Implement RegExp.prototype.compile()

JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai dbatyai.u-szeged@partner.samsung.com
This commit is contained in:
Dániel Bátyai
2015-08-17 15:16:44 +02:00
parent 63d3e02b59
commit 13941df8dd
8 changed files with 341 additions and 58 deletions
@@ -45,6 +45,173 @@
* @{
*/
#ifndef CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN
/**
* The RegExp.prototype object's 'compile' routine
*
* See also:
* ECMA-262 v5, B.2.5.1
*
* @return completion value
* Returned value must be freed with ecma_free_completion_value.
*/
static ecma_completion_value_t
ecma_builtin_regexp_prototype_compile (ecma_value_t this_arg, /**< this argument */
ecma_value_t pattern_arg, /**< pattern or RegExp object */
ecma_value_t flags_arg) /**< flags */
{
ecma_completion_value_t ret_value = ecma_make_empty_completion_value ();
ecma_object_t *this_obj_p = ecma_get_object_from_value (this_arg);
if (ecma_object_get_class_name (this_obj_p) != LIT_MAGIC_STRING_REGEXP_UL)
{
/* Compile can only be called on RegExp objects. */
ret_value = ecma_make_throw_obj_completion_value (ecma_new_standard_error (ECMA_ERROR_TYPE));
}
else
{
ecma_string_t *pattern_string_p = NULL;
uint8_t flags = 0;
if (ecma_is_value_object (pattern_arg)
&& ecma_object_get_class_name (ecma_get_object_from_value (pattern_arg)) == LIT_MAGIC_STRING_REGEXP_UL)
{
if (!ecma_is_value_undefined (flags_arg))
{
ret_value = ecma_raise_type_error ("Invalid argument of RegExp compile.");
}
else
{
/* Compile from existing RegExp pbject. */
ecma_object_t *target_p = ecma_get_object_from_value (pattern_arg);
/* Get source. */
ecma_string_t *magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_SOURCE);
ecma_property_t *prop_p = ecma_get_named_data_property (target_p, magic_string_p);
pattern_string_p = ecma_get_string_from_value (ecma_get_named_data_property_value (prop_p));
ecma_deref_ecma_string (magic_string_p);
/* Get flags. */
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_GLOBAL);
prop_p = ecma_get_named_data_property (target_p, magic_string_p);
if (ecma_is_value_true (ecma_get_named_data_property_value (prop_p)))
{
flags |= RE_FLAG_GLOBAL;
}
ecma_deref_ecma_string (magic_string_p);
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_IGNORECASE_UL);
prop_p = ecma_get_named_data_property (target_p, magic_string_p);
if (ecma_is_value_true (ecma_get_named_data_property_value (prop_p)))
{
flags |= RE_FLAG_IGNORE_CASE;
}
ecma_deref_ecma_string (magic_string_p);
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_MULTILINE);
prop_p = ecma_get_named_data_property (target_p, magic_string_p);
if (ecma_is_value_true (ecma_get_named_data_property_value (prop_p)))
{
flags |= RE_FLAG_MULTILINE;
}
ecma_deref_ecma_string (magic_string_p);
/* Get bytecode property. */
ecma_property_t *bc_prop_p = ecma_get_internal_property (this_obj_p,
ECMA_INTERNAL_PROPERTY_REGEXP_BYTECODE);
re_bytecode_t *old_bc_p = ECMA_GET_NON_NULL_POINTER (re_bytecode_t,
bc_prop_p->u.internal_property.value);
FIXME ("We currently have to re-compile the bytecode, because we can't copy it without knowing its length.")
