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 = '/home/alan/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);
142 _t.add_nsinject(true);
143 print(typeof(Seed.argv[2]));
144 if (Seed.argv[2] == null) {
149 _t.nsdownloadNext(Seed.argv[2] * 1)
153 print("onload: calling gather links");
154 TabbedBrowser.browser.current_tab().get_web_view().execute_script(
155 'BrowserMirror.gatherlinks();'
164 var load_committed = function (web_view, web_frame)
166 update_url(web_view, web_frame);
169 var clicked_link = function (web_view, web_frame, request,
170 action, decision, window)
172 if(action.get_reason() == WebKit.WebNavigationReason.LINK_CLICKED &&
173 action.get_button() == 2)
175 browser.new_tab(request.get_uri(), null);
185 this.browse = function (url)
187 if(url.search("://") < 0)
188 url = "http://" + url;
193 this.set_tab = function (new_tab)
198 this.get_tab = function ()
204 this.downloadqueue = function()
206 var filesList = File.list(downloaddir);
207 print("DOWNLOAD QUEUE LENGTH: " + filesList.length);
208 if (!maxQueue || maxQueue < filesList.length) {
209 maxQueue = filesList.length;
212 tab.get_toolbar().set_progress(filesList.length / Math.max(maxQueue, filesList.length ));
213 if (filesList == null || filesList.length == 0 ){
217 var url = decodeURIComponent(filesList[0]);
218 if (!this.checkdomain(url)) {
219 print("SKIP (external domain) : " + url);
220 File.remove(downloaddir + '/' + filesList[0]);
221 return this.downloadqueue();
224 this.downloadhead( url);
230 this.queuerun = function()
232 if (this.downloadqueue()) return true;
233 return this.parsequeue();
236 this.parsequeue = function(){
237 var filesList = File.list(parsedir);
238 if (filesList == null || filesList.length == 0 ){
242 browsePage = decodeURIComponent(filesList[0]);
243 print("parsing page:" + browsePage );
244 this.browse(browsePage);
254 this.downloadpage = function(link){
255 print("calling download page: " + link);
256 _t.add_inject(); // just in case..
257 //var url = File.read(__script_path__+"/downloadqueue/"+link);
259 "BrowserMirror.downloadpage(" + JSON.stringify(link) +");"
265 this.downloadhead = function(link){
266 print("calling download head: " + link);
267 _t.add_inject(); // just in case..
268 //var url = File.read(__script_path__+"/downloadqueue/"+link);
270 "BrowserMirror.downloadhead(" + JSON.stringify(link) +");"
275 this.nsdownloadNext = function()
277 /*if (nsqueue === false) {
279 //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,
280 //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,
281 //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,
282 //369,363,242,364,241,243,365,246,368,245,367,377,427,
283 426,250,447,372,249,371,248,370,442,397,129,441,440,116,122,123,125,126,136,115,117,127
287 if (!nsqueue.length) {
291 var pg = nsqueue.shift() ;
296 //print(typeof(Seed.argv[2]));
302 var pg = Seed.argv[2] * 1;
303 if ( File.exists( storedir+'/output/' + pg + '.csv')) {
304 print("DONE : " + storedir+'/output/' + pg + '.csv');
309 var pd = JSON.parse(File.read('/home/alan/.nspasswd'));
311 print("downloadnext : " + pg);
313 'NS.login(' + JSON.stringify(pd.username) + ',' + JSON.stringify(pd.password) + ',' + pg + ' );'
322 //this.set_scroll_adjustments(null, null);
324 this.signal.title_changed.connect(update_title);
325 this.signal.load_committed.connect(load_committed);
326 this.signal.load_finished.connect(load_finished);
327 this.signal.load_progress_changed.connect(update_progress);
329 // For some reason, this segfaults seed in the instance init closure handler
330 // Once that's fixed, uncommenting the next line will give middle-click-open-in-new tab
331 //this.signal.navigation_policy_decision_requested.connect(clicked_link);
333 this.signal.hovering_over_link.connect(hover_link);
335 this.signal.create_web_view.connect(create_new_tab);
338 print("ADDing console message sig handler");
343 this.signal.console_message.connect(function(wv, msg, line, sid) {
344 // print('BrowserView.js got ' + msg);
348 ret = JSON.parse(msg);
351 print("GOT INVALID message:" + msg)
356 print("got method : " + ret.method);
358 if (ret.method == 'exit') {
363 if (ret.method == 'nsdownloadpage'){
365 var mt = ret.contentType.split(';').shift();
367 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
374 print("CONSOLE GOT BLANK - call download next?");
376 //_t.nsdownloadNext();
377 // _t.moveToDone(ret.requesturl);
382 var info_f = _t.dupeCheck(ret.requesturl);
384 info = JSON.parse(File.read(info_f));
386 var target = storedir+'/output/' + ret.filename
393 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
394 print("GOT array sized: " + ret.data.length);
396 File.writeBinaryArray(target,ret.data);
397 print("Wrote to file: " + target);
401 //_t.nsdownloadNext();
410 if (ret.method == 'gatherlinks'){
411 // flag the page as parsed.
414 _t.moveToDone(browsePage);
417 var sourcePage = browsePage;
419 print(typeof(ret.data));
420 if (typeof(ret.data) != 'object' ) {
421 print("GOT INVALID DATA?:" + JSON.stringify(ret,null,4));
425 ret.data.forEach(function(ln) {
426 if (!ln.href.match(/^http[s]*:\/\//)) {
427 print("SKIP link: " + ln.href);
430 ln.href= ln.href.replace(/#.*$/,'');
432 if (!_t.checkdomain(ln.href) ) {
433 print("SKIP link (external domain): " + ln.href);
437 // this is just for our purposes..
