public void testFunctionalFromTarHierarchy() throws Exception {
log.debug3("in testFromTarHierarchy");
// load the tarballs
InputStream file_input = null;
try {
file_input = getResourceAsStream(realTARFile_A);
// UrlCacher uc = au.makeUrlCacher(TAR_A_BASE);
// uc.storeContent(file_input, tarHeader);
UrlCacher uc = tarAu.makeUrlCacher(new UrlData(file_input, tarHeader, TAR_A_BASE));
uc.storeContent();
IOUtil.safeClose(file_input);
file_input = getResourceAsStream(realTARFile_B);
// uc = au.makeUrlCacher(TAR_B_BASE);
// uc.storeContent(file_input, tarHeader);
uc = tarAu.makeUrlCacher(new UrlData(file_input, tarHeader, TAR_B_BASE));
uc.storeContent();
IOUtil.safeClose(file_input);
} catch (IOException e) {
// TODO Auto-generated catch block
e.printStackTrace();
} finally {
IOUtil.safeClose(file_input);
}
CachedUrlSet cus = tarAu.getAuCachedUrlSet();
for (CachedUrl cu : cus.getCuIterable()) {
log.debug3("AU - cu is: " + cu.getUrl());
cu.release();
}
// We need to start from the level of the ArticleMetadataExtractor
MyListEmitter emitter = new MyListEmitter();
ArticleMetadataExtractor amEx =
new ElsevierDeferredArticleMetadataExtractor(ArticleFiles.ROLE_ARTICLE_METADATA);
Iterator<ArticleFiles> it = tarAu.getArticleIterator(MetadataTarget.Any());
while (it.hasNext()) {
ArticleFiles af = it.next();
log.debug3("Metadata test - articlefiles " + af.toString());
// CachedUrl cu = af.getFullTextCu();
CachedUrl cu = af.getRoleCu(ArticleFiles.ROLE_ARTICLE_METADATA);
log.debug3("metadata cu is " + cu.getUrl());
// List<ArticleMetadata> mdlist = mle.extract(MetadataTarget.Any(), cu);
amEx.extract(MetadataTarget.Any(), af, emitter);
List<ArticleMetadata> returnList = emitter.getAmList();
assertNotNull(returnList);
log.debug3("size of returnList is " + returnList.size());
Iterator<ArticleMetadata> mdIt = returnList.iterator();
ArticleMetadata mdRecord = null;
while (mdIt.hasNext()) {
mdRecord = (ArticleMetadata) mdIt.next();
validateCompleteMetadataRecord(mdRecord);
}
}
}
public void testCreateArticleFiles() throws Exception {
PluginTestUtil.crawlSimAu(sau);
String pat0 = "001file[.]html";
String rep0 = "52/1/S1";
PluginTestUtil.copyAu(sau, au, ".*[.]html$", pat0, rep0);
String pat1 = "001file[.]pdf";
String rep1 = "52/1/S1.pdf";
PluginTestUtil.copyAu(sau, au, ".*[.]pdf$", pat1, rep1);
String pdfurl = "http://pediatrics.aappublications.org/cgi/reprint/52/1/S1.pdf";
String url = "http://pediatrics.aappublications.org/cgi/reprint/52/1/S1";
au.makeCachedUrl(url);
CachedUrl cu = au.makeCachedUrl(pdfurl);
assertNotNull(cu);
SubTreeArticleIterator artIter = createSubTreeIter();
assertNotNull(artIter);
ArticleFiles af = artIter.next();
assertNotNull(af);
System.out.println("article files::" + af);
assertEquals(url, af.getRoleCu(ArticleFiles.ROLE_FULL_TEXT_PDF_LANDING_PAGE).getUrl());
assertEquals(pdfurl, af.getRoleCu(ArticleFiles.ROLE_FULL_TEXT_PDF).getUrl());
}