2 GLib = imports.gi.GLib;
3 WebKit = imports.gi.WebKit;
5 TabbedBrowser = imports.TabbedBrowser;
6 File = imports.File.File;
8 base64 = imports.base64.base64;
10 BrowserView = new GType({
11 parent: WebKit.WebView.type,
21 var browsePage = false;
25 var storedir = '/home/alan/wkqueue';
26 if (!File.exists(storedir)) {
29 var parsedir = storedir + '/parse_queue';
30 var downloaddir = storedir + '/download_queue';
31 var donedir = storedir + '/downloaded_queue';
32 if (!File.exists(parsedir)) {
35 if (!File.exists(downloaddir)) {
36 File.mkdir(downloaddir);
38 if (!File.exists(donedir)) {
42 var update_title = function (web_view, web_frame, title)
45 title = title.slice(0,25) + "...";
47 tab.get_tab_label().label = title;
50 var update_url = function (web_view, web_frame)
52 var toolbar = tab.get_toolbar();
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());
59 var update_progress = function (bar, progress)
61 tab.get_toolbar().set_progress(progress / 100);
64 var create_new_tab = function (web_view, web_frame, new_web_view)
66 new_web_view = new BrowserView();
67 new_web_view.signal.web_view_ready.connect(show_new_tab);
71 var show_new_tab = function (new_web_view)
73 TabbedBrowser.browser.new_tab("", new_web_view);
78 var hover_link = function (web_view, link, url)
80 tab.get_statusbar().set_status(url);
86 this.add_inject = function(force)
89 if (force || (typeof(injected[this.uri]) == 'undefined' )) {
90 injected[this.uri] = 0;
92 if (injected[this.uri] > 2) {
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(
107 this.add_nsinject = function(force)
110 //if (force || (typeof(injected[this.uri]) == 'undefined' )) {
111 // injected[this.uri] = 0;
113 //if (injected[this.uri] > 2) {
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(
133 var load_finished = function ()
135 print("load finished");
137 GLib.timeout_add(GLib.PRIORITY_LOW, 1000, function() {
139 // _t.open("http://sg.lifestyleasia.com/videos/video-british-polo-day-singapore-2012/");
140 var mf = _t.get_main_frame();
142 var ar = Gtk.PaperSize.get_paper_sizes();
143 var psetup = new Gtk.PageSetup();
144 for(var i = 0; i < ar.length; i++) {
145 // print(ar[i].get_name());
146 if (ar[i].get_name() =='iso_a2') {
147 psetup.set_paper_size(ar[i]);
153 var p = new Gtk.PrintOperation({ export_filename : '/home/chris/test_pdf/test2.pdf' });
154 var s = new Gtk.PrintSettings();
157 //s.set_paper_size('iso_a4');
158 psetup.set_scale(0.5);
159 p.set_default_page_setup(psetup);
162 //var s = p.get_print_settings();
164 //p.set_print_settings(s);
165 mf.print_full(p, Gtk.PrintOperationAction.EXPORT)
170 tab.get_toolbar().set_progress(0);
174 _t.add_nsinject(true);
175 print(typeof(Seed.argv[2]));
176 if (Seed.argv[2] == null) {
181 _t.nsdownloadNext(Seed.argv[2] * 1)
185 print("onload: calling gather links");
186 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
187 'BrowserMirror.gatherlinks();'
196 var load_committed = function (web_view, web_frame)
198 update_url(web_view, web_frame);
201 var clicked_link = function (web_view, web_frame, request,
202 action, decision, window)
204 if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
205 action.get_button() == 2)
207 browser.new_tab(request.get_uri(), null);
217 this.browse = function (url)
219 if(url.search("://") < 0)
220 url = "http://" + url;
225 this.set_tab = function (new_tab)
230 this.get_tab = function ()
236 this.downloadqueue = function()
238 var filesList = File.list(downloaddir);
239 print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
240 if (!maxQueue || maxQueue < filesList.length) {
241 maxQueue = filesList.length;
244 tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
245 if (filesList == null || filesList.length == 0 ){
249 var url = decodeURIComponent(filesList[0]);
250 if (!this.checkdomain(url)) {
251 print("SKIP (external domain) : " + url);
252 File.remove(downloaddir + '/' + filesList[0]);
253 return this.downloadqueue();
256 this.downloadhead( url);
262 this.queuerun = function()
264 if (this.downloadqueue()) return true;
265 return this.parsequeue();
268 this.parsequeue = function(){
269 var filesList = File.list(parsedir);
270 if (filesList == null || filesList.length == 0 ){
274 browsePage = decodeURIComponent(filesList[0]);
275 print("parsing page:" + browsePage );
276 this.browse(browsePage);
286 this.downloadpage = function(link){
287 print("calling download page: " + link);
288 _t.add_inject(); // just in case..