439 if (ln.href.match(/\/pages\/[0-9]+$/)) {
440 print("SKIP link (ingore unnamed pages): " + ln.href);
446 var fn = encodeURIComponent(ln.href);
448 var dupe = _t.dupeCheck(ln.href);
451 if (dupe == downloaddir + '/' + fn) {
452 var info = JSON.parse(File.read(dupe));
453 if (info && info.fromUrl && info.fromUrl.length > sourcePage.length) {
454 print("SKIP link (in queue already): " + ln.href);
457 print("found a longer link for url")
459 print("SKIP link (in another queue): " + ln.href);
464 File.write(downloaddir + '/' + fn, JSON.stringify( {
467 })); // write an empyt file indicating it needs downloading..
469 var filesList = File.list(downloaddir);
471 maxQueue = filesList.length;
474 if (ret.method == 'downloadpage'){
475 // got the results from download page:
477 // remove from downloadqueue.
482 var mt = ret.contentType.split(';').shift();
484 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
491 _t.moveToDone(ret.requesturl);
496 var info_f = _t.dupeCheck(ret.requesturl);
498 info = JSON.parse(File.read(info_f));
501 _t.moveToDone(ret.requesturl);
505 var target = _t.toFilename(ret.requesturl);
506 if (info && info.fromURL) {
507 var bn = decodeURIComponent(File.basename(target));
508 target = _t.toFilename(info.fromURL+'/'+ bn);
514 //File.write(target ,decodeURIComponent(escape(base64.decode( ret.data))));
515 print("GOT array sized: " + ret.data.length);
517 File.writeBinaryArray(target,ret.data);
518 print("Wrote to file: " + target);
522 // if it's HTML then add it to parse queue
523 // otehrwise save it.. and run the queue again.
526 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
530 if (ret.method == 'downloadhead'){
531 // got the results from download page:
533 // remove from downloadqueue.
537 var mt = ret.contentType.split(';').shift();
539 print("INVALID CONTENT TYPE? \n" + JSON.stringify(ret));
545 // add to parse QUEUE..
546 print("moving to parse queue");
547 _t.moveToParse(ret.requesturl);
549 // stuf we do not care about..
550 case 'application/atom+xml':
552 print("moving to done queue");
553 _t.moveToDone(ret.requesturl);
557 print("calling download file");
558 _t.downloadpage( ret.requesturl );
559 // keep it on the queue..
560 // do not run the queue..
565 // if it's HTML then add it to parse queue
566 // otehrwise save it.. and run the queue again.
569 //File.write(storedir+"/parsequeue/"+Math.random(), msg);
577 this.toFilename = function(url)
579 url = url.replace(/^http[s]*:\/\//, '');
580 var p = url.split('/');
581 p.unshift(storedir+'/output');
582 for (var i =1 ;i < p.length; i++) {
583 p[i] = encodeURIComponent(p[i]);
586 p[p.length-1] = decodeURIComponent(p[p.length-1]);
588 var dir = File.dirname(ret);
593 this.checkdomain = function(comp)
595 var b = parseUri(this.uri);
596 var d = parseUri(comp);
597 return (d.host == b.host && d.protocol == b.protocol);
602 this.dupeCheck = function(url)
605 // order - return highest up the queue first..
606 if (File.exists(downloaddir +'/' + encodeURIComponent(url))) {
607 return downloaddir +'/' + encodeURIComponent(url);
609 if (File.exists(parsedir +'/' + encodeURIComponent(url))) {
610 return parsedir +'/' + encodeURIComponent(url);
612 if (File.exists(donedir +'/' + encodeURIComponent(url))) {
613 return donedir +'/' + encodeURIComponent(url);
619 this.moveToParse = function(url)
621 var old = this.dupeCheck(url);
622 var target =parsedir +'/' + encodeURIComponent(url);
626 File.write(target, old ? File.read(old) : '');
633 this.moveToDownload= function(url)
635 var old = this.dupeCheck(url);
636 var target =downloaddir +'/' + encodeURIComponent(url);
640 File.write(target, old ? File.read(old) : '');
646 this.moveToDone= function(url)
648 var old = this.dupeCheck(url);
649 var target = donedir +'/' + encodeURIComponent(url);
653 File.write(target, old ? File.read(old) : '');
663 function parseUri (str) {
664 var o = parseUri.options,
665 m = o.parser[o.strictMode ? "strict" : "loose"].exec(str),
669 while (i--) uri[o.key[i]] = m[i] || "";
672 uri[o.key[12]].replace(o.q.parser, function ($0, $1, $2) {
673 if ($1) uri[o.q.name][$1] = $2;
681 key: ["source","protocol","authority","userInfo","user","password","host","port","relative","path","directory","file","query","anchor"],
684 parser: /(?:^|&)([^&=]*)=?([^&]*)/g
687 strict: /^(?:([^:\/?#]+):)?(?:\/\/((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?))?((((?:[^?#\/]*\/)*)([^?#]*))(?:\?([^#]*))?(?:#(.*))?)/,
688 loose: /^(?:(?![^:@]+:[^:@\/]*@)([^:\/?#.]+):)?(?:\/\/)?((?:(([^:@]*)(?::([^:@]*))?)?@)?([^:\/?#]*)(?::(\d*))?)(((\/(?:[^?#](?![^?#\/]*\.[^?#\/.]+(?:[?#]|$)))*\/?)?([^?#\/]*))(?:\?([^#]*))?(?:#(.*))?)/