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