re_bytecode_t *new_bc_p = NULL;
ecma_completion_value_t bc_comp = re_compile_bytecode (&new_bc_p, pattern_string_p, flags);
/* Should always succeed, since we're compiling from a source that has been compiled previously. */
JERRY_ASSERT (ecma_is_completion_value_empty (bc_comp));
mem_heap_free_block (old_bc_p);
ECMA_SET_POINTER (bc_prop_p->u.internal_property.value, new_bc_p);
re_initialize_props (this_obj_p, pattern_string_p, flags);
ret_value = ecma_make_normal_completion_value (ecma_make_simple_value (ECMA_SIMPLE_VALUE_UNDEFINED));
}
}
else
{
/* Get source string. */
if (!ecma_is_value_undefined (pattern_arg))
{
ECMA_TRY_CATCH (regexp_str_value,
ecma_op_to_string (pattern_arg),
ret_value);
if (ecma_string_get_length (ecma_get_string_from_value (regexp_str_value)) == 0)
{
pattern_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_EMPTY_NON_CAPTURE_GROUP);
}
else
{
pattern_string_p = ecma_copy_or_ref_ecma_string (ecma_get_string_from_value (regexp_str_value));
}
ECMA_FINALIZE (regexp_str_value);
}
else
{
pattern_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_EMPTY_NON_CAPTURE_GROUP);
}
/* Parse flags. */
if (ecma_is_completion_value_empty (ret_value) && !ecma_is_value_undefined (flags_arg))
{
ECMA_TRY_CATCH (flags_str_value,
ecma_op_to_string (flags_arg),
ret_value);
ECMA_TRY_CATCH (flags_dummy,
re_parse_regexp_flags (ecma_get_string_from_value (flags_str_value), &flags),
ret_value);
ECMA_FINALIZE (flags_dummy);
ECMA_FINALIZE (flags_str_value);
}
if (ecma_is_completion_value_empty (ret_value))
{
ecma_property_t *bc_prop_p = ecma_get_internal_property (this_obj_p,
ECMA_INTERNAL_PROPERTY_REGEXP_BYTECODE);
re_bytecode_t *old_bc_p = ECMA_GET_NON_NULL_POINTER (re_bytecode_t,
bc_prop_p->u.internal_property.value);
/* Try to compile bytecode from new source. */
re_bytecode_t *new_bc_p = NULL;
ECMA_TRY_CATCH (bc_dummy, re_compile_bytecode (&new_bc_p, pattern_string_p, flags), ret_value);
/* Replace old bytecode with new one. */
mem_heap_free_block (old_bc_p);
ECMA_SET_POINTER (bc_prop_p->u.internal_property.value, new_bc_p);
re_initialize_props (this_obj_p, pattern_string_p, flags);
ret_value = ecma_make_normal_completion_value (ecma_make_simple_value (ECMA_SIMPLE_VALUE_UNDEFINED));
ECMA_FINALIZE (bc_dummy);
}
if (pattern_string_p != NULL)
{
ecma_deref_ecma_string (pattern_string_p);
}
}
}
return ret_value;
} /* ecma_builtin_regexp_prototype_compile */
#endif /* !CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN */
/**
* The RegExp.prototype object's 'exec' routine
*
@@ -86,6 +86,9 @@ NUMBER_VALUE (LIT_MAGIC_STRING_LASTINDEX_UL,
ECMA_PROPERTY_NOT_ENUMERABLE,
ECMA_PROPERTY_NOT_CONFIGURABLE)
#ifndef CONFIG_ECMA_COMPACT_PROFILE_DISABLE_ANNEXB_BUILTIN
ROUTINE (LIT_MAGIC_STRING_COMPILE, ecma_builtin_regexp_prototype_compile, 2, 1)
#endif
ROUTINE (LIT_MAGIC_STRING_EXEC, ecma_builtin_regexp_prototype_exec, 1, 1)
ROUTINE (LIT_MAGIC_STRING_TEST, ecma_builtin_regexp_prototype_test, 1, 1)
ROUTINE (LIT_MAGIC_STRING_TO_STRING_UL, ecma_builtin_regexp_prototype_to_string, 0, 0)
+102 -52
View File
@@ -62,7 +62,7 @@
* @return completion value
* Returned value must be freed with ecma_free_completion_value
*/
static ecma_completion_value_t
ecma_completion_value_t
re_parse_regexp_flags (ecma_string_t *flags_str_p, /**< Input string with flags */
uint8_t *flags_p) /**< Output: parsed flag bits */
{
@@ -119,6 +119,101 @@ re_parse_regexp_flags (ecma_string_t *flags_str_p, /**< Input string with flags
return ret_value;
} /* re_parse_regexp_flags */
/*
* Initializes the source, global, ignoreCase, multiline, and lastIndex properties of RegExp instance.
