public SearchingParser(
IPositionedMessagesReader owner,
CreateSearchingParserParams p,
TextStreamPositioningParams textStreamPositioningParams,
DejitteringParams?dejitteringParams,
Stream rawStream,
Encoding streamEncoding,
bool allowPlainTextSearchOptimization,
LoadedRegex headerRe,
ILogSourceThreads threads,
ITraceSourceFactory traceSourceFactory,
RegularExpressions.IRegexFactory regexFactory
)
{
this.owner = owner;
this.parserParams = p;
this.plainTextSearchOptimizationAllowed = allowPlainTextSearchOptimization && ((p.Flags & MessagesParserFlag.DisablePlainTextSearchOptimization) == 0);
this.threads = threads;
this.requestedRange = p.Range;
this.textStreamPositioningParams = textStreamPositioningParams;
this.dejitteringParams = dejitteringParams;
this.rawStream = rawStream;
this.streamEncoding = streamEncoding;
this.regexFactory = regexFactory;
this.trace = traceSourceFactory.CreateTraceSource("LogSource", "srchp." + GetHashCode().ToString("x"));
this.dummyFilter = new Filter(FilterAction.Include, "", true, new Search.Options(), null, regexFactory);
var continuationToken = p.ContinuationToken as ContinuationToken;
if (continuationToken != null)
{
this.requestedRange = new FileRange.Range(continuationToken.NextPosition, requestedRange.End);
}
this.aligmentTextAccess = new StreamTextAccess(rawStream, streamEncoding, textStreamPositioningParams);
this.aligmentSplitter = new MessagesSplitter(aligmentTextAccess, headerRe.Clone().Regex, headerRe.GetHeaderReSplitterFlags());
this.aligmentCapture = new TextMessageCapture();
this.progressAndCancellation = new ProgressAndCancellation()
{
progressHandler = p.ProgressHandler,
cancellationToken = p.Cancellation,
continuationToken = new ContinuationToken()
{
NextPosition = requestedRange.Begin
}
};
this.impl = Enum();
}
/// <summary>
/// Creates valid TextStreamPosition object that points to the charachter that starts at or contains
/// the byte defined by <paramref name="streamPosition"/>
/// </summary>
/// <param name="streamPosition">Stream position. In other words 0-based byte index in stream's data.</param>
/// <param name="streamEncoding">Manadatory encoding information of the stream</param>
/// <param name="stream"></param>
/// <returns>Valid TextStreamPosition object</returns>
public static async Task <TextStreamPosition> StreamPositionToTextStreamPosition(long streamPosition, Encoding streamEncoding, Stream stream, TextStreamPositioningParams textStreamPositioningParams)
{
if (streamEncoding == null)
{
throw new ArgumentNullException("streamEncoding");
}
TextStreamPosition tmp = new TextStreamPosition(streamPosition, textStreamPositioningParams);
#if !SILVERLIGHT
if (streamEncoding.IsSingleByte)
{
return(tmp);
}
#endif
if (streamEncoding == Encoding.Unicode || streamEncoding == Encoding.BigEndianUnicode)
{
return(new TextStreamPosition(tmp.StreamPositionAlignedToBlockSize, tmp.CharPositionInsideBuffer / 2, textStreamPositioningParams));
}
#if !SILVERLIGHT
if (streamEncoding == Encoding.UTF32)
{
return(new TextStreamPosition(tmp.StreamPositionAlignedToBlockSize, tmp.CharPositionInsideBuffer / 4, textStreamPositioningParams));
}
#endif
if (stream == null)
{
throw new ArgumentNullException("stream object is required to determine text stream position with given encoding", "stream");
}
var boundedStream = new BoundedStream();
boundedStream.SetStream(stream, false);
boundedStream.SetBounds(null, streamPosition);
