/// <summary>
/// Returns a new
/// <c>SemanticGraph</c>
/// constructed from a given
/// <see cref="Edu.Stanford.Nlp.Trees.Tree"/>
/// with given options.
/// This factory method is intended to replace a profusion of highly similar
/// factory methods, such as
/// <c>typedDependencies()</c>
/// ,
/// <c>typedDependenciesCollapsed()</c>
/// ,
/// <c>allTypedDependencies()</c>
/// ,
/// <c>allTypedDependenciesCollapsed()</c>
/// , etc.
/// For a fuller explanation of the meaning of the boolean arguments, see
/// <see cref="Edu.Stanford.Nlp.Trees.GrammaticalStructure"/>
/// .
/// </summary>
/// <param name="tree">A tree representing a phrase structure parse</param>
/// <param name="includeExtras">
/// Whether to include extra dependencies, which may
/// result in a non-tree
/// </param>
/// <param name="filter">A filter to exclude certain dependencies; ignored if null</param>
/// <param name="originalDependencies">
/// generate original Stanford dependencies instead of new
/// Universal Dependencies
/// </param>
/// <returns>A SemanticGraph</returns>
public static SemanticGraph MakeFromTree(Tree tree, SemanticGraphFactory.Mode mode, GrammaticalStructure.Extras includeExtras, IPredicate <TypedDependency> filter, bool originalDependencies, bool includePunctuationDependencies)
{
GrammaticalStructure gs;
if (originalDependencies)
{
IPredicate <string> wordFilt;
if (includePunctuationDependencies)
{
wordFilt = Filters.AcceptFilter();
}
else
{
wordFilt = new PennTreebankLanguagePack().PunctuationWordRejectFilter();
}
gs = new EnglishGrammaticalStructure(tree, wordFilt, new SemanticHeadFinder(true));
}
else
{
IPredicate <string> tagFilt;
if (includePunctuationDependencies)
{
tagFilt = Filters.AcceptFilter();
}
else
{
tagFilt = new PennTreebankLanguagePack().PunctuationTagRejectFilter();
}
gs = new UniversalEnglishGrammaticalStructure(tree, tagFilt, new UniversalSemanticHeadFinder(true));
}
return(MakeFromTree(gs, mode, includeExtras, filter));
}
/// <summary>Prints out all matches of a semgrex pattern on a file of dependencies.</summary>
/// <remarks>
/// Prints out all matches of a semgrex pattern on a file of dependencies.
/// <p>
/// Usage:<br />
/// java edu.stanford.nlp.semgraph.semgrex.SemgrexPattern [args]
/// <br />
/// See the help() function for a list of possible arguments to provide.
/// </remarks>
/// <exception cref="System.IO.IOException"/>
public static void Main(string[] args)
{
IDictionary <string, int> flagMap = Generics.NewHashMap();
flagMap[Pattern] = 1;
flagMap[TreeFile] = 1;
flagMap[Mode] = 1;
flagMap[Extras] = 1;
flagMap[ConlluFile] = 1;
flagMap[OutputFormatOption] = 1;
IDictionary <string, string[]> argsMap = StringUtils.ArgsToMap(args, flagMap);
// args = argsMap.get(null);
// TODO: allow patterns to be extracted from a file
if (!(argsMap.Contains(Pattern)) || argsMap[Pattern].Length == 0)
{
Help();
System.Environment.Exit(2);
}
Edu.Stanford.Nlp.Semgraph.Semgrex.SemgrexPattern semgrex = Edu.Stanford.Nlp.Semgraph.Semgrex.SemgrexPattern.Compile(argsMap[Pattern][0]);
string modeString = DefaultMode;
if (argsMap.Contains(Mode) && argsMap[Mode].Length > 0)
{
modeString = argsMap[Mode][0].ToUpper();
}
SemanticGraphFactory.Mode mode = SemanticGraphFactory.Mode.ValueOf(modeString);
string outputFormatString = DefaultOutputFormat;
if (argsMap.Contains(OutputFormatOption) && argsMap[OutputFormatOption].Length > 0)
{
outputFormatString = argsMap[OutputFormatOption][0].ToUpper();
}
SemgrexPattern.OutputFormat outputFormat = SemgrexPattern.OutputFormat.ValueOf(outputFormatString);
bool useExtras = true;
if (argsMap.Contains(Extras) && argsMap[Extras].Length > 0)
{
useExtras = bool.ValueOf(argsMap[Extras][0]);
}
IList <SemanticGraph> graphs = Generics.NewArrayList();
// TODO: allow other sources of graphs, such as dependency files
if (argsMap.Contains(TreeFile) && argsMap[TreeFile].Length > 0)
{
foreach (string treeFile in argsMap[TreeFile])
{
log.Info("Loading file " + treeFile);
MemoryTreebank treebank = new MemoryTreebank(new TreeNormalizer());
treebank.LoadPath(treeFile);
foreach (Tree tree in treebank)
{
// TODO: allow other languages... this defaults to English
