public static void RunTest(SemgrexPattern pattern, SemanticGraph graph, params string[] expectedMatches)
{
// results are not in the order I would expect. Using a counter
// allows them to be in any order
IntCounter <string> counts = new IntCounter <string>();
for (int i = 0; i < expectedMatches.Length; ++i)
{
counts.IncrementCount(expectedMatches[i]);
}
IntCounter <string> originalCounts = new IntCounter <string>(counts);
SemgrexMatcher matcher = pattern.Matcher(graph);
for (int i_1 = 0; i_1 < expectedMatches.Length; ++i_1)
{
if (!matcher.Find())
{
throw new AssertionFailedError("Expected " + expectedMatches.Length + " matches for pattern " + pattern + " on " + graph + ", only got " + i_1);
}
string match = matcher.GetMatch().ToString();
if (!counts.ContainsKey(match))
{
throw new AssertionFailedError("Unexpected match " + match + " for pattern " + pattern + " on " + graph);
}
counts.DecrementCount(match);
if (counts.GetCount(match) < 0)
{
throw new AssertionFailedError("Found too many matches for " + match + " for pattern " + pattern + " on " + graph);
}
}
if (matcher.FindNextMatchingNode())
{
throw new AssertionFailedError("Found more than " + expectedMatches.Length + " matches for pattern " + pattern + " on " + graph + "... extra match is " + matcher.GetMatch());
}
}
public virtual void TestNamedRelation()
{
SemanticGraph graph = SemanticGraph.ValueOf("[ate subj>Bill dobj>[muffins compound>blueberry]]");
SemgrexPattern pattern = SemgrexPattern.Compile("{idx:0}=gov >>=foo {idx:3}=dep");
SemgrexMatcher matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual("ate", matcher.GetNode("gov").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("dep").ToString());
NUnit.Framework.Assert.AreEqual("compound", matcher.GetRelnString("foo"));
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{idx:3}=dep <<=foo {idx:0}=gov");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual("ate", matcher.GetNode("gov").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("dep").ToString());
NUnit.Framework.Assert.AreEqual("dobj", matcher.GetRelnString("foo"));
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{idx:3}=dep <=foo {idx:2}=gov");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("gov").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("dep").ToString());
NUnit.Framework.Assert.AreEqual("compound", matcher.GetRelnString("foo"));
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{idx:2}=gov >=foo {idx:3}=dep");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("gov").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("dep").ToString());
NUnit.Framework.Assert.AreEqual("compound", matcher.GetRelnString("foo"));
NUnit.Framework.Assert.IsFalse(matcher.Find());
}
public static void OutputResults(SemgrexPattern pattern, SemanticGraph graph, params string[] ignored)
{
System.Console.Out.WriteLine("Matching pattern " + pattern + " to\n" + graph + " :" + (pattern.Matcher(graph).Matches() ? "matches" : "doesn't match"));
System.Console.Out.WriteLine();
pattern.PrettyPrint();
System.Console.Out.WriteLine();
SemgrexMatcher matcher = pattern.Matcher(graph);
while (matcher.Find())
{
System.Console.Out.WriteLine(" " + matcher.GetMatch());
ICollection <string> nodeNames = matcher.GetNodeNames();
if (nodeNames != null && nodeNames.Count > 0)
{
foreach (string name in nodeNames)
{
System.Console.Out.WriteLine(" " + name + ": " + matcher.GetNode(name));
}
}
ICollection <string> relNames = matcher.GetRelationNames();
if (relNames != null)
{
foreach (string name in relNames)
{
System.Console.Out.WriteLine(" " + name + ": " + matcher.GetRelnString(name));
}
}
}
}
public virtual void TestEnv()
{
SemanticGraph h = SemanticGraph.ValueOf("[married/VBN nsubjpass>Hughes/NNP auxpass>was/VBD nmod:to>Gracia/NNP]");
h.GetFirstRoot().Set(typeof(PatternsAnnotations.PatternLabel1), "YES");
//SemanticGraph t = SemanticGraph
// .valueOf("[loved/VBD\nnsubj:Hughes/NNP\ndobj:[wife/NN poss:his/PRP$ appos:Gracia/NNP]\nconj_and:[obsessed/JJ\ncop:was/VBD\nadvmod:absolutely/RB\nprep_with:[Elicia/NN poss:his/PRP$ amod:little/JJ nn:daughter/NN]]]");
string macro = "macro WORD = married";
Env env = new Env();
