public DescriptionPattern(Relation newRelation, Edu.Stanford.Nlp.Trees.Tregex.DescriptionPattern oldPattern)
{
this.rel = newRelation;
this.negDesc = oldPattern.negDesc;
this.isLink = oldPattern.isLink;
this.linkedName = oldPattern.linkedName;
this.stringDesc = oldPattern.stringDesc;
this.descriptionMode = oldPattern.descriptionMode;
this.descPattern = oldPattern.descPattern;
this.exactMatch = oldPattern.exactMatch;
this.stringFilter = oldPattern.stringFilter;
this.name = oldPattern.name;
this.SetChild(oldPattern.child);
this.basicCatFunction = oldPattern.basicCatFunction;
this.variableGroups = oldPattern.variableGroups;
}
public DescriptionPattern(Relation rel, bool negDesc, string desc, string name, bool useBasicCat, Func <string, string> basicCatFunction, IList <Pair <int, string> > variableGroups, bool isLink, string linkedName)
{
// what size string matchers to use before switching to regex for
// disjunction matches
// todo: conceptually final, but we'd need to rewrite TregexParser
// to make it so.
// also conceptually final, but it depends on the child
// specifies the groups in a regex that are captured as matcher-global string variables
// for example, /^:$/
// for example, /^[$]$/
// for example, /^-NONE-$/
// for example, /^JJ/
this.rel = rel;
this.negDesc = negDesc;
this.isLink = isLink;
this.linkedName = linkedName;
if (desc != null)
{
stringDesc = desc;
// TODO: factor out some of these blocks of code
if (desc.Equals("__") || desc.Equals("/.*/") || desc.Equals("/^.*$/"))
{
descriptionMode = DescriptionPattern.DescriptionMode.Anything;
descPattern = null;
exactMatch = null;
stringFilter = null;
}
else
{
if (SingleWordPattern.Matcher(desc).Matches())
{
// Expressions are written like this to put special characters
// in the tregex matcher, but a regular expression is less
// efficient than a simple string match
descriptionMode = DescriptionPattern.DescriptionMode.Exact;
descPattern = null;
Java.Util.Regex.Matcher matcher = SingleWordPattern.Matcher(desc);
matcher.Matches();
string matchedGroup = null;
for (int i = 1; i <= matcher.GroupCount(); ++i)
{
if (matcher.Group(i) != null)
{
matchedGroup = matcher.Group(i);
break;
}
}
exactMatch = matchedGroup;
stringFilter = null;
}
else
{
//log.info("DescriptionPattern: converting " + desc + " to " + exactMatch);
if (MultiWordPattern.Matcher(desc).Matches())
{
Java.Util.Regex.Matcher matcher = MultiWordPattern.Matcher(desc);
matcher.Matches();
string matchedGroup = null;
for (int i = 1; i <= matcher.GroupCount(); ++i)
{
if (matcher.Group(i) != null)
{
matchedGroup = matcher.Group(i);
break;
}
}
matchedGroup = matchedGroup.ReplaceAll("\\\\", string.Empty);
if (matchedGroup.Split("[|]").Length > MaxStringMatcherSize)
{
descriptionMode = DescriptionPattern.DescriptionMode.Pattern;
descPattern = Pattern.Compile(Sharpen.Runtime.Substring(desc, 1, desc.Length - 1));
exactMatch = null;
stringFilter = null;
}
else
{
//log.info("DescriptionPattern: not converting " + desc);
descriptionMode = DescriptionPattern.DescriptionMode.Strings;
descPattern = null;
exactMatch = null;
stringFilter = new ArrayStringFilter(ArrayStringFilter.Mode.Exact, matchedGroup.Split("[|]"));
}
}
else
{
//log.info("DescriptionPattern: converting " + desc + " to " + stringFilter);
if (CaseInsensitivePattern.Matcher(desc).Matches())
{