StreamTextAccess tmpTextAccess = new StreamTextAccess(boundedStream, streamEncoding, textStreamPositioningParams);
await tmpTextAccess.BeginReading(tmp.StreamPositionAlignedToBlockSize, TextAccessDirection.Forward);
tmp = tmpTextAccess.CharIndexToPosition(tmpTextAccess.BufferString.Length);
tmpTextAccess.EndReading();
return(tmp);
}
public static MessageFilteringResult GetFilteringResultFromPostprocessorResult(object obj)
{
var f = (IFilter)obj;
if (f == null)
{
return new MessageFilteringResult {
Action = FilterAction.Exclude
}
}
;
if (f == dummyFilter)
{
return new MessageFilteringResult {
Action = FilterAction.Include
}
}
;
return(new MessageFilteringResult
{
Action = f.Action,
Filter = f
});
}
};
IEnumerable <SearchResultMessage> Enum()
{
using (var threadsBulkProcessing = threads.UnderlyingThreadsContainer.StartBulkProcessing())
{
Func <IMessagesPostprocessor> postprocessor =
() => new MessagesPostprocessor(parserParams.SearchParams, trace);
long searchableRangesLength = 0;
int searchableRangesCount = 0;
long totalMessagesCount = 0;
long totalHitsCount = 0;
foreach (var currentSearchableRange in EnumSearchableRanges())
{
searchableRangesLength += currentSearchableRange.Length;
++searchableRangesCount;
using (var parser = CreateParserForSearchableRange(currentSearchableRange, postprocessor))
{
long messagesCount = 0;
long hitsCount = 0;
for (;;)
{
var tmp = parser.ReadNextAndPostprocess();
if (tmp.Message == null)
{
break;
}
++messagesCount;
var msg = tmp.Message;
var filteringResult = MessagesPostprocessor.GetFilteringResultFromPostprocessorResult(
tmp.PostprocessingResult);
if (filteringResult.Action != FilterAction.Exclude)
{
++hitsCount;
yield return(new SearchResultMessage(msg, filteringResult));
}
progressAndCancellation.HandleMessageReadingProgress(msg.Position);
progressAndCancellation.continuationToken.NextPosition = msg.EndPosition;
progressAndCancellation.CheckTextIterationCancellation();
}
PrintPctStats(string.Format("hits pct in range {0}", currentSearchableRange),
hitsCount, messagesCount);
totalMessagesCount += messagesCount;
totalHitsCount += hitsCount;
}
}
trace.Info("Stats: searchable ranges count: {0}", searchableRangesCount);
trace.Info("Stats: ave searchable range len: {0}",
searchableRangesCount != 0 ? searchableRangesLength / searchableRangesCount : 0);
PrintPctStats("searchable ranges coverage pct", searchableRangesLength, requestedRange.Length);
PrintPctStats("hits pct overall", totalHitsCount, totalMessagesCount);
}
yield return(new SearchResultMessage(null, new MessageFilteringResult()));
}
void PrintPctStats(string name, long num, long denum)
{
trace.Info("Stats: {0}: {1:F4}%", name, denum != 0 ? num * 100d / denum : 0d);
}
IEnumerable <FileRange.Range> EnumSearchableRanges()
{
var matcher = new PlainTextMatcher(parserParams, textStreamPositioningParams, plainTextSearchOptimizationAllowed);
if (!matcher.PlainTextSearchOptimizationPossible)
{
yield return(requestedRange);
yield break;
}
long?skipRangesDownThisPosition = null;
foreach (var currentRange in EnumSearchableRangesCore(matcher))
{
if (skipRangesDownThisPosition == null)
{
yield return(currentRange);
}
else
{
long skipRangesDownThisPositionVal = skipRangesDownThisPosition.Value;
if (currentRange.End < skipRangesDownThisPositionVal) // todo: < or <= ?
{
continue;
}
skipRangesDownThisPosition = null;
if (currentRange.Begin < skipRangesDownThisPositionVal) // todo: < or <= ?
