| public AddError ( string index, string key, string error ) : void | ||
| index | string | |
| key | string | |
| error | string | |
| return | void | |
public override void Remove(string[] keys, WorkContext context)
{
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), name));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
keys.Apply(
key => batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, key),
exception);
context.AddError(name, key, exception.Message);
},
trigger => trigger.OnIndexEntryDeleted(key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
batcher => batcher.Dispose());
return(keys.Length);
});
}
public override void Remove(string[] keys, WorkContext context)
{
Write(context, (writer, analyzer) =>
{
if (logIndexing.IsDebugEnabled)
{
logIndexing.DebugFormat("Deleting ({0}) from {1}", string.Format(", ", keys), name);
}
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
keys.Apply(
key => batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, key);
context.AddError(name, key, exception.Message);
},
trigger => trigger.OnIndexEntryDeleted(name, key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k)).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
batcher => batcher.Dispose());
return(true);
});
}
protected void Write(WorkContext context, Func <IndexWriter, Analyzer, bool> action)
{
if (disposed)
{
throw new ObjectDisposedException("Index " + name + " has been disposed");
}
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List <Action>();
Analyzer analyzer = null;
try
{
try
{
analyzer = CreateAnalyzer(new LowerCaseAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
{
indexWriter = new IndexWriter(directory, new StopAnalyzer(Version.LUCENE_29), IndexWriter.MaxFieldLength.UNLIMITED);
}
try
{
shouldRecreateSearcher = action(indexWriter, analyzer);
foreach (var indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocumented);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
}
finally
{
currentlyIndexDocumented.Clear();
if (analyzer != null)
{
analyzer.Close();
}
foreach (var dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
{
RecreateSearcher();
}
}
}
public override void Remove(string[] keys, WorkContext context)
{
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), indexId));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(
key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.DocumentIdFieldName, key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(indexId, null, e.Message, "Dispose Trigger");
},
batcher => batcher.Dispose());
IndexStats currentIndexStats = null;
context.TransactionalStorage.Batch(accessor => currentIndexStats = accessor.Indexing.GetIndexStats(indexId));
return(new IndexedItemsInfo
{
ChangedDocs = keys.Length,
HighestETag = currentIndexStats.LastIndexedEtag,
DeletedKeys = keys
});
});
}
public override void Remove(string[] keys, WorkContext context)
{
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
if (logIndexing.IsDebugEnabled)
{
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), PublicName));
}
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(
key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.DocumentIdFieldName, key.ToLowerInvariant())));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger in " + PublicName, e);
context.AddError(indexId, PublicName, null, e, "Dispose Trigger");
},
batcher => batcher.Dispose());
return(new IndexedItemsInfo(GetLastEtagFromStats())
{
ChangedDocs = keys.Length,
DeletedKeys = keys
});
});
}
protected IEnumerable <object> RobustEnumerationReduce(IEnumerable <object> input, IndexingFunc func,
IStorageActionsAccessor actions, WorkContext context)
{
// not strictly accurate, but if we get that many errors, probably an error anyway.
return(new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = actions.Indexing.IncrementReduceIndexingAttempt,
CancelMoveNext = actions.Indexing.DecrementReduceIndexingAttempt,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
try
{
actions.Indexing.IncrementReduceIndexingFailure();
}
catch (Exception e)
{
// we don't care about error here, because it is an error on error problem
logIndexing.WarnException(
String.Format("Could not increment indexing failure rate for {0}", name),
e);
}
}
}.RobustEnumeration(input, func));
}
// we don't use the usual GroupBy, because that isn't streaming
// we rely on the fact that all values from the same docs are always outputed at
// the same time, so we can take advantage of this fact
private IEnumerable <IGrouping <object, dynamic> > GroupByDocumentId(WorkContext context, IEnumerable <object> docs)
{
var enumerator = docs.GetEnumerator();
if (enumerator.MoveNext() == false)
{
yield break;
}
while (true)
{
object documentId;
try
{
documentId = GetDocumentId(enumerator.Current);
}
catch (Exception e)
{
context.AddError(name, null, e.Message);
if (enumerator.MoveNext() == false)
{
yield break;
}
continue;
}
var groupByDocumentId = new Grouping(documentId, enumerator);
yield return(groupByDocumentId);
if (groupByDocumentId.Done)
{
break;
}
}
}
private bool?MoveNext(IEnumerator en, StatefulEnumerableWrapper <object> innerEnumerator, WorkContext context,
IStorageActionsAccessor actions)
{
try
{
actions.Indexing.IncrementIndexingAttempt();
var moveNext = en.MoveNext();
if (moveNext == false)
{
actions.Indexing.DecrementIndexingAttempt();
}
return(moveNext);
}
catch (Exception e)
{
actions.Indexing.IncrementIndexingFailure();
context.AddError(name,
TryGetDocKey(innerEnumerator.Current),
e.Message
);
log.WarnFormat(e, "Failed to execute indexing function on {0} on {1}", name,
GetDocId(innerEnumerator));
}
return(null);
}
protected IEnumerable <object> RobustEnumerationReduce(IEnumerable <object> input, IndexingFunc func, IStorageActionsAccessor actions, WorkContext context)
{
return(new RobustEnumerator
{
BeforeMoveNext = actions.Indexing.IncrementReduceIndexingAttempt,
CancelMoveNext = actions.Indexing.DecrementReduceIndexingAttempt,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnFormat(exception, "Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o));
try
{
actions.Indexing.IncrementReduceIndexingFailure();
}
catch (Exception e)
{
// we don't care about error here, because it is an error on error problem
logIndexing.WarnFormat(e, "Could not increment indexing failure rate for {0}", name);
}
}
}.RobustEnumeration(input, func));
}
public override void Remove(string[] keys, WorkContext context)
{
context.TransactionalStorage.Batch(actions =>
{
var reduceKeyAndBuckets = new Dictionary <ReduceKeyAndBucket, int>();
foreach (var key in keys)
{
context.CancellationToken.ThrowIfCancellationRequested();
actions.MapReduce.DeleteMappedResultsForDocumentId(key, indexId, reduceKeyAndBuckets);
}
actions.MapReduce.UpdateRemovedMapReduceStats(indexId, reduceKeyAndBuckets, context.CancellationToken);
foreach (var reduceKeyAndBucket in reduceKeyAndBuckets)
{
context.CancellationToken.ThrowIfCancellationRequested();
actions.MapReduce.ScheduleReductions(indexId, 0, reduceKeyAndBucket.Key);
}
});
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
if (logIndexing.IsDebugEnabled)
{
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), PublicName));
}
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(
key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.ReduceKeyFieldName, key.ToLowerInvariant())));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.ReduceKeyFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger in " + PublicName, e);
context.AddError(indexId, PublicName, null, e, "Dispose Trigger");
},
batcher => batcher.Dispose());
return(new IndexedItemsInfo(null)
{
ChangedDocs = keys.Length
});
});
}
// we don't care about tracking map/reduce stats here, since it is merely
// an optimization step
protected IEnumerable <object> RobustEnumerationReduceDuringMapPhase(IEnumerable <object> input, IndexingFunc func,
IStorageActionsAccessor actions, WorkContext context)
{
// not strictly accurate, but if we get that many errors, probably an error anyway.
