Search results scraping for LexisNexis

This commit is contained in:
Simon Kornblith 2006-06-25 20:09:27 +00:00
parent a48ea7dabf
commit 6a627fad0a

View File

@ -1200,12 +1200,13 @@ for (var i = 0; i < elmts.length; i++) {
}
model.addStatement(uri, prefixRDF + "type", prefixDummy + "journalArticle", false);');
REPLACE INTO "scrapers" VALUES('b047a13c-fe5c-6604-c997-bef15e502b09', '2006-06-18 10:13:00', 'LexisNexis Scraper', 'Simon Kornblith', '^http://web\.lexis-nexis\.com/universe/document', NULL,
REPLACE INTO "scrapers" VALUES('b047a13c-fe5c-6604-c997-bef15e502b09', '2006-06-18 10:13:00', 'LexisNexis Scraper', 'Simon Kornblith', '^http://web\.lexis-nexis\.com/universe/(?:document|doclist)', NULL,
'var prefixRDF = ''http://www.w3.org/1999/02/22-rdf-syntax-ns#'';
var prefixDC = ''http://purl.org/dc/elements/1.1/'';
var prefixDCMI = ''http://purl.org/dc/dcmitype/'';
var prefixDummy = ''http://chnm.gmu.edu/firefox-scholar/'';
function scrape(doc) {
var uri = doc.location.href;
var citationDataDiv;
@ -1268,6 +1269,29 @@ if(m) {
for(i in authors) {
model.addStatement(uri, prefixDC + "creator", authors[i].replace(" *", ""), true);
}
}
}
var detailRe = new RegExp("^http://[^/]+/universe/document");
if(detailRe.test(doc.location.href)) {
scrape(doc);
} else {
var items = utilities.getItemArray(doc, doc, "^http://[^/]+/universe/document");
items = utilities.selectItems(items);
if(!items) {
return true;
}
var uris = new Array();
for(i in items) {
uris.push(i);
}
utilities.processDocuments(browser, null, uris, function(browser) { scrape(browser.contentDocument) },
function() { done(); }, function() {});
wait();
}');
REPLACE INTO "scrapers" VALUES('cf87eca8-041d-b954-795a-2d86348999d5', '2006-06-23 13:34:00', 'Aleph Scraper', 'Simon Kornblith', '^http://[^/]+/F(?:/[A-Z0-9\-]+(?:\?.*)?$|\?func=find)',