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(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 Tuple<IndexToWorkOn, IndexingBatch>[indexesToWorkOn.Count];
var actions = new Action<IStorageActionsAccessor>[indexesToWorkOn.Count];
BackgroundTaskExecuter.Instance.ExecuteAll(context, 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)
{
if (FilterDocuments(item.Doc))
continue;
// did we already indexed this document in this index?
var etag = item.Doc.Etag;
if (etag == null)
continue;
if (indexLastInedexEtag.CompareTo(new ComparableByteArray(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;
}
if (Log.IsDebugEnabled)
{
Log.Debug("Going to index {0} documents in {1}: ({2})", batch.Ids.Count, indexToWorkOn, string.Join(", ", batch.Ids));
}
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 IEnumerable<Tuple<IndexToWorkOn, IEnumerable<JsonDocument>>> FilterIndexes(IEnumerable<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());
Action<IStorageActionsAccessor> action = null;
foreach (var indexToWorkOn in indexesToWorkOn)
{
var indexLastInedexEtag = new ComparableByteArray(indexToWorkOn.LastIndexedEtag.ToByteArray());
if (indexLastInedexEtag.CompareTo(lastIndexedEtag) >= 0)
continue;
var filteredDocs = jsonDocs.Where(doc => indexLastInedexEtag.CompareTo(new ComparableByteArray(doc.Etag.Value.ToByteArray())) < 0);
var indexName = indexToWorkOn.IndexName;
var viewGenerator = context.IndexDefinitionStorage.GetViewGenerator(indexName);
if(viewGenerator == null)
continue; // probably deleted
if (viewGenerator.ForEntityNames.Count != 0) // limit for the items that it care for
{
filteredDocs = filteredDocs.Where(x => viewGenerator.ForEntityNames.Contains(x.Metadata.Value<string>(Constants.RavenEntityName)));
}
List<JsonDocument> jsonDocuments = filteredDocs.ToList();
if(jsonDocuments.Count == 0)
{
// we use it this way to batch all the updates together
action += accessor => accessor.Indexing.UpdateLastIndexed(indexName, lastEtag, lastModified);
continue;
}
yield return Tuple.Create<IndexToWorkOn, IEnumerable<JsonDocument>>(indexToWorkOn, jsonDocuments);
}
if (action != null)
{
transactionalStorage.Batch(action);
}
}