return(new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = () => { }, // don't care
CancelMoveNext = () => { }, // don't care
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
}
}.RobustEnumeration(input, func));
}
protected IEnumerable <object> RobustEnumerationIndex(IEnumerable <object> input, IEnumerable <IndexingFunc> funcs,
IStorageActionsAccessor actions, WorkContext context, IndexingWorkStats stats)
{
return(new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = () => stats.IndexingAttempts++,
CancelMoveNext = () => stats.IndexingAttempts--,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
stats.IndexingErrors++;
}
}.RobustEnumeration(input, funcs));
}
protected void Write(WorkContext context, Func<IndexWriter, Analyzer, bool> action)
{
if (disposed)
throw new ObjectDisposedException("Index " + name + " has been disposed");
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List<Action>();
Analyzer analyzer = null;
try
{
try
{
analyzer = CreateAnalyzer(new LowerCaseAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
indexWriter = new IndexWriter(directory, new StopAnalyzer(Version.LUCENE_29), IndexWriter.MaxFieldLength.UNLIMITED);
try
{
shouldRecreateSearcher = action(indexWriter, analyzer);
foreach (IIndexExtension indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocumented);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
}
finally
{
currentlyIndexDocumented.Clear();
if (analyzer != null)
analyzer.Close();
foreach (Action dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
RecreateSearcher();
}
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IEnumerable<object> documents, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var count = 0;
Write(context, (indexWriter, analyzer) =>
{
bool madeChanges = false;
PropertyDescriptorCollection properties = null;
var processedKeys = new HashSet<string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
var documentsWrapped = documents.Select((dynamic doc) =>
{
if(doc.__document_id == null)
throw new ArgumentException("Cannot index something which doesn't have a document id, but got: " + doc);
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
return doc;
madeChanges = true;
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId),
exception);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, documentId));
indexWriter.DeleteDocuments(new Term(Constants.DocumentIdFieldName, documentId.ToLowerInvariant()));
return doc;
});
foreach (var doc in RobustEnumerationIndex(documentsWrapped, viewGenerator.MapDefinition, actions, context))
{
count++;
IndexingResult indexingResult;
if (doc is DynamicJsonObject)
indexingResult = ExtractIndexDataFromDocument((DynamicJsonObject)doc);
else
indexingResult = ExtractIndexDataFromDocument(properties, doc);
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
var luceneDoc = new Document();
luceneDoc.Add(new Field(Constants.DocumentIdFieldName, indexingResult.NewDocId.ToLowerInvariant(), Field.Store.YES,
Field.Index.NOT_ANALYZED_NO_NORMS));
madeChanges = true;
CopyFieldsToDocument(luceneDoc, indexingResult.Fields);
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.Warn(
string.Format( "Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId),
exception);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, indexingResult.NewDocId, luceneDoc));
logIndexing.Debug("Index '{0}' resulted in: {1}", name, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
actions.Indexing.IncrementSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return madeChanges;
});
logIndexing.Debug("Indexed {0} documents for {1}", count, name);
}
public override void Remove(string[] keys, WorkContext context)
{
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), name));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
keys.Apply(
key => batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, key),
exception);
context.AddError(name, key, exception.Message, "OnIndexEntryDeleted Trigger");
},
trigger => trigger.OnIndexEntryDeleted(key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message, "Dispose Trigger");
},
batcher => batcher.Dispose());
IndexStats currentIndexStats = null;
context.TransactionalStorage.Batch(accessor => currentIndexStats = accessor.Indexing.GetIndexStats(name));
return new IndexedItemsInfo
{
ChangedDocs = keys.Length,
HighestETag = currentIndexStats.LastIndexedEtag,
DeletedKeys = keys
};
});
}
protected IEnumerable<object> RobustEnumerationReduce(IEnumerable<object> input, IndexingFunc func,
IStorageActionsAccessor actions, WorkContext context)
{
// not strictly accurate, but if we get that many errors, probably an error anyway.
