suzanne.soy/zotero
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

-Adds Inter-Research translator.

Browse Source
This commit is contained in:
Michael Berkowitz 2008-04-22 17:24:39 +00:00
parent f0dc5ecba6
commit 9763eb428a
1 changed files with 95 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

96
scrapers.sql
View File

@ -22,7 +22,7 @@
-- Set the following timestamp to the most recent scraper update date
REPLACE INTO version VALUES ('repository', STRFTIME('%s', '2008-04-22 16:00:00'));
REPLACE INTO version VALUES ('repository', STRFTIME('%s', '2008-04-22 17:30:00'));
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) {
@ -1087,6 +1087,100 @@ REPLACE INTO translators VALUES ('88915634-1af6-c134-0171-56fd198235ed', '1.0.0b
Zotero.wait();
}');
REPLACE INTO translators VALUES ('0eeb2ac0-fbaf-4994-b98f-203d273eb9fa', '1.0.0b4.r5', '', '2008-04-22 17:30:00', '0', '100', '4', 'Inter-Research Science Center', 'Michael Berkowitz', 'http://www.int-res.com/',
'function detectWeb(doc, url) {
if (doc.evaluate(''//div[@class="journal-index"]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext() ||
doc.evaluate(''//div[@class="tx-indexedsearch-res"]//tr[1]/td[2]//a'', doc, null, XPathResult.ANY_TYPE, null).iterateNext()) {
return "multiple";
} else if (doc.evaluate(''//a[@class="citeexport"]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext()) {
return "journalArticle";
}
}',
'var journals = {
meps:["Marine Ecology Progress Series", "Mar Ecol Prog Ser"],
ab:["Aquatic Biology", "Aquat Biol"],
ame:["Aquatic Microbial Ecology", "Aquat Microb Ecol"],
dao:["Diseases of Aquatic Organisms", "Dis Aquat Org"],
cr:["Climate Research", "Clim Res"],
esr:["Endangered Species Research", "Endang Species Res"]
};
function doWeb(doc, url) {
var arts = new Array();
if (detectWeb(doc, url) == "multiple") {
var items = new Object();
if (doc.evaluate(''//div[@class="tx-indexedsearch-res"]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext()) {
var titlesx = doc.evaluate(''//div[@class="tx-indexedsearch-res"]//tr[2]/td[2]'', doc, null, XPathResult.ANY_TYPE, null);
var linksx = doc.evaluate(''//div[@class="tx-indexedsearch-res"]//tr[1]/td[2]//a'', doc, null, XPathResult.ANY_TYPE, null);
var title;
var link;
while ((title = titlesx.iterateNext()) && (link = linksx.iterateNext())) {
items[link.href] = Zotero.Utilities.trimInternal(title.textContent).match(/doi:\s+[^\s]+\s+(.*)$/)[1];
}
} else {
var stuff = doc.evaluate(''//div[@class="journal-index"]/*[a[contains(text(), "pdf format")]]'', doc, null, XPathResult.ANY_TYPE, null);
var thing;
var titles = "";
while (thing = stuff.iterateNext()) {
titles += thing.textContent;
}
titles = titles.split(/\n/);
Zotero.debug(titles);
var names = new Array();
for (var i = 0; i < titles.length; i++) {
if (((i-1)%2 == 0) && (titles[i].match(/\w+/))) {
names.push(titles[i]);
}
}
Zotero.debug(names);
var links = doc.evaluate(''//div[@class="journal-index"]/*[a[contains(text(), "pdf format")]]/a[1]'', doc, null, XPathResult.ANY_TYPE, null);
var link;
while (link = links.iterateNext()) {
items[link.href] = names.shift();
}
}
items = Zotero.selectItems(items);
for (var i in items) {
arts.push(i);
}
} else {
arts = [url];
}
Zotero.Utilities.processDocuments(arts, function(doc) {
var item = new Zotero.Item("journalArticle");
item.title = Zotero.Utilities.trimInternal(doc.evaluate(''//div[@class="bb"]/h2'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().textContent);
item.url = doc.location.href;
var voliss = item.url.match(/v(\d+)\/n(\d+)\/p([^/]+)\//);
item.volume = voliss[1];
item.issue = voliss[2];
item.pages = voliss[3];
var jour = item.url.match(/abstracts\/([^/]+)\//)[1];
item.publicationTitle = journals[jour][0];
item.journalAbbreviation = journals[jour][1];
item.abstractNote = Zotero.Utilities.trimInternal(doc.evaluate(''//p[@class="abstract_block"]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().textContent);
var authors = Zotero.Utilities.trimInternal(doc.evaluate(''//div[@class="bb"]/h3'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().textContent).split(/,\s+/);
for each (var aut in authors) {
aut = aut.replace(/[^\w^\s^\.]/g, "").replace(/\d/g, "");
item.creators.push(Zotero.Utilities.cleanAuthor(aut, "author"));
}
item.date = doc.evaluate(''//div[@class="abs-footer"]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().textContent.match(/date:\s+(.*)P/)[1];
item.DOI = Zotero.Utilities.trimInternal(doc.evaluate(''//h1[@class="csc-firstHeader"]/span'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().textContent).match(/doi:\s+(.*)/)[1];
var tags = doc.evaluate(''//div[@class="box"]/p/a'', doc, null, XPathResult.ANY_TYPE, null);
var tag;
while (tag = tags.iterateNext()) {
item.tags.push(tag.textContent);
}
var pdfurl = doc.evaluate(''//a[contains(@href, ".pdf")]'', doc, null, XPathResult.ANY_TYPE, null).iterateNext().href;
item.attachments = [
{url:item.url, title:item.publicationTitle + " Snapshot", mimeType:"text/html"},
{url:pdfurl, title:item.publicationTitle + " Full Text PDF", mimeType:"application/pdf"}
];
item.complete();
}, function() {Zotero.done;});
Zotero.wait();
}');
REPLACE INTO translators VALUES ('70295509-4c29-460f-81a3-16d4ddbb93f6', '1.0.0b4.r5', '', '2008-04-21 21:30:00', '0', '100', '4', 'GSA Journals Online', 'Michael Berkowitz', 'http://www.gsajournals.org/',
'function detectWeb(doc, url) {
if (url.indexOf("request=search") != -1 || url.indexOf("request=get-toc") != -1) {