// this function will incrementally build an index from the gallery using a high water mark stored in the commit metadata
// this function is useful for building a fresh index as in that case it is more efficient than diff-ing approach
public static void RebuildIndex(string sqlConnectionString, Lucene.Net.Store.Directory directory, FrameworksList frameworks, TextWriter log = null, PerfEventTracker perfTracker = null)
{
perfTracker = perfTracker ?? new PerfEventTracker();
log = log ?? DefaultTraceWriter;
Stopwatch sw = new Stopwatch();
sw.Start();
using (perfTracker.TrackEvent("RebuildIndex", String.Empty))
{
// Empty the index, we're rebuilding
CreateNewEmptyIndex(directory);
var projectFxs = frameworks.Load();
log.WriteLine("get curated feeds by PackageRegistration");
IDictionary <int, IEnumerable <string> > feeds = GalleryExport.GetFeedsByPackageRegistration(sqlConnectionString, log, verbose: false);
int highestPackageKey = 0;
while (true)
{
log.WriteLine("get the checksums from the gallery");
IDictionary <int, int> checksums = GalleryExport.FetchGalleryChecksums(sqlConnectionString, highestPackageKey);
log.WriteLine("get packages from gallery where the Package.Key > {0}", highestPackageKey);
List <Package> packages = GalleryExport.GetPublishedPackagesSince(sqlConnectionString, highestPackageKey, log, verbose: false);
if (packages.Count == 0)
{
break;
}
log.WriteLine("associate the feeds and checksum data with each packages");
List <IndexDocumentData> indexDocumentData = MakeIndexDocumentData(packages, feeds, checksums);
highestPackageKey = indexDocumentData.Max(d => d.Package.Key);
AddPackagesToIndex(indexDocumentData, directory, log, projectFxs, perfTracker);
// Summarize performance
// (Save some time by not bothering if the log is "null")
if (!ReferenceEquals(TextWriter.Null, log) && !ReferenceEquals(PerfEventTracker.Null, perfTracker))
{
SummarizePerf(log, perfTracker);
}
}
}
SummarizePerf(log, perfTracker);
sw.Stop();
log.WriteLine("all done, took {0}", sw.Elapsed);
}
public override void Execute()
{
var manager = GetSearcherManager();
IDictionary <int, int> databaseChecksums = GalleryExport.FetchGalleryChecksums(SqlConnectionString);
Log.WriteLine("fetched {0} keys from database", databaseChecksums.Count);
Tuple <int, int> minMax = GalleryExport.FindMinMaxKey(databaseChecksums);
Log.WriteLine("min = {0}, max = {1}", minMax.Item1, minMax.Item2);
// For now, use the in-memory Searcher client. But eventually this will use the original Search Service call below
IDictionary <int, int> index = ParseRangeResult(
Searcher.KeyRangeQuery(manager, minMax.Item1, minMax.Item2));
Log.WriteLine("fetched {0} keys from index", index.Count);
List <int> adds = new List <int>();
List <int> updates = new List <int>();
List <int> deletes = new List <int>();
SortIntoAddsUpdateDeletes(databaseChecksums, index, adds, updates, deletes);
Log.WriteLine("{0} adds", adds.Count);
Log.WriteLine("{0} updates", updates.Count);
Log.WriteLine("{0} deletes", deletes.Count);
if (adds.Count == 0 && updates.Count == 0)
{
return;
}
IDictionary <int, IEnumerable <string> > feeds = GalleryExport.GetFeedsByPackageRegistration(SqlConnectionString, Log, verbose: false);
IDictionary <int, IndexDocumentData> packages = PackageIndexing.LoadDocumentData(SqlConnectionString, adds, updates, deletes, feeds, databaseChecksums, Log);
Lucene.Net.Store.Directory directory = manager.Directory;
Func <int, IndexDocumentData> packageFetcher = (key) =>
{
IndexDocumentData knownDoc;
if (packages.TryGetValue(key, out knownDoc))
{
return(knownDoc);
}
else
{
// We're modifying a different document
var pkgs = GalleryExport.GetPackages(SqlConnectionString, new List <int>()
{
key
}, Log, verbose: false);
var docs = PackageIndexing.MakeIndexDocumentData(pkgs, feeds, databaseChecksums);
packages[key] = docs[0];
return(docs[0]);
}
};
var perfTracker = new PerfEventTracker();
PackageIndexing.UpdateIndex(WhatIf, adds, updates, deletes, packageFetcher, directory, Log, perfTracker, manager.Frameworks.Load());
}