return new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = actions.Indexing.IncrementReduceIndexingAttempt,
CancelMoveNext = actions.Indexing.DecrementReduceIndexingAttempt,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
try
{
actions.Indexing.IncrementReduceIndexingFailure();
}
catch (Exception e)
{
// we don't care about error here, because it is an error on error problem
logIndexing.WarnException(
String.Format("Could not increment indexing failure rate for {0}", name),
e);
}
}
}.RobustEnumeration(input, func);
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IEnumerable<object> documents, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
var count = 0;
Write(context, (indexWriter, analyzer, stats) =>
{
var processedKeys = new HashSet<string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
var documentsWrapped = documents.Select((dynamic doc) =>
{
if(doc.__document_id == null)
throw new ArgumentException(string.Format("Cannot index something which doesn't have a document id, but got: '{0}'", doc));
count++;
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
return doc;
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId),
exception);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(documentId));
indexWriter.DeleteDocuments(new Term(Constants.DocumentIdFieldName, documentId.ToLowerInvariant()));
return doc;
});
var anonymousObjectToLuceneDocumentConverter = new AnonymousObjectToLuceneDocumentConverter(indexDefinition);
var luceneDoc = new Document();
var documentIdField = new Field(Constants.DocumentIdFieldName, "dummy", Field.Store.YES, Field.Index.ANALYZED_NO_NORMS);
foreach (var doc in RobustEnumerationIndex(documentsWrapped, viewGenerator.MapDefinitions, actions, context, stats))
{
count++;
float boost;
var indexingResult = GetIndexingResult(doc, anonymousObjectToLuceneDocumentConverter, out boost);
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
count += 1;
luceneDoc.GetFields().Clear();
luceneDoc.SetBoost(boost);
documentIdField.SetValue(indexingResult.NewDocId.ToLowerInvariant());
luceneDoc.Add(documentIdField);
foreach (var field in indexingResult.Fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format( "Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId),
exception);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(indexingResult.NewDocId, luceneDoc));
LogIndexedDocument(indexingResult.NewDocId, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
stats.IndexingSuccesses++;
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return count;
});
logIndexing.Debug("Indexed {0} documents for {1}", count, name);
}
protected void Write(WorkContext context, Func <IndexWriter, Analyzer, bool> action)
{
if (disposed)
{
throw new ObjectDisposedException("Index " + name + " has been disposed");
}
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List <Action>();
Analyzer searchAnalyzer = null;
try
{
try
{
searchAnalyzer = CreateAnalyzer(new LowerCaseKeywordAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
{
indexWriter = CreateIndexWriter(directory);
}
try
{
shouldRecreateSearcher = action(indexWriter, searchAnalyzer);
foreach (IIndexExtension indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocuments);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
WriteTempIndexToDiskIfNeeded(context);
}
finally
{
currentlyIndexDocuments.Clear();
if (searchAnalyzer != null)
{
searchAnalyzer.Close();
}
foreach (Action dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
{
RecreateSearcher();
}
}
}
private bool? MoveNext(IEnumerator en, StatefulEnumerableWrapper<object> innerEnumerator, WorkContext context,
DocumentStorageActions actions)
{
try
{
actions.IncrementIndexingAttempt();
var moveNext = en.MoveNext();
if (moveNext == false)
actions.DecrementIndexingAttempt();
return moveNext;
}
catch (Exception e)
{
actions.IncrementIndexingFailure();
context.AddError(name,
TryGetDocKey(innerEnumerator.Current),
e.Message
);
log.WarnFormat(e, "Failed to execute indexing function on {0} on {1}", name,
GetDocId(innerEnumerator));
}
return null;
}
protected void Write(WorkContext context, Func <IndexWriter, Analyzer, IndexingWorkStats, int> action)
{
if (disposed)
{
throw new ObjectDisposedException("Index " + name + " has been disposed");
}
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List <Action>();
Analyzer searchAnalyzer = null;
try
{
try
{
searchAnalyzer = CreateAnalyzer(new LowerCaseKeywordAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
{
indexWriter = CreateIndexWriter(directory);
}
var stats = new IndexingWorkStats();
try
{
var changedDocs = action(indexWriter, searchAnalyzer, stats);
docCountSinceLastOptimization += changedDocs;
shouldRecreateSearcher = changedDocs > 0;
foreach (IIndexExtension indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocuments);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
UpdateIndexingStats(context, stats);
WriteTempIndexToDiskIfNeeded(context);
if (configuration.TransactionMode == TransactionMode.Safe)
{
Flush(); // just make sure changes are flushed to disk
}
}
finally
{
currentlyIndexDocuments.Clear();
if (searchAnalyzer != null)
{
searchAnalyzer.Close();
}
foreach (Action dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
{
RecreateSearcher();
}
}
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IEnumerable <object> documents, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var count = 0;
Write(context, (indexWriter, analyzer) =>
{
bool madeChanges = false;
PropertyDescriptorCollection properties = null;
var processedKeys = new HashSet <string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
var documentsWrapped = documents.Select((dynamic doc) =>
{
if (doc.__document_id == null)
{
