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));
}
}
// Break the line at commas, return a map of the resulting strings
// broken at equals sign. (<i>Ie</i>, name value pairs.)
Map getRow(String line) {
Map map = new HashMap();
for (Iterator iter = StringUtil.breakAt(line, ',').iterator(); iter.hasNext(); ) {
String item = (String) iter.next();
List pair = StringUtil.breakAt(item, '=');
map.put(pair.get(0), pair.get(1));
}
return map;
}
// test completion & callback
public void testDone() throws Exception {
HashQueue q = new HashQueue();
final List cookieList = new LinkedList();
final List eList = new LinkedList();
HashService.Callback cb =
new HashService.Callback() {
public void hashingFinished(
CachedUrlSet urlset,
long timeUsed,
Object cookie,
CachedUrlSetHasher hasher,
Exception e) {
cookieList.add(cookie);
eList.add(e);
}
};
HashQueue.Request r1, r2, r3, r4, r5;
r1 = req(2000, 0, 100, cb);
r2 = req(10000, 0, 200, cb);
r3 = req(20000, 0, 0, cb);
r4 = req(50000, 0, 1, cb);
assertTrue(q.insert(r1));
assertTrue(q.insert(r2));
assertTrue(q.insert(r4));
assertEquals(0, cookieList.size());
q.removeCompleted();
assertEquals(0, cookieList.size());
// make r1 timeout
r1.deadline.expire();
q.removeCompleted();
List exp = ListUtil.list(r1);
assertEquals(exp, cookieList);
assertEquals(exp, q.getCompletedSnapshot());
// make r2 timeout
TimeBase.step(11000);
// r3 is finished
assertTrue(q.insert(r3));
Exception r4e = new Exception();
// make r4 error
r4.e = r4e;
q.removeCompleted();
// check that they all finished, and in the right order
Object exp2[] = {r1, r2, r3, r4};
assertIsomorphic(exp2, cookieList);
assertIsomorphic(exp2, q.getCompletedSnapshot());
// check their exceptions
assertTrue(eList.get(0) instanceof HashService.Timeout);
assertTrue(eList.get(1) instanceof HashService.Timeout);
assertSame(null, eList.get(2));
assertSame(r4e, eList.get(3));
}
/**
* 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));
}
public void testStepperThrows() {
final List finished = new ArrayList();
TaskCallback cb =
new TaskCallback() {
public void taskEvent(SchedulableTask task, Schedule.EventType event) {
if (event == Schedule.EventType.FINISH) {
finished.add(task);
}
}
};
MyMockStepper stepper = new MyMockStepper(10, -10);
stepper.setWhenToThrow(5);
StepTask t1 = task(100, 200, 100, cb, stepper);
Schedule s = sched(ListUtil.list(t1));
fact.setResult(s);
assertTrue(tr.addToSchedule(t1));
TimeBase.setSimulated(101);
assertTrue(tr.findTaskToRun());
Interrupter intr = null;
try {
intr = interruptMeIn(TIMEOUT_SHOULDNT, true);
tr.runSteps(new MutableBoolean(true), null);
intr.cancel();
} catch (Exception e) {
log.error("runSteps threw:", e);
} finally {
if (intr.did()) {
fail("runSteps looped");
}
}
assertSame(t1, finished.get(0));
assertTrue(t1.e instanceof ExpectedRuntimeException);
assertEquals(5, stepper.nSteps);
}
// remove task-ending chunk, past task deadline, s.b. Timeout error.
