mirrors/zotero
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Search results scraping for SIRSI (old versions)

Browse source
This commit is contained in:
Simon Kornblith 2006-06-24 14:35:05 +00:00
parent 97940c7470
commit 06cf9e7853
1 changed files with 90 additions and 42 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

70
scrapers.sql
View file

@ -1,7 +1,7 @@
-- 14
-- Set the following timestamp to the most recent scraper update date
REPLACE INTO "version" VALUES ('repository', STRFTIME('%s', '2006-06-23 16:53:00'));
REPLACE INTO "version" VALUES ('repository', STRFTIME('%s', '2006-06-24 10:34:00'));
REPLACE INTO "scrapers" VALUES('96b9f483-c44d-5784-cdad-ce21b984fe01', '2006-06-22 22:58:00', 'Amazon.com Scraper', 'Simon Kornblith', '^http://www\.amazon\.com/(?:gp/(?:product|search)/|exec/obidos/search-handle-url/)', NULL, 'var prefixRDF = ''http://www.w3.org/1999/02/22-rdf-syntax-ns#'';
var prefixDC = ''http://purl.org/dc/elements/1.1/'';
@ -1266,6 +1266,7 @@ if(marcs.length == 1) {
for(var j=0; j<links.length; j++) {
if(tagRegexp.test(links[j].href)) {
url = links[j].href;
break;
}
}
if(url) {
@ -1461,7 +1462,7 @@ utilities.processDocuments(browser, null, uris, function(newBrowser) {
wait();');
REPLACE INTO "scrapers" VALUES('5287d20c-8a13-6004-4dcb-5bb2b66a9cc9', '2006-06-18 11:19:00', 'SIRSI -2003 Scraper', 'Simon Kornblith', '/uhtbin/cgisirsi',
REPLACE INTO "scrapers" VALUES('5287d20c-8a13-6004-4dcb-5bb2b66a9cc9', '2006-06-24 10:34:00', 'SIRSI -2003 Scraper', 'Simon Kornblith', '/uhtbin/cgisirsi',
'var namespace = doc.documentElement.namespaceURI;
var nsResolver = namespace ? function(prefix) {
if (prefix == ''x'') return namespace; else return null;
@ -1469,10 +1470,15 @@ var nsResolver = namespace ? function(prefix) {
var elmts = utilities.gatherElementsOnXPath(doc, doc, ''/html/body/form/p/text()[1]'', nsResolver);
for(i in elmts) {
if(elmts[i].nodeValue == "\n\nViewing record\n") {
if(utilities.superCleanString(elmts[i].nodeValue) == "Viewing record") {
return true;
}
}
var xpath = ''//form[@name="hitlist"]/table/tbody/tr'';
var elmts = utilities.gatherElementsOnXPath(doc, doc, xpath, nsResolver);
if(elmts.length) {
return true;
}
return false;',
'var prefixRDF = ''http://www.w3.org/1999/02/22-rdf-syntax-ns#'';
var prefixDC = ''http://purl.org/dc/elements/1.1/'';
@ -1485,27 +1491,68 @@ var nsResolver = namespace ? function(prefix) {
} : null;
var uri = doc.location.href;
var recNumbers = new Array();
var xpath = ''//form[@name="hitlist"]/table/tbody/tr'';
var elmts = utilities.gatherElementsOnXPath(doc, doc, xpath, nsResolver);
if(elmts.length) { // Search results page
var uriRegexp = /^http:\/\/[^\/]+/;
var m = uriRegexp.exec(uri);
var postAction = doc.forms.namedItem("hitlist").getAttribute("action");
var newUri = m[0]+postAction.substr(0, postAction.length-1)+"40"
var titleRe = /<br>\s*(.*[^\s])\s*<br>/i;
var items = new Array();
for(var i=0; i<elmts.length; i++) {
var links = utilities.gatherElementsOnXPath(doc, elmts[i], ''.//a'', nsResolver);
// Collect title
var myTd = utilities.getNode(doc, elmts[i], "./td[2]", nsResolver);
var m = titleRe.exec(myTd.innerHTML);
var title = m[1];
items[i] = title;
}
items = utilities.selectItems(items);
if(!items) {
return true;
}
for(i in items) {
recNumbers.push(parseInt(i)+1);
}
} else { // Normal page
var uriRegexp = /^(.*)(\/[0-9]+)$/;
var m = uriRegexp.exec(uri);
var newUri = m[1]+"/40";
var newUri = m[1]+"/40"
var elmts = utilities.gatherElementsOnXPath(doc, doc, ''/html/body/form/p'', nsResolver);
for(i in elmts) {
var elmt = elmts[i];
var initialText = utilities.getNode(doc, elmt, ''./text()[1]'', nsResolver);
if(initialText.nodeValue == "\n\nViewing record\n") {
var recNumber = utilities.getNode(doc, elmt, ''./b[1]/text()[1]'', nsResolver).nodeValue;
if(initialText && initialText.nodeValue && utilities.superCleanString(initialText.nodeValue) == "Viewing record") {
recNumbers.push(utilities.getNode(doc, elmt, ''./b[1]/text()[1]'', nsResolver).nodeValue);
break;
}
}
}
utilities.HTTPUtilities.doPost(newUri, ''marks=''+recNumber+''&shadow=NO&format=FLAT+ASCII&sort=TITLE&vopt_elst=ALL&library=ALL&display_rule=ASCENDING&duedate_code=l&holdcount_code=t&DOWNLOAD_x=22&DOWNLOAD_y=12&address=&form_type='', null, function(text) {
utilities.HTTPUtilities.doGet(newUri+''?marks=''+recNumbers.join(",")+''&shadow=NO&format=FLAT+ASCII&sort=TITLE&vopt_elst=ALL&library=ALL&display_rule=ASCENDING&duedate_code=l&holdcount_code=t&DOWNLOAD_x=22&DOWNLOAD_y=12&address=&form_type='', null, function(text) {
utilities.debugPrint(text);
var texts = text.split("<PRE>");
texts = texts[1].split("</PRE>");
text = texts[0];
var lines = text.split("\n");
var documents = text.split("*** DOCUMENT BOUNDARY ***");
for(var j=1; j<documents.length; j++) {
var uri = newUri+"?marks="+recNumbers[j]+"&shadow=NO&format=FLAT+ASCII&sort=TITLE&vopt_elst=ALL&library=ALL&display_rule=ASCENDING&duedate_code=l&holdcount_code=t&DOWNLOAD_x=22&DOWNLOAD_y=12&address=&form_type=";
var lines = documents[j].split("\n");
var record = new MARC_Record();
var tag, ind1, ind2, content;
for(var i=0; i<lines.length; i++) {
var line = lines[i];
@ -1532,10 +1579,11 @@ utilities.HTTPUtilities.doPost(newUri, ''marks=''+recNumber+''&shadow=NO&format=
content = line.substring(6);
}
}
utilities.importMARCRecord(record, uri, model);
}
done();
})
});
wait();');
REPLACE INTO "scrapers" VALUES('0f9fc2fc-306e-5204-1117-25bca009dffc', '2006-06-18 11:19:00', 'TLC/YouSeeMore Scraper', 'Simon Kornblith', 'TLCScripts/interpac\.dll\?.*LabelDisplay.*RecordNumber=[0-9]', NULL,