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