public void testRemoveChunkTaskEndTimeout() {
final List finished = new ArrayList();
StepTask t1 =
task(
100,
200,
100,
new TaskCallback() {
public void taskEvent(SchedulableTask task, Schedule.EventType event) {
if (log.isDebug2()) {
log.debug2("testRemoveChunkTaskEndTimeout callback");
}
if (event == Schedule.EventType.FINISH) {
finished.add(task);
}
}
});
Schedule s = sched(ListUtil.list(t1));
fact.setResult(s);
assertTrue(tr.addToSchedule(t1));
Schedule.Chunk chunk = (Schedule.Chunk) s.getEvents().get(0);
assertTrue(tr.getCurrentSchedule().getEvents().contains(chunk));
chunk.setTaskEnd();
TimeBase.setSimulated(201);
tr.removeChunk(chunk);
assertFalse(tr.getCurrentSchedule().getEvents().contains(chunk));
assertSame(t1, finished.get(0));
assertNotNull(t1.e);
assertTrue(t1.e.toString(), t1.e instanceof SchedService.Timeout);
assertEmpty(tr.getAcceptedTasks());
}
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");
}
protected void assertEqualTables(Object[][] a1, List lines) {
assertEquals("numrows", a1.length, lines.size() - NUM_HEADER_LINES);
for (int irow = 0; irow <= a1.length - 1; irow++) {
Object expRow[] = a1[irow];
List row = StringUtil.breakAt((String) lines.get(irow + NUM_HEADER_LINES), ',');
assertEquals("numcols", expRow.length, row.size());
assertEquals(("row " + irow), SetUtil.fromArray(expRow), new HashSet(row));
}
}
public void testText() throws Exception {
MockStatusAccessor statusAccessor =
MockStatusAccessor.generateStatusAccessor(colArray1, rowArray1);
statusAccessor.setTitle("testtbl", null);
statSvc.registerStatusAccessor("testtbl", statusAccessor);
WebResponse resp = getTable("testtbl", true);
assertResponseOk(resp);
assertEquals("Content type", "text/plain", resp.getContentType());
log.debug(resp.getText());
List lines = getLines(resp);
assertEquals(rowArray1.length + 3, lines.size());
Map row0 = getRow((String) lines.get(0));
assertEquals("2.4.6.8", row0.get("host"));
Map row2 = getRow((String) lines.get(2));
assertEquals("testtbl", row2.get("table"));
assertEqualTables(table1, lines);
}
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));
}
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 testFindTaskToRunRemovesExpiredOverrunners() {
assertFalse(tr.findTaskToRun());
StepTask t1 = task(100, 200, 100);
StepTask t2 = task(100, 300, 50);
StepTask texp1 = task(0, 0, 50);
StepTask texp2 = task(0, 0, 49);
Schedule s = sched(ListUtil.list(t1, t2));
fact.setResults(s, s);
assertTrue(tr.addToSchedule(t1));
assertTrue(tr.addToSchedule(t2));
tr.addOverrunner(texp1);
tr.addOverrunner(texp2);
// if this fails, it might be because the sorted list/set is treating
// sort-order equivalence as object equality, which we don't want
assertEquals(2, tr.getOverrunTasks().size());
assertFalse(tr.findTaskToRun());
assertEquals(0, removedChunks.size());
assertEquals(2, removedTasks.size());
assertEquals(
SetUtil.set(texp1, texp2),
SetUtil.set((StepTask) removedTasks.get(0), (StepTask) removedTasks.get(1)));
}
/**
* 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 testFindTaskToRunRemovesExpiredChunks() {
assertFalse(tr.findTaskToRun());
StepTask t1 = task(100, 200, 100);
StepTask t2 = task(100, 300, 50);
StepTask texp1 = task(0, 0, 50);
StepTask texp2 = task(0, 0, 50);
Schedule s = sched(ListUtil.list(texp1, texp2, t1, t2));
fact.setResults(s, s);
assertTrue(tr.addToSchedule(t1));
assertTrue(tr.addToSchedule(t2));
assertFalse(tr.findTaskToRun());
assertEquals(2, removedChunks.size());
assertEquals(
SetUtil.set(texp1, texp2),
SetUtil.set(
((Schedule.Chunk) removedChunks.get(0)).getTask(),
((Schedule.Chunk) removedChunks.get(1)).getTask()));
}
public void testStorePollHistories() throws Exception {
TimeBase.setSimulated(123321);
MockCachedUrlSetSpec mspec = new MockCachedUrlSetSpec("http://www.example.com", null);
CachedUrlSet mcus = new MockCachedUrlSet(mau, mspec);
NodeStateImpl nodeState = new NodeStateImpl(mcus, -1, null, null, repository);
List histories =
ListUtil.list(
createPollHistoryBean(3),
createPollHistoryBean(3),
createPollHistoryBean(3),
createPollHistoryBean(3),
createPollHistoryBean(3));
/*
* CASTOR: [summary] Rewrite test in non-Castor way
* This is obviously not an appropriate way of writing this test,
* Right now it creates sample data in Castor format, from legacy
* code back when Castor was the built-in serialization engine.