throw new ArgumentException("Cannot index something which doesn't have a document id, but got: " + doc);
}
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
{
return(doc);
}
madeChanges = true;
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, documentId));
indexWriter.DeleteDocuments(new Term(Constants.DocumentIdFieldName, documentId.ToLowerInvariant()));
return(doc);
});
foreach (var doc in RobustEnumerationIndex(documentsWrapped, viewGenerator.MapDefinition, actions, context))
{
count++;
IndexingResult indexingResult;
if (doc is DynamicJsonObject)
{
indexingResult = ExtractIndexDataFromDocument((DynamicJsonObject)doc);
}
else
{
indexingResult = ExtractIndexDataFromDocument(properties, doc);
}
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
var luceneDoc = new Document();
luceneDoc.Add(new Field(Constants.DocumentIdFieldName, indexingResult.NewDocId.ToLowerInvariant(), Field.Store.YES,
Field.Index.NOT_ANALYZED));
madeChanges = true;
CopyFieldsToDocument(luceneDoc, indexingResult.Fields);
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, indexingResult.NewDocId, luceneDoc));
logIndexing.DebugFormat("Index '{0}' resulted in: {1}", name, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
actions.Indexing.IncrementSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return(madeChanges);
});
logIndexing.DebugFormat("Indexed {0} documents for {1}", count, name);
}
public void ReduceDocuments(AbstractViewGenerator viewGenerator,
IEnumerable <object> mappedResults,
WorkContext context,
IStorageActionsAccessor actions,
string[] reduceKeys)
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var count = 0;
Write(context, (indexWriter, analyzer) =>
{
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
foreach (var reduceKey in reduceKeys)
{
var entryKey = reduceKey;
indexWriter.DeleteDocuments(new Term(Raven.Abstractions.Data.Constants.ReduceKeyFieldName, entryKey.ToLowerInvariant()));
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, entryKey);
context.AddError(name,
entryKey,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, entryKey));
}
PropertyDescriptorCollection properties = null;
foreach (var doc in RobustEnumerationReduce(mappedResults, viewGenerator.ReduceDefinition, actions, context))
{
count++;
var fields = GetFields(doc, ref properties).ToList();
string reduceKeyAsString = ExtractReduceKey(viewGenerator, doc);
var luceneDoc = new Document();
luceneDoc.Add(new Field(Raven.Abstractions.Data.Constants.ReduceKeyFieldName, reduceKeyAsString.ToLowerInvariant(), Field.Store.NO, Field.Index.NOT_ANALYZED));
foreach (var field in fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, reduceKeyAsString);
context.AddError(name,
reduceKeyAsString,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, reduceKeyAsString, luceneDoc));
logIndexing.DebugFormat("Reduce key {0} result in index {1} gave document: {2}", reduceKeyAsString, name, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
actions.Indexing.IncrementReduceSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return(true);
});
if (logIndexing.IsDebugEnabled)
{
logIndexing.DebugFormat("Reduce resulted in {0} entries for {1} for reduce keys: {2}", count, name, string.Join(", ", reduceKeys));
}
}
protected IEnumerable<object> RobustEnumeration(IEnumerable<object> input, IndexingFunc func, IStorageActionsAccessor actions, WorkContext context)
{
return new RobustEnumerator
{
BeforeMoveNext = actions.Indexing.IncrementIndexingAttempt,
CancelMoveNext = actions.Indexing.DecrementIndexingAttempt,
OnError = (exception, o) =>
{
actions.Indexing.IncrementIndexingFailure();
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnFormat(exception, "Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o));
}
}.RobustEnumeration(input, func);
}
public void ReduceDocuments(AbstractViewGenerator viewGenerator,
IEnumerable<object> mappedResults,
WorkContext context,
IStorageActionsAccessor actions,
string[] reduceKeys)
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var count = 0;
Write(context, indexWriter =>
{
var batchers = context.IndexUpdateTriggers.Select(x=>x.CreateBatcher(name))
.Where(x=>x!=null)
.ToList();
foreach (var reduceKey in reduceKeys)
{
var entryKey = reduceKey;
indexWriter.DeleteDocuments(new Term("__reduce_key", entryKey.ToLowerInvariant()));
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, entryKey);
context.AddError(name,
entryKey,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, entryKey));
}
PropertyDescriptorCollection properties = null;
foreach (var doc in RobustEnumeration(mappedResults, viewGenerator.ReduceDefinition, actions, context))
{
count++;
var fields = GetFields(doc, ref properties);
dynamic reduceKey = viewGenerator.GroupByExtraction(doc);
if (reduceKey == null)
{
throw new InvalidOperationException("Could not find reduce key for " + name + " in the result: " + doc);
}
string reduceKeyAsString = ReduceKeyToString(reduceKey);
var luceneDoc = new Document();
luceneDoc.Add(new Field("__reduce_key", reduceKeyAsString.ToLowerInvariant(), Field.Store.NO, Field.Index.NOT_ANALYZED));
foreach (var field in fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, reduceKeyAsString);
context.AddError(name,
reduceKeyAsString,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, reduceKeyAsString, luceneDoc));
logIndexing.DebugFormat("Reduce key {0} result in index {1} gave document: {2}", reduceKeyAsString, name, luceneDoc);
indexWriter.AddDocument(luceneDoc);
actions.Indexing.IncrementSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return true;
});
if (logIndexing.IsDebugEnabled)
{
logIndexing.DebugFormat("Reduce resulted in {0} entries for {1} for reduce keys: {2}", count, name, string.Join(", ", reduceKeys));
