Allow unescaped ] as leading char in char class (fixes #6)

slevithan · Nov 26, 2024 · 1c49b25 · 1c49b25
1 parent df8a9ec
commit 1c49b25
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 17 deletions.
diff --git a/src/parse.js b/src/parse.js
@@ -248,9 +248,10 @@ function parseCharacterClassHyphen(context, state) {
 
 function parseCharacterClassOpen(context, state) {
   const {token, tokens, verbose, walk} = context;
+  const firstClassToken = tokens[context.current];
   let node = createCharacterClass({negate: token.negate});
   const intersection = node.elements[0];
-  let nextToken = throwIfUnclosedCharacterClass(tokens[context.current]);
+  let nextToken = throwIfUnclosedCharacterClass(firstClassToken);
   while (nextToken.type !== TokenTypes.CharacterClassClose) {
     if (nextToken.type === TokenTypes.CharacterClassIntersector) {
       intersection.classes.push(createCharacterClass({negate: false, baseOnly: true}));
@@ -260,7 +261,7 @@ function parseCharacterClassOpen(context, state) {
       const cc = intersection.classes.at(-1);
       cc.elements.push(walk(cc, state));
     }
-    nextToken = throwIfUnclosedCharacterClass(tokens[context.current]);
+    nextToken = throwIfUnclosedCharacterClass(tokens[context.current], firstClassToken);
   }
   if (!verbose) {
     optimizeCharacterClassIntersection(intersection);
@@ -699,8 +700,12 @@ function optimizeCharacterClassIntersection(intersection) {
   }
 }
 
-function throwIfUnclosedCharacterClass(token) {
-  return throwIfNot(token, 'Unclosed character class');
+function throwIfUnclosedCharacterClass(token, firstClassToken) {
+  return throwIfNot(
+    token,
+    // Easier to understand error when applicable
+    `${firstClassToken?.value === 93 ? 'Empty' : 'Unclosed'} character class`
+  );
 }
 
 function throwIfUnclosedGroup(token) {

diff --git a/src/tokenize.js b/src/tokenize.js
@@ -58,7 +58,7 @@ const EscapeCharCodes = new Map([
   ['v', 11], // vertical tab
 ]);
 
-const charClassOpenPattern = r`\[\^?\]?`;
+const charClassOpenPattern = r`\[\^?`;
 const sharedEscapesPattern = `${
   // Control char
   'c.? | C(?:-.?)?'
@@ -408,7 +408,6 @@ function getTokenWithDetails(context, pattern, m, lastIndex) {
 }
 
 function getAllTokensForCharClass(pattern, opener, lastIndex) {
-  assertNonEmptyCharClass(opener);
   const tokens = [createToken(TokenTypes.CharacterClassOpen, opener, {
     negate: opener[1] === '^',
   })];
@@ -420,16 +419,22 @@ function getAllTokensForCharClass(pattern, opener, lastIndex) {
     // Start of nested char class
     // POSIX classes are handled as a single token; not as a nested char class
     if (m[0] === '[' && m[1] !== ':') {
-      assertNonEmptyCharClass(m);
       numCharClassesOpen++;
       tokens.push(createToken(TokenTypes.CharacterClassOpen, m, {
         negate: m[1] === '^',
       }));
     } else if (m === ']') {
-      numCharClassesOpen--;
-      tokens.push(createToken(TokenTypes.CharacterClassClose, m));
-      if (!numCharClassesOpen) {
-        break;
+      if (tokens.at(-1).type === TokenTypes.CharacterClassOpen) {
+        // Allow unescaped `]` as leading char
+        tokens.push(createToken(TokenTypes.Character, m, {
+          value: 93,
+        }));
+      } else {
+        numCharClassesOpen--;
+        tokens.push(createToken(TokenTypes.CharacterClassClose, m));
+        if (!numCharClassesOpen) {
+          break;
+        }
       }
     } else {
       const result = createTokenForAnyTokenWithinCharClass(m);
@@ -739,12 +744,6 @@ function splitEscapedNumToken(token, numCaptures) {
   return tokens;
 }
 
-function assertNonEmptyCharClass(raw) {
-  if (raw.endsWith(']')) {
-    throw new Error(`Empty character class "${raw}" unsupported in Oniguruma`);
-  }
-}
-
 function assertSingleCodePoint(raw) {
   if ([...raw].length !== 1) {
     throw new Error(`Expected "${raw}" to be a single code point`);