|
public static CreateIndexWriter ( Lucene directory, bool create ) : |
||
| directory | Lucene | |
| create | bool | |
| return | ||
static string IndexBatch(string path, string connectionString, IDictionary <int, List <string> > packageFrameworks, int beginKey, int endKey)
{
var folder = string.Format(@"{0}\index_{1}_{2}", path, beginKey, endKey);
var directoryInfo = new DirectoryInfo(folder);
directoryInfo.Create();
using (var connection = new SqlConnection(connectionString))
{
connection.Open();
var cmdText = @"
SELECT
Packages.[Key] 'key',
PackageRegistrations.Id 'id',
Packages.[Version] 'verbatimVersion',
Packages.NormalizedVersion 'version',
Packages.Title 'title',
Packages.Tags 'tags',
Packages.[Description] 'description',
Packages.FlattenedAuthors 'authors',
Packages.Summary 'summary',
Packages.IconUrl 'iconUrl',
Packages.ProjectUrl 'projectUrl',
Packages.MinClientVersion 'minClientVersion',
Packages.ReleaseNotes 'releaseNotes',
Packages.Copyright 'copyright',
Packages.[Language] 'language',
Packages.LicenseUrl 'licenseUrl',
Packages.RequiresLicenseAcceptance 'requireLicenseAcceptance',
Packages.[Hash] 'packageHash',
Packages.HashAlgorithm 'packageHashAlgorithm',
Packages.PackageFileSize 'packageSize',
Packages.FlattenedDependencies 'flattenedDependencies',
Packages.Created 'created',
Packages.LastEdited 'lastEdited',
Packages.Published 'published',
Packages.Listed 'listed',
Packages.SemVerLevelKey 'semVerLevelKey'
FROM Packages
INNER JOIN PackageRegistrations ON Packages.PackageRegistrationKey = PackageRegistrations.[Key]
AND Packages.[Key] >= @BeginKey
AND Packages.[Key] < @EndKey
WHERE Packages.PackageStatusKey = 0
ORDER BY Packages.[Key]
";
var command = new SqlCommand(cmdText, connection);
command.CommandTimeout = (int)TimeSpan.FromMinutes(15).TotalSeconds;
command.Parameters.AddWithValue("BeginKey", beginKey);
command.Parameters.AddWithValue("EndKey", endKey);
var reader = command.ExecuteReader();
var batch = 0;
var directory = new SimpleFSDirectory(directoryInfo);
using (var writer = DocumentCreator.CreateIndexWriter(directory, true))
{
while (reader.Read())
{
var document = CreateDocument(reader, packageFrameworks);
writer.AddDocument(document);
if (batch++ == 1000)
{
writer.Commit();
batch = 0;
}
}
if (batch > 0)
{
writer.Commit();
}
}
}
return(folder);
}
public static void Export(string sourceConnectionString, Uri catalogIndexUrl, string destinationPath, ILoggerFactory loggerFactory)
{
var logger = loggerFactory.CreateLogger <Sql2Lucene>();
var stopwatch = new Stopwatch();
// Get the commit timestamp from catalog index page for lucene index
var initTime = GetCommitTimestampFromCatalogAsync(catalogIndexUrl, logger).Result;
stopwatch.Start();
var batches = CalculateBatches(sourceConnectionString);
logger.LogInformation("Calculated {BatchCount} batches (took {BatchCalculationTime} seconds)", batches.Count, stopwatch.Elapsed.TotalSeconds);
stopwatch.Restart();
var packageFrameworks = LoadPackageFrameworks(sourceConnectionString);
logger.LogInformation("Loaded package frameworks (took {PackageFrameworksLoadTime} seconds)", stopwatch.Elapsed.TotalSeconds);
stopwatch.Restart();
var tasks = new List <Task <string> >();
foreach (var batch in batches)
{
tasks.Add(Task.Run(() => { return(IndexBatch(destinationPath + @"\batches", sourceConnectionString, packageFrameworks, batch.Item1, batch.Item2)); }));
}
try
{
Task.WaitAll(tasks.ToArray());
}
catch (AggregateException ex)
{
logger.LogError("An AggregateException occurred while running batches.", ex);
throw;
}
logger.LogInformation("Partition indexes generated (took {PartitionIndexGenerationTime} seconds", stopwatch.Elapsed.TotalSeconds);
stopwatch.Restart();
using (var directory = new SimpleFSDirectory(new DirectoryInfo(destinationPath)))
{
using (var writer = DocumentCreator.CreateIndexWriter(directory, true))
{
NuGetMergePolicyApplyer.ApplyTo(writer);
var partitions = tasks.Select(t => new SimpleFSDirectory(new DirectoryInfo(t.Result))).ToArray();
writer.AddIndexesNoOptimize(partitions);
foreach (var partition in partitions)
{
partition.Dispose();
}
writer.Commit(DocumentCreator.CreateCommitMetadata(initTime, "from SQL", writer.NumDocs(), Guid.NewGuid().ToString())
.ToDictionary());
}
}
logger.LogInformation("Sql2Lucene.Export done (took {Sql2LuceneExportTime} seconds)", stopwatch.Elapsed.TotalSeconds);
stopwatch.Reset();
}