| 1 | REPLACE INTO translators VALUES ('57a00950-f0d1-4b41-b6ba-44ff0fc30289', '1.0.0b3.r1', '', '2007-08-30 15:50:40', '1', '100', '4', 'Google Scholar', 'Simon Kornblith', '^http://scholar\.google\.[a-z]+/scholar', |
|---|
| 2 | 'function detectWeb(doc, url) { |
|---|
| 3 | return "multiple"; |
|---|
| 4 | }', |
|---|
| 5 | 'var haveEndNoteLinks; |
|---|
| 6 | |
|---|
| 7 | function scrape(doc) { |
|---|
| 8 | var nsResolver = doc.createNSResolver(doc.documentElement); |
|---|
| 9 | |
|---|
| 10 | var items = new Array(); |
|---|
| 11 | var itemGrabLinks = new Array(); |
|---|
| 12 | var links = new Array(); |
|---|
| 13 | var types = new Array(); |
|---|
| 14 | |
|---|
| 15 | var itemTypes = new Array(); |
|---|
| 16 | var attachments = new Array(); |
|---|
| 17 | |
|---|
| 18 | var elmts = doc.evaluate(''//p[@class="g"]'', doc, nsResolver, |
|---|
| 19 | XPathResult.ANY_TYPE, null); |
|---|
| 20 | var elmt; |
|---|
| 21 | var i=0; |
|---|
| 22 | Zotero.debug("get elms"); |
|---|
| 23 | while(elmt = elmts.iterateNext()) { |
|---|
| 24 | var isCitation = doc.evaluate("./font[1]/b[1]/text()[1]", elmt, nsResolver, |
|---|
| 25 | XPathResult.ANY_TYPE, null).iterateNext(); |
|---|
| 26 | // use EndNote links if available |
|---|
| 27 | if(haveEndNoteLinks) { |
|---|
| 28 | var itemGrabLink = doc.evaluate(''.//a[text() = "Import into EndNote"]'', |
|---|
| 29 | elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); |
|---|
| 30 | } else { |
|---|
| 31 | var itemGrabLink = doc.evaluate(''.//a[text() = "Related Articles"]'', |
|---|
| 32 | elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); |
|---|
| 33 | } |
|---|
| 34 | |
|---|
| 35 | var noLinkRe = /^\[[^\]]+\]$/; |
|---|
| 36 | if(itemGrabLinks) { |
|---|
| 37 | itemGrabLinks[i] = itemGrabLink.href; |
|---|
| 38 | if(isCitation && noLinkRe.test(isCitation.textContent)) { |
|---|
| 39 | // get titles for [BOOK] or [CITATION] entries |
|---|
| 40 | items[i] = Zotero.Utilities.getNodeString(doc, elmt, ''./text()|./b/text()'', nsResolver); |
|---|
| 41 | } else { |
|---|
| 42 | // get titles for articles |
|---|
| 43 | var link = doc.evaluate(''.//a'', elmt, nsResolver, |
|---|
| 44 | XPathResult.ANY_TYPE, null).iterateNext(); |
|---|
| 45 | if(link) { |
|---|
| 46 | items[i] = link.textContent; |
|---|
| 47 | links[i] = link.href; |
|---|
| 48 | } |
|---|
| 49 | } |
|---|
| 50 | |
|---|
| 51 | if(items[i]) { |
|---|
| 52 | i++; |
|---|
| 53 | } |
|---|
| 54 | } |
|---|
| 55 | } |
|---|
| 56 | |
|---|
| 57 | items = Zotero.selectItems(items); |
|---|
| 58 | if(!items) { |
|---|
| 59 | if(Zotero.done) Zotero.done(true); |
|---|
| 60 | return true; |
|---|
| 61 | } |
|---|
| 62 | |
|---|
| 63 | var relatedMatch = /[&?]q=related:([^&]+)/; |
|---|
| 64 | |
|---|
| 65 | var urls = new Array(); |
|---|
| 66 | for(var i in items) { |
|---|
| 67 | // get url |
|---|
| 68 | if(haveEndNoteLinks) { |
|---|
| 69 | urls.push(itemGrabLinks[i]); |
|---|
| 70 | } else { |
|---|
| 71 | var m = relatedMatch.exec(itemGrabLinks[i]); |
|---|
| 72 | urls.push("http://scholar.google.com/scholar.ris?hl=en&lr=&q=info:"+m[1]+"&oe=UTF-8&output=citation&oi=citation"); |
|---|
| 73 | } |
|---|
| 74 | |
|---|
| 75 | if(links[i]) { |
|---|
| 76 | attachments.push([{title:"Google Scholar Linked Page", type:"text/html", |
|---|
| 77 | url:links[i]}]); |
|---|
| 78 | } else { |
|---|
| 79 | attachments.push([]); |
|---|
| 80 | } |
|---|
| 81 | } |
|---|
| 82 | |
|---|
| 83 | var translator = Zotero.loadTranslator("import"); |
|---|
| 84 | translator.setTranslator("881f60f2-0802-411a-9228-ce5f47b64c7d"); |
|---|
| 85 | translator.setHandler("itemDone", function(obj, item) { |
|---|
| 86 | item.attachments = attachments.shift(); |
|---|
| 87 | item.complete(); |
|---|
| 88 | }); |
|---|
| 89 | |
|---|
| 90 | Zotero.Utilities.HTTP.doGet(urls, function(text) { |
|---|
| 91 | translator.setString(text); |
|---|
| 92 | translator.translate(); |
|---|
| 93 | }, function() { Zotero.done() }); |
|---|
| 94 | } |
|---|
| 95 | |
|---|
| 96 | function doWeb(doc, url) { |
|---|
| 97 | var prefsurl; |
|---|
| 98 | var nsResolver = doc.createNSResolver(doc.documentElement); |
|---|
| 99 | |
|---|
| 100 | //SR:Will use preference setting url instead of cookie to get EndNote links (works with ezproxy, doesn''t overwrite other prefs) |
|---|
| 101 | //doc.cookie = "GSP=ID=deadbeefdeadbeef:IN=ebe89f7e83a8fe75+7e6cc990821af63:CF=3; domain=.scholar.google.com"; |
|---|
| 102 | |
|---|
| 103 | // determine if we need to reload the page |
|---|
| 104 | |
|---|
| 105 | // first check for EndNote links |
|---|
| 106 | Zotero.debug("get links"); |
|---|
| 107 | haveEndNoteLinks = doc.evaluate(''//a[text() = "Import into EndNote"]'', |
|---|
| 108 | doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); |
|---|
| 109 | if(!haveEndNoteLinks) { |
|---|
| 110 | // SR:Commenting out this bit as code for retreiving citations from "Related" links is unreliable and unnecessary |
|---|
| 111 | //// next check if there are docs with no related articles |
|---|
| 112 | //if(doc.evaluate(''//p[@class="g"][not(descendant-or-self::text() = "Related Articles")]'', |
|---|
| 113 | // doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext()) { |
|---|
| 114 | // SR:Set preferences to show links and do page reload |
|---|
| 115 | prefsurl = url.replace("scholar?", "scholar_setprefs?scis=yes&scisf=3&submit=Save+Preferences&"); |
|---|
| 116 | Zotero.Utilities.loadDocument(prefsurl, scrape); |
|---|
| 117 | haveEndNoteLinks = true; |
|---|
| 118 | return; |
|---|
| 119 | //} |
|---|
| 120 | } |
|---|
| 121 | |
|---|
| 122 | scrape(doc, url); |
|---|
| 123 | }'); |
|---|