2 WebKit = imports.gi.WebKit;
4 TabbedBrowser = imports.TabbedBrowser;
5 File = imports.File.File;
7 base64 = imports.base64.base64;
9 BrowserView = new GType({
10 parent: WebKit.WebView.type,
20 var browsePage = false;
24 var storedir = '/tmp/wkqueue';
25 if (!File.exists(storedir)) {
28 var parsedir = storedir + '/parse_queue';
29 var downloaddir = storedir + '/download_queue';
30 var donedir = storedir + '/downloaded_queue';
31 if (!File.exists(parsedir)) {
34 if (!File.exists(downloaddir)) {
35 File.mkdir(downloaddir);
37 if (!File.exists(donedir)) {
41 var update_title = function (web_view, web_frame, title)
44 title = title.slice(0,25) + "...";
46 tab.get_tab_label().label = title;
49 var update_url = function (web_view, web_frame)
51 var toolbar = tab.get_toolbar();
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());
58 var update_progress = function (bar, progress)
60 tab.get_toolbar().set_progress(progress / 100);
63 var create_new_tab = function (web_view, web_frame, new_web_view)
65 new_web_view = new BrowserView();
66 new_web_view.signal.web_view_ready.connect(show_new_tab);
70 var show_new_tab = function (new_web_view)
72 TabbedBrowser.browser.new_tab("", new_web_view);
77 var hover_link = function (web_view, link, url)
79 tab.get_statusbar().set_status(url);
85 this.add_inject = function(force)
88 if (force || (typeof(injected[this.uri]) == 'undefined' )) {
89 injected[this.uri] = 0;
91 if (injected[this.uri] > 2) {
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(
106 this.add_nsinject = function(force)
109 //if (force || (typeof(injected[this.uri]) == 'undefined' )) {
110 // injected[this.uri] = 0;
112 //if (injected[this.uri] > 2) {
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(
132 var load_finished = function ()
134 print("load finished");
135 tab.get_toolbar().set_progress(0);
138 var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
142 _t.add_nsinject(true);
143 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
144 'NS.login(' + JSON.stringify(pd.username) + JSON.stringify(pd.password) + ' );'
149 print("onload: calling gather links");
150 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
151 'BrowserMirror.gatherlinks();'
160 var load_committed = function (web_view, web_frame)
162 update_url(web_view, web_frame);
165 var clicked_link = function (web_view, web_frame, request,
166 action, decision, window)
168 if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
169 action.get_button() == 2)
171 browser.new_tab(request.get_uri(), null);
181 this.browse = function (url)
183 if(url.search("://") < 0)
184 url = "http://" + url;
189 this.set_tab = function (new_tab)
194 this.get_tab = function ()
200 this.downloadqueue = function()
202 var filesList = File.list(downloaddir);
203 print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
204 if (!maxQueue || maxQueue < filesList.length) {
205 maxQueue = filesList.length;
208 tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
209 if (filesList == null || filesList.length == 0 ){
213 var url = decodeURIComponent(filesList[0]);
214 if (!this.checkdomain(url)) {
215 print("SKIP (external domain) : " + url);
216 File.remove(downloaddir + '/' + filesList[0]);
217 return this.downloadqueue();
220 this.downloadhead( url);
226 this.queuerun = function()
228 if (this.downloadqueue()) return true;
229 return this.parsequeue();
232 this.parsequeue = function(){
233 var filesList = File.list(parsedir);
234 if (filesList == null || filesList.length == 0 ){
238 browsePage = decodeURIComponent(filesList[0]);
239 print("parsing page:" + browsePage );
240 this.browse(browsePage);
250 this.downloadpage = function(link){
251 print("calling download page: " + link);
252 _t.add_inject(); // just in case..
253 //var url = File.read(__script_path__+"/downloadqueue/"+link);
255 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
261 this.downloadhead = function(link){
262 print("calling download head: " + link);
263 _t.add_inject(); // just in case..
264 //var url = File.read(__script_path__+"/downloadqueue/"+link);
266 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
271 this.nsdownloadNext = function()
273 if (nsqueue === false) {
275 //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,
276 //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,
277 //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,
278 //369,363,242,364,241,243,365,246,368,245,367,377,427,
279 426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
283 if (!nsqueue.length) {
287 var pg = nsqueue.shift() ;
289 if ( File.exists( storedir+'/output/' + pg + '.csv')) {
290 this.nsdownloadNext();
295 "NS.rungrab(" + pg +");"
302 //this.set_scroll_adjustments(null, null);
304 this.signal.title_changed.connect(update_title);
305 this.signal.load_committed.connect(load_committed);
306 this.signal.load_finished.connect(load_finished);
307 this.signal.load_progress_changed.connect(update_progress);
309 // For some reason, this segfaults seed in the instance init closure handler
310 // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
311 //this.signal.navigation_policy_decision_requested.connect(clicked_link);
313 this.signal.hovering_over_link.connect(hover_link);
315 this.signal.create_web_view.connect(create_new_tab);
318 print("ADDing console message sig handler");
323 this.signal.console_message.connect(function(wv, msg, line, sid) {
324 // print('BrowserView.js got ' + msg);
328 ret = JSON.parse(msg);
331 print("GOT INVALID message:" + msg)
336 print("got method : " + ret.method);
338 if (ret.method == 'nsdownloadpage'){
340 var mt = ret.contentType.split(';').shift();
342 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
350 // _t.moveToDone(ret.requesturl);
355 var info_f = _t.dupeCheck(ret.requesturl);
357 info = JSON.parse(File.read(info_f));
359 var target = storedir+'/output/' + ret.filename
366 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
367 print("GOT array sized: " + ret.data.length);
369 File.writeBinaryArray(target,ret.data);
370 print("Wrote to file: " + target);
381 if (ret.method == 'gatherlinks'){
382 // flag the page as parsed.