env.Bind("pattern1", typeof(PatternsAnnotations.PatternLabel1));
string pattern = "({pattern1:YES}=parent >>nsubjpass {}=node)";
IList <SemgrexPattern> pats = SemgrexBatchParser.CompileStream(new ByteArrayInputStream(Sharpen.Runtime.GetBytesForString((macro + "\n" + pattern), StandardCharsets.Utf8)), env);
SemgrexPattern pat3 = pats[0];
bool ignoreCase = true;
SemgrexMatcher mat3 = pat3.Matcher(h, ignoreCase);
if (mat3.Find())
{
string parent = mat3.GetNode("parent").Word();
string node = mat3.GetNode("node").Word();
System.Console.Out.WriteLine("Result: parent is " + parent + " and node is " + node);
NUnit.Framework.Assert.AreEqual(parent, "married");
NUnit.Framework.Assert.AreEqual(node, "Hughes");
}
else
{
throw new Exception("failed!");
}
}
public virtual void TestEqualsRelation()
{
SemanticGraph graph = SemanticGraph.ValueOf("[ate subj>Bill dobj>[muffins compound>blueberry]]");
SemgrexPattern pattern = SemgrexPattern.Compile("{} >> ({}=a == {}=b)");
SemgrexMatcher matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
// This split pattern should also work
pattern = SemgrexPattern.Compile("{} >> {}=a >> {}=b : {}=a == {}=b");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
}
public virtual void TestMatchAll()
{
SemanticGraph graph = SemanticGraph.ValueOf("[ate subj>Bill dobj>[muffins compound>blueberry]]");
ICollection <IndexedWord> words = graph.VertexSet();
SemgrexPattern pattern = SemgrexPattern.Compile("{}");
SemgrexMatcher matcher = pattern.Matcher(graph);
string[] expectedMatches = new string[] { "ate", "Bill", "muffins", "blueberry" };
for (int i = 0; i < expectedMatches.Length; ++i)
{
NUnit.Framework.Assert.IsTrue(matcher.FindNextMatchingNode());
}
NUnit.Framework.Assert.IsFalse(matcher.FindNextMatchingNode());
}
public virtual void TestNamedNode()
{
SemanticGraph graph = MakeComplicatedGraph();
RunTest("{} >dobj ({} >expl {})", graph, "A");
SemgrexPattern pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo)");
SemgrexMatcher matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(1, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("foo").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo) >mod {}");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(1, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("foo").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo) >mod ({} >mark {})");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(1, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("foo").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo) >mod ({} > {})");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(1, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("foo").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo) >mod ({} > {}=foo)");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(1, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("foo").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
pattern = SemgrexPattern.Compile("{} >dobj ({} >expl {}=foo) >mod ({}=foo > {})");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsFalse(matcher.Find());
}
public virtual void TestInitialConditions()
{
SemanticGraph graph = MakeComplicatedGraph();
SemgrexPattern pattern = SemgrexPattern.Compile("{}=a >> {}=b : {}=a >> {}=c");
IDictionary <string, IndexedWord> variables = new Dictionary <string, IndexedWord>();
variables["b"] = graph.GetNodeByIndex(5);
variables["c"] = graph.GetNodeByIndex(2);
SemgrexMatcher matcher = pattern.Matcher(graph, variables);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(3, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("A", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("E", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.AreEqual("B", matcher.GetNode("c").ToString());
NUnit.Framework.Assert.AreEqual("A", matcher.GetMatch().ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
}
private void ResetChild()
{
if (childMatcher == null)
{
if (myNode.child == null)
{
matchedOnce = false;
}
else
{
childMatcher = myNode.child.Matcher(sg, alignment, sg_aligned, (myNode.reln is GraphRelation.ALIGNMENT) ? !hyp : hyp, nextMatch, namesToNodes, namesToRelations, variableStrings, ignoreCase);
