public static string GetExtract(BinaryData binaryData, Node node)
{
using (var op = SnTrace.Index.StartOperation("Getting text extract, VId:{0}, Path:{1}", node.VersionId, node.Path))
{
var extractor = ResolveExtractor(binaryData);
if (extractor == null)
{
op.Successful = true;
return(string.Empty);
}
var result = string.Empty;
using (var stream = binaryData.GetStream())
{
if (stream == null || stream.Length == 0)
{
op.Successful = true;
return(String.Empty);
}
try
{
var ctx = new TextExtractorContext(node.VersionId);
// async
Action <TimeboxedActivity> timeboxedFunctionCall = activity =>
{
var x = (Stream)activity.InArgument;
var extract = extractor.Extract(x, ctx);
activity.OutArgument = extract;
};
var act = new TimeboxedActivity();
act.InArgument = stream;
act.Activity = timeboxedFunctionCall;
act.Context = HttpContext.Current;
var finishedWithinTime = act.ExecuteAndWait(Configuration.Indexing.TextExtractTimeout * 1000);
if (!finishedWithinTime)
{
act.Abort();
var msg = String.Format("Text extracting timeout. Version: {0}, path: {1}", node.Version, node.Path);
SnTrace.Index.Write(msg);
SnLog.WriteWarning(msg);
op.Successful = true;
return(String.Empty);
}
else if (act.ExecutionException != null)
{
WriteError(act.ExecutionException, node);
}
else
{
result = (string)act.OutArgument;
}
}
catch (Exception e)
{
WriteError(e, node);
}
}
if (result == null)
{
SnLog.WriteWarning(string.Format(CultureInfo.InvariantCulture, @"Couldn't extract text. VersionId: {0}, path: '{1}' ", node.VersionId, node.Path));
}
else
{
result = result.Replace('\0', '.');
}
if (result == null)
{
SnTrace.Index.Write("Couldn't extract text");
}
else
{
SnTrace.Index.Write("Extracted length length: {0}.", result.Length);
}
op.Successful = true;
return(result);
}
}
public static string GetExtract(BinaryData binaryData, Node node)
{
if (binaryData == null)
{
return(string.Empty);
}
var fname = binaryData.FileName;
if (fname == null)
{
return(string.Empty);
}
var ext = fname.Extension;
if (String.IsNullOrEmpty(ext))
{
return(string.Empty);
}
ITextExtractor extractor = null;
var result = string.Empty;
switch (ext.ToLower())
{
case "contenttype":
case "xml": extractor = new XmlTextExtractor(); break;
case "doc": extractor = new DocTextExtractor(); break;
case "xls": extractor = new XlsTextExtractor(); break;
case "pdf": extractor = new PdfTextExtractor(); break;
case "docx": extractor = new DocxTextExtractor(); break;
case "xlsx": extractor = new XlsxTextExtractor(); break;
case "pptx": extractor = new PptxTextExtractor(); break;
case "txt": extractor = new PlainTextExtractor(); break;
default:
return(String.Empty);
}
var stream = binaryData.GetStream();
if (stream == null)
{
return(String.Empty);
}
if (stream.Length == 0)
{
return(String.Empty);
}
try
{
////-- sync
//result = extractor.Extract(stream);
//-- async
Action <TimeboxedActivity> timeboxedFunctionCall = activity =>
{
var x = (Stream)activity.InArgument;
var extract = extractor.Extract(x);
activity.OutArgument = extract;
};
var act = new TimeboxedActivity();
act.InArgument = stream;
act.Activity = timeboxedFunctionCall;
act.Context = HttpContext.Current;
var finishedWithinTime = act.ExecuteAndWait(Repository.TextExtractTimeout * 1000);
if (!finishedWithinTime)
{
act.Abort();
var msg = String.Format("Text extracting timeout. Version: {0}, path: {1}", node.Version, node.Path);
Logger.WriteWarning(msg);
return(String.Empty);
}
else if (act.ExecutionException != null)
{
WriteError(act.ExecutionException, node);
}
else
{
result = (string)act.OutArgument;
}
}
catch (Exception e)
{
WriteError(e, node);
}
if (String.IsNullOrEmpty(result))
{
var format = @"Couldn't extract text. VersionId: {0}, path: '{1}' ";
var inf = String.Format(CultureInfo.InvariantCulture, format, node.VersionId, node.Path);
Logger.WriteWarning(inf);
}
result = result.Replace('\0', '.');
return(result);
}
public static string GetExtract(BinaryData binaryData, Node node)
{
var extractor = ResolveExtractor(binaryData);
if (extractor == null)
{
return(string.Empty);
}
var result = string.Empty;
var stream = binaryData.GetStream();
if (stream == null)
{
return(String.Empty);
}
if (stream.Length == 0)
{
return(String.Empty);
}
try
{
var ctx = new TextExtractorContext(node.VersionId);
//-- async
Action <TimeboxedActivity> timeboxedFunctionCall = activity =>
{
var x = (Stream)activity.InArgument;
var extract = extractor.Extract(x, ctx);
activity.OutArgument = extract;
};
var act = new TimeboxedActivity();
act.InArgument = stream;
act.Activity = timeboxedFunctionCall;
act.Context = HttpContext.Current;
var finishedWithinTime = act.ExecuteAndWait(Repository.TextExtractTimeout * 1000);
if (!finishedWithinTime)
{
act.Abort();
var msg = String.Format("Text extracting timeout. Version: {0}, path: {1}", node.Version, node.Path);
Logger.WriteWarning(Logger.EventId.NotDefined, msg);
return(String.Empty);
}
else if (act.ExecutionException != null)
{
WriteError(act.ExecutionException, node);
}
else
{
result = (string)act.OutArgument;
}
}
catch (Exception e)
{
WriteError(e, node);
}
if (result == null)
{
Logger.WriteWarning(Logger.EventId.NotDefined, String.Format(CultureInfo.InvariantCulture, @"Couldn't extract text. VersionId: {0}, path: '{1}' ", node.VersionId, node.Path));
}
else
{
result = result.Replace('\0', '.');
}
return(result);
}
