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