git.roojs.org Git - app.wkmirror/blob - BrowserView.js

   1 Gtk = imports.gi.Gtk;
   2 GLib = imports.gi.GLib;
   3 WebKit = imports.gi.WebKit;
   4
   5 TabbedBrowser = imports.TabbedBrowser;
   6 File = imports.File.File;
   7
   8 base64 = imports.base64.base64;
   9
  10 BrowserView = new GType({
  11     parent: WebKit.WebView.type,
  12     name: "BrowserView",
  13     init: function ()
  14     {
  15         // Private
  16
  17
  18         var _t = this;
  19
  20         var tab;
  21         var browsePage = false;
  22         var maxQueue = 0;
  23         var injected = {};
  24
  25         var storedir = '/home/alan/wkqueue';
  26         if (!File.exists(storedir)) {
  27             File.mkdir(storedir);
  28         }
  29         var parsedir = storedir + '/parse_queue';
  30         var downloaddir = storedir + '/download_queue';
  31         var donedir = storedir + '/downloaded_queue';
  32         if (!File.exists(parsedir)) {
  33             File.mkdir(parsedir);
  34         }
  35         if (!File.exists(downloaddir)) {
  36             File.mkdir(downloaddir);
  37         }
  38         if (!File.exists(donedir)) {
  39             File.mkdir(donedir);
  40         }
  41
  42         var update_title = function (web_view, web_frame, title)
  43         {
  44             if(title.length > 25)
  45                 title = title.slice(0,25) + "...";
  46
  47             tab.get_tab_label().label = title;
  48         };
  49
  50         var update_url = function (web_view, web_frame)
  51         {
  52             var toolbar = tab.get_toolbar();
  53
  54             toolbar.set_url(web_frame.get_uri());
  55             toolbar.set_can_go_back(web_view.can_go_back());
  56             toolbar.set_can_go_forward(web_view.can_go_forward());
  57         };
  58
  59         var update_progress = function (bar, progress)
  60         {
  61             tab.get_toolbar().set_progress(progress / 100);
  62         };
  63
  64         var create_new_tab = function (web_view, web_frame, new_web_view)
  65         {
  66             new_web_view = new BrowserView();
  67             new_web_view.signal.web_view_ready.connect(show_new_tab);
  68             return new_web_view;
  69         };
  70
  71         var show_new_tab = function (new_web_view)
  72         {
  73             TabbedBrowser.browser.new_tab("", new_web_view);
  74
  75             return false;
  76         };
  77
  78         var hover_link = function (web_view, link, url)
  79         {
  80             tab.get_statusbar().set_status(url);
  81         };
  82
  83
  84
  85
  86         this.add_inject = function(force)
  87         {
  88
  89             if (force || (typeof(injected[this.uri]) == 'undefined' )) {
  90                 injected[this.uri] = 0;
  91             }
  92             if (injected[this.uri] > 2) {
  93                 return;
  94             }
  95             injected[this.uri]++;
  96             var fn = __script_path__ + "/inject.js";
  97             if (File.exists(fn)) {
  98                 print("Adding inject");
  99                 var newjs = File.read(__script_path__ + "/inject.js");
 100                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 101                     newjs
 102
 103                 );
 104             }
 105
 106         }
 107          this.add_nsinject = function(force)
 108         {
 109
 110             //if (force || (typeof(injected[this.uri]) == 'undefined' )) {
 111             //    injected[this.uri] = 0;
 112             //}
 113             //if (injected[this.uri] > 2) {
 114             //    return;
 115             //}
 116             //injected[this.uri]++;
 117             var fn = __script_path__ + "/nsinject.js";
 118             //if (File.exists(fn)) {
 119                 print("Adding inject");
 120                 var newjs = File.read(__script_path__ + "/nsinject.js");
 121                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 122                     newjs
 123
 124                 );
 125             //}
 126
 127         }
 128
 129
 130
 131
 132
 133         var load_finished = function ()
 134         {
 135             print("load finished");
 136
 137             GLib.timeout_add(GLib.PRIORITY_LOW, 1000, function() {
 138
 139 //                _t.open("http://sg.lifestyleasia.com/videos/video-british-polo-day-singapore-2012/");
 140                 var mf = _t.get_main_frame();
 141
