Fix the semgrex parser usage of includes

stanfordnlp · Feb 27, 2025 · 3063b26 · 3063b26
1 parent 33de4e1
commit 3063b26
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 30 deletions.
diff --git a/src/edu/stanford/nlp/semgraph/semgrex/SemgrexParser.java b/src/edu/stanford/nlp/semgraph/semgrex/SemgrexParser.java
@@ -530,11 +530,11 @@ final public SemgrexPattern Root() throws ParseException {// Root pattern for th
         Token value = null;
         Token attrType = null;
     switch ((jj_ntk==-1)?jj_ntk_f():jj_ntk) {
-    case ALIGNRELN:
     case IDENTIFIER:{
+      attr = jj_consume_token(IDENTIFIER);
       switch ((jj_ntk==-1)?jj_ntk_f():jj_ntk) {
-      case IDENTIFIER:{
-        attr = jj_consume_token(IDENTIFIER);
+      case 10:
+      case 22:{
         switch ((jj_ntk==-1)?jj_ntk_f():jj_ntk) {
         case 10:{
           attrType = jj_consume_token(10);
@@ -563,6 +563,10 @@ final public SemgrexPattern Root() throws ParseException {// Root pattern for th
           jj_consume_token(-1);
           throw new ParseException();
         }
+if (attr != null && value != null) {
+               boolean negated = attrType.image.equals("!:");
+               attributes.setAttribute(attr.image, value.image, negated);
+             }
         break;
         }
       case ALIGNRELN:{
@@ -583,23 +587,20 @@ final public SemgrexPattern Root() throws ParseException {// Root pattern for th
           jj_consume_token(-1);
           throw new ParseException();
         }
+if (attrType.image.equals("@")) {
+               if (attr == null || key == null || value == null) {
+                 {if (true) throw new SemgrexParseException("null while parsing semgrex expression: attr=" + attr +
+                                                 " key=" + key + " value=" + value);}
+               }
+               attributes.addContains(attr.image, key.image, value.image);
+             }
         break;
         }
       default:
         jj_la1[26] = jj_gen;
         jj_consume_token(-1);
         throw new ParseException();
       }
-if (attrType.image.equals("@")) {
-             if (attr == null || key == null || value == null) {
-               {if (true) throw new SemgrexParseException("null while parsing semgrex expression: attr=" + attr +
-                                               " key=" + key + " value=" + value);}
-             }
-             attributes.addContains(attr.image, key.image, value.image);
-           } else if (attr != null && value != null) {
-             boolean negated = attrType.image.equals("!:");
-             attributes.setAttribute(attr.image, value.image, negated);
-           }
       break;
       }
     case ROOT:{
@@ -625,7 +626,6 @@ final public SemgrexPattern Root() throws ParseException {// Root pattern for th
         NodePattern pat;
     jj_consume_token(23);
     switch ((jj_ntk==-1)?jj_ntk_f():jj_ntk) {
-    case ALIGNRELN:
     case IDENTIFIER:
     case EMPTY:
     case ROOT:{
@@ -690,7 +690,7 @@ final public SemgrexPattern Root() throws ParseException {// Root pattern for th
 	   jj_la1_init_0();
 	}
 	private static void jj_la1_init_0() {
-	   jj_la1_0 = new int[] {0x400,0x828808,0x3801c,0x3801c,0x828800,0x2000,0x3c01c,0x4000,0x3801c,0x2001c,0x80000,0x10,0x110,0x110,0x100000,0x200000,0x1c,0x828800,0x2000,0x82c000,0x4000,0x828000,0x820000,0x400400,0x110,0x110,0x18,0xd8,0x1000000,0xd8,0x200000,};
+	   jj_la1_0 = new int[] {0x400,0x828808,0x3801c,0x3801c,0x828800,0x2000,0x3c01c,0x4000,0x3801c,0x2001c,0x80000,0x10,0x110,0x110,0x100000,0x200000,0x1c,0x828800,0x2000,0x82c000,0x4000,0x828000,0x820000,0x400400,0x110,0x110,0x400408,0xd0,0x1000000,0xd0,0x200000,};
 	}
 
   /** Constructor with InputStream. */

diff --git a/src/edu/stanford/nlp/semgraph/semgrex/SemgrexParser.jj b/src/edu/stanford/nlp/semgraph/semgrex/SemgrexParser.jj
@@ -275,21 +275,25 @@ void AddAttribute(NodeAttributes attributes) : {
 	Token value = null;
         Token attrType = null;
 } {
-        ((attr = <IDENTIFIER>
-           ( (attrType = ":" | attrType = "!:") (value = <IDENTIFIER> | value = <REGEX>) ) |
-           ( (attrType = "@") (key = <IDENTIFIER>) "=" (value = <IDENTIFIER> | value = <REGEX>) ) )
-         {
-           if (attrType.image.equals("@")) {
-             if (attr == null || key == null || value == null) {
-               throw new SemgrexParseException("null while parsing semgrex expression: attr=" + attr +
-                                               " key=" + key + " value=" + value);
+        (attr = <IDENTIFIER>
+           (( (attrType = ":" | attrType = "!:") (value = <IDENTIFIER> | value = <REGEX>) {
+             if (attr != null && value != null) {
+               boolean negated = attrType.image.equals("!:");
+               attributes.setAttribute(attr.image, value.image, negated);
              }
-             attributes.addContains(attr.image, key.image, value.image);
-           } else if (attr != null && value != null) {
-             boolean negated = attrType.image.equals("!:");
-             attributes.setAttribute(attr.image, value.image, negated);
-           }
-         })
+           }) 
+           |
+           (attrType = "@") (key = <IDENTIFIER>) "=" (value = <IDENTIFIER> | value = <REGEX>)
+           {
+             if (attrType.image.equals("@")) {
+               if (attr == null || key == null || value == null) {
+                 throw new SemgrexParseException("null while parsing semgrex expression: attr=" + attr +
+                                                 " key=" + key + " value=" + value);
+               }
+               attributes.addContains(attr.image, key.image, value.image);
+             }
+           })
+        )
 	|
 	( attr = <ROOT> { attributes.setRoot(true); } )
 	|

diff --git a/test/src/edu/stanford/nlp/semgraph/semgrex/SemgrexTest.java b/test/src/edu/stanford/nlp/semgraph/semgrex/SemgrexTest.java
@@ -234,7 +234,7 @@ public void testBrokenContainsExpression() {
 
   public void testContainsExpression() {
     // morphofeatures is a Map, so this should work
-    //SemgrexPattern pattern = SemgrexPattern.compile("{morphofeatures@foo=bar}");
+    SemgrexPattern pattern = SemgrexPattern.compile("{morphofeatures@foo=bar}");
   }
 
   public void testReferencedRegex() {