}
}
else
{
childMatcher.ResetChildIter(nextMatch);
}
}
public virtual void TestMacro()
{
SemanticGraph h = SemanticGraph.ValueOf("[married/VBN nsubjpass>Hughes/NNP auxpass>was/VBD nmod:to>Gracia/NNP]");
string macro = "macro WORD = married";
string pattern = "({word:${WORD}}=parent >>nsubjpass {}=node)";
IList <SemgrexPattern> pats = SemgrexBatchParser.CompileStream(new ByteArrayInputStream(Sharpen.Runtime.GetBytesForString((macro + "\n" + pattern), StandardCharsets.Utf8)));
SemgrexPattern pat3 = pats[0];
bool ignoreCase = true;
SemgrexMatcher mat3 = pat3.Matcher(h, ignoreCase);
if (mat3.Find())
{
string parent = mat3.GetNode("parent").Word();
string node = mat3.GetNode("node").Word();
System.Console.Out.WriteLine("Result: parent is " + parent + " and node is " + node);
NUnit.Framework.Assert.AreEqual(parent, "married");
NUnit.Framework.Assert.AreEqual(node, "Hughes");
}
else
{
throw new Exception("failed!");
}
}
public virtual void TestNotEquals()
{
SemanticGraph graph = SemanticGraph.ValueOf("[ate subj>Bill dobj>[muffins compound>blueberry]]");
SemgrexPattern pattern = SemgrexPattern.Compile("{} >> {}=a >> {}=b : {}=a !== {}=b");
SemgrexMatcher matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
// same as the first test, essentially, but with a more compact expression
pattern = SemgrexPattern.Compile("{} >> {}=a >> ({}=b !== {}=a)");
matcher = pattern.Matcher(graph);
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("Bill", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsTrue(matcher.Find());
NUnit.Framework.Assert.AreEqual(2, matcher.GetNodeNames().Count);
NUnit.Framework.Assert.AreEqual("ate", matcher.GetMatch().ToString());
NUnit.Framework.Assert.AreEqual("blueberry", matcher.GetNode("a").ToString());
NUnit.Framework.Assert.AreEqual("muffins", matcher.GetNode("b").ToString());
NUnit.Framework.Assert.IsFalse(matcher.Find());
}
/// <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 virtual void TestFind()
{
SemanticGraph h = SemanticGraph.ValueOf("[married/VBN nsubjpass>Hughes/NNP auxpass>was/VBD prep_to>Gracia/NNP]");
SemanticGraph t = SemanticGraph.ValueOf("[loved/VBD\nnsubj>Hughes/NNP\ndobj>[wife/NN poss>his/PRP$ appos>Gracia/NNP]\nconj_and>[obsessed/JJ\ncop>was/VBD\nadvmod>absolutely/RB\nprep_with>[Elicia/NN poss>his/PRP$ amod>little/JJ compound>daughter/NN]]]"
);
string s = "(ROOT\n(S\n(NP (DT The) (NN chimney) (NNS sweeps))\n(VP (VBP do) (RB not)\n(VP (VB like)\n(S\n(VP (VBG working)\n(PP (IN on)\n(NP (DT an) (JJ empty) (NN stomach)))))))\n(. .)))";
Tree tree = Tree.ValueOf(s);
SemanticGraph sg = SemanticGraphFactory.MakeFromTree(tree, SemanticGraphFactory.Mode.Collapsed, GrammaticalStructure.Extras.Maximal, null);
SemgrexPattern pat = SemgrexPattern.Compile("{}=gov ![>det {}] & > {word:/^(?!not).*$/}=dep");
sg.PrettyPrint();
// SemgrexPattern pat =
// SemgrexPattern.compile("{} [[<prep_to ({word:married} >nsubjpass {})] | [<nsubjpass ({word:married} >prep_to {})]]");
pat.PrettyPrint();
SemgrexMatcher mat = pat.Matcher(sg);
while (mat.Find())
{
// String match = mat.getMatch().word();
string gov = mat.GetNode("gov").Word();
// String reln = mat.getRelnString("reln");
string dep = mat.GetNode("dep").Word();
// System.out.println(match);
System.Console.Out.WriteLine(dep + ' ' + gov);
}
SemgrexPattern pat2 = SemgrexPattern.Compile("{} [[>/nn|appos/ ({lemma:/wife|husband|partner/} >/poss/ {}=txtPartner)] | [<poss ({}=txtPartner >/nn|appos/ {lemma:/wife|husband|partner/})]" + "| [<nsubj ({$} >> ({word:/wife|husband|partner/} >poss {word:/his|her/} >/nn|appos/ {}))]]"
);
SemgrexMatcher mat2 = pat2.Matcher(t);
while (mat2.Find())
{
string match = mat2.GetMatch().Word();
// String gov = mat.getNode("gov").word();
// String reln = mat.getRelnString("reln");
// String dep = mat.getNode("dep").word();
System.Console.Out.WriteLine(match);
}
// System.out.println(dep + " " + gov);
Dictionary <IndexedWord, IndexedWord> map = new Dictionary <IndexedWord, IndexedWord>();
map[h.GetNodeByWordPattern("Hughes")] = t.GetNodeByWordPattern("Hughes");
map[h.GetNodeByWordPattern("Gracia")] = t.GetNodeByWordPattern("Gracia");
Alignment alignment = new Alignment(map, 0, string.Empty);
