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

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