/// <summary>
/// Converts the specified string representation of a tree element to its <see cref="AdTreeElement"/>
/// equivalent and returns a value that indicates whether the conversion succeeded.
/// </summary>
/// <param name="element">
/// When this method returns, contains the <see cref="AdTreeElement"/> value equivalent to the element
/// contained in <paramref name="line"/>, if the conversion succeeded, or <c>null</c> if the conversion
/// failed. The conversion fails if the <paramref name="line"/> parameter is null, is an empty string (""),
/// or does not contain a valid string representation of a AdElement. This parameter is passed
/// uninitialized.
/// </param>
/// <param name="line">The string representation of the element.</param>
/// <param name="safeParse">if set to <c>true</c> the invalid sentences will be ignored.</param>
/// <param name="monitor">The evaluation monitor.</param>
/// <returns><c>true</c> if the s parameter was converted successfully; otherwise, <c>false</c>.</returns>
private static bool TryParseElement(out AdTreeElement element, string line, bool safeParse, Monitor monitor)
{
var m = nodePattern.Match(line);
if (m.Success)
{
element = new AdNode {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value
};
return(true);
}
m = leafPattern.Match(line);
if (m.Success)
{
element = new AdLeaf {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value,
FunctionalTag = m.Groups[3].Value,
Lemma = m.Groups[4].Value,
SecondaryTag = m.Groups[5].Value,
MorphologicalTag = m.Groups[6].Value,
Lexeme = m.Groups[7].Value
};
return(true);
}
m = punctuationPattern.Match(line);
if (m.Success)
{
element = new AdLeaf {
Level = m.Groups[1].Length + 1,
Lexeme = m.Groups[2].Value
};
return(true);
}
if (safeParse)
{
element = null;
return(false);
}
// Knuppe: The most bizarre cases I found, were invalid data (like HTML, inside the sentences)
// so I decided to implement the safeParse attribute, to ignore this junk...
//
// I think any program should adapt to an error in a file. otherwise the files will never
// be fixed...
// process the bizarre cases.
if (line.Equals("_") || line.StartsWith("<lixo") || line.StartsWith("pause"))
{
element = null;
return(false);
}
if (line.StartsWith("="))
{
m = bizarreLeafPattern.Match(line);
if (m.Success)
{
var leaf = new AdLeaf {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value,
Lemma = m.Groups[3].Value,
MorphologicalTag = m.Groups[4].Value,
Lexeme = m.Groups[5].Value
};
if (!string.IsNullOrEmpty(leaf.Lemma) && leaf.Lemma.Length > 2)
{
leaf.Lemma = leaf.Lemma.Substring(1);
}
element = leaf;
return(true);
}
var level = line.LastIndexOf("=", StringComparison.InvariantCulture) + 1;
if (level > 0 && level < line.Length - 2 && Regex.IsMatch(line.Substring(level + 1), "\\w.*?[\\.<>].*"))
{
element = new AdLeaf {
Level = level + 1,
Lexeme = line.Substring(level + 1)
};
return(true);
}
}
if (monitor != null)
{
monitor.OnWarning("Couldn't parse leaf: " + line);
}
element = null;
return(false);
}
/// <summary>
/// Adds an element.
/// </summary>
/// <param name="element">The element.</param>
public void AddElement(AdTreeElement element) {
Elements.Add(element);
}
/// <summary>
/// Converts the specified string representation of a tree element to its <see cref="AdTreeElement"/>
/// equivalent and returns a value that indicates whether the conversion succeeded.
/// </summary>
/// <param name="element">
/// When this method returns, contains the <see cref="AdTreeElement"/> value equivalent to the element
/// contained in <paramref name="line"/>, if the conversion succeeded, or <c>null</c> if the conversion
/// failed. The conversion fails if the <paramref name="line"/> parameter is null, is an empty string (""),
/// or does not contain a valid string representation of a AdElement. This parameter is passed
/// uninitialized.
/// </param>
/// <param name="line">The string representation of the element.</param>
/// <param name="safeParse">if set to <c>true</c> the invalid sentences will be ignored.</param>
/// <param name="monitor">The evaluation monitor.</param>
/// <returns><c>true</c> if the s parameter was converted successfully; otherwise, <c>false</c>.</returns>
private static bool TryParseElement(out AdTreeElement element, string line, bool safeParse, Monitor monitor) {
var m = nodePattern.Match(line);
if (m.Success) {
element = new AdNode {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value
};
return true;
}
m = leafPattern.Match(line);
if (m.Success) {
element = new AdLeaf {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value,
FunctionalTag = m.Groups[3].Value,
Lemma = m.Groups[4].Value,
SecondaryTag = m.Groups[5].Value,
MorphologicalTag = m.Groups[6].Value,
Lexeme = m.Groups[7].Value
};
return true;
}
m = punctuationPattern.Match(line);
if (m.Success) {
element = new AdLeaf {
Level = m.Groups[1].Length + 1,
Lexeme = m.Groups[2].Value
};
return true;
}
if (safeParse) {
element = null;
return false;
}
// Knuppe: The most bizarre cases I found, were invalid data (like HTML, inside the sentences)
// so I decided to implement the safeParse attribute, to ignore this junk...
//
// I think any program should adapt to an error in a file. otherwise the files will never
// be fixed...
// process the bizarre cases.
if (line.Equals("_") || line.StartsWith("<lixo") || line.StartsWith("pause")) {
element = null;
return false;
}
if (line.StartsWith("=")) {
m = bizarreLeafPattern.Match(line);
if (m.Success) {
var leaf = new AdLeaf {
Level = m.Groups[1].Length + 1,
SyntacticTag = m.Groups[2].Value,
Lemma = m.Groups[3].Value,
MorphologicalTag = m.Groups[4].Value,
Lexeme = m.Groups[5].Value
};
if (!string.IsNullOrEmpty(leaf.Lemma) && leaf.Lemma.Length > 2) {
leaf.Lemma = leaf.Lemma.Substring(1);
}
element = leaf;
return true;
}
var level = line.LastIndexOf("=", StringComparison.InvariantCulture) + 1;
if (level > 0 && level < line.Length - 2 && Regex.IsMatch(line.Substring(level + 1), "\\w.*?[\\.<>].*")) {
element = new AdLeaf {
Level = level + 1,
Lexeme = line.Substring(level + 1)
};
return true;
}
}
if (monitor != null) {
monitor.OnWarning("Couldn't parse leaf: " + line);
}
element = null;
return false;
}
/// <summary>
/// Adds an element.
/// </summary>
/// <param name="element">The element.</param>
public void AddElement(AdTreeElement element)
{
Elements.Add(element);
}