/**
* Method that creates a simulated Cached URL from the source code provided by the goodContent
* String. It then asserts that the metadata extracted, by using the
* MetaPressRisMetadataExtractorFactory, match the metadata in the source code.
*
* @throws Exception
*/
public void testExtractGoodRisContent() throws Exception {
String goodContent = createGoodRisContent();
log.debug3(goodContent);
List<ArticleMetadata> mdlist = setupContentForAU(bau1, RIS_URL, goodContent, false);
assertNotEmpty(mdlist);
ArticleMetadata md = mdlist.get(0);
assertNotNull(md);
assertEquals(goodVolume, md.get(MetadataField.FIELD_VOLUME));
assertEquals(goodIssue, md.get(MetadataField.FIELD_ISSUE));
assertEquals(goodStartPage, md.get(MetadataField.FIELD_START_PAGE));
assertEquals(goodEndPage, md.get(MetadataField.FIELD_END_PAGE));
assertEquals(goodIssn, md.get(MetadataField.FIELD_ISSN));
Iterator<String> actAuthIter = md.getList(MetadataField.FIELD_AUTHOR).iterator();
for (String expAuth : goodAuthors) {
assertEquals(expAuth, actAuthIter.next());
}
assertEquals(goodTitle, md.get(MetadataField.FIELD_ARTICLE_TITLE));
assertEquals(goodJournal, md.get(MetadataField.FIELD_PUBLICATION_TITLE));
assertEquals(goodDate, md.get(MetadataField.FIELD_DATE));
assertEquals(goodPublisher, md.get(MetadataField.FIELD_PUBLISHER));
assertEquals(goodDOI, md.get(MetadataField.FIELD_DOI));
// This shouldn't get set. It will default later to fuill_text_cu
assertNotEquals(doiURL, md.get(MetadataField.FIELD_ACCESS_URL));
}
private void assertEqualMessages(V3LcapMessage a, V3LcapMessage b) throws Exception {
assertTrue(a.getOriginatorId() == b.getOriginatorId());
assertEquals(a.getOpcode(), b.getOpcode());
assertEquals(a.getTargetUrl(), b.getTargetUrl());
assertEquals(a.getArchivalId(), b.getArchivalId());
assertEquals(a.getProtocolVersion(), b.getProtocolVersion());
assertEquals(a.getPollerNonce(), b.getPollerNonce());
assertEquals(a.getVoterNonce(), b.getVoterNonce());
assertEquals(a.getVoterNonce2(), b.getVoterNonce2());
assertEquals(a.getPluginVersion(), b.getPluginVersion());
assertEquals(a.getHashAlgorithm(), b.getHashAlgorithm());
assertEquals(a.isVoteComplete(), b.isVoteComplete());
assertEquals(a.getRepairDataLength(), b.getRepairDataLength());
assertEquals(a.getLastVoteBlockURL(), b.getLastVoteBlockURL());
assertIsomorphic(a.getNominees(), b.getNominees());
List aBlocks = new ArrayList();
List bBlocks = new ArrayList();
for (VoteBlocksIterator iter = a.getVoteBlockIterator(); iter.hasNext(); ) {
aBlocks.add(iter.next());
}
for (VoteBlocksIterator iter = b.getVoteBlockIterator(); iter.hasNext(); ) {
bBlocks.add(iter.next());
}
assertTrue(aBlocks.equals(bBlocks));
// TODO: Figure out how to test time.
