Skip to content

Commit 971fbfd

Browse files
committedDec 12, 2013
Fixes Bloomberg, Slideshare;
Small cleaning fixes for FP, Ingental EEBO; Switch out tests for DOI and NPG; Test updates Guardian, IoP, ScienceDirect
1 parent c212889 commit 971fbfd

11 files changed

+93
-74
lines changed
 

‎Bloomberg.js

+7-7
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@
99
"inRepository": true,
1010
"translatorType": 4,
1111
"browserSupport": "gcsibv",
12-
"lastUpdated": "2013-04-15 18:04:37"
12+
"lastUpdated": "2013-12-12 11:55:06"
1313
}
1414

1515
/* FW LINE 57:6869c32952b1 */ function flatten(c){var b=new Array();for(var d in c){var e=c[d];if(e instanceof Array){b=b.concat(flatten(e))}else{b.push(e)}}return b}var FW={_scrapers:new Array()};FW._Base=function(){this.callHook=function(b,c,e,a){if(typeof this["hooks"]==="object"){var d=this["hooks"][b];if(typeof d==="function"){d(c,e,a)}}};this.evaluateThing=function(f,e,c){var b=typeof f;if(b==="object"){if(f instanceof Array){var d=this.evaluateThing;var a=f.map(function(g){return d(g,e,c)});return flatten(a)}else{return f.evaluate(e,c)}}else{if(b==="function"){return f(e,c)}else{return f}}}};FW.Scraper=function(a){FW._scrapers.push(new FW._Scraper(a))};FW._Scraper=function(a){for(x in a){this[x]=a[x]}this._singleFieldNames=["abstractNote","applicationNumber","archive","archiveLocation","artworkMedium","artworkSize","assignee","audioFileType","audioRecordingType","billNumber","blogTitle","bookTitle","callNumber","caseName","code","codeNumber","codePages","codeVolume","committee","company","conferenceName","country","court","date","dateDecided","dateEnacted","dictionaryTitle","distributor","docketNumber","documentNumber","DOI","edition","encyclopediaTitle","episodeNumber","extra","filingDate","firstPage","forumTitle","genre","history","institution","interviewMedium","ISBN","ISSN","issue","issueDate","issuingAuthority","journalAbbreviation","label","language","legalStatus","legislativeBody","letterType","libraryCatalog","manuscriptType","mapType","medium","meetingName","nameOfAct","network","number","numberOfVolumes","numPages","pages","patentNumber","place","postType","presentationType","priorityNumbers","proceedingsTitle","programTitle","programmingLanguage","publicLawNumber","publicationTitle","publisher","references","reportNumber","reportType","reporter","reporterVolume","rights","runningTime","scale","section","series","seriesNumber","seriesText","seriesTitle","session","shortTitle","studio","subject","system","thesisType","title","type","university","url","version","videoRecordingType","volume","websiteTitle","websiteType"];this._makeAttachments=function(p,b,g,t){if(g instanceof Array){g.forEach(function(k){this._makeAttachments(p,b,k,t)},this)}else{if(typeof g==="object"){var o=g.urls||g.url;var m=g.types||g.type;var f=g.titles||g.title;var q=g.snapshots||g.snapshot;var j=this.evaluateThing(o,p,b);var n=this.evaluateThing(f,p,b);var s=this.evaluateThing(m,p,b);var d=this.evaluateThing(q,p,b);if(!(j instanceof Array)){j=[j]}for(var l in j){var c=j[l];var h;var e;var r;if(s instanceof Array){h=s[l]}else{h=s}if(n instanceof Array){e=n[l]}else{e=n}if(d instanceof Array){r=d[l]}else{r=d}t.attachments.push({url:c,title:e,type:h,snapshot:r})}}}};if(this.itemTrans!