385 _t.moveToDone(browsePage);
388 var sourcePage = browsePage;
390 print(typeof(ret.data));
391 if (typeof(ret.data) != 'object' ) {
392 print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
396 ret.data.forEach(function(ln) {
397 if (!ln.href.match(/^http[s]*:\/\//)) {
398 print("SKIP link: " + ln.href);
401 ln.href= ln.href.replace(/#.*$/,'');
403 if (!_t.checkdomain(ln.href) ) {
404 print("SKIP link (external domain): " + ln.href);
408 // this is just for our purposes..
410 if (ln.href.match(/\/pages\/[0-9]+$/)) {
411 print("SKIP link (ingore unnamed pages): " + ln.href);
417 var fn = encodeURIComponent(ln.href);
419 var dupe = _t.dupeCheck(ln.href);
422 if (dupe == downloaddir + '/' + fn) {
423 var info = JSON.parse(File.read(dupe));
424 if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
425 print("SKIP link (in queue already): " + ln.href);
428 print("found a longer link for url")
430 print("SKIP link (in another queue): " + ln.href);
435 File.write(downloaddir + '/' + fn, JSON.stringify( {
438 })); // write an empyt file indicating it needs downloading..
440 var filesList = File.list(downloaddir);
442 maxQueue = filesList.length;
445 if (ret.method == 'downloadpage'){
446 // got the results from download page:
448 // remove from downloadqueue.
453 var mt = ret.contentType.split(';').shift();
455 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
462 _t.moveToDone(ret.requesturl);
467 var info_f = _t.dupeCheck(ret.requesturl);
469 info = JSON.parse(File.read(info_f));
472 _t.moveToDone(ret.requesturl);
476 var target = _t.toFilename(ret.requesturl);
477 if (info && info.fromURL) {
478 var bn = decodeURIComponent(File.basename(target));
479 target = _t.toFilename(info.fromURL+'/'+ bn);
485 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
486 print("GOT array sized: " + ret.data.length);
488 File.writeBinaryArray(target,ret.data);
489 print("Wrote to file: " + target);
493 // if it's HTML then add it to parse queue
494 // otehrwise save it.. and run the queue again.
497 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
501 if (ret.method == 'downloadhead'){
502 // got the results from download page:
504 // remove from downloadqueue.
508 var mt = ret.contentType.split(';').shift();
510 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
516 // add to parse QUEUE..
517 print("moving to parse queue");
518 _t.moveToParse(ret.requesturl);
520 // stuf we do not care about..
521 case 'application/atom+xml':
523 print("moving to done queue");
524 _t.moveToDone(ret.requesturl);
528 print("calling download file");
529 _t.downloadpage( ret.requesturl );
530 // keep it on the queue..
531 // do not run the queue..
536 // if it's HTML then add it to parse queue
537 // otehrwise save it.. and run the queue again.
540 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
548 this.toFilename = function(url)
550 url = url.replace(/^http[s]*:\/\//, '');
551 var p = url.split('/');
552 p.unshift(storedir+'/output');
553 for (var i =1 ;i < p.length; i++) {
554 p[i] = encodeURIComponent(p[i]);
557 p[p.length-1] = decodeURIComponent(p[p.length-1]);
559 var dir = File.dirname(ret);
564 this.checkdomain = function(comp)
566 var b = parseUri(this.uri);
567 var d = parseUri(comp);
568 return (d.host == b.host && d.protocol == b.protocol);
573 this.dupeCheck = function(url)
576 // order - return highest up the queue first..
577 if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
578 return downloaddir +'/' + encodeURIComponent(url);
580 if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
581 return parsedir +'/' + encodeURIComponent(url);
583 if (File.exists(donedir +'/' + encodeURIComponent(url))) {
584 return donedir +'/' + encodeURIComponent(url);
590 this.moveToParse = function(url)
592 var old = this.dupeCheck(url);
593 var target =parsedir +'/' + encodeURIComponent(url);
597 File.write(target, old ? File.read(old) : '');
604 this.moveToDownload= function(url)
606 var old = this.dupeCheck(url);
607 var target =downloaddir +'/' + encodeURIComponent(url);
611 File.write(target, old ? File.read(old) : '');
617 this.moveToDone= function(url)
619 var old = this.dupeCheck(url);
620 var target = donedir +'/' + encodeURIComponent(url);
624 File.write(target, old ? File.read(old) : '');
634 function parseUri (str) {
635 var o = parseUri.options,
636 m = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
640 while (i--) uri[o.key[i]] = m[i] || "";
643 uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
644 if ($1) uri[o.q.name][$1] = $2;
652 key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
655 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
658 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
659 loose: /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/