private IEnumerable<IndexingBatchForIndex> FilterIndexes(IList<IndexToWorkOn> indexesToWorkOn, List<JsonDocument> jsonDocs, Etag highestETagInBatch)
{
var last = jsonDocs.Last();
Debug.Assert(last.Etag != null);
Debug.Assert(last.LastModified != null);
var lastEtag = last.Etag;
var lastModified = last.LastModified.Value;
var documentRetriever = new DocumentRetriever(null, context.ReadTriggers, context.Database.InFlightTransactionalState);
var filteredDocs =
BackgroundTaskExecuter.Instance.Apply(context, jsonDocs, doc =>
{
var filteredDoc = documentRetriever.ExecuteReadTriggers(doc, null, ReadOperation.Index);
return filteredDoc == null ? new
{
Doc = doc,
Json = (object)new FilteredDocument(doc)
} : new
{
Doc = filteredDoc,
Json = JsonToExpando.Convert(doc.ToJson())
};
});
Log.Debug("After read triggers executed, {0} documents remained", filteredDocs.Count);
var results = new IndexingBatchForIndex[indexesToWorkOn.Count];
var actions = new Action<IStorageActionsAccessor>[indexesToWorkOn.Count];
BackgroundTaskExecuter.Instance.ExecuteAll(context, indexesToWorkOn, (indexToWorkOn, i) =>
{
var indexName = indexToWorkOn.IndexName;
var viewGenerator = context.IndexDefinitionStorage.GetViewGenerator(indexName);
if (viewGenerator == null)
return; // probably deleted
var batch = new IndexingBatch(highestETagInBatch);
foreach (var item in filteredDocs)
{
if (prefetchingBehavior.FilterDocuments(item.Doc) == false)
continue;
// did we already indexed this document in this index?
var etag = item.Doc.Etag;
if (etag == null)
continue;
// is the Raven-Entity-Name a match for the things the index executes on?
if (viewGenerator.ForEntityNames.Count != 0 &&
viewGenerator.ForEntityNames.Contains(item.Doc.Metadata.Value<string>(Constants.RavenEntityName)) == false)
{
continue;
}
batch.Add(item.Doc, item.Json, prefetchingBehavior.ShouldSkipDeleteFromIndex(item.Doc));
if (batch.DateTime == null)
batch.DateTime = item.Doc.LastModified;
else
batch.DateTime = batch.DateTime > item.Doc.LastModified
? item.Doc.LastModified
: batch.DateTime;
}
if (batch.Docs.Count == 0)
{
Log.Debug("All documents have been filtered for {0}, no indexing will be performed, updating to {1}, {2}", indexName,
lastEtag, lastModified);
// we use it this way to batch all the updates together
actions[i] = accessor => accessor.Indexing.UpdateLastIndexed(indexName, lastEtag, lastModified);
return;
}
if (Log.IsDebugEnabled)
{
Log.Debug("Going to index {0} documents in {1}: ({2})", batch.Ids.Count, indexToWorkOn, string.Join(", ", batch.Ids));
}
results[i] = new IndexingBatchForIndex
{
Batch = batch,
IndexName = indexToWorkOn.IndexName,
Index = indexToWorkOn.Index,
LastIndexedEtag = indexToWorkOn.LastIndexedEtag
};
});
transactionalStorage.Batch(actionsAccessor =>
{
foreach (var action in actions)
{
if (action != null)
action(actionsAccessor);
}
});
return results.Where(x => x != null);
}
private IEnumerable<Tuple<IndexToWorkOn, IndexingBatch>> FilterIndexes(IList<IndexToWorkOn> indexesToWorkOn, JsonDocument[] jsonDocs)
{
var last = jsonDocs.Last();
Debug.Assert(last.Etag != null);
Debug.Assert(last.LastModified != null);
var lastEtag = last.Etag.Value;
var lastModified = last.LastModified.Value;
var lastIndexedEtag = new ComparableByteArray(lastEtag.ToByteArray());
var documentRetriever = new DocumentRetriever(null, context.ReadTriggers);
var filteredDocs =
BackgroundTaskExecuter.Instance.Apply(jsonDocs, doc =>
{
doc = documentRetriever.ExecuteReadTriggers(doc, null, ReadOperation.Index);
return doc == null ? null : new {Doc = doc, Json = JsonToExpando.Convert(doc.ToJson())};
});
log.Debug("After read triggers executed, {0} documents remained", filteredDocs.Count);
var results = new Tuple<IndexToWorkOn, IndexingBatch>[indexesToWorkOn.Count];
var actions = new Action<IStorageActionsAccessor>[indexesToWorkOn.Count];
BackgroundTaskExecuter.Instance.ExecuteAll(context.Configuration, scheduler, indexesToWorkOn, (indexToWorkOn, i) =>
{
var indexLastInedexEtag = new ComparableByteArray(indexToWorkOn.LastIndexedEtag.ToByteArray());
if (indexLastInedexEtag.CompareTo(lastIndexedEtag) >= 0)
return;
var indexName = indexToWorkOn.IndexName;
var viewGenerator = context.IndexDefinitionStorage.GetViewGenerator(indexName);
if (viewGenerator == null)
return; // probably deleted
var batch = new IndexingBatch();
foreach (var item in filteredDocs)
{
// did we already indexed this document in this index?