==undefined){this.makeItems=this.itemTrans.makeItems}else{this.makeItems=function(o,b,m,c,l){var q=new Zotero.Item(this.itemType);q.url=b;for(var h in this._singleFieldNames){var n=this._singleFieldNames[h];if(this[n]){var g=this.evaluateThing(this[n],o,b);if(g instanceof Array){q[n]=g[0]}else{q[n]=g}}}var r=["creators","tags"];for(var f in r){var p=r[f];var d=this.evaluateThing(this[p],o,b);if(d){for(var e in d){q[p].push(d[e])}}}this._makeAttachments(o,b,this["attachments"],q);c(q,this,o,b);l([q])}}};FW._Scraper.prototype=new FW._Base;FW.MultiScraper=function(a){FW._scrapers.push(new FW._MultiScraper(a))};FW._MultiScraper=function(a){for(x in a){this[x]=a[x]}this._mkSelectItems=function(e,d){var b=new Object;for(var c in e){b[d[c]]=e[c]}return b};this._selectItems=function(d,c,e){var b=new Array();Zotero.selectItems(this._mkSelectItems(d,c),function(f){for(var g in f){b.push(g)}e(b)})};this._mkAttachments=function(g,d,f){var b=this.evaluateThing(this["attachments"],g,d);var c=new Object();if(b){for(var e in f){c[f[e]]=b[e]}}return c};this._makeChoices=function(f,p,c,d,h){if(f instanceof Array){f.forEach(function(k){this._makeTitlesUrls(k,p,c,d,h)},this)}else{if(typeof f==="object"){var m=f.urls||f.url;var e=f.titles||f.title;var n=this.evaluateThing(m,p,c);var j=this.evaluateThing(e,p,c);var l=(j instanceof Array);if(!(n instanceof Array)){n=[n]}for(var g in n){var b=n[g];var o;if(l){o=j[g]}else{o=j}h.push(b);d.push(o)}}}};this.makeItems=function(j,b,g,c,f){if(this.beforeFilter){var k=this.beforeFilter(j,b);if(k!=b){this.makeItems(j,k,g,c,f);return}}var e=[];var h=[];this._makeChoices(this["choices"],j,b,e,h);var d=this._mkAttachments(j,b,h);this._selectItems(e,h,function(m){if(!m){f([])}else{var l=[];var n=this.itemTrans;Zotero.Utilities.processDocuments(m,function(q){var p=q.documentURI;var o=n;if(o===undefined){o=FW.getScraper(q,p)}if(o===undefined){}else{o.makeItems(q,p,d[p],function(r){l.push(r);c(r,o,q,p)},function(){})}},function(){f(l)})}})}};FW._MultiScraper.prototype=new FW._Base;FW.DelegateTranslator=function(a){return new FW._DelegateTranslator(a)};FW._DelegateTranslator=function(a){for(x in a){this[x]=a[x]}this._translator=Zotero.loadTranslator(this.translatorType);this._translator.setTranslator(this.translatorId);this.makeItems=function(g,d,b,f,c){var e;Zotero.Utilities.HTTP.doGet(d,function(h){this._translator.setHandler("itemDone",function(k,j){e=j;if(b){j.attachments=b}});if(this.preProcess){h=this.preProcess(h)}this._translator.setString(h);this._translator.translate();f(e)},function(){c([e])})}};FW.DelegateTranslator.prototype=new FW._Scraper;FW._StringMagic=function(){this._filters=new Array();this.addFilter=function(a){this._filters.push(a);return this};this.split=function(a){return this.addFilter(function(b){return b.split(a).filter(function(c){return(c!