}
}
public override void Remove(string[] keys, WorkContext context)
{
DeletionBatchInfo deletionBatchInfo = null;
try
{
deletionBatchInfo = context.ReportDeletionBatchStarted(PublicName, keys.Length);
Write((writer, analyzer, stats) =>
{
var indexUpdateTriggersDuration = new Stopwatch();
stats.Operation = IndexingWorkStats.Status.Ignore;
if (logIndexing.IsDebugEnabled)
{
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), PublicName));
}
List <AbstractIndexUpdateTriggerBatcher> batchers;
using (StopwatchScope.For(indexUpdateTriggersDuration))
{
batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.DocumentIdFieldName, key.ToLowerInvariant())));
}
var deleteDocumentsDuration = new Stopwatch();
using (StopwatchScope.For(deleteDocumentsDuration))
{
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
}
deletionBatchInfo.PerformanceStats.Add(PerformanceStats.From(IndexingOperation.Delete_Documents, deleteDocumentsDuration.ElapsedMilliseconds));
using (StopwatchScope.For(indexUpdateTriggersDuration))
{
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger in " + PublicName, e);
context.AddError(indexId, PublicName, null, e, "Dispose Trigger");
},
batcher => batcher.Dispose());
}
deletionBatchInfo.PerformanceStats.Add(PerformanceStats.From(IndexingOperation.Delete_IndexUpdateTriggers, indexUpdateTriggersDuration.ElapsedMilliseconds));
return(new IndexedItemsInfo(GetLastEtagFromStats())
{
ChangedDocs = keys.Length,
DeletedKeys = keys
});
}, deletionBatchInfo.PerformanceStats);
}
finally
{
if (deletionBatchInfo != null)
{
context.ReportDeletionBatchCompleted(deletionBatchInfo);
}
}
}
protected IEnumerable<object> RobustEnumerationIndex(IEnumerable<object> input, IEnumerable<IndexingFunc> funcs,
IStorageActionsAccessor actions, WorkContext context, IndexingWorkStats stats)
{
return new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = () => stats.IndexingAttempts++,
CancelMoveNext = () => stats.IndexingAttempts--,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
stats.IndexingErrors++;
}
}.RobustEnumeration(input, funcs);
}
protected void Write(WorkContext context, Func<IndexWriter, bool> action)
{
if (disposed)
throw new ObjectDisposedException("Index " + name + " has been disposed");
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List<Action>();
Analyzer analyzer = null;
try
{
analyzer = CreateAnalyzer(toDispose);
var indexWriter = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
try
{
shouldRecreateSearcher = action(indexWriter);
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
finally
{
indexWriter.Close();
}
}
finally
{
if (analyzer != null)
analyzer.Close();
foreach (var dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
RecreateSearcher();
}
}
// This method may be called concurrently, by both the ReduceTask (for removal)
// and by the ReducingExecuter (for add/modify). This is okay with us, since the
// Write() call is already handling locking properly
public void ReduceDocuments(AbstractViewGenerator viewGenerator,
IEnumerable<object> mappedResults,
WorkContext context,
IStorageActionsAccessor actions,
string[] reduceKeys)
{
var count = 0;
Write(context, (indexWriter, analyzer) =>
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
foreach (var reduceKey in reduceKeys)
{
var entryKey = reduceKey;
indexWriter.DeleteDocuments(new Term(Abstractions.Data.Constants.ReduceKeyFieldName, entryKey.ToLowerInvariant()));
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, entryKey),
exception);
context.AddError(name,
entryKey,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, entryKey));
}
PropertyDescriptorCollection properties = null;
foreach (var doc in RobustEnumerationReduce(mappedResults, viewGenerator.ReduceDefinition, actions, context))
{
count++;
var fields = GetFields(doc, ref properties).ToList();
string reduceKeyAsString = ExtractReduceKey(viewGenerator, doc);
var luceneDoc = new Document();
luceneDoc.Add(new Field(Abstractions.Data.Constants.ReduceKeyFieldName, reduceKeyAsString.ToLowerInvariant(),
Field.Store.NO, Field.Index.NOT_ANALYZED_NO_NORMS));
foreach (var field in fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, reduceKeyAsString),
exception);
context.AddError(name,
reduceKeyAsString,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, reduceKeyAsString, luceneDoc));
logIndexing.Debug("Reduce key {0} result in index {1} gave document: {2}", reduceKeyAsString, name, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
actions.Indexing.IncrementReduceSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return true;
});
logIndexing.Debug(() => string.Format("Reduce resulted in {0} entries for {1} for reduce keys: {2}", count, name,
string.Join(", ", reduceKeys)));
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IndexingBatch batch, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
var count = 0;
var sourceCount = 0;
var sw = Stopwatch.StartNew();
var start = SystemTime.UtcNow;
Write((indexWriter, analyzer, stats) =>
{
var processedKeys = new HashSet<string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
try
{
var docIdTerm = new Term(Constants.DocumentIdFieldName);
var documentsWrapped = batch.Docs.Select((doc,i) =>
{
Interlocked.Increment(ref sourceCount);
if (doc.__document_id == null)
throw new ArgumentException(
string.Format("Cannot index something which doesn't have a document id, but got: '{0}'", doc));
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
return doc;
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId),
exception);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(documentId));
if (batch.SkipDeleteFromIndex[i] == false ||
context.ShouldRemoveFromIndex(documentId)) // maybe it is recently deleted?