if (indexLastInedexEtag.CompareTo(new ComparableByteArray(item.Doc.Etag.Value.ToByteArray())) >= 0)
continue;
// is the Raven-Entity-Name a match for the things the index executes on?
if (viewGenerator.ForEntityNames.Count != 0 &&
viewGenerator.ForEntityNames.Contains(item.Doc.Metadata.Value<string>(Constants.RavenEntityName)) == false)
{
continue;
}
batch.Add(item.Doc, item.Json);
if (batch.DateTime == null)
batch.DateTime = item.Doc.LastModified;
else
batch.DateTime = batch.DateTime > item.Doc.LastModified
? item.Doc.LastModified
: batch.DateTime;
}
if (batch.Docs.Count == 0)
{
log.Debug("All documents have been filtered for {0}, no indexing will be performed, updating to {1}, {2}", indexName,
lastEtag, lastModified);
// we use it this way to batch all the updates together
actions[i] = accessor => accessor.Indexing.UpdateLastIndexed(indexName, lastEtag, lastModified);
return;
}
log.Debug("Going to index {0} documents in {1}", batch.Ids.Count, indexToWorkOn);
results[i] = Tuple.Create(indexToWorkOn, batch);
});
transactionalStorage.Batch(actionsAccessor =>
{
foreach (var action in actions)
{
if (action != null)
action(actionsAccessor);
}
});
return results.Where(x => x != null);
}
private void IndexDocuments(IStorageActionsAccessor actions, string index, JsonDocument[] jsonDocs)
{
var viewGenerator = context.IndexDefinitionStorage.GetViewGenerator(index);
if (viewGenerator == null)
return; // index was deleted, probably
var dateTime = jsonDocs.Min(x => x.LastModified) ?? DateTime.MinValue;
var documentRetriever = new DocumentRetriever(null, context.ReadTriggers);
try
{
log.DebugFormat("Indexing {0} documents for index: {1}", jsonDocs.Length, index);
context.IndexStorage.Index(index, viewGenerator,
jsonDocs
.Select(doc => documentRetriever
.ExecuteReadTriggers(doc, null, ReadOperation.Index))
.Where(doc => doc != null)
.Select(x => JsonToExpando.Convert(x.ToJson())), context, actions, dateTime);
}
catch (Exception e)
{
if (actions.IsWriteConflict(e))
return;
log.WarnFormat(e, "Failed to index documents for index: {0}", index);
}
}
public void GetDocuments(int start, int pageSize, Etag etag, CancellationToken token, Action<RavenJObject> addDocument)
{
TransactionalStorage.Batch(actions =>
{
bool returnedDocs = false;
while (true)
{
var documents = etag == null
? actions.Documents.GetDocumentsByReverseUpdateOrder(start, pageSize)
: actions.Documents.GetDocumentsAfter(etag, pageSize, WorkContext.CancellationToken);
var documentRetriever = new DocumentRetriever(actions, Database.ReadTriggers, Database.InFlightTransactionalState);
int docCount = 0;
foreach (var doc in documents)
{
docCount++;
token.ThrowIfCancellationRequested();
if (etag != null)
etag = doc.Etag;
DocumentRetriever.EnsureIdInMetadata(doc);
var nonAuthoritativeInformationBehavior = Database.InFlightTransactionalState.GetNonAuthoritativeInformationBehavior<JsonDocument>(null, doc.Key);
var document = nonAuthoritativeInformationBehavior == null ? doc : nonAuthoritativeInformationBehavior(doc);
document = documentRetriever
.ExecuteReadTriggers(document, null, ReadOperation.Load);
if (document == null)
continue;
addDocument(document.ToJson());
returnedDocs = true;
}
if (returnedDocs || docCount == 0)
break;
start += docCount;
}
});
}
public void GetDocumentsWithIdStartingWith(string idPrefix, string matches, string exclude, int start, int pageSize,
CancellationToken token, ref int nextStart, Action<RavenJObject> addDoc,
string transformer = null, Dictionary<string, RavenJToken> transformerParameters = null,
string skipAfter = null)
{
if (idPrefix == null)
throw new ArgumentNullException("idPrefix");
idPrefix = idPrefix.Trim();
var canPerformRapidPagination = nextStart > 0 && start == nextStart;
var actualStart = canPerformRapidPagination ? start : 0;
var addedDocs = 0;
var matchedDocs = 0;
TransactionalStorage.Batch(
actions =>
{
var docsToSkip = canPerformRapidPagination ? 0 : start;
int docCount;