 142                 var ar = Gtk.PaperSize.get_paper_sizes();
 143                 var psetup = new Gtk.PageSetup();
 144                 for(var i = 0; i < ar.length; i++) {
 145 //                    print(ar[i].get_name());
 146                     if (ar[i].get_name() =='iso_a2') {
 147                         psetup.set_paper_size(ar[i]);
 148                     }
 149                 }
 150
 151
 152
 153                 var p = new Gtk.PrintOperation({ export_filename : '/home/chris/test_pdf/test2.pdf' });
 154
 155                 //s.set_paper_size('iso_a4');
 156                 psetup.set_scale(0.5);
 157                 p.set_default_page_setup(psetup);
 158
 159                 p.set_print_settings
 160                 //var s = p.get_print_settings();
 161                 //print(s);
 162                 //p.set_print_settings(s);
 163                 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
 164                 //Seed.quit();
 165             });
 166
 167             return;
 168             tab.get_toolbar().set_progress(0);
 169
 170             _t.add_inject(true);
 171
 172             _t.add_nsinject(true);
 173               print(typeof(Seed.argv[2]));
 174             if (Seed.argv[2] == null) {
 175                 print("NEED ID!");
 176                 Seed.quit();
 177
 178             }
 179             _t.nsdownloadNext(Seed.argv[2]  * 1)
 180
 181
 182             if (browsePage) {
 183                 print("onload: calling gather links");
 184                 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
 185                     'BrowserMirror.gatherlinks();'
 186
 187                 );
 188
 189             }
 190
 191
 192         };
 193
 194         var load_committed = function (web_view, web_frame)
 195         {
 196             update_url(web_view, web_frame);
 197         };
 198
 199         var clicked_link = function (web_view, web_frame, request,
 200                                      action, decision, window)
 201         {
 202             if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
 203                action.get_button() == 2)
 204             {
 205                 browser.new_tab(request.get_uri(), null);
 206                 return true;
 207             }
 208
 209             return false;
 210         };
 211
 212         // Public
 213
 214
 215         this.browse = function (url)
 216         {
 217             if(url.search("://") < 0)
 218                 url = "http://" + url;
 219
 220             this.open(url);
 221         };
 222
 223         this.set_tab = function (new_tab)
 224         {
 225             tab = new_tab;
 226         };
 227
 228         this.get_tab = function ()
 229         {
 230             return tab;
 231         };
 232
 233
 234         this.downloadqueue = function()
 235         {
 236             var filesList = File.list(downloaddir);
 237             print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
 238             if (!maxQueue || maxQueue < filesList.length) {
 239                 maxQueue  = filesList.length;
 240             }
 241
 242             tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
 243             if (filesList == null || filesList.length == 0  ){
 244
 245                 return false;
 246             }
 247             var url = decodeURIComponent(filesList[0]);
 248             if (!this.checkdomain(url)) {
 249                 print("SKIP (external domain) : " + url);
 250                 File.remove(downloaddir + '/' + filesList[0]);
 251                 return this.downloadqueue();
 252
 253             }
 254             this.downloadhead( url);
 255             return true;
 256
 257         };
 258
 259
 260         this.queuerun = function()
 261         {
 262             if (this.downloadqueue()) return true;
 263             return this.parsequeue();
 264         };
 265
 266         this.parsequeue = function(){
 267             var filesList = File.list(parsedir);
 268             if (filesList == null || filesList.length == 0  ){
 269                 return false;
 270             }
 271
 272             browsePage = decodeURIComponent(filesList[0]);
 273             print("parsing page:" + browsePage );
 274             this.browse(browsePage);
 275
 276
 277
 278
 279
 280             return true;
 281         };
 282
 283
 284         this.downloadpage = function(link){
 285             print("calling download page: " + link);
 286             _t.add_inject(); // just in case..
