Invalid regexp literals should throw syntax error in ES11 (jerryscript-project#4506)

JerryScript-DCO-1.0-Signed-off-by: Robert Fancsik [email protected]

Author: rerobika

jerry-core/include/jerryscript-snapshot.h

@@ -30,7 +30,7 @@ extern "C"
 /**
  * Jerry snapshot format version.
  */
-#define JERRY_SNAPSHOT_VERSION (62u)
+#define JERRY_SNAPSHOT_VERSION (63u)
 
 /**
  * Flags for jerry_generate_snapshot and jerry_generate_function_snapshot.

jerry-core/parser/js/byte-code.c

@@ -27,7 +27,7 @@ JERRY_STATIC_ASSERT ((sizeof (cbc_uint16_arguments_t) % sizeof (jmem_cpointer_t)
  */
 JERRY_STATIC_ASSERT (CBC_END == 238,
                      number_of_cbc_opcodes_changed);
-JERRY_STATIC_ASSERT (CBC_EXT_END == 149,
+JERRY_STATIC_ASSERT (CBC_EXT_END == 148,
                      number_of_cbc_ext_opcodes_changed);
 
 #if ENABLED (JERRY_PARSER) || ENABLED (JERRY_PARSER_DUMP_BYTE_CODE)

jerry-core/parser/js/byte-code.h

@@ -610,8 +610,6 @@
               VM_OC_LINE) \
   CBC_OPCODE (CBC_EXT_THROW_REFERENCE_ERROR, CBC_NO_FLAG, 1, \
               VM_OC_THROW_REFERENCE_ERROR) \
-  CBC_OPCODE (CBC_EXT_THROW_SYNTAX_ERROR, CBC_HAS_LITERAL_ARG, 1, \
-              VM_OC_THROW_SYNTAX_ERROR | VM_OC_GET_LITERAL) \
   CBC_OPCODE (CBC_EXT_THROW_ASSIGN_CONST_ERROR, CBC_NO_FLAG, 0, \
               VM_OC_THROW_CONST_ERROR) \
   CBC_OPCODE (CBC_EXT_REQUIRE_OBJECT_COERCIBLE, CBC_NO_FLAG, 0, \

jerry-core/parser/js/js-lexer.c

@@ -95,42 +95,6 @@ lexer_hex_to_code_point (const uint8_t *source_p, /**< current source position *
 
 #if ENABLED (JERRY_ESNEXT)
 
-/**
- * Find a string literal in the literal pool matching with the given buffer's content
- *
- * @return PARSER_INVALID_LITERAL_INDEX - if the literal is not present in the literal pool
- *         literal's index in the pool - otherwise
- */
-static uint16_t
-parser_find_string_literal (parser_context_t *context_p, /**< context */
-                            lexer_literal_t **out_literal_p, /**< [out] found literal */
-                            uint8_t *buffer_p, /**< character buffer */
-                            lit_utf8_size_t size) /**< buffer's size */
-{
-  JERRY_ASSERT (out_literal_p != NULL);
-  JERRY_ASSERT (buffer_p != NULL);
-
-  uint16_t literal_index = 0;
-  lexer_literal_t *literal_p;
-  parser_list_iterator_t literal_iterator;
-  parser_list_iterator_init (&context_p->literal_pool, &literal_iterator);
-
-  while ((literal_p = (lexer_literal_t *) parser_list_iterator_next (&literal_iterator)) != NULL)
-  {
-    if (literal_p->type == LEXER_STRING_LITERAL
-        && literal_p->prop.length == size
-        && memcmp (literal_p->u.char_p, buffer_p, size) == 0)
-    {
-      *out_literal_p = literal_p;
-      return literal_index;
-    }
-
-    literal_index++;
-  }
-
-  return PARSER_INVALID_LITERAL_INDEX;
-} /* parser_find_string_literal */
-
 /**
  * Parse hexadecimal character sequence enclosed in braces
  *
@@ -3104,56 +3068,14 @@ lexer_construct_regexp_object (parser_context_t *context_p, /**< context */
   re_compiled_code_t *re_bytecode_p = re_compile_bytecode (pattern_str_p, current_flags);
   ecma_deref_ecma_string (pattern_str_p);
 