}
public void testSimpleMainXML() throws Exception {
log.debug3("testSimpleMainXML");
String xml_url = TAR_A_BASE + SUBDIR + "01420615/v64sC/S0142061514004608/main.xml";
List<ArticleMetadata> mdList =
extractFromContent(xml_url, "text/xml", simpleMain, nocheck_mle, null);
assertEquals(1, mdList.size());
validateSingleMainMetadataRecord(mdList.get(0), "10.1016/j.jidx.2014.07.028", "article");
}
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);
}
}
}
List<String> auUrls(ArchivalUnit au) {
List<String> res = new ArrayList<String>();
for (Iterator iter = au.getAuCachedUrlSet().contentHashIterator(); iter.hasNext(); ) {
CachedUrlSetNode cusn = (CachedUrlSetNode) iter.next();
if (cusn.hasContent()) {
res.add(cusn.getUrl());
}
}
return res;
}
void assertPubKs(File file, String pass, List<String> hosts) throws Exception {
KeyStore ks = loadKeyStore("jceks", file, pass);
List aliases = ListUtil.fromIterator(new EnumerationIterator(ks.aliases()));
assertEquals(hosts.size(), aliases.size());
for (String host : hosts) {
String alias = host + ".crt";
Certificate cert = ks.getCertificate(alias);
assertNotNull(cert);
assertEquals("X.509", cert.getType());
}
}
public void testDOIExtraction() throws Exception {
List<ArticleMetadata> mdlist =
setupContentForAU(bau1, ABS_URL, goodHtmlContentNoDOIorPublisher, true);
assertNotEmpty(mdlist);
ArticleMetadata md = mdlist.get(0);
assertNotNull(md);
// gets pulled from the URL if not set in the metadata
assertEquals("10.1175/2010WCAS1063.1", md.get(MetadataField.FIELD_DOI));
// gets set manually if not in the metadata
// first it would try the TDB
assertEquals(goodPublisher, md.get(MetadataField.FIELD_PUBLISHER));
}
void assertPrivateKs(File file, String pass, String alias) throws Exception {
KeyStore ks = loadKeyStore("jceks", file, alias);
List aliases = ListUtil.fromIterator(new EnumerationIterator(ks.aliases()));
assertEquals(2, aliases.size());
Certificate cert = ks.getCertificate(alias + ".crt");
assertNotNull(cert);
assertEquals("X.509", cert.getType());
assertTrue(ks.isKeyEntry(alias + ".key"));
assertTrue(ks.isCertificateEntry(alias + ".crt"));
Key key = ks.getKey(alias + ".key", pass.toCharArray());
assertNotNull(key);
assertEquals("RSA", key.getAlgorithm());
}
public List findExistingRepositoriesFor(String auid) {
List res = null;
for (Iterator iter = getRepositoryList().iterator(); iter.hasNext(); ) {
String repoName = (String) iter.next();
String path = LockssRepositoryImpl.getLocalRepositoryPath(repoName);
if (LockssRepositoryImpl.doesAuDirExist(auid, path)) {
if (res == null) {
res = new ArrayList();
}
res.add(repoName);
}
}
return res == null ? Collections.EMPTY_LIST : res;
}
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));
}
public void testSimpleDatasetXML() throws Exception {
log.debug3("testSimpleDatasetXML");
String file_input = StringUtil.fromInputStream(getResourceAsStream(testDatasetFile));
String xml_url = TAR_A_BASE + SUBDIR + "dataset.xml";
List<ArticleMetadata> mdList =
extractFromContent(xml_url, "text/xml", file_input, nocheck_mle, null);
assertEquals(6, mdList.size());
Iterator<ArticleMetadata> mdIt = mdList.iterator();
ArticleMetadata mdRecord = null;
while (mdIt.hasNext()) {
mdRecord = (ArticleMetadata) mdIt.next();
validateDatasetMetadataRecord(mdRecord);
}
}
public void emitMetadata(ArticleFiles af, ArticleMetadata md) {
if (log.isDebug3()) log.debug3("emit(" + af + ", " + md + ")");
if (md != null) {
log.debug3("add " + md + " to amlist");
amlst.add(md);
}
;
}
public void testBaseUrlPath() throws Exception {
sau1 = setupSimAu(simAuConfig(tempDirPath));
createContent(sau1);
crawlContent(sau1);
CachedUrlSet cus1 = sau1.getAuCachedUrlSet();
tempDirPath2 = getTempDir().getAbsolutePath() + File.separator;
Configuration config2 = simAuConfig(tempDirPath2);
config2.put("base_url", "http://anotherhost.org/some/path/");
SimulatedArchivalUnit sau2 = setupSimAu(config2);
createContent(sau2);
crawlContent(sau2);
CachedUrlSet cus2 = sau1.getAuCachedUrlSet();
List urls1 = auUrls(sau1);
List urls2 = auUrls(sau2);
Pattern pat1 = Pattern.compile("http://www\\.example\\.com(/.*)$");
Pattern pat2 = Pattern.compile("http://anotherhost\\.org/some/path(/.*)$");
List<String> l1 = auUrls(sau1);
List<String> l2 = auUrls(sau2);
assertEquals(l1.size(), l2.size());
for (int ix = 0; ix < l1.size(); ix++) {
Matcher m1 = pat1.matcher(l1.get(ix));
assertTrue(m1.matches());
Matcher m2 = pat2.matcher(l2.get(ix));
assertTrue(m2.matches());
assertEquals(m1.group(1), m2.group(1));
}
}
/**
* Checks the consistency of the node, and continues with its children if it's consistent.