 287                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 288             this.execute_script(
 289                 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
 290             );
 291
 292
 293         };
 294
 295         this.downloadhead = function(link){
 296             print("calling download head: " + link);
 297             _t.add_inject(); // just in case..
 298                 //var url = File.read(__script_path__+"/downloadqueue/"+link);
 299             this.execute_script(
 300                 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
 301             );
 302
 303         };
 304         var nsqueue = false;
 305         this.nsdownloadNext = function()
 306         {
 307             /*if (nsqueue === false) {
 308                 nsqueue = [
 309                            //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,
 310                            //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,
 311                             //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,
 312                            //369,363,242,364,241,243,365,246,368,245,367,377,427,
 313                            426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
 314                         ];
 315             }
 316
 317             if (!nsqueue.length) {
 318                 print("DONE");
 319                 return;
 320             }
 321             var pg = nsqueue.shift() ;
 322
 323
 324             */
 325
 326             //print(typeof(Seed.argv[2]));
 327             if (!Seed.argv[2]) {
 328                 print("NEED ID!");
 329                 Seed.quit();
 330
 331             }
 332             var pg = Seed.argv[2]  * 1;
 333             if ( File.exists( storedir+'/output/' + pg + '.csv')) {
 334                 print("DONE : " + storedir+'/output/' + pg + '.csv');
 335                 Seed.quit();
 336                 return;
 337             }
 338
 339             var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
 340
 341             print("downloadnext : " + pg);
 342             this.execute_script(
 343                 'NS.login(' +  JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
 344
 345             );
 346
 347
 348
 349         }
 350
 351         // Implementation
 352         //this.set_scroll_adjustments(null, null);
 353
 354         this.signal.title_changed.connect(update_title);
 355         this.signal.load_committed.connect(load_committed);
 356         this.signal.load_finished.connect(load_finished);
 357         this.signal.load_progress_changed.connect(update_progress);
 358
 359         // For some reason, this segfaults seed in the instance init closure handler
 360         // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
 361         //this.signal.navigation_policy_decision_requested.connect(clicked_link);
 362
 363         this.signal.hovering_over_link.connect(hover_link);
 364
 365         this.signal.create_web_view.connect(create_new_tab);
 366
 367
 368
 369         print("ADDing console message sig handler");
 370
 371
 372
 373
 374         this.signal.console_message.connect(function(wv, msg, line, sid) {
 375             // print('BrowserView.js got ' + msg);
 376             var methodname;
 377             var ret;
 378             try {
 379                 ret = JSON.parse(msg);
 380
 381             } catch(e) {
 382                 print("GOT INVALID message:" + msg)
 383                 return true;
 384
 385
 386             }
 387             print("got method : " + ret.method);
 388
 389             if (ret.method == 'exit') {
 390                 Seed.quit();
 391             }
 392
 393
 394             if (ret.method == 'nsdownloadpage'){
 395                 try {
 396                     var mt = ret.contentType.split(';').shift();
 397                 } catch( e) {
 398                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 399                     mt='';
 400                 }
 401
 402                 switch(mt) {
 403
 404                     case '':
 405                         print("CONSOLE GOT BLANK - call download next?");
 406                           Seed.quit();
 407                          //_t.nsdownloadNext();
 408                        // _t.moveToDone(ret.requesturl);
 409                         break;
 410
 411                     default:
 412                         var info = false;
 413                         var info_f = _t.dupeCheck(ret.requesturl);
 414                         if (info_f) {
 415                             info  = JSON.parse(File.read(info_f));
 416                         }
 417                          var target  = storedir+'/output/' + ret.filename
 418                          // flag it as done..
 419
 420
 421
 422
 423
 424                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 425                         print("GOT array sized: " + ret.data.length);
 426
 427                         File.writeBinaryArray(target,ret.data);
 428                         print("Wrote to file: " + target);
 429                         // get next..
 430                         return true;
 431                         //Seed.quit();
 432                         //_t.nsdownloadNext();
 433
 434
 435
 436                         break;
 437                 }
 438                 return true; //???
