From 2dce530f1be4d780cf1f52eefe64b8e2285955cc Mon Sep 17 00:00:00 2001 From: Michael Berkowitz Date: Tue, 20 May 2008 20:26:09 +0000 Subject: [PATCH] -Adds Wildlife Biology in Practice translator. --- scrapers.sql | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 72 insertions(+), 1 deletion(-) diff --git a/scrapers.sql b/scrapers.sql index 7030d88dc..b879582ed 100644 --- a/scrapers.sql +++ b/scrapers.sql @@ -22,7 +22,7 @@ -- Set the following timestamp to the most recent scraper update date -REPLACE INTO version VALUES ('repository', STRFTIME('%s', '2008-05-20 19:00:00')); +REPLACE INTO version VALUES ('repository', STRFTIME('%s', '2008-05-20 21:30:00')); REPLACE INTO translators VALUES ('96b9f483-c44d-5784-cdad-ce21b984fe01', '1.0.0b4.r1', '', '2008-03-21 20:00:00', '1', '100', '4', 'Amazon.com', 'Sean Takats and Michael Berkowitz', '^https?://(?:www\.)?amazon', 'function detectWeb(doc, url) { @@ -1149,6 +1149,77 @@ REPLACE INTO translators VALUES ('83538f48-906f-40ef-bdb3-e94f63676307', '1.0.0b }, function() {Zotero.done;}); }'); +REPLACE INTO translators VALUES ('b33af0e1-d122-45b2-b144-4b4eedd12d5d', '1.0.0b4.r5', '', '2008-05-20 21:30:00', '0', '100', '4', 'Wildlife Biology in Practice', 'Michael Berkowitz', 'http://www.socpvs.org/journals/index.php/wbp', +'function detectWeb(doc, url) { + if (url.match(/showToc/) || url.match(/advancedResults/)) { + return "multiple"; + } else if (url.match(/article/)) { + return "journalArticle"; + } +}', +'function doWeb(doc, url) { + var n = doc.documentElement.namespaceURI; + var ns = n ? function(prefix) { + if (prefix == ''x'') return n; else return null; + } : null; + + var arts = new Array(); + if (detectWeb(doc, url) == "multiple") { + var items = new Object(); + var xpath = ''//tr[td/a[2]]''; + if (url.match(/issue/)) { + var linkx = ''./td[2]/a[1]''; + var titlex = ''./td[1]''; + } else if (url.match(/advanced/)) { + var linkx = ''./td[3]/a[1]''; + var titlex = ''./td[2]''; + } + var results = doc.evaluate(xpath, doc, ns, XPathResult.ANY_TYPE, null); + var result; + while (result = results.iterateNext()) { + var title = doc.evaluate(titlex, result, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent; + var link = doc.evaluate(linkx, result, ns, XPathResult.ANY_TYPE, null).iterateNext().href; + items[link] = Zotero.Utilities.trimInternal(title); + } + items = Zotero.selectItems(items); + for (var i in items) { + arts.push(i.replace(/view/, "viewArticle")); + } + } else { + arts = [url.replace(/viewRST/, "viewArticle")]; + } + Zotero.Utilities.processDocuments(arts, function(doc) { + var item = new Zotero.Item("journalArticle"); + var voliss = Zotero.Utilities.trimInternal(doc.evaluate(''//div[@id="main"]/h2'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent); + voliss = voliss.match(/^([^,]+),\s+([^;]+);\s+(\d+)\((\d+)\);\s+([^;]+)/); + item.journalAbbreviation = voliss[1]; + item.date = voliss[2]; + item.issue = voliss[3]; + item.volume = voliss[4]; + item.pages = voliss[5]; + var authors = doc.evaluate(''//div[@id="authorDetails"]/ul[@class="lista"]/li/strong/a'', doc, ns, XPathResult.ANY_TYPE, null); + var author; + while (author = authors.iterateNext()) { + item.creators.push(Zotero.Utilities.cleanAuthor(author.title.match(/^\w+\b\s+(.*)\s+\b\w+$/)[1], "author")); + } + item.publicationTitle = "Wildlife Biology in Practice"; + item.ISSN = "1646-2742"; + item.DOI = doc.evaluate(''//div[@id="copyArticle"]/a[1]'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent.match(/doi:\s+([^\s]+)/)[1]; + item.url = doc.location.href; + item.title = Zotero.Utilities.trimInternal(doc.evaluate(''//div[@id="content"]/h3'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent); + item.abstractNote = Zotero.Utilities.trimInternal(doc.evaluate(''//div[@id="abstract"]/blockquote/p'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent); + item.tags = doc.evaluate(''//div[@id="abstract"]'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().textContent.match(/Keywords:\s+([^\.]+)/)[1].split(/,\s+/); + + var pdfurl = doc.evaluate(''//div[@id="rt"]/a[@class="action noarrow"]'', doc, ns, XPathResult.ANY_TYPE, null).iterateNext().href; + item.attachments = [ + {url:item.url, title:item.publicationTitle + " Snapshot", mimeType:"text/html"}, + {url:pdfurl, title:item.publicationTitle + " PDF", mimeType:"application/pdf"} + ]; + item.complete(); + }, function() {Zotero.done;}); + Zotero.wait(); +}'); + REPLACE INTO translators VALUES ('d2416f31-4f24-4e18-8c66-06122af5bc2c', '1.0.0b4.r5', '', '2008-05-20 19:10:00', '0', '100', '4', 'Women in Judaism', 'Michael Berkowitz', 'http://jps.library.utoronto.ca/', 'function detectWeb(doc, url) { if (doc.evaluate(''//tr[td/a[2]]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext()) {