This commit is contained in:
Avram Lyon 2011-03-06 19:09:46 +00:00
parent 4eab4bbbae
commit e44e44b222

View File

@ -2,17 +2,17 @@
"translatorID": "c54d1932-73ce-dfd4-a943-109380e06574", "translatorID": "c54d1932-73ce-dfd4-a943-109380e06574",
"label": "Project MUSE", "label": "Project MUSE",
"creator": "Simon Kornblith, Avram Lyon", "creator": "Simon Kornblith, Avram Lyon",
"target":"https?://[^/]*muse\\.jhu\\.edu[^/]*/(?:journals/[^/]+/[^/]+/[^/]+\\.html|search/results)", "target": "^https?://[^/]*muse\\.jhu\\.edu[^/]*/(?:journals/[^/]+/[^/]+/[^/]+\\.html|search/results)",
"minVersion": "1.0.0b4.r1", "minVersion": "1.0.0b4.r1",
"maxVersion": "", "maxVersion": "",
"priority": 100, "priority": 100,
"inRepository": "1", "inRepository": "1",
"translatorType": 4, "translatorType": 4,
"lastUpdated":"2010-11-10 10:15:00" "lastUpdated": "2011-03-02 23:38:17"
} }
function detectWeb(doc, url) { function detectWeb(doc, url) {
var searchRe = new RegExp("(^https?://[^/]+/search/results|/toc/)"); var searchRe = new RegExp("(^https?://[^/]+/search/results|/search/save|/toc/)");
if(searchRe.test(url)) { if(searchRe.test(url)) {
return "multiple"; return "multiple";
} else { } else {
@ -26,7 +26,7 @@ function doWeb(doc, url) {
if (prefix == 'x') return namespace; else return null; if (prefix == 'x') return namespace; else return null;
} : null; } : null;
var searchRe = new RegExp("^https?://[^/]+/search/results"); var searchRe = new RegExp("^https?://[^/]+/search/results|/search/save");
if(detectWeb(doc, url) == "multiple") { if(detectWeb(doc, url) == "multiple") {
var items = new Array(); var items = new Array();
var attachments = new Array(); var attachments = new Array();
@ -34,15 +34,15 @@ function doWeb(doc, url) {
var htmlRe = /HTML/; var htmlRe = /HTML/;
if (searchRe.test(url)) { if (searchRe.test(url)) {
// Search results // Search results
var tableRows = doc.evaluate('//div[@id="advancedsearch"]/save_form/table//tr', var tableRows = doc.evaluate('//save_form//tr[@class="resultsrow"]',
doc, nsResolver, XPathResult.ANY_TYPE, null); doc, nsResolver, XPathResult.ANY_TYPE, null);
var tableRow; var tableRow;
// Go through table rows // Go through table rows
while(tableRow = tableRows.iterateNext()) { while(tableRow = tableRows.iterateNext()) {
var input = doc.evaluate('.//input[@name="aid"]', tableRow, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); var input = doc.evaluate('.//div[@class="links"]//a[last()]', tableRow, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var title = doc.evaluate('.//div[@class="title"]', tableRow, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); var title = doc.evaluate('.//div[@class="title"]', tableRow, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
if(input && input.value && title && title.textContent) { if(input && input.href && title && title.textContent) {
items[input.value] = title.textContent; items[input.href] = title.textContent;
} }
} }
} else if (url.match(/\/toc\//)) { } else if (url.match(/\/toc\//)) {
@ -52,7 +52,7 @@ function doWeb(doc, url) {
var result; var result;
while(result = results.iterateNext()) { while(result = results.iterateNext()) {
//Zotero.debug(result.textContent); //Zotero.debug(result.textContent);
var link = doc.evaluate('.//div[@class="links"]/p//a[3]', result, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); var link = doc.evaluate('.//div[@class="links"]/p//a[last()]', result, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var title = doc.evaluate('.//div[@class="title"]', result, nsResolver, XPathResult.ANY_TYPE, null).iterateNext(); var title = doc.evaluate('.//div[@class="title"]', result, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
//Zotero.debug(link.textContent); //Zotero.debug(link.textContent);
if(link && link.href && title && title.textContent) { if(link && link.href && title && title.textContent) {
@ -60,6 +60,21 @@ function doWeb(doc, url) {
//Zotero.debug(link.href); //Zotero.debug(link.href);
} }
} }
// Some journals have old-style TOCs for back issues
// Ex. http://muse.jhu.edu/journals/eighteenth-century_studies/toc/ecs33.4.html
if (items.length == 0) {
var articles = doc.evaluate('//ul', doc, nsResolver, XPathResult.ANY_TYPE, null);
var article;
while (article = articles.iterateNext()) {
var link = doc.evaluate('./li/a[contains(@href,".html")]', article, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var title = doc.evaluate('./li/i', article, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
//Zotero.debug(link.textContent);
if(link && link.href && title && title.textContent) {
items[link.href] = title.textContent;
//Zotero.debug(link.href);
}
}
}
} }
items = Zotero.selectItems(items); items = Zotero.selectItems(items);
if(!items) { if(!items) {
@ -67,9 +82,9 @@ function doWeb(doc, url) {
} }
var i; var i;
var urls = []; var urls = [];
for (i in items) urls.push(i); for (i in items) {urls.push(i);};
Zotero.Utilities.processDocuments(urls, scrapeOne); Zotero.Utilities.processDocuments(urls, scrapeOne, function() {Zotero.done();});
} else scrapeOne(doc); } else scrapeOne(doc);
Zotero.wait(); Zotero.wait();
} }
@ -77,6 +92,7 @@ function doWeb(doc, url) {
// Given an article page, get the RIS and open it // Given an article page, get the RIS and open it
function scrapeOne(doc) { function scrapeOne(doc) {
var url = doc.location.href; var url = doc.location.href;
//Zotero.debug("scrapeOne has "+doc.location.href)
var namespace = doc.documentElement.namespaceURI; var namespace = doc.documentElement.namespaceURI;
var nsResolver = namespace ? function(prefix) { var nsResolver = namespace ? function(prefix) {
if (prefix == 'x') return namespace; else return null; if (prefix == 'x') return namespace; else return null;