Add support for '\q{}' escape sequence in regular expressions.

Napalys · Napalys · commit ed418be97a07 · 2025-03-03T14:37:20.000+01:00
diff --git a/javascript/extractor/src/com/semmle/js/ast/regexp/CharacterClassQuotedString.java b/javascript/extractor/src/com/semmle/js/ast/regexp/CharacterClassQuotedString.java
@@ -0,0 +1,25 @@
+package com.semmle.js.ast.regexp;
+
+import com.semmle.js.ast.SourceLocation;
+
+/**
+ * A '\q{}' escape sequence in a regular expression, which is a special extension
+ * to standard regular expressions.
+ */
+public class CharacterClassQuotedString extends RegExpTerm {
+  private final RegExpTerm term;
+
+  public CharacterClassQuotedString(SourceLocation loc, RegExpTerm term) {
+    super(loc, "CharacterClassQuotedString");
+    this.term = term;
+  }
+
+  public RegExpTerm getTerm() {
+    return term;
+  }
+
+  @Override
+  public void accept(Visitor v) {
+    v.visit(this);
+  }
+}
diff --git a/javascript/extractor/src/com/semmle/js/ast/regexp/Visitor.java b/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(CharacterClassQuotedString nd);
 }
diff --git a/javascript/extractor/src/com/semmle/js/extractor/RegExpExtractor.java b/javascript/extractor/src/com/semmle/js/extractor/RegExpExtractor.java
@@ -10,6 +10,7 @@
 import com.semmle.js.ast.regexp.Caret;
 import com.semmle.js.ast.regexp.CharacterClass;
 import com.semmle.js.ast.regexp.CharacterClassEscape;
+import com.semmle.js.ast.regexp.CharacterClassQuotedString;
 import com.semmle.js.ast.regexp.CharacterClassRange;
 import com.semmle.js.ast.regexp.Constant;
 import com.semmle.js.ast.regexp.ControlEscape;
@@ -92,6 +93,7 @@ public RegExpExtractor(TrapWriter trapwriter, LocationManager locationManager) {
     termkinds.put("ZeroWidthPositiveLookbehind", 25);
     termkinds.put("ZeroWidthNegativeLookbehind", 26);
     termkinds.put("UnicodePropertyEscape", 27);
+    termkinds.put("CharacterClassQuotedString", 28);
   }
 
   private static final String[] errmsgs =
@@ -344,6 +346,12 @@ public void visit(CharacterClassRange nd) {
       visit(nd.getLeft(), lbl, 0);
       visit(nd.getRight(), lbl, 1);
     }
+
+    @Override
+    public void visit(CharacterClassQuotedString nd) {
+      Label lbl = extractTerm(nd, parent, idx);
+      visit(nd.getTerm(), lbl, 0);
+    }
   }
 
   public void extract(String src, SourceMap sourceMap, Node parent, boolean isSpeculativeParsing, String flags) {
diff --git a/javascript/extractor/src/com/semmle/js/parser/RegExpParser.java b/javascript/extractor/src/com/semmle/js/parser/RegExpParser.java
@@ -6,6 +6,7 @@
 import com.semmle.js.ast.regexp.Caret;
 import com.semmle.js.ast.regexp.CharacterClass;
 import com.semmle.js.ast.regexp.CharacterClassEscape;
+import com.semmle.js.ast.regexp.CharacterClassQuotedString;
 import com.semmle.js.ast.regexp.CharacterClassRange;
 import com.semmle.js.ast.regexp.Constant;
 import com.semmle.js.ast.regexp.ControlEscape;
@@ -283,6 +284,45 @@ private RegExpTerm parseTerm() {
     return this.finishTerm(this.parseQuantifierOpt(loc, this.parseAtom()));
   }
 
+  private RegExpTerm parseDisjunctionInsideQuotedString() {
+    SourceLocation loc = new SourceLocation(pos());
+    List<RegExpTerm> disjuncts = new ArrayList<>();
+    disjuncts.add(this.parseAlternativeInsideQuotedString());
+    while (this.match("|")) {
+        disjuncts.add(this.parseAlternativeInsideQuotedString());
+    }
+    if (disjuncts.size() == 1) return disjuncts.get(0);
+        return this.finishTerm(new Disjunction(loc, disjuncts));
+  }
+
+  private RegExpTerm parseAlternativeInsideQuotedString() {
+    SourceLocation loc = new SourceLocation(pos());
+    StringBuilder sb = new StringBuilder();
+    boolean escaped = false;
+    while (true) {
+      // If we're at the end of the string, something went wrong.
+      if (this.atEOS()) {
+        this.error(Error.UNEXPECTED_EOS);
+        break;
+      }
+      // We can end parsing if we're not escaped and we see a `|` which would mean Alternation
+      // or `}` which would mean the end of the Quoted String.
+      if(!escaped && this.lookahead(null, "|", "}")){
+        break;
+      }
+      char c = this.nextChar();
+      // Track whether the character is an escape character. 
+      escaped = !escaped && (c == '\\');
+      sb.append(c);
+    }
+    
+    String literal = sb.toString();
+    loc.setEnd(pos());
+    loc.setSource(literal);
+    
+    return new Constant(loc, literal);
+  }
+
   private RegExpTerm parseQuantifierOpt(SourceLocation loc, RegExpTerm atom) {
     if (this.match("*")) return this.finishTerm(new Star(loc, atom, !this.match("?")));
     if (this.match("+")) return this.finishTerm(new Plus(loc, atom, !this.match("?")));
@@ -427,6 +467,12 @@ private RegExpTerm parseAtomEscape(SourceLocation loc, boolean inCharClass) {
       return this.finishTerm(new NamedBackReference(loc, name, "\\k<" + name + ">"));
     }
 
+    if (this.match("q{")) {
+      RegExpTerm term = parseDisjunctionInsideQuotedString();
+      this.expectRBrace();
+      return this.finishTerm(new CharacterClassQuotedString(loc, term));
+    }
+
     if (this.match("p{", "P{")) {
       String name = this.readIdentifier();
       if (this.match("=")) {
diff --git a/javascript/extractor/tests/es2024/output/trap/regex_quoted_string.js.trap b/javascript/extractor/tests/es2024/output/trap/regex_quoted_string.js.trap

Original file line number	Diff line number	Diff line change
`@@ -61,4 +61,6 @@ public interface Visitor {`
`61`	`61`	`public void visit(ZeroWidthNegativeLookbehind nd);`
`62`	`62`
`63`	`63`	`public void visit(UnicodePropertyEscape nd);`
	`64`	`+`
	`65`	`+ public void visit(CharacterClassQuotedString nd);`
`64`	`66`	`}`