="")})})};this.replace=function(c,b,a){return this.addFilter(function(d){if(d.match(c)){return d.replace(c,b,a)}else{return d}})};this.prepend=function(a){return this.replace(/^/,a)};this.append=function(a){return this.replace(/$/,a)};this.remove=function(b,a){return this.replace(b,"",a)};this.trim=function(){return this.addFilter(function(a){return Zotero.Utilities.trim(a)})};this.trimInternal=function(){return this.addFilter(function(a){return Zotero.Utilities.trimInternal(a)})};this.match=function(a,b){if(!b){b=0}return this.addFilter(function(d){var c=d.match(a);if(c===undefined||c===null){return undefined}else{return c[b]}})};this.cleanAuthor=function(b,a){return this.addFilter(function(c){return Zotero.Utilities.cleanAuthor(c,b,a)})};this.key=function(a){return this.addFilter(function(b){return b[a]})};this.capitalizeTitle=function(){if(arguments.length>0&&arguments[0]==true){return this.addFilter(function(a){return Zotero.Utilities.capitalizeTitle(a,true)})}else{return this.addFilter(function(a){return Zotero.Utilities.capitalizeTitle(a)})}};this.unescapeHTML=function(){return this.addFilter(function(a){return Zotero.Utilities.unescapeHTML(a)})};this.unescape=function(){return this.addFilter(function(a){return unescape(a)})};this._applyFilters=function(c,e){for(i in this._filters){c=flatten(c);c=c.filter(function(a){return((a!==undefined)&&(a!==null))});for(var d=0;d<c.length;d++){try{if((c[d]===undefined)||(c[d]===null)){continue}else{c[d]=this._filters[i](c[d],e)}}catch(b){c[d]=undefined;Zotero.debug("Caught exception "+b+"on filter: "+this._filters[i])}}c=c.filter(function(a){return((a!==undefined)&&(a!==null))})}return flatten(c)}};FW.PageText=function(){return new FW._PageText()};FW._PageText=function(){this._filters=new Array();this.evaluate=function(c){var b=[c.documentElement.innerHTML];b=this._applyFilters(b,c);if(b.length==0){return false}else{return b}}};FW._PageText.prototype=new FW._StringMagic();FW.Url=function(){return new FW._Url()};FW._Url=function(){this._filters=new Array();this.evaluate=function(d,c){var b=[c];b=this._applyFilters(b,d);if(b.length==0){return false}else{return b}}};FW._Url.prototype=new FW._StringMagic();FW.Xpath=function(a){return new FW._Xpath(a)};FW._Xpath=function(a){this._xpath=a;this._filters=new Array();this.text=function(){var b=function(c){if(typeof c==="object"&&c.textContent){return c.textContent}else{return c}};this.addFilter(b);return this};this.sub=function(b){var c=function(f,e){var d=e.evaluate(b,f,null,XPathResult.ANY_TYPE,null);if(d){return d.iterateNext()}else{return undefined}};this.addFilter(c);return this};this.evaluate=function(f){var e=f.evaluate(this._xpath,f,null,XPathResult.ANY_TYPE,null);var d=e.resultType;var c=new Array();if(d==XPathResult.STRING_TYPE){c.push(e.stringValue)}else{if(d==XPathResult.ORDERED_NODE_ITERATOR_TYPE||d==XPathResult.UNORDERED_NODE_ITERATOR_TYPE){var b;while((b=e.iterateNext())){c.push(b)}}}c=this._applyFilters(c,f);if(c.length==0){return false}else{return c}}};FW._Xpath.prototype=new FW._StringMagic();FW.detectWeb=function(e,b){for(var c in FW._scrapers){var d=FW._scrapers[c];var f=d.evaluateThing(d.itemType,e,b);var a=d.evaluateThing(d.detect,e,b);if(a.length>0&&a[0]){return f}}return undefined};FW.getScraper=function(b,a){var c=FW.detectWeb(b,a);return FW._scrapers.filter(function(d){return(d.evaluateThing(d.itemType,b,a)==c)&&(d.evaluateThing(d.detect,b,a))})[0]};FW.doWeb=function(c,a){var b=FW.getScraper(c,a);b.makeItems(c,a,[],function(f,e,g,d){e.callHook("scraperDone",f,g,d);if(!f.title){f.title=""}f.complete()},function(){Zotero.done()});Zotero.wait()};