-  lexer_literal_t *literal_p = NULL;
-  uint8_t literal_type = LEXER_REGEXP_LITERAL;
-
   if (JERRY_UNLIKELY (re_bytecode_p == NULL))
   {
-#if ENABLED (JERRY_ESNEXT)
-    ecma_value_t error = jcontext_take_exception ();
-    ecma_property_t *prop_p = ecma_find_named_property (ecma_get_object_from_value (error),
-                                                        ecma_get_magic_string (LIT_MAGIC_STRING_MESSAGE));
-    const char default_msg[] = "Invalid regular expression";
-    lit_utf8_byte_t *buffer_p = (lit_utf8_byte_t *) default_msg;
-    lit_utf8_size_t size = sizeof (buffer_p) - 1;
-
-    if (prop_p != NULL)
-    {
-      ecma_string_t *message_p = ecma_get_string_from_value (ECMA_PROPERTY_VALUE_PTR (prop_p)->value);
-      JERRY_ASSERT (!ECMA_IS_DIRECT_STRING (message_p));
-      JERRY_ASSERT (ECMA_STRING_GET_CONTAINER (message_p) == ECMA_STRING_CONTAINER_HEAP_ASCII_STRING);
-      buffer_p = ECMA_ASCII_STRING_GET_BUFFER (message_p);
-      size = ECMA_ASCII_STRING_GET_SIZE (message_p);
-    }
-
-    uint16_t literal_index = parser_find_string_literal (context_p, &literal_p, buffer_p, size);
-
-    if (literal_index != PARSER_INVALID_LITERAL_INDEX)
-    {
-      ecma_free_value (error);
-      context_p->lit_object.literal_p = literal_p;
-      context_p->lit_object.index = literal_index;
-      return;
-    }
-
-    literal_p = (lexer_literal_t *) parser_list_append (context_p, &context_p->literal_pool);
-    literal_p->u.char_p = (uint8_t *) jmem_heap_alloc_block (size);
-    memcpy ((uint8_t *) literal_p->u.char_p, buffer_p, size);
-    literal_type = LEXER_STRING_LITERAL;
-    length = size;
-
-    ecma_free_value (error);
-#else /* !ENABLED (JERRY_ESNEXT) */
     parser_raise_error (context_p, PARSER_ERR_INVALID_REGEXP);
-#endif /* ENABLED (JERRY_ESNEXT) */
-  }
-  else
-  {
-    literal_p = (lexer_literal_t *) parser_list_append (context_p, &context_p->literal_pool);
-    literal_p->u.bytecode_p = (ecma_compiled_code_t *) re_bytecode_p;
   }
 
-  literal_p->type = literal_type;
+  lexer_literal_t *literal_p = (lexer_literal_t *) parser_list_append (context_p, &context_p->literal_pool);
+  literal_p->u.bytecode_p = (ecma_compiled_code_t *) re_bytecode_p;
+  literal_p->type = LEXER_REGEXP_LITERAL;
   literal_p->prop.length = (prop_length_t) length;
   literal_p->status_flags = 0;
 

jerry-core/parser/js/js-parser-expr.c

@@ -2177,15 +2177,6 @@ parser_parse_unary_expression (parser_context_t *context_p, /**< context */
     case LEXER_ASSIGN_DIVIDE:
     {
       lexer_construct_regexp_object (context_p, false);
-
-#if ENABLED (JERRY_ESNEXT)
-      if (JERRY_UNLIKELY (context_p->lit_object.literal_p->type == LEXER_STRING_LITERAL))
-      {
-        parser_emit_cbc_ext_literal (context_p, CBC_EXT_THROW_SYNTAX_ERROR, context_p->lit_object.index);
-        break;
-      }
-#endif /* ENABLED (JERRY_ESNEXT) */
-
       uint16_t literal_index = (uint16_t) (context_p->literal_count - 1);
 
       if (context_p->last_cbc_opcode == CBC_PUSH_LITERAL)

jerry-core/parser/js/js-parser.c

@@ -2832,7 +2832,6 @@ parser_parse_script (const uint8_t *arg_list_p, /**< function argument list */
 #if ENABLED (JERRY_ERROR_MESSAGES)
     ecma_string_t *err_str_p;
 