*
* @param node RepositoryNodeImpl the node to check
*/
private void recurseConsistencyCheck(RepositoryNodeImpl node) {
logger.debug2("Checking node '" + node.getNodeUrl() + "'...");
// check consistency at each node
// correct/deactivate as necessary
// 'checkNodeConsistency()' will repair if possible
if (node.checkNodeConsistency()) {
logger.debug3("Node consistent; recursing on children...");
List children = node.getNodeList(null, false);
Iterator iter = children.iterator();
while (iter.hasNext()) {
RepositoryNodeImpl child = (RepositoryNodeImpl) iter.next();
recurseConsistencyCheck(child);
}
} else {
logger.debug3("Node inconsistent; deactivating...");
deactivateInconsistentNode(node);
}
}
/**
* Method that creates a simulated Cached URL from the source code provided by the goodContent
* String. It then asserts that the metadata extracted, by using the
* MetaPressRisMetadataExtractorFactory, match the metadata in the source code.
*
* @throws Exception
*/
public void testExtractAlternateRisContent() throws Exception {
String goodContent = createAlternateRisContent();
log.debug3(goodContent);
List<ArticleMetadata> mdlist = setupContentForAU(bau1, RIS_URL, goodContent, false);
assertNotEmpty(mdlist);
ArticleMetadata md = mdlist.get(0);
assertNotNull(md);
Iterator<String> actAuthIter = md.getList(MetadataField.FIELD_AUTHOR).iterator();
for (String expAuth : goodAuthors) {
assertEquals(expAuth, actAuthIter.next());
}
assertEquals(goodTitle, md.get(MetadataField.FIELD_ARTICLE_TITLE));
assertEquals(goodJournal, md.get(MetadataField.FIELD_PUBLICATION_TITLE));
assertEquals(goodDate, md.get(MetadataField.FIELD_DATE));
assertEquals(goodPublisher, md.get(MetadataField.FIELD_PUBLISHER));
}
public void testRequestMessageCreation() throws Exception {
V3LcapMessage reqMsg =
new V3LcapMessage(
"ArchivalID_2",
"key",
"Plug42",
m_testBytes,
m_testBytes,
V3LcapMessage.MSG_REPAIR_REQ,
987654321,
m_testID,
tempDir,
theDaemon);
reqMsg.setTargetUrl("http://foo.com/");
for (Iterator ix = m_testVoteBlocks.iterator(); ix.hasNext(); ) {
reqMsg.addVoteBlock((VoteBlock) ix.next());
}
assertEquals(3, reqMsg.getProtocolVersion());
assertEquals("Plug42", reqMsg.getPluginVersion());
assertTrue(m_testID == reqMsg.getOriginatorId());
assertEquals(V3LcapMessage.MSG_REPAIR_REQ, reqMsg.getOpcode());
assertEquals("ArchivalID_2", reqMsg.getArchivalId());
assertEquals("http://foo.com/", reqMsg.getTargetUrl());
assertEquals(m_testBytes, reqMsg.getPollerNonce());
assertEquals(m_testBytes, reqMsg.getVoterNonce());
assertEquals(null, reqMsg.getVoterNonce2());
List aBlocks = new ArrayList();
List bBlocks = new ArrayList();
for (VoteBlocksIterator iter = m_testMsg.getVoteBlockIterator(); iter.hasNext(); ) {
aBlocks.add(iter.next());
}
for (VoteBlocksIterator iter = reqMsg.getVoteBlockIterator(); iter.hasNext(); ) {
bBlocks.add(iter.next());
}
assertEquals(aBlocks, bBlocks);
// Actual size of test vote blocks is unpredictable
assertTrue(reqMsg.getEstimatedEncodedLength() > V3LcapMessage.EST_ENCODED_HEADER_LENGTH);
}
protected void recordExportFile(File file) {
if (fileSet.add(file)) {
fileList.add(file);
}
}
protected void recordError(String msg) {
log.error(msg);
errors.add(msg);
}
protected void recordError(String msg, Throwable t) {
log.error(msg, t);
errors.add(msg + ": " + t.toString());
}
public void setConfig(
Configuration config, Configuration oldConfig, Configuration.Differences changedKeys) {
// Build list of repositories from list of disk (fs) paths). Needs to
// be generalized if ever another repository implementation.