@@ -42,13 +42,13 @@ function doWeb(doc, url) { return FW.doWeb(doc, url); }
4242
/**Webpage*/
4343
FW.Scraper({
4444
itemType : 'webpage',
45-
detect : FW.Xpath('//div[@id="disqus_title"]/h1'),
46-
title : FW.Xpath('//div[@id="disqus_title"]/h1').text().trim(),
45+
detect : FW.Xpath('//h1[@class="article_title"]'),
46+
title : FW.Xpath('//h1[@class="article_title"]').text().trim(),
4747
attachments : [{ url: FW.Url().replace(/news\//, "news/print/"),
4848
title: "Bloomberg Printable",
4949
type: "text/html" }],
50-
creators : FW.Xpath('//div[@id="primary_content"]//cite[@class="byline"]/span[not(@class="datestamp")]').text().replace(/^\s*By/, "").trimInternal().replace(/ -$/, "").split(/ and /).cleanAuthor("author"),
51-
date : FW.Xpath('//p[@class="sByline"]').text().replace(/.+\//, "").replace(/By/, ""),
50+
creators : FW.Xpath('//div[@class="byline"]/span[@class="author"]').text().replace(/^\s*By/, "").trimInternal().replace(/ -$/, "").split(/ and /).cleanAuthor("author"),
51+
date : FW.Xpath('//div[@class="byline"]/span[@class="date"]').text(),
5252
abstractNote : FW.Xpath('//meta[@name="description"]/@content').text(),
5353
websiteTitle : "Bloomberg",
5454
publicationTitle : FW.Xpath('//div[@id="mainColumn"]/div[1]/div/a/img/@title').text().prepend("CSM Blog: ")
@@ -92,13 +92,13 @@ var testCases = [
9292
"seeAlso": [],
9393
"attachments": [
9494
{
95-
"url": "http://www.bloomberg.com/news/print/2012-01-04/bank-earnings-increase-57-in-analyst-forecasts-which-proved-wrong-in-2011.html",
9695
"title": "Bloomberg Printable",
9796
"type": "text/html"
9897
}
9998
],
10099
"url": "http://www.bloomberg.com/news/2012-01-04/bank-earnings-increase-57-in-analyst-forecasts-which-proved-wrong-in-2011.html",
101100
"abstractNote": "Analysts’ failure to foresee declining earnings per share for the biggest U.S. banks last year hasn’t stopped them from predicting an even bigger profit surge for 2012.",
101+
"date": "Jan 3, 2012 5:00 PM MT",
102102
"title": "Bank Earnings Jump 57% in Analyst Forecasts",
103103
"websiteTitle": "Bloomberg",
104104
"libraryCatalog": "Bloomberg",
@@ -118,13 +118,13 @@ var testCases = [
118118
"seeAlso": [],
119119
"attachments": [
120120
{
121-
"url": "http://www.bloomberg.com/news/print/2012-01-05/four-economists-come-together-to-say-we-agree-business-class.html",
122121
"title": "Bloomberg Printable",
123122
"type": "text/html"
124123
}
125124
],
126125
"url": "http://www.bloomberg.com/news/2012-01-05/four-economists-come-together-to-say-we-agree-business-class.html",
127126
"abstractNote": "“If you laid all the economists in the world end to end, they still wouldn’t reach a conclusion.” This old joke still works because it reflects a common belief that economists can’t agree on anything important. Yet the four of us are part of a project that we believe will demonstrate that this proposition is wrong.",
127+
"date": "Jan 4, 2012 5:01 PM MT",
128128
"title": "Four Economists Come Together to Say ‘We Agree’: Business Class",
129129
"websiteTitle": "Bloomberg",
130130
"libraryCatalog": "Bloomberg",

