOT array sized: " + ret.data.length);
 405
 406                         File.writeBinaryArray(target,ret.data);
 407                         print("Wrote to file: " + target);
 408                         // get next..
 409                         return true;
 410                         //Seed.quit();
 411                         //_t.nsdownloadNext();
 412
 413
 414
 415                         break;
 416                 }
 417                 return true; //???
 418             }
 419
 420             if (ret.method == 'gatherlinks'){
 421                 // flag the page as parsed.
 422
 423                 if (browsePage) {
 424                     _t.moveToDone(browsePage);
 425
 426                 }
 427                 var sourcePage = browsePage;
 428                 browsePage = false;
 429                 print(typeof(ret.data));
 430                 if (typeof(ret.data) != 'object' ) {
 431                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 432                     ret.data= [];
 433
 434                 }
 435                 ret.data.forEach(function(ln) {
 436                     if (!ln.href.match(/^http[s]*:\/\//)) {
 437                         print("SKIP link: " + ln.href);
 438                         return;
 439                     }
 440                     ln.href= ln.href.replace(/#.*$/,'');
 441
 442                     if (!_t.checkdomain(ln.href) ) {
 443                         print("SKIP link (external domain): " + ln.href);
 444                         return;
 445                     }
 446
 447                     // this is just for our purposes..
 448
 449                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 450                         print("SKIP link (ingore unnamed pages): " + ln.href);
 451                         return;
 452                     }
 453
 454
 455
 456                     var fn = encodeURIComponent(ln.href);
 457
 458                     var dupe  = _t.dupeCheck(ln.href);
 459                     if (dupe) {
 460
 461                         if (dupe == downloaddir + '/'  + fn) {
 462                             var info = JSON.parse(File.read(dupe));
 463                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 464                                 print("SKIP link (in queue already): " + ln.href);
 465                                 return;
 466                             }
 467                             print("found a longer link for url")
 468                         } else {
 469                             print("SKIP link (in another queue): " + ln.href);
 470                             return;
 471                         }
 472                     }
 473
 474                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 475                         label : ln.label,
 476                         fromURL : sourcePage
 477                     })); // write an empyt file indicating it needs downloading..
 478                 });
 479                 var filesList = File.list(downloaddir);
 480
 481                 maxQueue =  filesList.length;
 482             }
 483
 484             if (ret.method == 'downloadpage'){
 485                 // got the results from download page:
 486                 // requesturl
 487                 // remove from downloadqueue.
 488
 489
 490                 //
 491                 try {
 492                     var mt = ret.contentType.split(';').shift();
 493                 } catch( e) {
 494                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 495                     mt='';
 496                 }
 497
 498                 switch(mt) {
 499
 500                     case '':
 501                         _t.moveToDone(ret.requesturl);
 502                         break;
 503
 504                     default:
 505                         var info = false;
 506                         var info_f = _t.dupeCheck(ret.requesturl);
 507                         if (info_f) {
 508                             info  = JSON.parse(File.read(info_f));
 509                         }
 510
 511                         _t.moveToDone(ret.requesturl);
 512                         // flag it as done..
 513
 514                         // write it...
 515                         var target  = _t.toFilename(ret.requesturl);
 516                         if (info && info.fromURL) {
 517                             var bn = decodeURIComponent(File.basename(target));
 518                             target = _t.toFilename(info.fromURL+'/'+ bn);
 519                         }
 520
 521
 522
 523
 524                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 525                         print("GOT array sized: " + ret.data.length);
 526
 527                         File.writeBinaryArray(target,ret.data);
 528                         print("Wrote to file: " + target);
 529                         break;
 530                 }
 531
 532                 // if it's HTML then add it to parse queue
 533                 // otehrwise save it.. and run the queue again.
 534
 535
 536                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 537             }
 538
 539
 540             if (ret.method == 'downloadhead'){
 541                 // got the results from download page:
 542                 // requesturl
 543                 // remove from downloadqueue.