* TODO: Rewrite test in non-Castor way
*/
// nodeState.setPollHistoryBeanList(histories);
nodeState.setPollHistoryList(NodeHistoryBean.fromBeanListToList(histories));
repository.storePollHistories(nodeState);
String filePath = LockssRepositoryImpl.mapAuToFileLocation(tempDirPath, mau);
filePath =
LockssRepositoryImpl.mapUrlToFileLocation(
filePath, "http://www.example.com/" + HistoryRepositoryImpl.HISTORY_FILE_NAME);
File xmlFile = new File(filePath);
assertTrue(xmlFile.exists());
nodeState.setPollHistoryList(new ArrayList());
repository.loadPollHistories(nodeState);
List loadedHistory = nodeState.getPollHistoryList();
assertEquals(histories.size(), loadedHistory.size());
// CASTOR: some Castor-tailored stuff here
// PollHistoryBean expect1 = (PollHistoryBean)histories.get(0);
// PollHistoryBean elem1 = (PollHistoryBean)loadedHistory.get(0);
PollHistory expect1 = (PollHistory) histories.get(0);
PollHistory elem1 = (PollHistory) loadedHistory.get(0);
assertEquals(expect1.type, elem1.type);
assertEquals(expect1.lwrBound, elem1.lwrBound);
assertEquals(expect1.uprBound, elem1.uprBound);
assertEquals(expect1.status, elem1.status);
assertEquals(expect1.startTime, elem1.startTime);
assertEquals(expect1.duration, elem1.duration);
// CASTOR: some Castor-tailored stuff here
// List expectBeans = (List)expect1.getVoteBeans();
// List elemBeans = (List)elem1.getVoteBeans();
Iterator expectIter = (Iterator) expect1.getVotes();
Iterator elemIter = (Iterator) elem1.getVotes();
while (expectIter.hasNext() && elemIter.hasNext()) {
Vote expectVote = (Vote) expectIter.next();
Vote elemVote = (Vote) elemIter.next();
assertEquals(
expectVote.getVoterIdentity().getIdString(), elemVote.getVoterIdentity().getIdString());
assertEquals(expectVote.isAgreeVote(), elemVote.isAgreeVote());
assertEquals(expectVote.getChallengeString(), elemVote.getChallengeString());
assertEquals(expectVote.getVerifierString(), elemVote.getVerifierString());
assertEquals(expectVote.getHashString(), elemVote.getHashString());
}
assertFalse(expectIter.hasNext());
assertFalse(expectIter.hasNext());
TimeBase.setReal();
}
protected void initResultMap() throws PluginException.InvalidDefinition {
HttpResultMap hResultMap = new HttpResultMap();
// XXX Currently this only allows a CacheResultHandler class to
// initialize the result map. Instead, don't use a CacheResultMap
// directly, use either the plugin's CacheResultHandler, if specified,
// or a default one that wraps the CacheResultMap
String handler_class = null;
handler_class = definitionMap.getString(KEY_EXCEPTION_HANDLER, null);
if (handler_class != null) {
try {
resultHandler = (CacheResultHandler) newAuxClass(handler_class, CacheResultHandler.class);
resultHandler.init(hResultMap);
} catch (Exception ex) {
throw new PluginException.InvalidDefinition(
mapName + " has invalid Exception handler: " + handler_class, ex);
} catch (LinkageError le) {
throw new PluginException.InvalidDefinition(
mapName + " has invalid Exception handler: " + handler_class, le);
}
} else {
// Expect a list of mappings from either result code or exception
// name to CacheException name
Collection<String> mappings = definitionMap.getCollection(KEY_EXCEPTION_LIST, null);
if (mappings != null) {
// add each entry
for (String entry : mappings) {
if (log.isDebug2()) {
log.debug2("initMap(" + entry + ")");
}
String first;
String ceName;
try {
List<String> pair = StringUtil.breakAt(entry, '=', 2, true, true);
first = pair.get(0);
ceName = pair.get(1);
} catch (Exception ex) {
throw new PluginException.InvalidDefinition(
"Invalid syntax: " + entry + "in " + mapName);
}
Object val;
// Value should be either a CacheException or CacheResultHandler
// class name.
PluginFetchEventResponse resp =
(PluginFetchEventResponse) newAuxClass(ceName, PluginFetchEventResponse.class, null);
if (resp instanceof CacheException) {
val = resp.getClass();
} else if (resp instanceof CacheResultHandler) {
val = WrapperUtil.wrap((CacheResultHandler) resp, CacheResultHandler.class);
} else {
throw new PluginException.InvalidDefinition(
"Second arg not a "
+ "CacheException or "
+ "CacheResultHandler class: "
+ entry
+ ", in "
+ mapName);
}
try {
int code = Integer.parseInt(first);
// If parseable as an integer, it's a result code.
hResultMap.storeMapEntry(code, val);
} catch (NumberFormatException e) {
try {
Class eClass = Class.forName(first);
// If a class name, it should be an exception class
if (Exception.class.isAssignableFrom(eClass)) {
hResultMap.storeMapEntry(eClass, val);
} else {
throw new PluginException.InvalidDefinition(
"First arg not an " + "Exception class: " + entry + ", in " + mapName);
}
} catch (Exception ex) {
throw new PluginException.InvalidDefinition(
"First arg not a " + "number or class: " + entry + ", in " + mapName);
} catch (LinkageError le) {
throw new PluginException.InvalidDefinition("Can't load " + first, le);
}
}
}
}
}
resultMap = hResultMap;
}
