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