SemgrexPattern fullPat = SemgrexPattern.Compile("({}=partnerOne [[<prep_to ({word:married} >nsubjpass {}=partnerTwo)] | [<nsubjpass ({word:married} >prep_to {}=partnerTwo)]]) @ ({} [[>/nn|appos/ ({lemma:/wife|husband|partner/} >/poss/ {}=txtPartner)] | [<poss ({}=txtPartner >/nn|appos/ {lemma:/wife|husband|partner/})]"
+ "| [<nsubj ({$} >> ({word:/wife|husband|partner/} >poss {word:/his|her/} >/nn|appos/ {}=txtPartner))]])");
fullPat.PrettyPrint();
SemgrexMatcher fullMat = fullPat.Matcher(h, alignment, t);
if (fullMat.Find())
{
System.Console.Out.WriteLine("woo: " + fullMat.GetMatch().Word());
System.Console.Out.WriteLine(fullMat.GetNode("txtPartner"));
System.Console.Out.WriteLine(fullMat.GetNode("partnerOne"));
System.Console.Out.WriteLine(fullMat.GetNode("partnerTwo"));
}
else
{
System.Console.Out.WriteLine("boo");
}
SemgrexPattern pat3 = SemgrexPattern.Compile("({word:LIKE}=parent >>/aux.*/ {word:/do/}=node)");
System.Console.Out.WriteLine("pattern is ");
pat3.PrettyPrint();
System.Console.Out.WriteLine("tree is ");
sg.PrettyPrint();
//checking if ignoring case or not
SemgrexMatcher mat3 = pat3.Matcher(sg, true);
if (mat3.Find())
{
string parent = mat3.GetNode("parent").Word();
string node = mat3.GetNode("node").Word();
System.Console.Out.WriteLine("Result: parent is " + parent + " and node is " + node);
NUnit.Framework.Assert.AreEqual(parent, "like");
NUnit.Framework.Assert.AreEqual(node, "do");
}
else
{
NUnit.Framework.Assert.Fail();
}
}
public virtual void TestSiblingPatterns()
{
SemanticGraph sg = SemanticGraph.ValueOf("[loved/VBD-2\nnsubj>Hughes/NNP-1\ndobj>[wife/NN-4 nmod:poss>his/PRP$-3 appos>Gracia/NNP-5]\nconj:and>[obsessed/JJ-9\ncop>was/VBD-7\nadvmod>absolutely/RB-8\nnmod:with>[Elicia/NN-14 nmod:poss>his/PRP$-11 amod>little/JJ-12 compound>daughter/NN-13]]]"
);
/* Test "." */
SemgrexPattern pat1 = SemgrexPattern.Compile("{tag:NNP}=w1 . {tag:VBD}=w2");
SemgrexMatcher matcher = pat1.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
string w2 = matcher.GetNode("w2").Word();
NUnit.Framework.Assert.AreEqual("Hughes", w1);
NUnit.Framework.Assert.AreEqual("loved", w2);
}
else
{
throw new Exception("failed!");
}
/* Test "$+" */
SemgrexPattern pat2 = SemgrexPattern.Compile("{word:was}=w1 $+ {}=w2");
matcher = pat2.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
string w2 = matcher.GetNode("w2").Word();
NUnit.Framework.Assert.AreEqual("was", w1);
NUnit.Framework.Assert.AreEqual("absolutely", w2);
}
else
{
throw new Exception("failed!");
}
/* Test "$-" */
SemgrexPattern pat3 = SemgrexPattern.Compile("{word:absolutely}=w1 $- {}=w2");
matcher = pat3.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
string w2 = matcher.GetNode("w2").Word();
NUnit.Framework.Assert.AreEqual("absolutely", w1);
NUnit.Framework.Assert.AreEqual("was", w2);
}
else
{
throw new Exception("failed!");
}
/* Test "$++" */
SemgrexPattern pat4 = SemgrexPattern.Compile("{word:his}=w1 $++ {tag:NN}=w2");
matcher = pat4.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
string w2 = matcher.GetNode("w2").Word();
NUnit.Framework.Assert.AreEqual("his", w1);
NUnit.Framework.Assert.AreEqual("daughter", w2);
}
else
{
throw new Exception("failed!");
}
/* Test "$--" */
SemgrexPattern pat6 = SemgrexPattern.Compile("{word:daughter}=w1 $-- {tag:/PRP./}=w2");
matcher = pat6.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
string w2 = matcher.GetNode("w2").Word();
NUnit.Framework.Assert.AreEqual("daughter", w1);
NUnit.Framework.Assert.AreEqual("his", w2);
}
else
{
throw new Exception("failed!");
}
/* Test for not matching. */
SemgrexPattern pat5 = SemgrexPattern.Compile("{word:his}=w1 $-- {}=w2");
matcher = pat5.Matcher(sg);
if (matcher.Find())
{
throw new Exception("failed!");
}
/* Test for negation. */
SemgrexPattern pat7 = SemgrexPattern.Compile("{word:his}=w1 !$-- {}");
matcher = pat7.Matcher(sg);
if (matcher.Find())
{
string w1 = matcher.GetNode("w1").Word();
NUnit.Framework.Assert.AreEqual("his", w1);
}
else
{
throw new Exception("failed!");
}
SemgrexPattern pat8 = SemgrexPattern.Compile("{word:his}=w1 !$++ {}");
matcher = pat8.Matcher(sg);
if (matcher.Find())
{
throw new Exception("failed!");
}
}