{
yield return(new FileRange.Range(skipRangesDownThisPositionVal, currentRange.End));
}
else
{
yield return(currentRange);
}
}
}
}
IPositionedMessagesParser CreateParserForSearchableRange(
FileRange.Range searchableRange,
Func <IMessagesPostprocessor> messagesPostprocessor)
{
bool disableMultithreading = false;
return(owner.CreateParser(new CreateParserParams(
searchableRange.Begin, searchableRange,
MessagesParserFlag.HintParserWillBeUsedForMassiveSequentialReading
| (disableMultithreading ? MessagesParserFlag.DisableMultithreading : MessagesParserFlag.None),
MessagesParserDirection.Forward,
messagesPostprocessor)));
}
IEnumerable <FileRange.Range> EnumSearchableRangesCore(PlainTextMatcher matcher)
{
ITextAccess ta = new StreamTextAccess(rawStream, streamEncoding, textStreamPositioningParams);
using (var tai = ta.OpenIterator(requestedRange.Begin, TextAccessDirection.Forward))
{
var lastRange = new FileRange.Range();
foreach (var r in
IterateMatchRanges(
EnumCheckpoints(tai, matcher, progressAndCancellation, trace),
// todo: tune next parameter to find the value giving max performance.
// On one sample log bigger block was better than many small ones.
// Hence quite big threshold.
textStreamPositioningParams.AlignmentBlockSize * 8,
progressAndCancellation
)
.Select(r => PostprocessHintRange(r, lastRange))
)
{
lastRange = r;
yield return(r);
}
}
}
FileRange.Range PostprocessHintRange(FileRange.Range r, FileRange.Range lastRange)
{
long fixedBegin = r.Begin;
long fixedEnd = r.End;
int?inflateRangeBy = null;
if (dejitteringParams != null && (parserParams.Flags & MessagesParserFlag.DisableDejitter) == 0)
{
inflateRangeBy = dejitteringParams.Value.JitterBufferSize;
}
aligmentSplitter.BeginSplittingSession(requestedRange, r.End, MessagesParserDirection.Forward);
if (aligmentSplitter.GetCurrentMessageAndMoveToNextOne(aligmentCapture))
{
fixedEnd = aligmentCapture.EndPosition;
if (inflateRangeBy != null)
{
for (int i = 0; i < inflateRangeBy.Value; ++i)
{
if (!aligmentSplitter.GetCurrentMessageAndMoveToNextOne(aligmentCapture))
{
break;
}
fixedEnd = aligmentCapture.EndPosition;
}
}
}
else
{
fixedEnd = requestedRange.End;
}
aligmentSplitter.EndSplittingSession();
aligmentSplitter.BeginSplittingSession(requestedRange, fixedBegin, MessagesParserDirection.Backward);
if (aligmentSplitter.GetCurrentMessageAndMoveToNextOne(aligmentCapture))
{
fixedBegin = aligmentCapture.BeginPosition;
if (inflateRangeBy != null)
{
for (int i = 0; i < inflateRangeBy.Value; ++i)
{
if (!aligmentSplitter.GetCurrentMessageAndMoveToNextOne(aligmentCapture))
{
break;
}
fixedBegin = aligmentCapture.BeginPosition;
}
}
}
aligmentSplitter.EndSplittingSession();
var ret = new FileRange.Range(fixedBegin, fixedEnd);
ret = FileRange.Range.Intersect(ret, requestedRange).Common;
var lastRangeIntersection = FileRange.Range.Intersect(ret, lastRange);
if (lastRangeIntersection.RelativePosition == 0)
{
ret = lastRangeIntersection.Leftover1Right;
}
return(ret);
}
private TextStreamPosition DetectEndPositionFromMediaSize()
{
return(StreamTextAccess.StreamPositionToTextStreamPosition(mediaSize, StreamEncoding, VolatileStream, textStreamPositioningParams));
}
public TextAccessIterator(StreamTextAccess impl)
{
this.impl = impl;
}