*/
void
re_initialize_props (ecma_object_t *re_obj_p, /**< RegExp obejct */
ecma_string_t *source_p, /**< source string */
uint8_t flags) /**< flags */
{
/* Set source property. ECMA-262 v5, 15.10.7.1 */
ecma_string_t *magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_SOURCE);
ecma_property_t *prop_p = ecma_find_named_property (re_obj_p, magic_string_p);
if (prop_p == NULL)
{
prop_p = ecma_create_named_data_property (re_obj_p,
magic_string_p,
false, false, false);
}
ecma_deref_ecma_string (magic_string_p);
JERRY_ASSERT (prop_p->type == ECMA_PROPERTY_NAMEDDATA);
ecma_named_data_property_assign_value (re_obj_p,
prop_p,
ecma_make_string_value (source_p));
ecma_simple_value_t prop_value;
/* Set global property. ECMA-262 v5, 15.10.7.2 */
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_GLOBAL);
prop_p = ecma_find_named_property (re_obj_p, magic_string_p);
if (prop_p == NULL)
{
prop_p = ecma_create_named_data_property (re_obj_p,
magic_string_p,
false, false, false);
}
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_GLOBAL ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
JERRY_ASSERT (prop_p->type == ECMA_PROPERTY_NAMEDDATA);
ecma_set_named_data_property_value (prop_p, ecma_make_simple_value (prop_value));
/* Set ignoreCase property. ECMA-262 v5, 15.10.7.3 */
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_IGNORECASE_UL);
prop_p = ecma_find_named_property (re_obj_p, magic_string_p);
if (prop_p == NULL)
{
prop_p = ecma_create_named_data_property (re_obj_p,
magic_string_p,
false, false, false);
}
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_IGNORE_CASE ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
JERRY_ASSERT (prop_p->type == ECMA_PROPERTY_NAMEDDATA);
ecma_set_named_data_property_value (prop_p, ecma_make_simple_value (prop_value));
/* Set multiline property. ECMA-262 v5, 15.10.7.4 */
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_MULTILINE);
prop_p = ecma_find_named_property (re_obj_p, magic_string_p);
if (prop_p == NULL)
{
prop_p = ecma_create_named_data_property (re_obj_p,
magic_string_p,
false, false, false);
}
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_MULTILINE ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
JERRY_ASSERT (prop_p->type == ECMA_PROPERTY_NAMEDDATA);
ecma_set_named_data_property_value (prop_p, ecma_make_simple_value (prop_value));
/* Set lastIndex property. ECMA-262 v5, 15.10.7.5 */
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_LASTINDEX_UL);
prop_p = ecma_find_named_property (re_obj_p, magic_string_p);
if (prop_p == NULL)
{
prop_p = ecma_create_named_data_property (re_obj_p,
magic_string_p,
true, false, false);
}
ecma_deref_ecma_string (magic_string_p);
ecma_number_t *lastindex_num_p = ecma_alloc_number ();
*lastindex_num_p = ECMA_NUMBER_ZERO;
JERRY_ASSERT (prop_p->type == ECMA_PROPERTY_NAMEDDATA);
ecma_named_data_property_assign_value (re_obj_p, prop_p, ecma_make_number_value (lastindex_num_p));
ecma_dealloc_number (lastindex_num_p);
} /* re_initialize_props */
/**
* RegExp object creation operation.