-#if !ENABLED (JERRY_ESNEXT)
     if (parser_error.error == PARSER_ERR_INVALID_REGEXP)
     {
       ecma_value_t error = jcontext_take_exception ();
@@ -2844,7 +2843,6 @@ parser_parse_script (const uint8_t *arg_list_p, /**< function argument list */
       ecma_ref_ecma_string (err_str_p);
     }
     else
-#endif /* !ENABLED (JERRY_ESNEXT) */
     {
       const lit_utf8_byte_t *err_bytes_p = (const lit_utf8_byte_t *) parser_error_to_string (parser_error.error);
       lit_utf8_size_t err_bytes_size = lit_zt_utf8_string_size (err_bytes_p);
@@ -2865,12 +2863,12 @@ parser_parse_script (const uint8_t *arg_list_p, /**< function argument list */
     ecma_free_value (line_str_val);
     ecma_deref_ecma_string (err_str_p);
 #else /* !ENABLED (JERRY_ERROR_MESSAGES) */
-#if !ENABLED (JERRY_ESNEXT)
-    if (parser_error.error != PARSER_ERR_INVALID_REGEXP)
-#endif /* !ENABLED (JERRY_ESNEXT) */
+    if (parser_error.error == PARSER_ERR_INVALID_REGEXP)
     {
-      ecma_raise_syntax_error ("");
+      jcontext_release_exception ();
     }
+
+    ecma_raise_syntax_error ("");
 #endif /* ENABLED (JERRY_ERROR_MESSAGES) */
 
     return NULL;

jerry-core/vm/vm.c

@@ -1740,14 +1740,6 @@ vm_loop (vm_frame_ctx_t *frame_ctx_p) /**< frame context */
           result = ecma_raise_type_error (ECMA_ERR_MSG ("Constant bindings cannot be reassigned."));
           goto error;
         }
-        case VM_OC_THROW_SYNTAX_ERROR:
-        {
-          ecma_string_t *msg_p = ecma_get_string_from_value (left_value);
-          ecma_object_t *error_obj_p = ecma_new_standard_error (ECMA_ERROR_SYNTAX, msg_p);
-          jcontext_raise_exception (ecma_make_object_value (error_obj_p));
-          result = ECMA_VALUE_ERROR;
-          goto error;
-        }
         case VM_OC_COPY_TO_GLOBAL:
         {
           uint32_t literal_index;

jerry-core/vm/vm.h

@@ -243,7 +243,6 @@ typedef enum
   VM_OC_ASSIGN_LET_CONST,        /**< assign values to let/const declarations */
   VM_OC_INIT_BINDING,            /**< create and intialize a binding */
   VM_OC_THROW_CONST_ERROR,       /**< throw invalid assignment to const variable error */
-  VM_OC_THROW_SYNTAX_ERROR,      /**< throw syntax error */
   VM_OC_COPY_TO_GLOBAL,          /**< copy value to global lex env */
   VM_OC_COPY_FROM_ARG,           /**< copy value from arg lex env */
   VM_OC_CLONE_CONTEXT,           /**< clone lexical environment with let/const declarations */
@@ -331,7 +330,6 @@ typedef enum
   VM_OC_ASSIGN_LET_CONST = VM_OC_NONE,        /**< assign values to let/const declarations */
   VM_OC_INIT_BINDING = VM_OC_NONE,            /**< create and intialize a binding */
   VM_OC_THROW_CONST_ERROR = VM_OC_NONE,       /**< throw invalid assignment to const variable error */
-  VM_OC_THROW_SYNTAX_ERROR = VM_OC_NONE,      /**< throw syntax error */
   VM_OC_COPY_TO_GLOBAL = VM_OC_NONE,          /**< copy value to global lex env */
   VM_OC_COPY_FROM_ARG = VM_OC_NONE,           /**< copy value from arg lex env */
   VM_OC_CLONE_CONTEXT = VM_OC_NONE,           /**< clone lexical environment with let/const declarations */

tests/jerry/es.next/regression-test-issue-2058.js

@@ -13,7 +13,7 @@
 // limitations under the License.
 
 try {
-  /?:/
+  eval('/?:/');
   assert(false);
 } catch (e) {
   assert(e instanceof SyntaxError);

tests/jerry/es.next/regression-test-issue-4408.js

@@ -13,13 +13,13 @@
 // limitations under the License.
 
 try {
-  /(?<=^abc)def/;
+  eval('/(?<=^abc)def/');
 } catch(e) {
   assert(e instanceof SyntaxError);
 }
 
 try {
-  /(?a)/;
+  eval('/(?a)/;')
 } catch(e) {
   assert(e instanceof SyntaxError);
 }