Add the capacity to check variable strings on an exact match or global match by treating any variable as the entire text

Author: AngledLuffa

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

@@ -166,6 +166,21 @@ private Attribute buildRegexAttribute(String key, String value, boolean negated,
     }
   }
 
+  private static boolean checkVarMatch(String key, String matchedString,
+                                       VariableStrings variableStrings, VariableStrings tempVariableStrings) {
+    String existingString = variableStrings.getString(key);
+    if (existingString == null) {
+      existingString = tempVariableStrings.getString(key);
+    }
+    if (existingString != null && !existingString.equals(matchedString)) {
+      return false;
+    }
+    if (matchedString != null) {
+      tempVariableStrings.setVar(key, matchedString);
+    }
+    return true;
+  }
+
   private boolean checkMatch(Attribute attr, boolean ignoreCase, String nodeValue,
                              VariableStrings variableStrings, VariableStrings tempVariableStrings) {
     if (nodeValue == null) {
@@ -179,29 +194,47 @@ private boolean checkMatch(Attribute attr, boolean ignoreCase, String nodeValue,
     boolean matches;
     if (toMatch instanceof Boolean) {
       matches = ((Boolean) toMatch);
+
+      if (matches) {
+        for (Pair<Integer, String> varGroup : attr.variableGroups) {
+          // TODO possibly a bug here - it is not honoring ignoreCase
+          String matchedString = nodeValue;
+          String key = varGroup.second();
+          if (!checkVarMatch(key, matchedString, variableStrings, tempVariableStrings)) {
+            matches = false;
+            break;
+          }
+        }
+      }
     } else if (toMatch instanceof String) {
       if (ignoreCase) {
         matches = nodeValue.equalsIgnoreCase(toMatch.toString());
       } else {
         matches = nodeValue.equals(toMatch.toString());
       }
+
+      if (matches) {
+        for (Pair<Integer, String> varGroup : attr.variableGroups) {
+          // TODO possibly a bug here - it is not honoring ignoreCase
+          String matchedString = nodeValue;
+          String key = varGroup.second();
+          if (!checkVarMatch(key, matchedString, variableStrings, tempVariableStrings)) {
+            matches = false;
+            break;
+          }
+        }
+      }
     } else if (toMatch instanceof Pattern) {
       Matcher matcher = ((Pattern) toMatch).matcher(nodeValue);
       if (matcher.matches()) {
         matches = true;
         for (Pair<Integer, String> varGroup : attr.variableGroups) {
-          String existingString = variableStrings.getString(varGroup.second());
-          if (existingString == null) {
-            existingString = tempVariableStrings.getString(varGroup.second());
-          }
           String matchedString = matcher.group(varGroup.first());
-          if (existingString != null && !existingString.equals(matchedString)) {
+          String key = varGroup.second();
+          if (!checkVarMatch(key, matchedString, variableStrings, tempVariableStrings)) {
             matches = false;
             break;
           }
-          if (matchedString != null) {
-            tempVariableStrings.setVar(varGroup.second(), matchedString);
-          }
         }
       } else {
         matches = false;

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

@@ -1572,7 +1572,7 @@ public void testBatchUniq() {
     assertEquals(1, matches.get(2).second().size());
   }
 
-  public void testVariableGroups() {
+  public void testRegexVariableGroups() {
     // first, a basic test that it is capturing the variable groups correctly
     SemgrexPattern pattern = SemgrexPattern.compile("{word:/(.*ill.*)/#1%name}");
     SemanticGraph graph = SemanticGraph.valueOf("[ate-2 subj> Bill-1 obj>[muffins-6 compound> Blueberry-3 compound> Flueberry-4 compound> filled-5]]");
@@ -1612,6 +1612,25 @@ public void testVariableGroups() {
     assertEquals(expectedMatches, matches);
   }
 
+  public void testExactVariableGroups() {
+    SemgrexPattern pattern = SemgrexPattern.compile("{word:__#1%name} .. {word:__#1%name}");
+    SemanticGraph graph = SemanticGraph.valueOf("[ate-2 subj> Bill-1 obj>[muffins-6 compound> Blueberry-3 compound> Bill-4 compound> filled-5]]");
+
+    // This should match exactly once, for Bill & Bill
+    SemgrexMatcher matcher = pattern.matcher(graph);
+    assertTrue(matcher.find());
+    assertEquals("Bill", matcher.variableStrings.getString("name"));
+    assertFalse(matcher.find());
+
+    pattern = SemgrexPattern.compile("{word:Bill#1%name} .. {word:__#1%name}");
+
+    // This should match exactly once, for Bill & Bill
+    matcher = pattern.matcher(graph);
+    assertTrue(matcher.find());
+    assertEquals("Bill", matcher.variableStrings.getString("name"));
+    assertFalse(matcher.find());
+  }
+
   public static void outputBatchResults(SemgrexPattern pattern, List<CoreMap> sentences) {
     List<Pair<CoreMap, List<SemgrexMatch>>> matches = pattern.matchSentences(sentences, false);
     for (Pair<CoreMap, List<SemgrexMatch>> sentenceMatch : matches) {