*
@@ -155,64 +250,19 @@ ecma_op_create_regexp_object (ecma_string_t *pattern_p, /**< input pattern */
ecma_property_t *class_prop_p = ecma_create_internal_property (obj_p, ECMA_INTERNAL_PROPERTY_CLASS);
class_prop_p->u.internal_property.value = LIT_MAGIC_STRING_REGEXP_UL;
/* Set source property. ECMA-262 v5, 15.10.7.1 */
ecma_string_t *magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_SOURCE);
ecma_property_t *source_prop_p = ecma_create_named_data_property (obj_p,
magic_string_p,
false, false, false);
ecma_deref_ecma_string (magic_string_p);
ecma_set_named_data_property_value (source_prop_p,
ecma_make_string_value (ecma_copy_or_ref_ecma_string (pattern_p)));
ecma_simple_value_t prop_value;
/* Set global property. ECMA-262 v5, 15.10.7.2*/
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_GLOBAL);
ecma_property_t *global_prop_p = ecma_create_named_data_property (obj_p,
magic_string_p,
false, false, false);
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_GLOBAL ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
ecma_set_named_data_property_value (global_prop_p, ecma_make_simple_value (prop_value));
/* Set ignoreCase property. ECMA-262 v5, 15.10.7.3*/
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_IGNORECASE_UL);
ecma_property_t *ignorecase_prop_p = ecma_create_named_data_property (obj_p,
magic_string_p,
false, false, false);
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_IGNORE_CASE ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
ecma_set_named_data_property_value (ignorecase_prop_p, ecma_make_simple_value (prop_value));
/* Set multiline property. ECMA-262 v5, 15.10.7.4*/
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_MULTILINE);
ecma_property_t *multiline_prop_p = ecma_create_named_data_property (obj_p,
magic_string_p,
false, false, false);
ecma_deref_ecma_string (magic_string_p);
prop_value = flags & RE_FLAG_MULTILINE ? ECMA_SIMPLE_VALUE_TRUE : ECMA_SIMPLE_VALUE_FALSE;
ecma_set_named_data_property_value (multiline_prop_p, ecma_make_simple_value (prop_value));
/* Set lastIndex property. ECMA-262 v5, 15.10.7.5*/
magic_string_p = ecma_get_magic_string (LIT_MAGIC_STRING_LASTINDEX_UL);
ecma_property_t *lastindex_prop_p = ecma_create_named_data_property (obj_p,
magic_string_p,
true, false, false);
ecma_deref_ecma_string (magic_string_p);
ecma_number_t *lastindex_num_p = ecma_alloc_number ();
*lastindex_num_p = ECMA_NUMBER_ZERO;
ecma_named_data_property_assign_value (obj_p, lastindex_prop_p, ecma_make_number_value (lastindex_num_p));
ecma_dealloc_number (lastindex_num_p);
re_initialize_props (obj_p, pattern_p, flags);
/* Set bytecode internal property. */
ecma_property_t *bytecode_prop_p;
bytecode_prop_p = ecma_create_internal_property (obj_p, ECMA_INTERNAL_PROPERTY_REGEXP_BYTECODE);
/* Compile bytecode. */
ECMA_TRY_CATCH (empty, re_compile_bytecode (bytecode_prop_p, pattern_p, flags), ret_value);
re_bytecode_t *bc_p = NULL;
ECMA_TRY_CATCH (empty, re_compile_bytecode (&bc_p, pattern_p, flags), ret_value);
ECMA_SET_POINTER (bytecode_prop_p->u.internal_property.value, bc_p);
ret_value = ecma_make_normal_completion_value (ecma_make_object_value (obj_p));
ECMA_FINALIZE (empty);
if (ecma_is_completion_value_throw (ret_value))
@@ -65,6 +65,14 @@ re_set_result_array_properties (ecma_object_t *array_obj_p,
uint32_t num_of_elements,
int32_t index);
extern ecma_completion_value_t