SemanticGraph graph = SemanticGraphFactory.MakeFromTree(tree, mode, useExtras ? GrammaticalStructure.Extras.Maximal : GrammaticalStructure.Extras.None);
graphs.Add(graph);
}
}
}
if (argsMap.Contains(ConlluFile) && argsMap[ConlluFile].Length > 0)
{
CoNLLUDocumentReader reader = new CoNLLUDocumentReader();
foreach (string conlluFile in argsMap[ConlluFile])
{
log.Info("Loading file " + conlluFile);
IEnumerator <SemanticGraph> it = reader.GetIterator(IOUtils.ReaderFromString(conlluFile));
while (it.MoveNext())
{
SemanticGraph graph = it.Current;
graphs.Add(graph);
}
}
}
foreach (SemanticGraph graph_1 in graphs)
{
SemgrexMatcher matcher = semgrex.Matcher(graph_1);
if (!matcher.Find())
{
continue;
}
if (outputFormat == SemgrexPattern.OutputFormat.List)
{
log.Info("Matched graph:" + Runtime.LineSeparator() + graph_1.ToString(SemanticGraph.OutputFormat.List));
int i = 1;
bool found = true;
while (found)
{
log.Info("Match " + i + " at: " + matcher.GetMatch().ToString(CoreLabel.OutputFormat.ValueIndex));
IList <string> nodeNames = Generics.NewArrayList();
Sharpen.Collections.AddAll(nodeNames, matcher.GetNodeNames());
nodeNames.Sort();
foreach (string name in nodeNames)
{
log.Info(" " + name + ": " + matcher.GetNode(name).ToString(CoreLabel.OutputFormat.ValueIndex));
}
log.Info(" ");
found = matcher.Find();
}
}
else
{
if (outputFormat == SemgrexPattern.OutputFormat.Offset)
{
if (graph_1.VertexListSorted().IsEmpty())
{
continue;
}
System.Console.Out.Printf("+%d %s%n", graph_1.VertexListSorted()[0].Get(typeof(CoreAnnotations.LineNumberAnnotation)), argsMap[ConlluFile][0]);
}
}
}
}
public static SemanticGraph MakeFromTree(Tree tree, SemanticGraphFactory.Mode mode, bool includeExtras)
{
return(MakeFromTree(tree, mode, includeExtras ? GrammaticalStructure.Extras.Maximal : GrammaticalStructure.Extras.None));
}
public static SemanticGraph MakeFromTree(Tree tree, SemanticGraphFactory.Mode mode, GrammaticalStructure.Extras includeExtras)
{
return(MakeFromTree(tree, mode, includeExtras, null, false));
}
public static SemanticGraph MakeFromTree(Tree tree, SemanticGraphFactory.Mode mode, bool includeExtras, IPredicate <TypedDependency> filter)
{
return(MakeFromTree(tree, mode, includeExtras ? GrammaticalStructure.Extras.Maximal : GrammaticalStructure.Extras.None, filter, false));
}
// TODO: these booleans would be more readable as enums similar to Mode.
// Then the arguments would make more sense
public static SemanticGraph MakeFromTree(GrammaticalStructure gs, SemanticGraphFactory.Mode mode, GrammaticalStructure.Extras includeExtras, IPredicate <TypedDependency> filter)
{
ICollection <TypedDependency> deps;
switch (mode)
{
case SemanticGraphFactory.Mode.Enhanced:
{
deps = gs.TypedDependenciesEnhanced();
break;
}
case SemanticGraphFactory.Mode.EnhancedPlusPlus:
{
deps = gs.TypedDependenciesEnhancedPlusPlus();
break;
}
case SemanticGraphFactory.Mode.CollapsedTree:
{
deps = gs.TypedDependenciesCollapsedTree();
break;
}
case SemanticGraphFactory.Mode.Collapsed:
{
deps = gs.TypedDependenciesCollapsed(includeExtras);
break;
}
case SemanticGraphFactory.Mode.Ccprocessed:
{
deps = gs.TypedDependenciesCCprocessed(includeExtras);
break;
}
case SemanticGraphFactory.Mode.Basic:
{
deps = gs.TypedDependencies(includeExtras);
break;
}
default:
{
throw new ArgumentException("Unknown mode " + mode);
}
}
if (filter != null)
{
IList <TypedDependency> depsFiltered = Generics.NewArrayList();
foreach (TypedDependency td in deps)
{
if (filter.Test(td))
{
depsFiltered.Add(td);
}
}
deps = depsFiltered;
}
// there used to be an if clause that filtered out the case of empty
// dependencies. However, I could not understand (or replicate) the error
// it alluded to, and it led to empty dependency graphs for very short fragments,
// which meant they were ignored by the RTE system. Changed. (pado)
// See also the SemanticGraph constructor.
//log.info(deps.toString());
return(new SemanticGraph(deps));
}
public static SemanticGraph MakeFromTree(Tree tree, SemanticGraphFactory.Mode mode, GrammaticalStructure.Extras includeExtras, IPredicate <TypedDependency> filter, bool originalDependencies)
{
return(MakeFromTree(tree, mode, includeExtras, filter, originalDependencies, IncludePunctuationDependencies));
}