AbstractTransformer storedTransformer = null;
if (transformer != null)
{
storedTransformer = IndexDefinitionStorage.GetTransformer(transformer);
if (storedTransformer == null)
throw new InvalidOperationException("No transformer with the name: " + transformer);
}
do
{
docCount = 0;
var docs = actions.Documents.GetDocumentsWithIdStartingWith(idPrefix, actualStart, pageSize, string.IsNullOrEmpty(skipAfter) ? null : skipAfter);
var documentRetriever = new DocumentRetriever(actions, Database.ReadTriggers, Database.InFlightTransactionalState, transformerParameters);
foreach (var doc in docs)
{
token.ThrowIfCancellationRequested();
docCount++;
var keyTest = doc.Key.Substring(idPrefix.Length);
if (!WildcardMatcher.Matches(matches, keyTest) || WildcardMatcher.MatchesExclusion(exclude, keyTest))
continue;
DocumentRetriever.EnsureIdInMetadata(doc);
var nonAuthoritativeInformationBehavior = Database.InFlightTransactionalState.GetNonAuthoritativeInformationBehavior<JsonDocument>(null, doc.Key);
var document = nonAuthoritativeInformationBehavior != null ? nonAuthoritativeInformationBehavior(doc) : doc;
document = documentRetriever.ExecuteReadTriggers(document, null, ReadOperation.Load);
if (document == null)
continue;
matchedDocs++;
if (matchedDocs <= docsToSkip)
continue;
token.ThrowIfCancellationRequested();
if (storedTransformer != null)
{
using (new CurrentTransformationScope(Database, documentRetriever))
{
var transformed =
storedTransformer.TransformResultsDefinition(new[] { new DynamicJsonObject(document.ToJson()) })
.Select(x => JsonExtensions.ToJObject(x))
.ToArray();
if (transformed.Length == 0)
{
throw new InvalidOperationException("The transform results function failed on a document: " + document.Key);
}
var transformedJsonDocument = new JsonDocument
{
Etag = document.Etag.HashWith(storedTransformer.GetHashCodeBytes()).HashWith(documentRetriever.Etag),
NonAuthoritativeInformation = document.NonAuthoritativeInformation,
LastModified = document.LastModified,
DataAsJson = new RavenJObject { { "$values", new RavenJArray(transformed) } },
};
addDoc(transformedJsonDocument.ToJson());
}
}
else
{
addDoc(document.ToJson());
}
addedDocs++;
if (addedDocs >= pageSize)
break;
}
actualStart += pageSize;
}
while (docCount > 0 && addedDocs < pageSize && actualStart > 0 && actualStart < int.MaxValue);
});
if (addedDocs != pageSize)
nextStart = start; // will mark as last page
else if (canPerformRapidPagination)
nextStart = start + matchedDocs;
else
nextStart = actualStart;
}
public Etag GetDocumentsWithIdStartingWith(string idPrefix, int pageSize, Etag etag, CancellationToken token, Func<JsonDocument, bool> addDocument)
{
Etag lastDocumentReadEtag = null;
TransactionalStorage.Batch(actions =>
{
bool returnedDocs = false;
while (true)
{
var documents = actions.Documents.GetDocumentsAfterWithIdStartingWith(etag, idPrefix, pageSize, token, timeout: TimeSpan.FromSeconds(2), lastProcessedDocument: x => lastDocumentReadEtag = x );
var documentRetriever = new DocumentRetriever(Database.Configuration, actions, Database.ReadTriggers);
int docCount = 0;
foreach (var doc in documents)
{
docCount++;
token.ThrowIfCancellationRequested();
etag = doc.Etag;
JsonDocument.EnsureIdInMetadata(doc);
var nonAuthoritativeInformationBehavior = actions.InFlightStateSnapshot.GetNonAuthoritativeInformationBehavior<JsonDocument>(null, doc.Key);
var document = nonAuthoritativeInformationBehavior == null ? doc : nonAuthoritativeInformationBehavior(doc);
document = documentRetriever.ExecuteReadTriggers(document, null, ReadOperation.Load);
if (document == null)
continue;
returnedDocs = true;
Database.WorkContext.UpdateFoundWork();
bool canContinue = addDocument(document);
if (!canContinue)
break;
}
if (returnedDocs)
break;
// No document was found that matches the requested criteria
if ( docCount == 0 )
{
// If we had a failure happen, we update the etag as we don't need to process those documents again (no matches there anyways).
