Add a Semgrex feature to test for connection in either direction, either gov or dep. Greatly simplifies expressions which previously needed to look in both directions

Author: AngledLuffa

src/edu/stanford/nlp/semgraph/semgrex/GraphRelation.java

@@ -1012,6 +1012,84 @@ void advance() {
     private static final long serialVersionUID = 1L;
   };
 
+  /**
+   * Iterates over nodes which are either gov or dep of the current node.
+   *<br>
+   * In other words, this can take the place of a disjunction over &lt; and &gt;
+   */
+  static private class CONNECTED extends GraphRelation {
+    CONNECTED(String reln, String name, String edgeName) {
+      super("<>", reln, name);
+    }
+
+    @Override
+    boolean satisfies(IndexedWord l1, IndexedWord l2, SemanticGraph sg) {
+      if (l1.equals(IndexedWord.NO_WORD) || l2.equals(IndexedWord.NO_WORD) )
+        return false;
+      List<Pair<GrammaticalRelation, IndexedWord>> govs = sg.parentPairs(l1);
+      for (Pair<GrammaticalRelation, IndexedWord> gov : govs) {
+        if (this.type.test(gov.first().toString()) &&
+            gov.second().equals(l2)) return true;
+      }
+      List<Pair<GrammaticalRelation, IndexedWord>> deps = sg.childPairs(l1);
+      for (Pair<GrammaticalRelation, IndexedWord> dep : deps) {
+        if (this.type.test(dep.first().toString()) &&
+            dep.second().equals(l2)) {
+          return true;
+        }
+      }
+      return false;
+    }
+
+
+    @Override
+    Iterator<IndexedWord> searchNodeIterator(final IndexedWord node, final SemanticGraph sg) {
+      return new SearchNodeIterator() {
+          Iterator<SemanticGraphEdge> it1;
+          Iterator<SemanticGraphEdge> it2;
+
+          @Override
+          public void advance() {
+            if (node.equals(IndexedWord.NO_WORD)) {
+              next = null;
+              return;
+            }
+            if (it1 == null) {
+              it1 = sg.outgoingEdgeIterator(node);
+            }
+            while (it1.hasNext()) {
+              SemanticGraphEdge edge = it1.next();
+              relation = edge.getRelation().toString();
+              if (!type.test(relation)) {
+                continue;
+              }
+              this.next = edge.getTarget();
+              this.edge = edge;
+              return;
+            }
+            if (it2 == null) {
+              it2 = sg.incomingEdgeIterator(node);
+            }
+            while (it2.hasNext()) {
+              SemanticGraphEdge edge = it2.next();
+              relation = edge.getRelation().toString();
+              if (!type.test(relation)) {
+                continue;
+              }
+              this.next = edge.getSource();
+              this.edge = edge;
+              return;
+            }
+            this.next = null;
+            this.edge = null;
+          }
+      };
+    }
+
+    // automatically generated by button mashing
+    private static final long serialVersionUID = -413981378678L;
+  }
+
   static private class EQUALS extends GraphRelation {
     EQUALS(String reln, String name) {
       super("==", reln, name);
@@ -1362,6 +1440,7 @@ public void advance() {
   public static boolean isKnownRelation(String reln) {
     return (reln.equals(">") || reln.equals("<") ||
             reln.equals(">>") || reln.equals("<<") ||
+            reln.equals("<>") ||
             reln.equals("@") || reln.equals("==") ||
             reln.equals("$+") || reln.equals("$++") ||
             reln.equals("$-") || reln.equals("$--") ||
@@ -1393,6 +1472,8 @@ public static GraphRelation getRelation(String reln,
         return new DEPENDENT_RIGHT(type, name, edgeName);
       case "<--":
         return new DEPENDENT_LEFT(type, name, edgeName);
+      case "<>":
+        return new CONNECTED(type, name, edgeName);
     }
     if (edgeName != null) {
       throw new ParseException("Relation " + reln + " does not allow for named edges");

src/edu/stanford/nlp/semgraph/semgrex/SemgrexParser.jj

@@ -62,7 +62,7 @@ SPECIAL_TOKEN:
 