indexWriter.DeleteDocuments(docIdTerm.CreateTerm(documentId.ToLowerInvariant()));
return doc;
})
.Where(x => x is FilteredDocument == false)
.ToList();
BackgroundTaskExecuter.Instance.ExecuteAllBuffered(context, documentsWrapped, (partition) =>
{
var anonymousObjectToLuceneDocumentConverter = new AnonymousObjectToLuceneDocumentConverter(indexDefinition);
var luceneDoc = new Document();
var documentIdField = new Field(Constants.DocumentIdFieldName, "dummy", Field.Store.YES,
Field.Index.NOT_ANALYZED_NO_NORMS);
foreach (var doc in RobustEnumerationIndex(partition, viewGenerator.MapDefinitions, actions, stats))
{
float boost;
var indexingResult = GetIndexingResult(doc, anonymousObjectToLuceneDocumentConverter, out boost);
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
Interlocked.Increment(ref count);
luceneDoc.GetFields().Clear();
luceneDoc.Boost = boost;
documentIdField.SetValue(indexingResult.NewDocId.ToLowerInvariant());
luceneDoc.Add(documentIdField);
foreach (var field in indexingResult.Fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId),
exception);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(indexingResult.NewDocId, luceneDoc));
LogIndexedDocument(indexingResult.NewDocId, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
Interlocked.Increment(ref stats.IndexingSuccesses);
}
});
}
catch(Exception e)
{
batchers.ApplyAndIgnoreAllErrors(
ex =>
{
logIndexing.WarnException("Failed to notify index update trigger batcher about an error", ex);
context.AddError(name, null, ex.Message);
},
x => x.AnErrorOccured(e));
throw;
}
finally
{
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
}
return sourceCount;
});
AddindexingPerformanceStat(new IndexingPerformanceStats
{
OutputCount = count,
InputCount = sourceCount,
Duration = sw.Elapsed,
Operation = "Index",
Started = start
});
logIndexing.Debug("Indexed {0} documents for {1}", count, name);
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IndexingBatch batch, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
var count = 0;
var sourceCount = 0;
var sw = Stopwatch.StartNew();
var start = SystemTime.UtcNow;
Write(context, (indexWriter, analyzer, stats) =>
{
var processedKeys = new HashSet <string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
try
{
var docIdTerm = new Term(Constants.DocumentIdFieldName);
var documentsWrapped = batch.Docs.Select((doc, i) =>
{
Interlocked.Increment(ref sourceCount);
if (doc.__document_id == null)
{
throw new ArgumentException(
string.Format("Cannot index something which doesn't have a document id, but got: '{0}'", doc));
}
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
{
return(doc);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId),
exception);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(documentId));
if (batch.SkipDeleteFromIndex[i] == false)
{
indexWriter.DeleteDocuments(docIdTerm.CreateTerm(documentId.ToLowerInvariant()));
}
return(doc);
})
.Where(x => x is FilteredDocument == false)
.ToList();
BackgroundTaskExecuter.Instance.ExecuteAllBuffered(context, documentsWrapped, (partition) =>
{
var anonymousObjectToLuceneDocumentConverter = new AnonymousObjectToLuceneDocumentConverter(indexDefinition);
var luceneDoc = new Document();
var documentIdField = new Field(Constants.DocumentIdFieldName, "dummy", Field.Store.YES,
Field.Index.NOT_ANALYZED_NO_NORMS);
foreach (var doc in RobustEnumerationIndex(partition, viewGenerator.MapDefinitions, actions, stats))
{
float boost;
var indexingResult = GetIndexingResult(doc, anonymousObjectToLuceneDocumentConverter, out boost);
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
Interlocked.Increment(ref count);
luceneDoc.GetFields().Clear();
luceneDoc.Boost = boost;
documentIdField.SetValue(indexingResult.NewDocId.ToLowerInvariant());
luceneDoc.Add(documentIdField);
foreach (var field in indexingResult.Fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId),
exception);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(indexingResult.NewDocId, luceneDoc));
LogIndexedDocument(indexingResult.NewDocId, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
Interlocked.Increment(ref stats.IndexingSuccesses);
}
});
}
catch (Exception e)
{
batchers.ApplyAndIgnoreAllErrors(
ex =>
{
logIndexing.WarnException("Failed to notify index update trigger batcher about an error", ex);
context.AddError(name, null, ex.Message);
},
x => x.AnErrorOccured(e));
throw;
}
finally
{
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
}
return(sourceCount);
});
AddindexingPerformanceStat(new IndexingPerformanceStats
{
OutputCount = count,
InputCount = sourceCount,
Duration = sw.Elapsed,
Operation = "Index",
Started = start
});
logIndexing.Debug("Indexed {0} documents for {1}", count, name);
}
// This method may be called concurrently, by both the ReduceTask (for removal)
// and by the ReducingExecuter (for add/modify). This is okay with us, since the
// Write() call is already handling locking properly
public void ReduceDocuments(AbstractViewGenerator viewGenerator,
IEnumerable <object> mappedResults,
WorkContext context,
IStorageActionsAccessor actions,
string[] reduceKeys)
{
var count = 0;
Write(context, (indexWriter, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Reduce;
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
foreach (var reduceKey in reduceKeys)
{
var entryKey = reduceKey;
indexWriter.DeleteDocuments(new Term(Abstractions.Data.Constants.ReduceKeyFieldName, entryKey.ToLowerInvariant()));
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, entryKey),
exception);
context.AddError(name,
entryKey,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(entryKey));
}
PropertyDescriptorCollection properties = null;
var anonymousObjectToLuceneDocumentConverter = new AnonymousObjectToLuceneDocumentConverter(indexDefinition);
var luceneDoc = new Document();
var reduceKeyField = new Field(Constants.ReduceKeyFieldName, "dummy",
Field.Store.NO, Field.Index.NOT_ANALYZED_NO_NORMS);
foreach (var doc in RobustEnumerationReduce(mappedResults, viewGenerator.ReduceDefinition, actions, context, stats))
{
count++;
float boost;
var fields = GetFields(anonymousObjectToLuceneDocumentConverter, doc, ref properties, out boost).ToList();