 439             }
 440
 441             if (ret.method == 'gatherlinks'){
 442                 // flag the page as parsed.
 443
 444                 if (browsePage) {
 445                     _t.moveToDone(browsePage);
 446
 447                 }
 448                 var sourcePage = browsePage;
 449                 browsePage = false;
 450                 print(typeof(ret.data));
 451                 if (typeof(ret.data) != 'object' ) {
 452                     print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
 453                     ret.data= [];
 454
 455                 }
 456                 ret.data.forEach(function(ln) {
 457                     if (!ln.href.match(/^http[s]*:\/\//)) {
 458                         print("SKIP link: " + ln.href);
 459                         return;
 460                     }
 461                     ln.href= ln.href.replace(/#.*$/,'');
 462
 463                     if (!_t.checkdomain(ln.href) ) {
 464                         print("SKIP link (external domain): " + ln.href);
 465                         return;
 466                     }
 467
 468                     // this is just for our purposes..
 469
 470                     if (ln.href.match(/\/pages\/[0-9]+$/)) {
 471                         print("SKIP link (ingore unnamed pages): " + ln.href);
 472                         return;
 473                     }
 474
 475
 476
 477                     var fn = encodeURIComponent(ln.href);
 478
 479                     var dupe  = _t.dupeCheck(ln.href);
 480                     if (dupe) {
 481
 482                         if (dupe == downloaddir + '/'  + fn) {
 483                             var info = JSON.parse(File.read(dupe));
 484                             if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
 485                                 print("SKIP link (in queue already): " + ln.href);
 486                                 return;
 487                             }
 488                             print("found a longer link for url")
 489                         } else {
 490                             print("SKIP link (in another queue): " + ln.href);
 491                             return;
 492                         }
 493                     }
 494
 495                     File.write(downloaddir + '/'  + fn, JSON.stringify( {
 496                         label : ln.label,
 497                         fromURL : sourcePage
 498                     })); // write an empyt file indicating it needs downloading..
 499                 });
 500                 var filesList = File.list(downloaddir);
 501
 502                 maxQueue =  filesList.length;
 503             }
 504
 505             if (ret.method == 'downloadpage'){
 506                 // got the results from download page:
 507                 // requesturl
 508                 // remove from downloadqueue.
 509
 510
 511                 //
 512                 try {
 513                     var mt = ret.contentType.split(';').shift();
 514                 } catch( e) {
 515                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 516                     mt='';
 517                 }
 518
 519                 switch(mt) {
 520
 521                     case '':
 522                         _t.moveToDone(ret.requesturl);
 523                         break;
 524
 525                     default:
 526                         var info = false;
 527                         var info_f = _t.dupeCheck(ret.requesturl);
 528                         if (info_f) {
 529                             info  = JSON.parse(File.read(info_f));
 530                         }
 531
 532                         _t.moveToDone(ret.requesturl);
 533                         // flag it as done..
 534
 535                         // write it...
 536                         var target  = _t.toFilename(ret.requesturl);
 537                         if (info && info.fromURL) {
 538                             var bn = decodeURIComponent(File.basename(target));
 539                             target = _t.toFilename(info.fromURL+'/'+ bn);
 540                         }
 541
 542
 543
 544
 545                         //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
 546                         print("GOT array sized: " + ret.data.length);
 547
 548                         File.writeBinaryArray(target,ret.data);
 549                         print("Wrote to file: " + target);
 550                         break;
 551                 }
 552
 553                 // if it's HTML then add it to parse queue
 554                 // otehrwise save it.. and run the queue again.
 555
 556
 557                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 558             }
 559
 560
 561             if (ret.method == 'downloadhead'){
 562                 // got the results from download page:
 563                 // requesturl
 564                 // remove from downloadqueue.
 565
 566                 //
 567                 try {
 568                     var mt = ret.contentType.split(';').shift();
 569                 } catch( e) {
 570                     print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
 571                     mt='';
 572                 }
 573
 574                 switch(mt) {
 575                     case 'text/html':
 576                         // add to parse QUEUE..