if (lastDocumentReadEtag != null)
etag = lastDocumentReadEtag;
break;
}
}
});
return etag;
}
public Etag GetDocuments(int start, int pageSize, Etag etag, CancellationToken token, Func<JsonDocument, bool> addDocument)
{
Etag lastDocumentReadEtag = null;
TransactionalStorage.Batch(actions =>
{
bool returnedDocs = false;
while (true)
{
var documents = etag == null
? actions.Documents.GetDocumentsByReverseUpdateOrder(start, pageSize)
: actions.Documents.GetDocumentsAfter(etag, pageSize, token);
var documentRetriever = new DocumentRetriever(Database.Configuration, actions, Database.ReadTriggers, Database.InFlightTransactionalState);
int docCount = 0;
foreach (var doc in documents)
{
docCount++;
token.ThrowIfCancellationRequested();
if (etag != null)
etag = doc.Etag;
JsonDocument.EnsureIdInMetadata(doc);
var nonAuthoritativeInformationBehavior = Database.InFlightTransactionalState.GetNonAuthoritativeInformationBehavior<JsonDocument>(null, doc.Key);
var document = nonAuthoritativeInformationBehavior == null ? doc : nonAuthoritativeInformationBehavior(doc);
document = documentRetriever.ExecuteReadTriggers(document, null, ReadOperation.Load);
if (document == null)
continue;
returnedDocs = true;
Database.WorkContext.UpdateFoundWork();
bool canContinue = addDocument(document);
if (!canContinue)
break;
lastDocumentReadEtag = etag;
}
if (returnedDocs || docCount == 0)
break;
start += docCount;
}
});
return lastDocumentReadEtag;
}
public Etag GetDocuments(int start, int pageSize, Etag etag, CancellationToken token,
Func<JsonDocument, bool> addDocument, long? maxSize = null, TimeSpan? timeout = null)
{
Etag lastDocumentReadEtag = null;
using (DocumentCacher.SkipSetDocumentsInDocumentCache())
TransactionalStorage.Batch(actions =>
{
bool returnedDocs = false;
while (true)
{
var documents = etag == null
? actions.Documents.GetDocumentsByReverseUpdateOrder(start, pageSize)
: actions.Documents.GetDocumentsAfter(etag, pageSize, token, maxSize: maxSize, timeout: timeout);
var documentRetriever = new DocumentRetriever(Database.Configuration, actions, Database.ReadTriggers);
var docCount = 0;
foreach (var doc in documents)
{
docCount++;
token.ThrowIfCancellationRequested();
if (etag != null)
etag = doc.Etag;
JsonDocument.EnsureIdInMetadata(doc);
var nonAuthoritativeInformationBehavior = actions.InFlightStateSnapshot.GetNonAuthoritativeInformationBehavior<JsonDocument>(null, doc.Key);
var document = nonAuthoritativeInformationBehavior == null ? doc : nonAuthoritativeInformationBehavior(doc);
document = documentRetriever.ExecuteReadTriggers(document, null, ReadOperation.Load);
if (document == null)
continue;
returnedDocs = true;
Database.WorkContext.UpdateFoundWork();
bool canContinue = addDocument(document);
if (!canContinue)
break;
lastDocumentReadEtag = etag;
}
if (returnedDocs || docCount == 0)
break;
// No document was found that matches the requested criteria
// If we had a failure happen, we update the etag as we don't need to process those documents again (no matches there anyways).
if (lastDocumentReadEtag != null)
etag = lastDocumentReadEtag;
start += docCount;
}
});
return lastDocumentReadEtag;
}