re_parse_regexp_flags (ecma_string_t *flags_str_p, uint8_t *flags_p);
extern void
re_initialize_props (ecma_object_t *re_obj_p,
ecma_string_t *source_p,
uint8_t flags);
/**
* @}
* @}
+1
View File
@@ -211,6 +211,7 @@ LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_INVALID_DATE_UL, "Invalid Date")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_APPLY, "apply")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_CALL, "call")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_BIND, "bind")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_COMPILE, "compile")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_EXEC, "exec")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_TEST, "test")
LIT_MAGIC_STRING_DEF (LIT_MAGIC_STRING_NAME, "name")
+14 -5
View File
@@ -633,7 +633,7 @@ re_parse_alternative (re_compiler_ctx_t *re_ctx_p, /**< RegExp compiler context
* Returned value must be freed with ecma_free_completion_value
*/
ecma_completion_value_t
re_compile_bytecode (ecma_property_t *bytecode_p, /**< bytecode */
re_compile_bytecode (re_bytecode_t **out_bytecode_p, /**< out:pointer to bytecode */
ecma_string_t *pattern_str_p, /**< pattern */
uint8_t flags) /**< flags */
{
@@ -684,12 +684,21 @@ re_compile_bytecode (ecma_property_t *bytecode_p, /**< bytecode */
}
ECMA_FINALIZE (empty);
/* The RegExp bytecode contains at least a RE_OP_SAVE_AT_START opdoce, so it cannot be NULL. */
JERRY_ASSERT (bc_ctx.block_start_p != NULL);
ECMA_SET_POINTER (bytecode_p->u.internal_property.value, bc_ctx.block_start_p);
MEM_FINALIZE_LOCAL_ARRAY (pattern_start_p);
if (!ecma_is_completion_value_empty (ret_value))
{
/* Compilation failed, free bytecode. */
mem_heap_free_block (bc_ctx.block_start_p);
*out_bytecode_p = NULL;
}
else
{
/* The RegExp bytecode contains at least a RE_OP_SAVE_AT_START opdoce, so it cannot be NULL. */
JERRY_ASSERT (bc_ctx.block_start_p != NULL);
*out_bytecode_p = bc_ctx.block_start_p;
}
#ifdef JERRY_ENABLE_LOG
re_dump_bytecode (&bc_ctx);
#endif
+1 -1
View File
@@ -96,7 +96,7 @@ typedef struct
} re_compiler_ctx_t;
ecma_completion_value_t
re_compile_bytecode (ecma_property_t *bytecode_p, ecma_string_t *pattern_str_p, uint8_t flags);
re_compile_bytecode (re_bytecode_t **out_bytecode_p, ecma_string_t *pattern_str_p, uint8_t flags);
re_opcode_t
re_get_opcode (re_bytecode_t **bc_p);
+45
View File
@@ -66,3 +66,48 @@ result = re.exec("a");
assert (result.length === 1);
assert (result[0] === "a");
assert (re.lastIndex === 1);
var re1 = /foo/gim;
var re2 = /bar/g;
try {
re2.compile("asd", "ggim");
assert (false);
} catch (e) {
assert (e instanceof SyntaxError);
assert (re2 == "/bar/g");
}
try {
re2.compile("[", undefined);
assert (false);
} catch (e) {
assert (e instanceof SyntaxError);
assert (re2 == "/bar/g");
}
try {
re2.compile(re1, "im");
assert (false);
} catch (e) {
assert (e instanceof TypeError);
assert (re2 == "/bar/g");
}
re2.lastIndex = 2;
re2.compile("asd", "im");
assert (re2 == "/asd/im");
assert (re2.global === false);
assert (re2.ignoreCase === true);
assert (re2.multiline === true);
assert (re2.source === "asd");
assert (re2.lastIndex === 0);
re2.compile(re1);
assert (re2.toString() === re1.toString());
assert (re2.global === re1.global);
assert (re2.ignoreCase === re1.ignoreCase);
assert (re2.multiline === re1.multiline);
assert (re2.source === re1.source);
assert (re2.lastIndex === 0);