string reduceKeyAsString = ExtractReduceKey(viewGenerator, doc);
reduceKeyField.SetValue(reduceKeyAsString.ToLowerInvariant());
luceneDoc.GetFields().Clear();
luceneDoc.SetBoost(boost);
luceneDoc.Add(reduceKeyField);
foreach (var field in fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, reduceKeyAsString),
exception);
context.AddError(name,
reduceKeyAsString,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(reduceKeyAsString, luceneDoc));
LogIndexedDocument(reduceKeyAsString, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
stats.ReduceSuccesses++;
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return(count + reduceKeys.Length);
});
logIndexing.Debug(() => string.Format("Reduce resulted in {0} entries for {1} for reduce keys: {2}", count, name,
string.Join(", ", reduceKeys)));
}
public override void Remove(string[] keys, WorkContext context)
{
Write(context, (writer, analyzer,stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), name));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
keys.Apply(
key => batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, key),
exception);
context.AddError(name, key, exception.Message);
},
trigger => trigger.OnIndexEntryDeleted(key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
batcher => batcher.Dispose());
return keys.Length;
});
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IEnumerable <object> documents, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
var count = 0;
Write(context, (indexWriter, analyzer, stats) =>
{
var processedKeys = new HashSet <string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
var documentsWrapped = documents.Select((dynamic doc) =>
{
if (doc.__document_id == null)
{
throw new ArgumentException(string.Format("Cannot index something which doesn't have a document id, but got: '{0}'", doc));
}
count++;
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
{
return(doc);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId),
exception);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(documentId));
indexWriter.DeleteDocuments(new Term(Constants.DocumentIdFieldName, documentId.ToLowerInvariant()));
return(doc);
});
var anonymousObjectToLuceneDocumentConverter = new AnonymousObjectToLuceneDocumentConverter(indexDefinition);
var luceneDoc = new Document();
var documentIdField = new Field(Constants.DocumentIdFieldName, "dummy", Field.Store.YES, Field.Index.ANALYZED_NO_NORMS);
foreach (var doc in RobustEnumerationIndex(documentsWrapped, viewGenerator.MapDefinitions, actions, context, stats))
{
count++;
float boost;
var indexingResult = GetIndexingResult(doc, anonymousObjectToLuceneDocumentConverter, out boost);
if (indexingResult.NewDocId != null && indexingResult.ShouldSkip == false)
{
count += 1;
luceneDoc.GetFields().Clear();
luceneDoc.SetBoost(boost);
documentIdField.SetValue(indexingResult.NewDocId.ToLowerInvariant());
luceneDoc.Add(documentIdField);
foreach (var field in indexingResult.Fields)
{
luceneDoc.Add(field);
}
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnException(
string.Format("Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, indexingResult.NewDocId),
exception);
context.AddError(name,
indexingResult.NewDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(indexingResult.NewDocId, luceneDoc));
LogIndexedDocument(indexingResult.NewDocId, luceneDoc);
AddDocumentToIndex(indexWriter, luceneDoc, analyzer);
}
stats.IndexingSuccesses++;
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return(count);
});
logIndexing.Debug("Indexed {0} documents for {1}", count, name);
}
protected void Write(WorkContext context, Func<IndexWriter, Analyzer, bool> action)
{
if (disposed)
throw new ObjectDisposedException("Index " + name + " has been disposed");
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List<Action>();
Analyzer searchAnalyzer = null;
try
{
try
{
searchAnalyzer = CreateAnalyzer(new LowerCaseKeywordAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
{
indexWriter = CreateIndexWriter(directory);
}
try
{
shouldRecreateSearcher = action(indexWriter, searchAnalyzer);
foreach (IIndexExtension indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocuments);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
WriteTempIndexToDiskIfNeeded(context);
}
finally
{
currentlyIndexDocuments.Clear();
if (searchAnalyzer != null)
searchAnalyzer.Close();
foreach (Action dispose in toDispose)
{
dispose();
}
}
if (shouldRecreateSearcher)
RecreateSearcher();
}
}
public override void Remove(string[] keys, WorkContext context)
{
Write(context, writer =>
{
if (logIndexing.IsDebugEnabled)
{
logIndexing.DebugFormat("Deleting ({0}) from {1}", string.Format(", ", keys), name);
}
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
keys.Apply(
key => batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, key);
context.AddError(name, key, exception.Message );
},
trigger => trigger.OnIndexEntryDeleted(name, key)));
writer.DeleteDocuments(keys.Select(k => new Term("__document_id", k)).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message );
},
batcher => batcher.Dispose());
return true;
});
}
// we don't care about tracking map/reduce stats here, since it is merely
// an optimization step
protected IEnumerable<object> RobustEnumerationReduceDuringMapPhase(IEnumerable<object> input, IndexingFunc func,
IStorageActionsAccessor actions, WorkContext context)
{
// not strictly accurate, but if we get that many errors, probably an error anyway.
return new RobustEnumerator(context.Configuration.MaxNumberOfItemsToIndexInSingleBatch)
{
BeforeMoveNext = () => { }, // don't care
CancelMoveNext = () => { }, // don't care
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnException(
String.Format("Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o)),
exception);
}
}.RobustEnumeration(input, func);
}
public override void IndexDocuments(AbstractViewGenerator viewGenerator, IEnumerable<object> documents, WorkContext context, IStorageActionsAccessor actions, DateTime minimumTimestamp)
{
actions.Indexing.SetCurrentIndexStatsTo(name);
var count = 0;
Write(context, indexWriter =>
{
bool madeChanges = false;
PropertyDescriptorCollection properties = null;
var processedKeys = new HashSet<string>();