 TOKEN:
 {
-	< RELATION: "<" | ">" | ">>" | "<<" | "==" | "$+" | "$-" | "$++" | "$--" | "." | ".." | "-" | "--" | ">++" | ">--" | "<++" | "<--" >
+	< RELATION: "<" | ">" | ">>" | "<<" | "<>" | "==" | "$+" | "$-" | "$++" | "$--" | "." | ".." | "-" | "--" | ">++" | ">--" | "<++" | "<--" >
 |	< ALIGNRELN: "@" >
 |	< IDENTIFIER: (~[" ", "\n", "\r", "(", "/", "|", "@", "!", "#", "%", "&", ")", "=", "?", "[", "]", ">", "<", "~", ".", ",", "$", ":", ";", "{", "}", "+", "-"])+ >
 |	< NUMBER: ( ["0"-"9"] )+ >

src/edu/stanford/nlp/semgraph/semgrex/SemgrexParserTokenManager.java

@@ -171,9 +171,9 @@ else if ((0x100002200L & l) != 0L)
                      { jjCheckNAdd(0); }
                   }
                   else if (curChar == 60)
-                     { jjAddStates(7, 9); }
+                     { jjCheckNAddStates(7, 10); }
                   else if (curChar == 62)
-                     { jjAddStates(10, 12); }
+                     { jjCheckNAddStates(11, 13); }
                   else if (curChar == 45)
                      { jjCheckNAdd(6); }
                   else if (curChar == 46)
@@ -257,7 +257,7 @@ else if (curChar == 43)
                   break;
                case 15:
                   if (curChar == 62)
-                     { jjAddStates(10, 12); }
+                     { jjCheckNAddStates(11, 13); }
                   break;
                case 16:
                   if (curChar == 62 && kind > 2)
@@ -274,7 +274,7 @@ else if (curChar == 43)
                   break;
                case 20:
                   if (curChar == 60)
-                     { jjAddStates(7, 9); }
+                     { jjCheckNAddStates(7, 10); }
                   break;
                case 21:
                   if (curChar == 60 && kind > 2)
@@ -388,7 +388,7 @@ protected Token jjFillToken()
    return t;
 }
 static final int[] jjnextStates = {
-   17, 6, 25, 26, 12, 13, 14, 21, 22, 23, 16, 18, 19, 
+   17, 6, 25, 26, 12, 13, 14, 21, 16, 22, 23, 16, 18, 19, 
 };
 private static final boolean jjCanMove_0(int hiByte, int i1, int i2, long l1, long l2)
 {

src/edu/stanford/nlp/semgraph/semgrex/SemgrexPattern.java

@@ -80,6 +80,7 @@
  * <tr><td>A &gt;&gt;reln B <td>A is the governor of a relation reln in a chain to B following {@code gov->dep} paths
  * <tr><td>{@code A x,y<<reln B} <td>A is the dependent of a relation reln in a chain to B following {@code dep->gov} paths between distances of x and y
  * <tr><td>{@code A x,y>>reln B} <td>A is the governor of a relation reln in a chain to B following {@code gov->dep} paths between distances of x and y
+ * <tr><td>A &lt;&gt;reln B <td> A is connected (either dependent or governor) via relation reln with B
  * <tr><td>A == B <td>A and B are the same nodes in the same graph
  * <tr><td>A . B <td>A immediately precedes B, i.e. A.index() == B.index() - 1
  * <tr><td>A - B <td>A immediately succeeds B, i.e. A.index() == B.index() + 1

test/src/edu/stanford/nlp/semgraph/semgrex/SemgrexTest.java

@@ -118,6 +118,18 @@ public void testSimpleDependency() {
             "ate", "ate", "muffins");
   }
 
+  public void testConnected() {
+    // the root should connect to all its children
+    runTest("{} <> {word:ate}", "[ate subj>Bill obj>[muffins compound>blueberry]]",
+            "Bill", "muffins");
+    // a node in the middle should connect to both its children and its parent
+    runTest("{} <> {word:muffins}", "[ate subj>Bill obj>[muffins compound>blueberry]]",
+            "ate", "blueberry");
+    // a leaf should connect to its parent
+    runTest("{} <> {word:blueberry}", "[ate subj>Bill obj>[muffins compound>blueberry]]",
+            "muffins");
+  }
+
   public void testMultipleAttributes() {
     runTest("{} >> {word:Bill}",
             "[ate subj>Bill/NNP obj>[muffins compound>blueberry]]",