 544
 545                 //
 546                 try {
 547                     var mt = ret.contentType.split(';').shift();
 548                 } catch( e) {
 549                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 550                     mt='';
 551                 }
 552
 553                 switch(mt) {
 554                     case 'text/html':
 555                         // add to parse QUEUE..
 556                         print("moving to parse queue");
 557                         _t.moveToParse(ret.requesturl);
 558                         break;
 559                     // stuf we do not care about..
 560                     case 'application/atom+xml':
 561                     case '':
 562                         print("moving to done queue");
 563                          _t.moveToDone(ret.requesturl);
 564                         break;
 565
 566                     default:
 567                         print("calling download file");
 568                         _t.downloadpage( ret.requesturl );
 569                         // keep it on the queue..
 570                         // do not run the queue..
 571                         return true;
 572
 573                 }
 574
 575                 // if it's HTML then add it to parse queue
 576                 // otehrwise save it.. and run the queue again.
 577
 578
 579                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 580             }
 581             _t.queuerun();
 582
 583             return true;
 584         });
 585
 586
 587         this.toFilename = function(url)
 588         {
 589             url = url.replace(/^http[s]*:\/\//, '');
 590             var p = url.split('/');
 591             p.unshift(storedir+'/output');
 592             for (var i =1 ;i < p.length; i++) {
 593                 p[i] = encodeURIComponent(p[i]);
 594
 595             }
 596             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 597             ret = p.join('/');
 598             var dir = File.dirname(ret);
 599             File.mkdirall(dir);
 600             return ret;
 601
 602         }
 603         this.checkdomain = function(comp)
 604         {
 605             var b = parseUri(this.uri);
 606             var d = parseUri(comp);
 607             return (d.host == b.host && d.protocol == b.protocol);
 608
 609
 610         }
 611
 612         this.dupeCheck = function(url)
 613         {
 614
 615            // order - return highest up the queue first..
 616             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 617                 return downloaddir +'/' + encodeURIComponent(url);
 618             }
 619              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 620                 return parsedir +'/' + encodeURIComponent(url);
 621             }
 622             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 623                 return donedir +'/' + encodeURIComponent(url);
 624             }
 625             return  false;
 626
 627
 628         }
 629         this.moveToParse = function(url)
 630         {
 631             var old = this.dupeCheck(url);
 632             var target =parsedir +'/' + encodeURIComponent(url);
 633             if (old == target) {
 634                 return;
 635             }
 636             File.write(target, old ? File.read(old) : '');
 637             if (old) {
 638                 File.remove(old);
 639             }
 640
 641         }
 642
 643         this.moveToDownload= function(url)
 644         {
 645             var old = this.dupeCheck(url);
 646             var target =downloaddir +'/' + encodeURIComponent(url);
 647             if (old == target) {
 648                 return;
 649             }
 650             File.write(target, old ? File.read(old) : '');
 651             if (old) {
 652                 File.remove(old);
 653             }
 654
 655         }
 656         this.moveToDone= function(url)
 657         {
 658             var old = this.dupeCheck(url);
 659             var target = donedir +'/' + encodeURIComponent(url);
 660             if (old == target) {
 661                 return;
 662             }
 663             File.write(target, old ? File.read(old) : '');
 664             if (old) {
 665                 File.remove(old);
 666             }
 667
 668         }
 669
 670     }
 671 });
 672
 673 function parseUri (str) {
 674         var     o   = parseUri.options,
 675                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 676                 uri = {},
 677                 i   = 14;
 678
 679         while (i--) uri[o.key[i]] = m[i] || "";
 680
 681         uri[o.q.name] = {};
 682         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 683                 if ($1) uri[o.q.name][$1] = $2;
 684         });
 685
 686         return uri;
 687 };
 688
 689 parseUri.options = {
 690         strictMode: false,
 691         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 692         q:   {
 693                 name:   "queryKey",
 694                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 695         },
 696         parser: {
 697                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 698                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 699         }
 700 };