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(name))
.Where(x => x != null)
.ToList();
var documentsWrapped = documents.Select((dynamic doc) =>
{
string documentId = doc.__document_id.ToString();
if (processedKeys.Add(documentId) == false)
return doc;
madeChanges = true;
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryDeleted trigger for index '{0}', key: '{1}'",
name, documentId);
context.AddError(name,
documentId,
exception.Message
);
},
trigger => trigger.OnIndexEntryDeleted(name, documentId));
indexWriter.DeleteDocuments(new Term("__document_id", documentId.ToLowerInvariant()));
return doc;
});
foreach (var doc in RobustEnumeration(documentsWrapped, viewGenerator.MapDefinition, actions, context))
{
count++;
string newDocId;
IEnumerable<AbstractField> fields;
if (doc is DynamicJsonObject)
fields = ExtractIndexDataFromDocument((DynamicJsonObject)doc, out newDocId);
else
fields = ExtractIndexDataFromDocument(properties, doc, out newDocId);
if (newDocId != null)
{
var luceneDoc = new Document();
luceneDoc.Add(new Field("__document_id", newDocId.ToLowerInvariant(), Field.Store.YES, Field.Index.NOT_ANALYZED));
madeChanges = true;
CopyFieldsToDocument(luceneDoc, fields);
batchers.ApplyAndIgnoreAllErrors(
exception =>
{
logIndexing.WarnFormat(exception,
"Error when executed OnIndexEntryCreated trigger for index '{0}', key: '{1}'",
name, newDocId);
context.AddError(name,
newDocId,
exception.Message
);
},
trigger => trigger.OnIndexEntryCreated(name, newDocId, luceneDoc));
logIndexing.DebugFormat("Index '{0}' resulted in: {1}", name, luceneDoc);
indexWriter.AddDocument(luceneDoc);
}
actions.Indexing.IncrementSuccessIndexing();
}
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.Warn("Failed to dispose on index update trigger", e);
context.AddError(name, null, e.Message);
},
x => x.Dispose());
return madeChanges;
});
logIndexing.DebugFormat("Indexed {0} documents for {1}", count, name);
}
public override void Remove(string[] keys, WorkContext context)
{
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), indexId));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(
key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.DocumentIdFieldName, key)));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.DocumentIdFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger", e);
context.AddError(indexId, null, e.Message, "Dispose Trigger");
},
batcher => batcher.Dispose());
return new IndexedItemsInfo(GetLastEtagFromStats())
{
ChangedDocs = keys.Length,
DeletedKeys = keys
};
});
}
protected void Write(WorkContext context, Func<IndexWriter, Analyzer, IndexingWorkStats, int> action)
{
if (disposed)
throw new ObjectDisposedException("Index " + name + " has been disposed");
LastIndexTime = SystemTime.UtcNow;
lock (writeLock)
{
bool shouldRecreateSearcher;
var toDispose = new List<Action>();
Analyzer searchAnalyzer = null;
try
{
waitReason = "Write";
try
{
searchAnalyzer = CreateAnalyzer(new LowerCaseKeywordAnalyzer(), toDispose);
}
catch (Exception e)
{
context.AddError(name, "Creating Analyzer", e.ToString());
throw;
}
if (indexWriter == null)
{
indexWriter = CreateIndexWriter(directory);
}
var locker = directory.MakeLock("writing-to-index.lock");
try
{
var stats = new IndexingWorkStats();
try
{
var changedDocs = action(indexWriter, searchAnalyzer, stats);
docCountSinceLastOptimization += changedDocs;
shouldRecreateSearcher = changedDocs > 0;
foreach (IIndexExtension indexExtension in indexExtensions.Values)
{
indexExtension.OnDocumentsIndexed(currentlyIndexDocuments);
}
}
catch (Exception e)
{
context.AddError(name, null, e.ToString());
throw;
}
UpdateIndexingStats(context, stats);
WriteTempIndexToDiskIfNeeded(context);
Flush(); // just make sure changes are flushed to disk
}
finally
{
locker.Release();
}
}
finally
{
currentlyIndexDocuments.Clear();
if (searchAnalyzer != null)
searchAnalyzer.Close();
foreach (Action dispose in toDispose)
{
dispose();
}
waitReason = null;
LastIndexTime = SystemTime.UtcNow;
}
if (shouldRecreateSearcher)
RecreateSearcher();
}
}
public override void Remove(string[] keys, WorkContext context)
{
context.TransactionalStorage.Batch(actions =>
{
var reduceKeyAndBuckets = new Dictionary<ReduceKeyAndBucket, int>();
foreach (var key in keys)
{
actions.MapReduce.DeleteMappedResultsForDocumentId(key, indexId, reduceKeyAndBuckets);
}
actions.MapReduce.UpdateRemovedMapReduceStats(indexId, reduceKeyAndBuckets);
foreach (var reduceKeyAndBucket in reduceKeyAndBuckets)
{
actions.MapReduce.ScheduleReductions(indexId, 0, reduceKeyAndBucket.Key);
}
});
Write((writer, analyzer, stats) =>
{
stats.Operation = IndexingWorkStats.Status.Ignore;
if (logIndexing.IsDebugEnabled)
logIndexing.Debug(() => string.Format("Deleting ({0}) from {1}", string.Join(", ", keys), PublicName));
var batchers = context.IndexUpdateTriggers.Select(x => x.CreateBatcher(indexId))
.Where(x => x != null)
.ToList();
keys.Apply(
key =>
InvokeOnIndexEntryDeletedOnAllBatchers(batchers, new Term(Constants.ReduceKeyFieldName, key.ToLowerInvariant())));
writer.DeleteDocuments(keys.Select(k => new Term(Constants.ReduceKeyFieldName, k.ToLowerInvariant())).ToArray());
batchers.ApplyAndIgnoreAllErrors(
e =>
{
logIndexing.WarnException("Failed to dispose on index update trigger in " + PublicName, e);
context.AddError(indexId, PublicName, null, e, "Dispose Trigger");
},
batcher => batcher.Dispose());
return new IndexedItemsInfo(null)
{
ChangedDocs = keys.Length
};
});
}
protected IEnumerable<object> RobustEnumerationReduce(IEnumerable<object> input, IndexingFunc func,
IStorageActionsAccessor actions, WorkContext context)
{
return new RobustEnumerator
{
BeforeMoveNext = actions.Indexing.IncrementReduceIndexingAttempt,
CancelMoveNext = actions.Indexing.DecrementReduceIndexingAttempt,
OnError = (exception, o) =>
{
context.AddError(name,
TryGetDocKey(o),
exception.Message
);
logIndexing.WarnFormat(exception, "Failed to execute indexing function on {0} on {1}", name,
TryGetDocKey(o));
try
{
actions.Indexing.IncrementReduceIndexingFailure();
}
catch (Exception e)
{
// we don't care about error here, because it is an error on error problem
logIndexing.WarnFormat(e, "Could not increment indexing failure rate for {0}", name);
}
}
}.RobustEnumeration(input, func);
}