 577                         print("moving to parse queue");
 578                         _t.moveToParse(ret.requesturl);
 579                         break;
 580                     // stuf we do not care about..
 581                     case 'application/atom+xml':
 582                     case '':
 583                         print("moving to done queue");
 584                          _t.moveToDone(ret.requesturl);
 585                         break;
 586
 587                     default:
 588                         print("calling download file");
 589                         _t.downloadpage( ret.requesturl );
 590                         // keep it on the queue..
 591                         // do not run the queue..
 592                         return true;
 593
 594                 }
 595
 596                 // if it's HTML then add it to parse queue
 597                 // otehrwise save it.. and run the queue again.
 598
 599
 600                 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
 601             }
 602             _t.queuerun();
 603
 604             return true;
 605         });
 606
 607
 608         this.toFilename = function(url)
 609         {
 610             url = url.replace(/^http[s]*:\/\//, '');
 611             var p = url.split('/');
 612             p.unshift(storedir+'/output');
 613             for (var i =1 ;i < p.length; i++) {
 614                 p[i] = encodeURIComponent(p[i]);
 615
 616             }
 617             p[p.length-1] = decodeURIComponent(p[p.length-1]);
 618             ret = p.join('/');
 619             var dir = File.dirname(ret);
 620             File.mkdirall(dir);
 621             return ret;
 622
 623         }
 624         this.checkdomain = function(comp)
 625         {
 626             var b = parseUri(this.uri);
 627             var d = parseUri(comp);
 628             return (d.host == b.host && d.protocol == b.protocol);
 629
 630
 631         }
 632
 633         this.dupeCheck = function(url)
 634         {
 635
 636            // order - return highest up the queue first..
 637             if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
 638                 return downloaddir +'/' + encodeURIComponent(url);
 639             }
 640              if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
 641                 return parsedir +'/' + encodeURIComponent(url);
 642             }
 643             if (File.exists(donedir +'/' + encodeURIComponent(url))) {
 644                 return donedir +'/' + encodeURIComponent(url);
 645             }
 646             return  false;
 647
 648
 649         }
 650         this.moveToParse = function(url)
 651         {
 652             var old = this.dupeCheck(url);
 653             var target =parsedir +'/' + encodeURIComponent(url);
 654             if (old == target) {
 655                 return;
 656             }
 657             File.write(target, old ? File.read(old) : '');
 658             if (old) {
 659                 File.remove(old);
 660             }
 661
 662         }
 663
 664         this.moveToDownload= function(url)
 665         {
 666             var old = this.dupeCheck(url);
 667             var target =downloaddir +'/' + encodeURIComponent(url);
 668             if (old == target) {
 669                 return;
 670             }
 671             File.write(target, old ? File.read(old) : '');
 672             if (old) {
 673                 File.remove(old);
 674             }
 675
 676         }
 677         this.moveToDone= function(url)
 678         {
 679             var old = this.dupeCheck(url);
 680             var target = donedir +'/' + encodeURIComponent(url);
 681             if (old == target) {
 682                 return;
 683             }
 684             File.write(target, old ? File.read(old) : '');
 685             if (old) {
 686                 File.remove(old);
 687             }
 688
 689         }
 690
 691     }
 692 });
 693
 694 function parseUri (str) {
 695         var     o   = parseUri.options,
 696                 m   = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
 697                 uri = {},
 698                 i   = 14;
 699
 700         while (i--) uri[o.key[i]] = m[i] || "";
 701
 702         uri[o.q.name] = {};
 703         uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
 704                 if ($1) uri[o.q.name][$1] = $2;
 705         });
 706
 707         return uri;
 708 };
 709
 710 parseUri.options = {
 711         strictMode: false,
 712         key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
 713         q:   {
 714                 name:   "queryKey",
 715                 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
 716         },
 717         parser: {
 718                 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
 719                 loose:  /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/
 720         }
 721 };