/// <summary>
/// Sets the params.
/// Analysis component factory names may optionally include the "Factory" suffix.
/// </summary>
/// <param name="params">
/// analysis pipeline specification: name, (optional) positionIncrementGap,
/// (optional) offsetGap, 0+ CharFilterFactory's, 1 TokenizerFactory,
/// and 0+ TokenFilterFactory's
/// </param>
public override void SetParams(string @params)
{
base.SetParams(@params);
ArgType expectedArgType = ArgType.ANALYZER_ARG;
StreamTokenizer stok = new StreamTokenizer(new StringReader(@params));
stok.CommentChar('#');
stok.QuoteChar('"');
stok.QuoteChar('\'');
stok.IsEOLSignificant = false;
stok.OrdinaryChar('(');
stok.OrdinaryChar(')');
stok.OrdinaryChar(':');
stok.OrdinaryChar(',');
try
{
while (stok.NextToken() != StreamTokenizer.TT_EOF)
{
switch (stok.TokenType)
{
case ',':
{
// Do nothing
break;
}
case StreamTokenizer.TT_WORD:
{
if (expectedArgType.Equals(ArgType.ANALYZER_ARG))
{
string argName = stok.StringValue;
if (!argName.Equals("name", StringComparison.OrdinalIgnoreCase) &&
!argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase) &&
!argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Missing 'name' param to AnalyzerFactory: '" + @params + "'");
}
stok.NextToken();
if (stok.TokenType != ':')
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Missing ':' after '" + argName + "' param to AnalyzerFactory");
}
stok.NextToken();
string argValue = stok.StringValue;
switch (stok.TokenType)
{
case StreamTokenizer.TT_NUMBER:
{
argValue = stok.NumberValue.ToString(CultureInfo.InvariantCulture);
// Drop the ".0" from numbers, for integer arguments
argValue = TRAILING_DOT_ZERO_PATTERN.Replace(argValue, "", 1);
// Intentional fallthrough
if (argName.Equals("name", StringComparison.OrdinalIgnoreCase))
{
factoryName = argValue;
expectedArgType = ArgType.ANALYZER_ARG_OR_CHARFILTER_OR_TOKENIZER;
}
else
{
int intArgValue = 0;
try
{
intArgValue = int.Parse(argValue, CultureInfo.InvariantCulture);
}
catch (FormatException e)
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Exception parsing " + argName + " value '" + argValue + "'", e);
}
if (argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase))
{
positionIncrementGap = intArgValue;
}
else if (argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
offsetGap = intArgValue;
}
}
break;
}
case '"':
case '\'':
case StreamTokenizer.TT_WORD:
{
if (argName.Equals("name", StringComparison.OrdinalIgnoreCase))
{
factoryName = argValue;
expectedArgType = ArgType.ANALYZER_ARG_OR_CHARFILTER_OR_TOKENIZER;
}
else
{
int intArgValue = 0;
try
{
intArgValue = int.Parse(argValue, CultureInfo.InvariantCulture);
}
catch (FormatException e)
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Exception parsing " + argName + " value '" + argValue + "'", e);
}
if (argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase))
{
positionIncrementGap = intArgValue;
}
else if (argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
offsetGap = intArgValue;
}
}
break;
}
case StreamTokenizer.TT_EOF:
{
throw new Exception("Unexpected EOF: " + stok.ToString());
}
default:
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Unexpected token: " + stok.ToString());
}
}
}
else if (expectedArgType.Equals(ArgType.ANALYZER_ARG_OR_CHARFILTER_OR_TOKENIZER))
{
string argName = stok.StringValue;
if (argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase) ||
argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
stok.NextToken();
if (stok.TokenType != ':')
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Missing ':' after '" + argName + "' param to AnalyzerFactory");
}
stok.NextToken();
int intArgValue = (int)stok.NumberValue;
switch (stok.TokenType)
{
case '"':
case '\'':
case StreamTokenizer.TT_WORD:
{
intArgValue = 0;
try
{
intArgValue = int.Parse(stok.StringValue.Trim(), CultureInfo.InvariantCulture);
}
catch (FormatException e)
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Exception parsing " + argName + " value '" + stok.StringValue + "'", e);
}
// Intentional fall-through
if (argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase))
{
positionIncrementGap = intArgValue;
}
else if (argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
offsetGap = intArgValue;
}
break;
}
case StreamTokenizer.TT_NUMBER:
{
if (argName.Equals("positionIncrementGap", StringComparison.OrdinalIgnoreCase))
{
positionIncrementGap = intArgValue;
}
else if (argName.Equals("offsetGap", StringComparison.OrdinalIgnoreCase))
{
offsetGap = intArgValue;
}
break;
}
case StreamTokenizer.TT_EOF:
{
throw new Exception("Unexpected EOF: " + stok.ToString());
}
default:
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Unexpected token: " + stok.ToString());
}
}
break;
}
try
{
Type clazz;
clazz = LookupAnalysisClass(argName, typeof(CharFilterFactory));
CreateAnalysisPipelineComponent(stok, clazz);
}
catch (ArgumentException /*e*/)
{
try
{
Type clazz;
clazz = LookupAnalysisClass(argName, typeof(TokenizerFactory));
CreateAnalysisPipelineComponent(stok, clazz);
expectedArgType = ArgType.TOKENFILTER;
}
catch (ArgumentException e2)
{
throw new Exception("Line #" + GetLineNumber(stok) + ": Can't find class '"
+ argName + "' as CharFilterFactory or TokenizerFactory", e2);
}
}
}
else
{ // expectedArgType = ArgType.TOKENFILTER
string className = stok.StringValue;
Type clazz;
try
{
clazz = LookupAnalysisClass(className, typeof(TokenFilterFactory));
}
catch (ArgumentException e)
{
throw new Exception
("Line #" + GetLineNumber(stok) + ": Can't find class '" + className + "' as TokenFilterFactory", e);
}
CreateAnalysisPipelineComponent(stok, clazz);
}
break;
}
default:
{
throw new Exception("Line #" + GetLineNumber(stok) + ": Unexpected token: " + stok.ToString());
}
}
}
}
catch (Exception e)
{
if (e.Message.StartsWith("Line #", StringComparison.Ordinal))
{
throw e;
}
else
{
throw new Exception("Line #" + GetLineNumber(stok) + ": ", e);
}
}
AnalyzerFactory analyzerFactory = new AnalyzerFactory
(charFilterFactories, tokenizerFactory, tokenFilterFactories);
analyzerFactory.SetPositionIncrementGap(positionIncrementGap.GetValueOrDefault());
analyzerFactory.SetOffsetGap(offsetGap.GetValueOrDefault());
RunData.AnalyzerFactories[factoryName] = analyzerFactory;
}