private static void setRowKeyOffset(Filter filter, int offset) {
if (filter instanceof BooleanExpressionFilter) {
BooleanExpressionFilter boolFilter = (BooleanExpressionFilter) filter;
IndexUtil.setRowKeyExpressionOffset(boolFilter.getExpression(), offset);
} else if (filter instanceof SkipScanFilter) {
SkipScanFilter skipScanFilter = (SkipScanFilter) filter;
skipScanFilter.setOffset(offset);
}
}
/** {@inheritDoc} */
public boolean handleEvent(
final ResourceResolver resolver,
final String resourceType,
final String eventTopics,
final String nodePath,
final String[] resourceChangedAttributes,
final String[] resourceAddedAttributes,
final String[] resourceRemovedAttributes,
final Enum<EventTopicsType> eventTopicsType)
throws Exception {
final boolean isDelete = (eventTopicsType == EventTopicsType.DELETED) ? true : false;
if (isDelete) {
// Check to see if the root is the index root node or not
final String componentName = IndexUtil.getComponentName(resolver, nodePath);
if (componentName != null && componentName.endsWith(IndexUtil.INDEX_FOLDER)) {
return false;
}
} else if (resourceType == null) {
return false;
} else if (checkResourceType(resolver, nodePath, resourceType, addedResourceTypes)
&& eventTopics.equals(SlingConstants.TOPIC_RESOURCE_ADDED)) {
// A target resource type has been added to the repository.
} else if (checkResourceType(resolver, nodePath, resourceType, changedResourceTypes)
&& eventTopics.equals(SlingConstants.TOPIC_RESOURCE_CHANGED)) {
// A target attribute may have been modified.
if (!(checkAttributes(resourceChangedAttributes, targetModifiedAttributes)
|| checkAttributes(resourceAddedAttributes, targetAddedRemovedAttributes)
|| checkAttributes(resourceRemovedAttributes, targetAddedRemovedAttributes))) {
// This attribute does not trigger an index update.
return false;
}
} else {
return false;
}
indexExec(resolver, nodePath, eventTopicsType);
return true;
}
/**
* optimise
*
* @param structureName
* @param index
* @param statsCounter
* @param numEntries
* @throws IOException
*/
@SuppressWarnings("unchecked")
public static void optimise(
String structureName,
Index index,
LexiconBuilder.CollectionStatisticsCounter statsCounter,
int numEntries)
throws IOException {
final String mapFileFilename =
constructFilename(structureName, index.getPath(), index.getPrefix(), MAPFILE_EXT);
final FixedSizeWriteableFactory<Text> keyFactory =
(FixedSizeWriteableFactory<Text>) index.getIndexStructure(structureName + "-keyfactory");
final FixedSizeWriteableFactory<LexiconEntry> valueFactory =
(FixedSizeWriteableFactory<LexiconEntry>)
index.getIndexStructure(structureName + "-valuefactory");
// logger.info("Optimsing lexicon with "+ numEntries + " entries");
// term id lookups
boolean termIdsAligned = true;
int[] termid2index = new int[numEntries];
Arrays.fill(termid2index, -1);
int counter = 0;
int lastTermId = -1;
// bsearch reduction
int previousFirstChar = -1;
int firstChar = 0;
final TIntObjectHashMap<int[]> map = new TIntObjectHashMap<int[]>();
Iterator<Map.Entry<Text, LexiconEntry>> iterator =
new FSOrderedMapFile.EntryIterator<Text, LexiconEntry>(
mapFileFilename, keyFactory, valueFactory);
while (iterator.hasNext()) {
Map.Entry<Text, LexiconEntry> lee = iterator.next();
// System.err.println(lee.toString());
// term id
int termId = lee.getValue().getTermId();
if (!(termId == lastTermId + 1)) termIdsAligned = false;
if (termid2index[termId] != -1) {
throw new WrappedIOException(
new IllegalArgumentException(
"Termid "
+ termId
+ " is not unique - used at entries "
+ termid2index[termId]
+ " and"
+ counter));
}
termid2index[termId] = counter;
lastTermId = termId;
// bsearch reduction optimisaion
firstChar = lee.getKey().charAt(0);
if (firstChar != previousFirstChar) {
int[] boundaries = new int[] {counter, 0};
map.put(firstChar, boundaries);
previousFirstChar = firstChar;
}
// increments
statsCounter.count(lee.getValue());
counter++;
}
if (counter != numEntries) termIdsAligned = false;
IndexUtil.close(iterator);
// deal with termids
if (termIdsAligned) {
index.setIndexProperty("index." + structureName + ".termids", "aligned");
// logger.info("All ids for structure "+structureName+ " are aligned, skipping "
// +ID_EXT+ " file");
} else {
DataOutputStream dos =
new DataOutputStream(
Files.writeFileStream(
constructFilename(structureName, index.getPath(), index.getPrefix(), ID_EXT)));
for (int indexof : termid2index) dos.writeInt(indexof);
dos.close();
index.setIndexProperty(
"index." + structureName + ".termids", (numEntries > 15000000) ? "file" : "fileinmem");
}
int[] mapKeys = map.keys();
Arrays.sort(mapKeys);
final int mapKeysSize = mapKeys.length;
for (int i = 0; i < mapKeysSize - 1; i++) {
int nextLowerBoundary = (map.get(mapKeys[i + 1]))[0];
int[] currentBoundaries = map.get(mapKeys[i]);
currentBoundaries[1] = nextLowerBoundary;
map.put(mapKeys[i], currentBoundaries);
}
// do something about the last entry
int nextLowerBoundary = counter;
int[] currentBoundaries = (int[]) map.get(mapKeys[mapKeysSize - 1]);
currentBoundaries[1] = nextLowerBoundary;
map.put(mapKeys[mapKeysSize - 1], currentBoundaries);
final ObjectOutputStream oos =
new ObjectOutputStream(
Files.writeFileStream(
constructFilename(structureName, index.getPath(), index.getPrefix(), HASH_EXT)));
oos.writeObject(map);
oos.close();
index.setIndexProperty("index." + structureName + ".bsearchshortcut", "charmap");
index.flush();
}
