/**
* Expunge an entry from memMap while updating diskMap.
*
* @param entry a SoftEntry<V> obtained from refQueuePoll()
*/
private synchronized void pageOutStaleEntry(SoftEntry<V> entry) {
PhantomEntry<V> phantom = entry.phantom;
// Still in memMap? if not, was paged-out by earlier direct access
// before placed into reference-queue; just return
if (memMap.get(phantom.key) != entry) { // NOTE: intentional identity compare
return;
}
// recover hidden value
V phantomValue = phantom.doctoredGet();
// Expected value present? (should be; only clear is at end of
// this method, after entry removal from memMap)
if (phantomValue == null) {
logger.log(Level.WARNING, "unexpected null phantomValue", new Exception());
return; // nothing to do
}
// given instance entry still in memMap;
// we have the key and phantom Value,
// the diskMap can be updated.
diskMap.put(phantom.key, phantomValue); // unchecked cast
expungeStatsDiskPut.incrementAndGet();
// remove memMap entry
boolean removed = memMap.remove(phantom.key, entry);
if (!removed) {
logger.log(Level.WARNING, "expunge memMap.remove() ineffective", new Exception());
}
phantom.clear(); // truly allows GC of unreferenced V object
}
/**
* Copies entries from an existing environment db to a new one. If historyMap is not provided,
* only logs the entries that would have been copied.
*
* @param sourceDir existing environment database directory
* @param historyMap new environment db (or null for a dry run)
* @return number of records
* @throws DatabaseException
*/
private static int copyPersistEnv(File sourceDir, StoredSortedMap<String, Map> historyMap)
throws DatabaseException {
int count = 0;
// open the source env history DB, copying entries to target env
EnhancedEnvironment sourceEnv = setupCopyEnvironment(sourceDir, true);
StoredClassCatalog sourceClassCatalog = sourceEnv.getClassCatalog();
DatabaseConfig historyDbConfig = HISTORY_DB_CONFIG.toDatabaseConfig();
historyDbConfig.setReadOnly(true);
Database sourceHistoryDB = sourceEnv.openDatabase(null, URI_HISTORY_DBNAME, historyDbConfig);
StoredSortedMap<String, Map> sourceHistoryMap =
new StoredSortedMap<String, Map>(
sourceHistoryDB,
new StringBinding(),
new SerialBinding<Map>(sourceClassCatalog, Map.class),
true);
Iterator<Entry<String, Map>> iter = sourceHistoryMap.entrySet().iterator();
while (iter.hasNext()) {
Entry<String, Map> item = iter.next();
if (logger.isLoggable(Level.FINE)) {
logger.fine(item.getKey() + " " + new JSONObject(item.getValue()));
}
if (historyMap != null) {
historyMap.put(item.getKey(), item.getValue());
}
count++;
}
StoredIterator.close(iter);
sourceHistoryDB.close();
sourceEnv.close();
return count;
}
/**
* Populates an environment db from a persist log. If historyMap is not provided, only logs the
* entries that would have been populated.
*
* @param persistLogReader persist log
* @param historyMap new environment db (or null for a dry run)
* @return number of records
* @throws UnsupportedEncodingException
* @throws DatabaseException
*/
private static int populatePersistEnvFromLog(
BufferedReader persistLogReader, StoredSortedMap<String, Map> historyMap)
throws UnsupportedEncodingException, DatabaseException {
int count = 0;
Iterator<String> iter = new LineReadingIterator(persistLogReader);
while (iter.hasNext()) {
String line = iter.next();
if (line.length() == 0) {
continue;
}
String[] splits = line.split(" ");
if (splits.length != 2) {
logger.severe("bad line has " + splits.length + " fields (should be 2): " + line);
continue;
}
Map alist;
try {
alist =
(Map) SerializationUtils.deserialize(Base64.decodeBase64(splits[1].getBytes("UTF-8")));
} catch (Exception e) {
logger.severe("caught exception " + e + " deserializing line: " + line);
continue;
}
if (logger.isLoggable(Level.FINE)) {
logger.fine(splits[0] + " " + ArchiveUtils.prettyString(alist));
}
if (historyMap != null)
try {
historyMap.put(splits[0], alist);
} catch (Exception e) {
logger.log(
Level.SEVERE,
"caught exception after loading "
+ count
+ " urls from the persist log (perhaps crawl was stopped by user?)",
e);
IOUtils.closeQuietly(persistLogReader);
// seems to finish most cleanly when we return rather than throw something
return count;
}
count++;
}
IOUtils.closeQuietly(persistLogReader);
return count;
}
