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