private String applyRules(String st, SuffixTree <SavoyRule> rules)
{
int length = st.Length - 1;
if (length < rules.Properties["size"]) //If the word is smaller than the minimum stemming size of this step, ignore it
{
return(st);
}
List <Pair <String, SavoyRule> > res = rules.getLongestSuffixesAndValues(st);
for (int i = res.Count - 1; i >= 0; i--)
{
Pair <String, SavoyRule> r = res[i];
String suffix = r.First;
SavoyRule rule = r.Second;
if (length > rule.size)
{
return(st.Substring(0, st.Length - suffix.Length) + rule.replacement);
}
}
return(st);
}
public void readRulesFromXML()
{
XmlDocument doc = new XmlDocument();
try{
doc.Load(System.Reflection.Assembly.GetExecutingAssembly().GetManifestResourceStream("SavoyStemmerRules.xml"));
}catch(Exception e) {
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file.",e);}
XmlElement root = doc.DocumentElement;
XmlAttribute val,val2,val3;
foreach (XmlNode step in root.ChildNodes)
{
val = step.Attributes["name"];
if(val == null)
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Invalid step.");
String stepName = val.Value;
SuffixTree<SavoyRule> suffixes = new SuffixTree<SavoyRule>();
setProperty(suffixes,"size",0,step);
foreach (XmlNode rule in step.ChildNodes)
{
val = rule.Attributes["suffix"];
val2 = rule.Attributes["replacement"];
val3 = rule.Attributes["size"];
if(val == null || val2 == null || val3 == null)
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Invalid rule in "+stepName+".");
String suffix = val.Value;
String replacement = val2.Value;
int size = 0;
try{
size = Convert.ToInt32(val3.Value);
}catch(Exception e) {throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Missing or invalid rules properties on step "+stepName+".", e);}
SavoyRule r = new SavoyRule(size,replacement);
suffixes.addSuffix(suffix,r);
}
if(stepName.Equals("pluralreduction"))
pluralreductionrules = suffixes;
else if(stepName.Equals("femininereduction"))
femininereductionrules = suffixes;
else if(stepName.Equals("finalvowel"))
finalvowel = suffixes;
}
if(pluralreductionrules == null || femininereductionrules == null || finalvowel == null)
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Missing steps.");
}
public void readRulesFromXML()
{
XmlDocument doc = new XmlDocument();
try{
doc.Load(System.Reflection.Assembly.GetExecutingAssembly().GetManifestResourceStream("SavoyStemmerRules.xml"));
}catch (Exception e) {
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file.", e);
}
XmlElement root = doc.DocumentElement;
XmlAttribute val, val2, val3;
foreach (XmlNode step in root.ChildNodes)
{
val = step.Attributes["name"];
if (val == null)
{
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Invalid step.");
}
String stepName = val.Value;
SuffixTree <SavoyRule> suffixes = new SuffixTree <SavoyRule>();
setProperty(suffixes, "size", 0, step);
foreach (XmlNode rule in step.ChildNodes)
{
val = rule.Attributes["suffix"];
val2 = rule.Attributes["replacement"];
val3 = rule.Attributes["size"];
if (val == null || val2 == null || val3 == null)
{
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Invalid rule in " + stepName + ".");
}
String suffix = val.Value;
String replacement = val2.Value;
int size = 0;
try{
size = Convert.ToInt32(val3.Value);
}catch (Exception e) { throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Missing or invalid rules properties on step " + stepName + ".", e); }
SavoyRule r = new SavoyRule(size, replacement);
suffixes.addSuffix(suffix, r);
}
if (stepName.Equals("pluralreduction"))
{
pluralreductionrules = suffixes;
}
else if (stepName.Equals("femininereduction"))
{
femininereductionrules = suffixes;
}
else if (stepName.Equals("finalvowel"))
{
finalvowel = suffixes;
}
}
if (pluralreductionrules == null || femininereductionrules == null || finalvowel == null)
{
throw new PTStemmerException("Problem while parsing Savoy's XML stemming rules file: Missing steps.");
}
}