Java.Util.Regex.Matcher matcher = CaseInsensitivePattern.Matcher(desc);
matcher.Matches();
string matchedGroup = null;
for (int i = 1; i <= matcher.GroupCount(); ++i)
{
if (matcher.Group(i) != null)
{
matchedGroup = matcher.Group(i);
break;
}
}
matchedGroup = matchedGroup.ReplaceAll("\\\\", string.Empty);
if (matchedGroup.Split("[|]").Length > MaxStringMatcherSize)
{
descriptionMode = DescriptionPattern.DescriptionMode.Pattern;
descPattern = Pattern.Compile(Sharpen.Runtime.Substring(desc, 1, desc.Length - 1));
exactMatch = null;
stringFilter = null;
}
else
{
//log.info("DescriptionPattern: not converting " + desc);
descriptionMode = DescriptionPattern.DescriptionMode.Strings;
descPattern = null;
exactMatch = null;
stringFilter = new ArrayStringFilter(ArrayStringFilter.Mode.CaseInsensitive, matchedGroup.Split("[|]"));
}
}
else
{
//log.info("DescriptionPattern: converting " + desc + " to " + stringFilter);
if (PrefixPattern.Matcher(desc).Matches())
{
Java.Util.Regex.Matcher matcher = PrefixPattern.Matcher(desc);
matcher.Matches();
string matchedGroup = null;
for (int i = 1; i <= matcher.GroupCount(); ++i)
{
if (matcher.Group(i) != null)
{
matchedGroup = matcher.Group(i);
break;
}
}
if (matchedGroup.Split("\\|").Length > MaxStringMatcherSize)
{
descriptionMode = DescriptionPattern.DescriptionMode.Pattern;
descPattern = Pattern.Compile(Sharpen.Runtime.Substring(desc, 1, desc.Length - 1));
exactMatch = null;
stringFilter = null;
}
else
{
//log.info("DescriptionPattern: not converting " + desc);
descriptionMode = DescriptionPattern.DescriptionMode.Strings;
descPattern = null;
exactMatch = null;
stringFilter = new ArrayStringFilter(ArrayStringFilter.Mode.Prefix, matchedGroup.Split("[|]"));
}
}
else
{
//log.info("DescriptionPattern: converting " + desc + " to " + stringFilter);
if (desc.Matches("/.*/"))
{
descriptionMode = DescriptionPattern.DescriptionMode.Pattern;
descPattern = Pattern.Compile(Sharpen.Runtime.Substring(desc, 1, desc.Length - 1));
exactMatch = null;
stringFilter = null;
}
else
{
if (desc.IndexOf('|') >= 0)
{
// patterns which contain ORs are a special case; we either
// promote those to regex match or make a string matcher out
// of them. for short enough disjunctions, a simple string
// matcher can be more efficient than a regex.
string[] words = desc.Split("[|]");
if (words.Length <= MaxStringMatcherSize)
{
descriptionMode = DescriptionPattern.DescriptionMode.Strings;
descPattern = null;
exactMatch = null;
stringFilter = new ArrayStringFilter(ArrayStringFilter.Mode.Exact, words);
}
else
{
descriptionMode = DescriptionPattern.DescriptionMode.Pattern;
descPattern = Pattern.Compile("^(?:" + desc + ")$");
exactMatch = null;
stringFilter = null;
}
}
else
{
// raw description
descriptionMode = DescriptionPattern.DescriptionMode.Exact;
descPattern = null;
exactMatch = desc;
stringFilter = null;
}
}
}
}
}
}
}
}
else
{
if (name == null && linkedName == null)
{
throw new AssertionError("Illegal description pattern. Does not describe a node or link/name a variable");
}
stringDesc = " ";
descriptionMode = null;
descPattern = null;
exactMatch = null;
stringFilter = null;
}
this.name = name;
SetChild(null);
this.basicCatFunction = (useBasicCat ? basicCatFunction : null);
// System.out.println("Made " + (negDesc ? "negated " : "") + "DescNode with " + desc);
this.variableGroups = variableGroups;
}
