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

made arXiv translator work with eprintweb as well

Browse Source
This commit is contained in:
Simon Kornblith 2006-11-25 02:53:38 +00:00
parent 05b3cd8566
commit e201c3b580
1 changed files with 79 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

72
scrapers.sql
View File

@ -1,4 +1,4 @@
-- 107 -- 108
-- ***** BEGIN LICENSE BLOCK ***** -- ***** BEGIN LICENSE BLOCK *****
-- --
@ -22,7 +22,7 @@
-- Set the following timestamp to the most recent scraper update date -- Set the following timestamp to the most recent scraper update date
REPLACE INTO "version" VALUES ('repository', STRFTIME('%s', '2006-11-24 18:04:00')); REPLACE INTO "version" VALUES ('repository', STRFTIME('%s', '2006-11-24 18:50:00'));
REPLACE INTO "translators" VALUES ('96b9f483-c44d-5784-cdad-ce21b984fe01', '2006-11-21 22:30:00', 1, 100, 4, 'Amazon', 'Sean Takats', '^http://(?:www\.)amazon', REPLACE INTO "translators" VALUES ('96b9f483-c44d-5784-cdad-ce21b984fe01', '2006-11-21 22:30:00', 1, 100, 4, 'Amazon', 'Sean Takats', '^http://(?:www\.)amazon',
'function detectWeb(doc, url) { 'function detectWeb(doc, url) {
@ -4116,9 +4116,9 @@ function doWeb(doc, url) {
} }
}'); }');
REPLACE INTO "translators" VALUES ('ecddda2e-4fc6-4aea-9f17-ef3b56d7377a', '2006-11-24 16:32:00', 1, 100, 4, 'arXiv.org', 'Simon Kornblith', '^http://(?:www\.)?arxiv\.org/(?:find/\w|abs/[^/]+/[0-9]+)', REPLACE INTO "translators" VALUES ('ecddda2e-4fc6-4aea-9f17-ef3b56d7377a', '2006-11-24 18:50:00', 1, 100, 4, 'arXiv.org/eprintweb.org', 'Simon Kornblith', '^http://(?:www\.)?(?:arxiv\.org/(?:find/\w|abs/[^/]+/[0-9]+)|eprintweb.org/S/search)',
'function detectWeb(doc, url) { 'function detectWeb(doc, url) {
var searchRe = /http:\/\/(?:www\.)?arxiv\.org\/find/; var searchRe = /^http:\/\/(?:www\.)?(?:arxiv\.org\/find|eprintweb.org\/S\/search$)/;
if(searchRe.test(url)) { if(searchRe.test(url)) {
return "multiple"; return "multiple";
} else { } else {
@ -4160,21 +4160,70 @@ REPLACE INTO "translators" VALUES ('ecddda2e-4fc6-4aea-9f17-ef3b56d7377a', '2006
function doWeb(doc, url) { function doWeb(doc, url) {
var fetchIDs = new Array(); var fetchIDs = new Array();
var absRe = /http:\/\/(?:www\.)?arxiv\.org\/abs\/(.+)$/; var arxivAbsRe = /^http:\/\/(?:www\.)?arxiv\.org\/abs\/(.+)$/;
var m = absRe.exec(url); var eprintsAbsRe = /^http:\/\/(?:www\.)?eprintweb.org\/S\/search(.*)$/
var arxivM = arxivAbsRe.exec(url);
var eprintsM = eprintsAbsRe.exec(url);
if(m) {
// single
fetchIDs.push(m[1]);
} else{
// search
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;
} : null; } : null;
if(arxivM) {
// arxiv single
fetchIDs.push(arxivM[1]);
} else if(eprintsM && eprintsM[1]) {
// eprints single
if(url.indexOf("refs") != -1 || url.indexOf("cited") != -1) {
var id = doc.evaluate(''//td[@class="panel"]//td[@class="txt"]/b[2]'', doc, nsResolver,
Components.interfaces.nsIDOMXPathResult.ANY_TYPE, null).iterateNext().textContent;
} else {
var id = doc.evaluate(''//td[@class="panel"]//td[@class="txt"]/b'', doc, nsResolver,
Components.interfaces.nsIDOMXPathResult.ANY_TYPE, null).iterateNext().textContent;
id = id.replace("/ ", "/");
id = id.substr(0, id.indexOf(" "));
}
fetchIDs.push(id);
} else {
// search
var items = new Object(); var items = new Object();
if(eprintsM) {
// eprints search
// get ids and titles
var started = false;
var elmts = doc.evaluate(''//td[@class="panel"]/table/tbody/tr/td'', doc, nsResolver,
Components.interfaces.nsIDOMXPathResult.ANY_TYPE, null);
var elmt, title, id;
while(elmt = elmts.iterateNext()) {
if(!started && elmt.className == "lti") {
// wait until first title to process
started = true;
title = elmt.textContent;
} else if(started) {
if(elmt.className == "lti") {
// finish previous item
items[id] = title;
title = null;
// grab title
title = elmt.textContent;
} else if(elmt.className == "txt") {
// get id
var tags = elmt.getElementsByTagName("b");
id = tags[0].textContent;
}
}
}
if(title) {
items[id] = title;
}
} else {
// arxiv search
// get IDs and titles // get IDs and titles
var ids = doc.evaluate(''//div[@id="content"]/dl/dt'', doc, nsResolver, var ids = doc.evaluate(''//div[@id="content"]/dl/dt'', doc, nsResolver,
Components.interfaces.nsIDOMXPathResult.ANY_TYPE, null); Components.interfaces.nsIDOMXPathResult.ANY_TYPE, null);
@ -4190,6 +4239,7 @@ function doWeb(doc, url) {
items[realID] = realID + " - " + title.textContent; items[realID] = realID + " - " + title.textContent;
} }
}
items = Zotero.selectItems(items); items = Zotero.selectItems(items);