From 3dcadf1518b8c10026c578e3a3188964d135e41a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sa=C3=BAl=20Ibarra=20Corretg=C3=A9?= <s@saghul.net>
Date: Fri, 4 Oct 2024 11:55:14 +0200
Subject: [PATCH] Fix next token parsing after a function definition

Ref: https://github.com/bellard/quickjs/commit/c06c399f4f8f8292e65e848ff3468f302c18a55e
Fixes: https://github.com/quickjs-ng/quickjs/issues/572
---
 quickjs.c              | 70 +++++++++++++++++++++++++++++-------------
 tests/test_language.js | 23 ++++++++++++++
 2 files changed, 72 insertions(+), 21 deletions(-)

diff --git a/quickjs.c b/quickjs.c
index 14fbb6e..1eccf2b 100644
--- a/quickjs.c
+++ b/quickjs.c
@@ -19224,6 +19224,48 @@ static __exception int ident_realloc(JSContext *ctx, char **pbuf, size_t *psize,
     return 0;
 }
 
+/* convert a TOK_IDENT to a keyword when needed */
+static void update_token_ident(JSParseState *s)
+{
+    if (s->token.u.ident.atom <= JS_ATOM_LAST_KEYWORD ||
+        (s->token.u.ident.atom <= JS_ATOM_LAST_STRICT_KEYWORD &&
+         (s->cur_func->js_mode & JS_MODE_STRICT)) ||
+        (s->token.u.ident.atom == JS_ATOM_yield &&
+         ((s->cur_func->func_kind & JS_FUNC_GENERATOR) ||
+          (s->cur_func->func_type == JS_PARSE_FUNC_ARROW &&
+           !s->cur_func->in_function_body && s->cur_func->parent &&
+           (s->cur_func->parent->func_kind & JS_FUNC_GENERATOR)))) ||
+        (s->token.u.ident.atom == JS_ATOM_await &&
+         (s->is_module ||
+          (s->cur_func->func_kind & JS_FUNC_ASYNC) ||
+          s->cur_func->func_type == JS_PARSE_FUNC_CLASS_STATIC_INIT ||
+          (s->cur_func->func_type == JS_PARSE_FUNC_ARROW &&
+           !s->cur_func->in_function_body && s->cur_func->parent &&
+           ((s->cur_func->parent->func_kind & JS_FUNC_ASYNC) ||
+            s->cur_func->parent->func_type == JS_PARSE_FUNC_CLASS_STATIC_INIT))))) {
+        if (s->token.u.ident.has_escape) {
+            s->token.u.ident.is_reserved = TRUE;
+            s->token.val = TOK_IDENT;
+        } else {
+            /* The keywords atoms are pre allocated */
+            s->token.val = s->token.u.ident.atom - 1 + TOK_FIRST_KEYWORD;
+        }
+    }
+}
+
+/* if the current token is an identifier or keyword, reparse it
+   according to the current function type */
+static void reparse_ident_token(JSParseState *s)
+{
+    if (s->token.val == TOK_IDENT ||
+        (s->token.val >= TOK_FIRST_KEYWORD &&
+         s->token.val <= TOK_LAST_KEYWORD)) {
+        s->token.val = TOK_IDENT;
+        s->token.u.ident.is_reserved = FALSE;
+        update_token_ident(s);
+    }
+}
+
 /* 'c' is the first character. Return JS_ATOM_NULL in case of error */
 static JSAtom parse_ident(JSParseState *s, const uint8_t **pp,
                           BOOL *pident_has_escape, int c, BOOL is_private)
@@ -19442,29 +19484,9 @@ static __exception int next_token(JSParseState *s)
             (atom == JS_ATOM_arguments || atom == JS_ATOM_await)) {
             s->token.u.ident.is_reserved = TRUE;
             s->token.val = TOK_IDENT;
-        } else if (s->token.u.ident.atom <= JS_ATOM_LAST_KEYWORD ||
-            (s->token.u.ident.atom <= JS_ATOM_LAST_STRICT_KEYWORD &&
-             (s->cur_func->js_mode & JS_MODE_STRICT)) ||
-            (s->token.u.ident.atom == JS_ATOM_yield &&
-             ((s->cur_func->func_kind & JS_FUNC_GENERATOR) ||
-              (s->cur_func->func_type == JS_PARSE_FUNC_ARROW &&
-               !s->cur_func->in_function_body && s->cur_func->parent &&
-               (s->cur_func->parent->func_kind & JS_FUNC_GENERATOR)))) ||
-            (s->token.u.ident.atom == JS_ATOM_await &&
-             (s->is_module ||
-              (((s->cur_func->func_kind & JS_FUNC_ASYNC) ||
-                (s->cur_func->func_type == JS_PARSE_FUNC_ARROW &&
-                 !s->cur_func->in_function_body && s->cur_func->parent &&
-                 (s->cur_func->parent->func_kind & JS_FUNC_ASYNC))))))) {
-                  if (ident_has_escape) {
-                      s->token.u.ident.is_reserved = TRUE;
-                      s->token.val = TOK_IDENT;
-                  } else {
-                      /* The keywords atoms are pre allocated */
-                      s->token.val = s->token.u.ident.atom - 1 + TOK_FIRST_KEYWORD;
-                  }
         } else {
             s->token.val = TOK_IDENT;
+            update_token_ident(s);
         }
         break;
     case '#':
@@ -32903,6 +32925,12 @@ static __exception int js_parse_function_decl2(JSParseState *s,
 done:
     s->cur_func = fd->parent;
 
+    /* Reparse identifiers after the function is terminated so that
+       the token is parsed in the englobing function. It could be done
+       by just using next_token() here for normal functions, but it is
+       necessary for arrow functions with an expression body. */
+    reparse_ident_token(s);
+
     /* create the function object */
     {
         int idx;
diff --git a/tests/test_language.js b/tests/test_language.js
index 8ab74ae..ad2dfe8 100644
--- a/tests/test_language.js
+++ b/tests/test_language.js
@@ -691,6 +691,28 @@ function test_optional_chaining()
     assert((a?.["b"])().c, 42);
 }
 
+function test_parse_semicolon()
+{
+    /* 'yield' or 'await' may not be considered as a token if the
+       previous ';' is missing */
+    function *f()
+    {
+        function func() {
+        }
+        yield 1;
+        var h = x => x + 1
+        yield 2;
+    }
+    async function g()
+    {
+        function func() {
+        }
+        await 1;
+        var h = x => x + 1
+        await 2;
+    }
+}
+
 test_op1();
 test_cvt();
 test_eq();
@@ -714,3 +736,4 @@ test_reserved_names();
 test_number_literals();
 test_syntax();
 test_optional_chaining();
+test_parse_semicolon();