289 //var url = File.read(__script_path__+"/downloadqueue/"+link);
291 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
297 this.downloadhead = function(link){
298 print("calling download head: " + link);
299 _t.add_inject(); // just in case..
300 //var url = File.read(__script_path__+"/downloadqueue/"+link);
302 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
307 this.nsdownloadNext = function()
309 /*if (nsqueue === false) {
311 //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,
312 //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,
313 //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,
314 //369,363,242,364,241,243,365,246,368,245,367,377,427,
315 426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
319 if (!nsqueue.length) {
323 var pg = nsqueue.shift() ;
328 //print(typeof(Seed.argv[2]));
334 var pg = Seed.argv[2] * 1;
335 if ( File.exists( storedir+'/output/' + pg + '.csv')) {
336 print("DONE : " + storedir+'/output/' + pg + '.csv');
341 var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
343 print("downloadnext : " + pg);
345 'NS.login(' + JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
354 //this.set_scroll_adjustments(null, null);
356 this.signal.title_changed.connect(update_title);
357 this.signal.load_committed.connect(load_committed);
358 this.signal.load_finished.connect(load_finished);
359 this.signal.load_progress_changed.connect(update_progress);
361 // For some reason, this segfaults seed in the instance init closure handler
362 // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
363 //this.signal.navigation_policy_decision_requested.connect(clicked_link);
365 this.signal.hovering_over_link.connect(hover_link);
367 this.signal.create_web_view.connect(create_new_tab);
371 print("ADDing console message sig handler");
376 this.signal.console_message.connect(function(wv, msg, line, sid) {
377 // print('BrowserView.js got ' + msg);
381 ret = JSON.parse(msg);
384 print("GOT INVALID message:" + msg)
389 print("got method : " + ret.method);
391 if (ret.method == 'exit') {
396 if (ret.method == 'nsdownloadpage'){
398 var mt = ret.contentType.split(';').shift();
400 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
407 print("CONSOLE GOT BLANK - call download next?");
409 //_t.nsdownloadNext();
410 // _t.moveToDone(ret.requesturl);
415 var info_f = _t.dupeCheck(ret.requesturl);
417 info = JSON.parse(File.read(info_f));
419 var target = storedir+'/output/' + ret.filename
426 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
427 print("GOT array sized: " + ret.data.length);
429 File.writeBinaryArray(target,ret.data);
430 print("Wrote to file: " + target);
434 //_t.nsdownloadNext();
443 if (ret.method == 'gatherlinks'){
444 // flag the page as parsed.
447 _t.moveToDone(browsePage);
450 var sourcePage = browsePage;
452 print(typeof(ret.data));
453 if (typeof(ret.data) != 'object' ) {
454 print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
458 ret.data.forEach(function(ln) {
459 if (!ln.href.match(/^http[s]*:\/\//)) {
460 print("SKIP link: " + ln.href);
463 ln.href= ln.href.replace(/#.*$/,'');
465 if (!_t.checkdomain(ln.href) ) {
466 print("SKIP link (external domain): " + ln.href);
470 // this is just for our purposes..
