/// <summary>
/// Factory method to create a <see cref="TagServer"/>, uses the private Constructor <see cref="TagServer(List{Question})" />
/// </summary>
public static TagServer CreateFromScratchAndSaveToDisk(List <Question> rawQuestions, string intermediateFilesFolder)
{
var tagServer = new TagServer(rawQuestions);
var serializeTimer = Stopwatch.StartNew();
Logger.LogStartupMessage("Serialisation folder: {0}", intermediateFilesFolder);
foreach (QueryType type in (QueryType[])Enum.GetValues(typeof(QueryType)))
{
var tagLookupFileName = "intermediate-Lookup-" + type + ".bin";
Serialisation.SerialiseToDisk(tagLookupFileName, intermediateFilesFolder, tagServer.GetTagLookupForQueryType(type));
var bitMapIndex = tagServer.GetTagBitMapIndexForQueryType(type);
if (bitMapIndex.Count == 0)
{
continue;
}
var bitMapIndexFileName = String.Format("intermediate-BitMap-{0}.bin", type);
Serialisation.SerialiseBitMapIndexToDisk(bitMapIndexFileName, intermediateFilesFolder, bitMapIndex);
// Sanity-check, de-serialise the data we've just written to disk
Serialisation.DeserialiseFromDisk(bitMapIndexFileName, intermediateFilesFolder);
Logger.LogStartupMessage();
}
// Now write out the AllTags Lookup, Tag -> Count (i.e. "C#" -> 579,321, "Java" -> 560,432)
Serialisation.SerialiseToDisk(AllTagsFileName, intermediateFilesFolder, tagServer.AllTags);
serializeTimer.Stop();
Logger.LogStartupMessage("\nTook {0} (in TOTAL) to serialise the intermediate data TO disk\n", serializeTimer.Elapsed);
return(tagServer);
}
/// <summary>
/// Factory method to create a <see cref="TagServer"/>, uses the private Constructor
/// <see cref="TagServer(List{Question}, TagLookup, Dictionary{QueryType, TagByQueryLookup}, Dictionary{QueryType, TagByQueryBitMapLookup})"/>
/// </summary>
public static TagServer CreateFromSerialisedData(List <Question> rawQuestions, string intermediateFilesFolder, bool deserialiseBitMapsIndexes = true)
{
var deserializeTimer = Stopwatch.StartNew();
Logger.LogStartupMessage("Deserialisation folder: {0}", intermediateFilesFolder);
var queryTypes = (QueryType[])Enum.GetValues(typeof(QueryType));
var intermediateLookups = new Dictionary <QueryType, TagByQueryLookup>(queryTypes.Length);
var intermediateBitMapIndexes = new Dictionary <QueryType, TagByQueryBitMapLookup>(queryTypes.Length);
foreach (QueryType type in queryTypes)
{
if (deserialiseBitMapsIndexes)
{
var tagLookupFileName = "intermediate-Lookup-" + type + ".bin";
var tempLookup = Serialisation.DeserialiseFromDisk <TagByQueryLookup>(tagLookupFileName, intermediateFilesFolder);
Logger.LogStartupMessage("{0,20} contains {1:N0} Tag Lookups", type, tempLookup.Count);
intermediateLookups.Add(type, tempLookup);
var bitMapIndexFileName = String.Format("intermediate-BitMap-{0}.bin", type);
var tempBitMapIndexes = Serialisation.DeserialiseFromDisk(bitMapIndexFileName, intermediateFilesFolder);
Logger.LogStartupMessage("{0,20} contains {1:N0} Tag BitMap Indexes", type, tempBitMapIndexes.Count);
intermediateBitMapIndexes.Add(type, tempBitMapIndexes);
Logger.LogStartupMessage();
}
else
{
// Maybe don't check this in, it's just here to save time when we don't need to real data!!!
var tempLookup = new TagByQueryLookup();
intermediateLookups.Add(type, tempLookup);
var tempBitMapIndexes = new TagByQueryBitMapLookup();
intermediateBitMapIndexes.Add(type, tempBitMapIndexes);
}
}
// Now fetch from disk the AllTags Lookup, Tag -> Count (i.e. "C#" -> 579,321, "Java" -> 560,432)
var allTags = Serialisation.DeserialiseFromDisk <TagLookup>(AllTagsFileName, intermediateFilesFolder);
deserializeTimer.Stop();
Logger.LogStartupMessage("\nTook {0} (in TOTAL) to DE-serialise the intermediate data FROM disk\n", deserializeTimer.Elapsed);
return(new TagServer(rawQuestions, allTags, intermediateLookups, intermediateBitMapIndexes));
}