‎DOI.js

+22-25
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@
99
"inRepository": true,
1010
"translatorType": 4,
1111
"browserSupport": "gcsv",
12-
"lastUpdated": "2012-04-04 05:29:41"
12+
"lastUpdated": "2013-12-12 12:35:55"
1313
}
1414

1515
var items = {};
@@ -35,6 +35,7 @@ function getDOIs(doc) {
3535
var node, m, DOI;
3636
var results = doc.evaluate(DOIXPath, doc, null, XPathResult.ANY_TYPE, null);
3737
while(node = results.iterateNext()) {
38+
//Z.debug(node.nodeValue)
3839
DOIre.lastMatch = 0;
3940
while(m = DOIre.exec(node.nodeValue)) {
4041
DOI = m[0];
@@ -137,25 +138,25 @@ function doWeb(doc, url) {
137138
var testCases = [
138139
{
139140
"type": "web",
140-
"url": "http://www.crossref.org/help/Content/01_About_DOIs/What_is_a_DOI.htm",
141+
"url": "http://blog.apastyle.org/apastyle/digital-object-identifier-doi/",
142+
"items": "multiple"
143+
},
144+
{
145+
"type": "web",
146+
"url": "http://libguides.csuchico.edu/citingbusiness",
141147
"items": [
142148
{
143149
"itemType": "journalArticle",
144150
"creators": [
145151
{
146152
"creatorType": "author",
147-
"firstName": "Jianxun",
148-
"lastName": "Mou"
153+
"firstName": "Jean M.",
154+
"lastName": "Twenge"
149155
},
150156
{
151157
"creatorType": "author",
152-
"firstName": "Jie",
153-
"lastName": "Yang"
154-
},
155-
{
156-
"creatorType": "author",
157-
"firstName": "Zhifeng",
158-
"lastName": "Shao"
158+
"firstName": "Stacy M.",
159+
"lastName": "Campbell"
159160
}
160161
],
161162
"notes": [],
@@ -164,23 +165,19 @@ var testCases = [
164165
"attachments": [
165166
{}
166167
],
167-
"issue": "3",
168-
"ISSN": "00222836",
169-
"DOI": "10.1006/jmbi.1995.0238",
170-
"url": "http://www.crossref.org/help/Content/01_About_DOIs/What_is_a_DOI.htm",
168+
"publicationTitle": "Journal of Managerial Psychology",
169+
"volume": "23",
170+
"issue": "8",
171+
"ISSN": "0268-3946",
172+
"date": "2008",
173+
"pages": "862-877",
174+
"DOI": "10.1108/02683940810904367",
175+
"url": "http://libguides.csuchico.edu/citingbusiness",
176+
"title": "Generational differences in psychological traits and their impact on the workplace",
171177
"libraryCatalog": "CrossRef",
172-
"publicationTitle": "Journal of Molecular Biology",
173-
"volume": "248",
174-
"date": "5/1995",
175-
"pages": "507-512",
176-
"title": "Atomic Force Microscopy of Cholera Toxin B-oligomers Bound to Bilayers of Biologically Relevant Lipids"
178+
"accessDate": "CURRENT_TIMESTAMP"
177179
}
178180
]
179-
},
180-
{
181-
"type": "web",
182-
"url": "http://blog.apastyle.org/apastyle/digital-object-identifier-doi/",
183-
"items": "multiple"
184181
}
185182
]
186183
/** END TEST CASES **/

‎Early English Books Online.js

+3-4
Original file line numberDiff line numberDiff line change
@@ -9,7 +9,7 @@
99
"inRepository": true,
1010
"translatorType": 4,
1111
"browserSupport": "gcsb",
12-
"lastUpdated": "2013-12-07 14:52:55"
12+
"lastUpdated": "2013-12-12 12:41:36"
1313
}
1414

1515
function detectWeb(doc, url) {
@@ -73,15 +73,15 @@ function scrape(eeboIDs, host){
7373
var postString = 'cit_format=RIS&Print=Print&cit_eeboid=' + eeboIDs[i] + '&EeboId=' + eeboIDs[i];
7474
var new_eeboid = eeboIDs[i]
7575
Zotero.Utilities.HTTP.doPost(host+'/search/print', postString, function(text) {
76-
Z.debug(text)
76+
//Z.debug(text)
7777
// load translator for RIS
7878
var translator = Zotero.loadTranslator("import");
7979
translator.setTranslator("32d59d2d-b65a-4da4-b0a3-bdd3cfb979e7");
8080
translator.setString(text.substring(17));
8181
translator.setHandler("itemDone", function(obj, item) {
8282
//remove date for authors
8383
for (var i in item.creators){
84-
if (item.creators[i].firstName) item.creators[i].firstName = item.creators[i].firstName.replace(/[,\-\d\.]+$/, "")
84+
if (item.creators[i].firstName) item.creators[i].firstName = item.creators[i].firstName.replace(/[,\-\d\.\s]+$/g, "")
8585
}
8686
item.attachments.push(
8787
{url : host+'/search/full_rec?SOURCE=pgimages.cfg&ACTION=ByID&ID=' + new_eeboid + '&FILE=../session/1190302085_15129&SEARCHSCREEN=CITATIONS&SEARCHCONFIG=config.cfg&DISPLAY=ALPHA',
@@ -90,7 +90,6 @@ function scrape(eeboIDs, host){
9090
item.complete();
9191
});
9292
translator.translate();
93-
Zotero.done();
9493
});
9594
}
9695
}

0 commit comments

Comments
 (0)
Please sign in to comment.