472 if (ln.href.match(/\/pages\/[0-9]+$/)) {
473 print("SKIP link (ingore unnamed pages): " + ln.href);
479 var fn = encodeURIComponent(ln.href);
481 var dupe = _t.dupeCheck(ln.href);
484 if (dupe == downloaddir + '/' + fn) {
485 var info = JSON.parse(File.read(dupe));
486 if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
487 print("SKIP link (in queue already): " + ln.href);
490 print("found a longer link for url")
492 print("SKIP link (in another queue): " + ln.href);
497 File.write(downloaddir + '/' + fn, JSON.stringify( {
500 })); // write an empyt file indicating it needs downloading..
502 var filesList = File.list(downloaddir);
504 maxQueue = filesList.length;
507 if (ret.method == 'downloadpage'){
508 // got the results from download page:
510 // remove from downloadqueue.
515 var mt = ret.contentType.split(';').shift();
517 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
524 _t.moveToDone(ret.requesturl);
529 var info_f = _t.dupeCheck(ret.requesturl);
531 info = JSON.parse(File.read(info_f));
534 _t.moveToDone(ret.requesturl);
538 var target = _t.toFilename(ret.requesturl);
539 if (info && info.fromURL) {
540 var bn = decodeURIComponent(File.basename(target));
541 target = _t.toFilename(info.fromURL+'/'+ bn);
547 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
548 print("GOT array sized: " + ret.data.length);
550 File.writeBinaryArray(target,ret.data);
551 print("Wrote to file: " + target);
555 // if it's HTML then add it to parse queue
556 // otehrwise save it.. and run the queue again.
559 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
563 if (ret.method == 'downloadhead'){
564 // got the results from download page:
566 // remove from downloadqueue.
570 var mt = ret.contentType.split(';').shift();
572 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
578 // add to parse QUEUE..
579 print("moving to parse queue");
580 _t.moveToParse(ret.requesturl);
582 // stuf we do not care about..
583 case 'application/atom+xml':
585 print("moving to done queue");
586 _t.moveToDone(ret.requesturl);
590 print("calling download file");
591 _t.downloadpage( ret.requesturl );
592 // keep it on the queue..
593 // do not run the queue..
598 // if it's HTML then add it to parse queue
599 // otehrwise save it.. and run the queue again.
602 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
610 this.toFilename = function(url)
612 url = url.replace(/^http[s]*:\/\//, '');
613 var p = url.split('/');
614 p.unshift(storedir+'/output');
615 for (var i =1 ;i < p.length; i++) {
616 p[i] = encodeURIComponent(p[i]);
619 p[p.length-1] = decodeURIComponent(p[p.length-1]);
621 var dir = File.dirname(ret);
626 this.checkdomain = function(comp)
628 var b = parseUri(this.uri);
629 var d = parseUri(comp);
630 return (d.host == b.host && d.protocol == b.protocol);
635 this.dupeCheck = function(url)
638 // order - return highest up the queue first..
639 if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
640 return downloaddir +'/' + encodeURIComponent(url);
642 if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
643 return parsedir +'/' + encodeURIComponent(url);
645 if (File.exists(donedir +'/' + encodeURIComponent(url))) {
646 return donedir +'/' + encodeURIComponent(url);
652 this.moveToParse = function(url)
654 var old = this.dupeCheck(url);
655 var target =parsedir +'/' + encodeURIComponent(url);
659 File.write(target, old ? File.read(old) : '');
666 this.moveToDownload= function(url)
668 var old = this.dupeCheck(url);
669 var target =downloaddir +'/' + encodeURIComponent(url);
673 File.write(target, old ? File.read(old) : '');
679 this.moveToDone= function(url)
681 var old = this.dupeCheck(url);
682 var target = donedir +'/' + encodeURIComponent(url);
686 File.write(target, old ? File.read(old) : '');
696 function parseUri (str) {
697 var o = parseUri.options,
698 m = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
702 while (i--) uri[o.key[i]] = m[i] || "";
705 uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
706 if ($1) uri[o.q.name][$1] = $2;
714 key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
717 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
720 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
721 loose: /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/