public void testExtractGoodHtmlContent() throws Exception {
List<ArticleMetadata> mdlist = setupContentForAU(bau1, ABS_URL, goodHtmlContent, true);
assertNotEmpty(mdlist);
ArticleMetadata md = mdlist.get(0);
assertNotNull(md);
assertEquals(goodPublisher, md.get(MetadataField.FIELD_PUBLISHER));
assertEquals(goodTitle, md.get(MetadataField.DC_FIELD_TITLE));
assertEquals(goodDate, md.get(MetadataField.FIELD_DATE));
assertEquals(goodFormat, md.get(MetadataField.DC_FIELD_FORMAT));
assertEquals(goodType, md.get(MetadataField.DC_FIELD_TYPE));
assertEquals(Arrays.asList(goodAuthors), md.getList(MetadataField.FIELD_AUTHOR));
assertEquals(goodAuthors[0], md.get(MetadataField.DC_FIELD_CREATOR));
}
private void checkHashSet(
SimulatedArchivalUnit sau, boolean namesOnly, boolean filter, byte[] expected)
throws Exception {
enableFilter(sau, filter);
CachedUrlSet set = sau.getAuCachedUrlSet();
byte[] hash = getHash(set, namesOnly);
assertEquals(expected, hash);
String parent = sau.getUrlRoot() + "/branch1";
CachedUrlSetSpec spec = new RangeCachedUrlSetSpec(parent);
set = sau.makeCachedUrlSet(spec);
byte[] hash2 = getHash(set, namesOnly);
assertFalse(Arrays.equals(hash, hash2));
}