if (changedKeys.contains(ConfigManager.PARAM_PLATFORM_DISK_SPACE_LIST)) {
List lst = new ArrayList();
String dspace = config.get(ConfigManager.PARAM_PLATFORM_DISK_SPACE_LIST, "");
List paths = StringUtil.breakAt(dspace, ';');
if (paths != null) {
for (Iterator iter = paths.iterator(); iter.hasNext(); ) {
lst.add("local:" + (String) iter.next());
}
}
repoList = lst;
}
if (changedKeys.contains(PARAM_MAX_PER_AU_CACHE_SIZE)) {
paramNodeCacheSize =
config.getInt(PARAM_MAX_PER_AU_CACHE_SIZE, DEFAULT_MAX_PER_AU_CACHE_SIZE);
for (Iterator iter = getDaemon().getAllLockssRepositories().iterator(); iter.hasNext(); ) {
LockssRepository repo = (LockssRepository) iter.next();
if (repo instanceof LockssRepositoryImpl) {
LockssRepositoryImpl repoImpl = (LockssRepositoryImpl) repo;
repoImpl.setNodeCacheSize(paramNodeCacheSize);
}
}
}
if (changedKeys.contains(PARAM_MAX_SUSPECT_VERSIONS_CACHE_SIZE)) {
paramSuspectVersionsCacheSize =
config.getInt(
PARAM_MAX_SUSPECT_VERSIONS_CACHE_SIZE, DEFAULT_MAX_SUSPECT_VERSIONS_CACHE_SIZE);
suspectVersionsCache.setMaxSize(paramSuspectVersionsCacheSize);
}
if (changedKeys.contains(GLOBAL_CACHE_PREFIX)) {
paramIsGlobalNodeCache =
config.getBoolean(PARAM_GLOBAL_CACHE_ENABLED, DEFAULT_GLOBAL_CACHE_ENABLED);
if (paramIsGlobalNodeCache) {
paramGlobalNodeCacheSize =
config.getInt(PARAM_MAX_GLOBAL_CACHE_SIZE, DEFAULT_MAX_GLOBAL_CACHE_SIZE);
log.debug("global node cache size: " + paramGlobalNodeCacheSize);
globalNodeCache.setMaxSize(paramGlobalNodeCacheSize);
}
}
if (changedKeys.contains(DISK_PREFIX)) {
int minMB = config.getInt(PARAM_DISK_WARN_FRRE_MB, DEFAULT_DISK_WARN_FRRE_MB);
double minPer =
config.getPercentage(PARAM_DISK_WARN_FRRE_PERCENT, DEFAULT_DISK_WARN_FRRE_PERCENT);
paramDFWarn = PlatformUtil.DF.makeThreshold(minMB, minPer);
minMB = config.getInt(PARAM_DISK_FULL_FRRE_MB, DEFAULT_DISK_FULL_FRRE_MB);
minPer = config.getPercentage(PARAM_DISK_FULL_FRRE_PERCENT, DEFAULT_DISK_FULL_FRRE_PERCENT);
paramDFFull = PlatformUtil.DF.makeThreshold(minMB, minPer);
}
if (changedKeys.contains(PARAM_SIZE_CALC_MAX_LOAD)) {
sizeCalcMaxLoad = config.getPercentage(PARAM_SIZE_CALC_MAX_LOAD, DEFAULT_SIZE_CALC_MAX_LOAD);
}
if (changedKeys.contains(PREFIX)) {
maxUnusedDirSearch =
config.getInt(PARAM_MAX_UNUSED_DIR_SEARCH, DEFAULT_MAX_UNUSED_DIR_SEARCH);
isStatefulUnusedDirSearch =
config.getBoolean(
PARAM_IS_STATEFUL_UNUSED_DIR_SEARCH, DEFAULT_IS_STATEFUL_UNUSED_DIR_SEARCH);
enableLongComponents =
config.getBoolean(PARAM_ENABLE_LONG_COMPONENTS, DEFAULT_ENABLE_LONG_COMPONENTS);
enableLongComponentsCompatibility =
config.getBoolean(
PARAM_ENABLE_LONG_COMPONENTS_COMPATIBILITY,
DEFAULT_ENABLE_LONG_COMPONENTS_COMPATIBILITY);
maxComponentLength = config.getInt(PARAM_MAX_COMPONENT_LENGTH, DEFAULT_MAX_COMPONENT_LENGTH);
checkUnnormalized =
(CheckUnnormalizedMode)
config.getEnum(
CheckUnnormalizedMode.class,
PARAM_CHECK_UNNORMALIZED,
DEFAULT_CHECK_UNNORMALIZED);
}
}
