Skip to content

Instantly share code, notes, and snippets.

@pulipulichen
Created March 14, 2013 18:20
Show Gist options
  • Save pulipulichen/5163827 to your computer and use it in GitHub Desktop.
Save pulipulichen/5163827 to your computer and use it in GitHub Desktop.
STPI Government Research Bulletin Zotero Translator
{
"translatorID": "0d6fbeb8-a7b6-4eec-8421-d2cbf5061a88",
"label": "STPI Government Research Bulletin",
"creator": "Pulipuli Chen",
"target": "^http?://grbsearch\\.stpi.narl\\.org\\.tw/GRB/",
"minVersion": "1.0",
"maxVersion": "",
"priority": 100,
"inRepository": true,
"translatorType": 5,
"browserSupport": "gcsi",
"lastUpdated": "2013-03-11 15:41:19"
}
/* 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()};
function detectWeb(doc, url) {
return FW.detectWeb(doc, url);
}
function doWeb(doc, url) { return FW.doWeb(doc, url); }
/** Reportes */
FW.Scraper({ itemType : 'report',
detect : FW.Xpath('//*[@id="table1"]'),
title : FW.Xpath('//*[@id="table1"]/tbody/tr[1]/td[2]').text().trim(),
creators : FW.Xpath('//*[@id="table1"]/tbody/tr[9]/td[2]').text().trim().split(/,/).cleanAuthor("author"),
abstractNote : FW.Xpath('//*[@id="table1"]/tbody/tr[12]/td[2]').text().trim().replace(/查無中文摘要/, ''),
accessDate : "CURRENT_TIMESTAMP",
reportType : FW.Xpath('//*[@id="table1"]/tbody/tr[3]/td[4]').text().trim(),
date : FW.Xpath('//*[@id="table1"]/tbody/tr[7]/td[2]').text().trim()
.addFilter(function (_chinese_year) {
var _year = _chinese_year;
var _footer = "年";
if (_year.substr((_year.length-_footer.length) , _year.length) == _footer) {
_year = _year.split(" ");
_year = parseInt(_year, 10) + 1911;
}
return _year;
}),
reportNumber : FW.Xpath('//*[@id="table1"]/tbody/tr[4]/td[2]').text().trim(),
institution : FW.Xpath('//*[@id="table1"]/tbody/tr[6]/td[2]').text().trim(),
tags : [FW.Xpath('//*[@id="table1"]/tbody/tr[10]/td[2]').text().trim().replace(/--/, '').split(";")
, FW.Xpath('//*[@id="table1"]/tbody/tr[11]/td[2]').text().trim().replace(/--/, '').split(";")]
});
/** Search results */
/*
FW.MultiScraper({ itemType : "multiple",
detect : FW.Xpath('//table[@class="sbody"]'),
choices : {
titles : FW.Xpath('/html/body/form[2]/table[1]/tbody/tr/td/table[3]/tbody/tr[1]/td[1]/table/tbody/tr[1]/td[2]/a').text(),
urls : FW.Xpath('/html/body/form[2]/table[1]/tbody/tr/td/table[3]/tbody/tr[1]/td[1]/table/tbody/tr[1]/td[2]/a').key('href').text()
//.replace(/amp;/, '')
//.prepend("http://grbsearch.stpi.narl.org.tw/GRB/")
.addFilter(function (s) {
//return encodeURI(s);
var _needle = ";highStr";
var _pos;
for (var i = 0; i < s.length; i++)
{
_pos = s[i].indexOf(_needle) - 4;
s[i] = s[i].substr(0, _pos);
s[i] = "result.jsp?id=1680148&amp;plan_no=NSC97-2517-S166-001&amp;plan_year=97&amp;projkey=PF9709-1059&amp;target=plan";
}
//s = s.substr(0, _pos);
//return "http://grbsearch.stpi.narl.org.tw/GRB/result.jsp?id=995439&plan_no=NSC93-2411-H364-001&plan_year=93&projkey=PE9308-0844&target=plan&highStr=%E9%99%B3&check=0&pnchDesc=%E9%80%A2%E8%AE%8A%E4%B9%8B%E9%9A%9B%EF%BC%9A1940%E5%B9%B4%E4%BB%A3%E4%B8%AD%E8%A5%BF%E5%8F%B2%E5%AE%B6%E5%8F%8A%E5%8F%B2%E5%AD%B8%E6%80%9D%E6%83%B3%E4%B9%8B%E7%A0%94%E7%A9%B6---%E8%90%AC%E9%87%8C%E4%B9%BE%E5%9D%A4%E8%BF%B7%E5%8E%BB%E4%BD%8F%EF%BC%9A1940%E5%B9%B4%E4%BB%A3%E9%99%B3%E5%AF%85%E6%81%AA%E5%8F%B2%E5%AD%B8%E8%88%87%E4%BA%BA%E7%94%9F%E7%9A%84%E8%BD%89%E6%8A%98%28I%29";
//return encodeURI("http://grbsearch.stpi.narl.org.tw/GRB/result.jsp?id=995439&plan_no=NSC93-2411-H364-001&plan_year=93&projkey=PE9308-0844&target=plan");
return s;
})
}
});
*/
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment