JS: Added parser intersection support

Author: Napalys

javascript/extractor/src/com/semmle/js/ast/regexp/Intersection.java

@@ -0,0 +1,22 @@
+package com.semmle.js.ast.regexp;
+
+import com.semmle.js.ast.SourceLocation;
+import java.util.List;
+
+public class Intersection extends RegExpTerm {
+    private final List<RegExpTerm> intersections;
+
+    public Intersection(SourceLocation loc, List<RegExpTerm> intersections) {
+        super(loc, "Intersection");
+        this.intersections = intersections;
+    }
+
+    @Override
+    public void accept(Visitor v) {
+        v.visit(this);
+    }
+
+    public List<RegExpTerm> getIntersections() {
+        return intersections;
+    }
+}

javascript/extractor/src/com/semmle/js/ast/regexp/Visitor.java

@@ -61,4 +61,6 @@ public interface Visitor {
   public void visit(ZeroWidthNegativeLookbehind nd);
 
   public void visit(UnicodePropertyEscape nd);
+
+  public void visit(Intersection nd);
 }

javascript/extractor/src/com/semmle/js/extractor/RegExpExtractor.java

@@ -22,6 +22,7 @@
 import com.semmle.js.ast.regexp.Group;
 import com.semmle.js.ast.regexp.HexEscapeSequence;
 import com.semmle.js.ast.regexp.IdentityEscape;
+import com.semmle.js.ast.regexp.Intersection;
 import com.semmle.js.ast.regexp.Literal;
 import com.semmle.js.ast.regexp.NamedBackReference;
 import com.semmle.js.ast.regexp.NonWordBoundary;
@@ -92,6 +93,7 @@ public RegExpExtractor(TrapWriter trapwriter, LocationManager locationManager) {
     termkinds.put("ZeroWidthPositiveLookbehind", 25);
     termkinds.put("ZeroWidthNegativeLookbehind", 26);
     termkinds.put("UnicodePropertyEscape", 27);
+    termkinds.put("Intersection", 28);
   }
 
   private static final String[] errmsgs =
@@ -344,6 +346,14 @@ public void visit(CharacterClassRange nd) {
       visit(nd.getLeft(), lbl, 0);
       visit(nd.getRight(), lbl, 1);
     }
+
+    @Override
+    public void visit(Intersection nd) {
+      Label lbl = extractTerm(nd, parent, idx);
+      int i = 0;
+      for (RegExpTerm element : nd.getIntersections())
+        visit(element, lbl, i++);
+    }
   }
 
   public void extract(String src, SourceMap sourceMap, Node parent, boolean isSpeculativeParsing) {

javascript/extractor/src/com/semmle/js/parser/RegExpParser.java

@@ -18,6 +18,7 @@
 import com.semmle.js.ast.regexp.Group;
 import com.semmle.js.ast.regexp.HexEscapeSequence;
 import com.semmle.js.ast.regexp.IdentityEscape;
+import com.semmle.js.ast.regexp.Intersection;
 import com.semmle.js.ast.regexp.NamedBackReference;
 import com.semmle.js.ast.regexp.NonWordBoundary;
 import com.semmle.js.ast.regexp.OctalEscape;
@@ -225,20 +226,33 @@ private <T extends RegExpTerm> T finishTerm(T term) {
   private RegExpTerm parseDisjunction() {
     SourceLocation loc = new SourceLocation(pos());
     List<RegExpTerm> disjuncts = new ArrayList<>();
-    disjuncts.add(this.parseAlternative());
-    while (this.match("|")) disjuncts.add(this.parseAlternative());
+    disjuncts.add(this.parseIntersection());
+    while (this.match("|")) {
+        disjuncts.add(this.parseIntersection());
+    }
     if (disjuncts.size() == 1) return disjuncts.get(0);
     return this.finishTerm(new Disjunction(loc, disjuncts));
-  }
+}
 
   private RegExpTerm parseAlternative() {
     SourceLocation loc = new SourceLocation(pos());
     List<RegExpTerm> elements = new ArrayList<>();
-    while (!this.lookahead(null, "|", ")")) elements.add(this.parseTerm());
+    while (!this.lookahead(null, "|", "&&", ")")) elements.add(this.parseTerm());
     if (elements.size() == 1) return elements.get(0);
     return this.finishTerm(new Sequence(loc, elements));
   }
 
+  private RegExpTerm parseIntersection() {
+    SourceLocation loc = new SourceLocation(pos());
+    List<RegExpTerm> intersections = new ArrayList<>();
+    intersections.add(this.parseAlternative());
+    while (this.match("&&")) {
+        intersections.add(this.parseAlternative());
+    }
+    if (intersections.size() == 1) return intersections.get(0);
+    return this.finishTerm(new Intersection(loc, intersections));
+}
+  
   private RegExpTerm parseTerm() {
     SourceLocation loc = new SourceLocation(pos());
 

javascript/extractor/tests/es2024/output/trap/intersection.js.trap

@@ -39,7 +39,7 @@ enclosing_stmt(#20009,#20008)
 expr_containers(#20009,#20001)
 literals("/[[abc]&&[bcd]]/v","/[[abc]&&[bcd]]/v",#20009)
 #20010=*
-regexpterm(#20010,1,#20009,0,"[[abc]&&[bcd]]")
+regexpterm(#20010,28,#20009,0,"[[abc]&&[bcd]]")
 #20011=@"loc,{#10000},1,2,1,15"
 locations_default(#20011,#10000,1,2,1,15)
 hasLocation(#20010,#20011)
@@ -73,13 +73,12 @@ locations_default(#20021,#10000,1,6,1,6)
 hasLocation(#20020,#20021)
 regexp_const_value(#20020,"c")
 #20022=*
-regexpterm(#20022,14,#20010,1,"&&")
-#20023=@"loc,{#10000},1,8,1,9"
-locations_default(#20023,#10000,1,8,1,9)
+regexpterm(#20022,1,#20010,1,"[bcd]]")
+#20023=@"loc,{#10000},1,10,1,15"
+locations_default(#20023,#10000,1,10,1,15)
 hasLocation(#20022,#20023)
-regexp_const_value(#20022,"&&")
 #20024=*
-regexpterm(#20024,23,#20010,2,"[bcd]")
+regexpterm(#20024,23,#20022,0,"[bcd]")
 #20025=@"loc,{#10000},1,10,1,14"
 locations_default(#20025,#10000,1,10,1,14)
 hasLocation(#20024,#20025)
@@ -102,7 +101,7 @@ locations_default(#20031,#10000,1,13,1,13)
 hasLocation(#20030,#20031)
 regexp_const_value(#20030,"d")
 #20032=*
-regexpterm(#20032,14,#20010,3,"]")
+regexpterm(#20032,14,#20022,1,"]")
 #20033=@"loc,{#10000},1,15,1,15"
 locations_default(#20033,#10000,1,15,1,15)
 hasLocation(#20032,#20033)

javascript/ql/lib/semmlecode.javascript.dbscheme

@@ -859,7 +859,8 @@ case @regexpterm.kind of
 | 24 = @regexp_char_range
 | 25 = @regexp_positive_lookbehind
 | 26 = @regexp_negative_lookbehind
-| 27 = @regexp_unicode_property_escape;
+| 27 = @regexp_unicode_property_escape
+| 28 = @regexp_intersection;
 
 regexp_parse_errors (unique int id: @regexp_parse_error,
     int regexp: @regexpterm ref,