/*
* Test the functionality of the MetadataUtilities
*
*/
public void testNormalizeTitleValue() throws Exception {
assertEquals(
BaseAtyponMetadataUtil.normalizeTitle("The title goes here"),
BaseAtyponMetadataUtil.normalizeTitle("Title Goes Here"));
assertEquals(
BaseAtyponMetadataUtil.normalizeTitle("Title with random spaces"),
BaseAtyponMetadataUtil.normalizeTitle("Title with random spaces"));
assertEquals(
BaseAtyponMetadataUtil.normalizeTitle("Normalize -- hyphen"),
BaseAtyponMetadataUtil.normalizeTitle("normalize \u2013\u2013 hyphen"));
assertEquals(
BaseAtyponMetadataUtil.normalizeTitle("Title and title"),
BaseAtyponMetadataUtil.normalizeTitle("Title & title"));
assertEquals(
BaseAtyponMetadataUtil.normalizeTitle(" leading spaces"),
BaseAtyponMetadataUtil.normalizeTitle("leading spaces"));
// now checking the fall-back last ditch attempt
assertEquals(
BaseAtyponMetadataUtil.generateRawTitle("leading spaces:colon?"),
BaseAtyponMetadataUtil.generateRawTitle("leadingspacescolon"));
assertEquals(
BaseAtyponMetadataUtil.generateRawTitle("relapsing-remitting"),
BaseAtyponMetadataUtil.generateRawTitle("relapsing?remitting"));
assertEquals(
BaseAtyponMetadataUtil.generateRawTitle("foo\"blah"),
BaseAtyponMetadataUtil.generateRawTitle("foo-blah"));
}
// override this to do some additional attempts to get valid data before emitting
@Override
public void extract(MetadataTarget target, CachedUrl cu, FileMetadataExtractor.Emitter emitter)
throws IOException, PluginException {
ArticleMetadata am = extract(target, cu);
/*
* if, due to overcrawl, we got to a page that didn't have anything
* valid, eg "this page not found" html page
* don't emit empty metadata (because defaults would get put in
* Must do this after cooking, because it checks size of cooked info
*/
if (am.isEmpty()) {
return;
}
/*
* RIS data can be variable. We don't have any way to add priority to
* the cooking of data, so fallback to alternate values manually
*
* There are differences between books and journals, so fork for titles
* and metadata check
*/
if (am.get(MetadataField.FIELD_DATE) == null) {
if (am.getRaw("Y1") != null) { // if DA wasn't there, use Y1
am.put(MetadataField.FIELD_DATE, am.getRaw("Y1"));
}
}
/*
* Determine if this is a book item or a journal item.
* set the appropriate article type once the daemon passes along the TY
*/
String ris_type = am.getRaw("TY");
if (ris_type == null) {
// pre 1.69, do an alternate check because TY wasn't passed through
ris_type = "JOUR"; // set a default
if (am.get(MetadataField.FIELD_ISBN) != null) {
// it is a bad value, but it was recognized as an isbn because of TY type
ris_type = "BOOK"; // it could be a chapter but until TY is passed through...
}
}
// Modify or try alternate RIS tag values based after cooking
postCookProcess(cu, am, ris_type);
// Only emit if this item is likely to be from this AU
// protect against counting overcrawled articles by checking against
// values from the TDB file - differentiate between book items and journal itesm
ArchivalUnit au = cu.getArchivalUnit();
if (ris_type.contains("BOOK") || ris_type.contains("CHAP")) {
if (!BaseAtyponMetadataUtil.metadataMatchesBookTdb(au, am)) {
return;
}
} else {
// JOURNAL default is to assume it's a journal for backwards compatibility
if (!BaseAtyponMetadataUtil.metadataMatchesTdb(au, am)) {
return;
}
}
/*
* Fill in DOI, publisher, other information available from
* the URL or TDB
* CORRECT the access.url if it is not in the AU
*/
BaseAtyponMetadataUtil.completeMetadata(cu, am);
emitter.emitMetadata(cu, am);
}