YourWishes/jerryscript
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Rework RegExp engine and add support for proper unicode matching (#3746)

Browse Source 
This change includes several bugfixes, general improvements, and support
for additional features.
- Added full support for web compatibility syntax defined in Annex B
- Implemented parsing and matching patterns in unicode mode
- Fixed capture results when iterating with nested capturing groups
- Significantly reduced regexp bytecode size
- Reduced stack usage during regexp execution
- Improved matching performance

JerryScript-DCO-1.0-Signed-off-by: Dániel Bátyai dbatyai@inf.u-szeged.hu
This commit is contained in:
Dániel Bátyai
2020-05-26 15:28:54 +02:00
committed by GitHub
parent 908240ba62
commit 8f76a1f382
30 changed files with 3641 additions and 2647 deletions
Download Patch File Download Diff File Expand all files Collapse all files
jerry-core/ecma/base/ecma-gc.c
+1 -1
View File
@@ -1467,7 +1467,7 @@ ecma_gc_run (void)
#if ENABLED (JERRY_BUILTIN_REGEXP)
/* Free RegExp bytecodes stored in cache */
re_cache_gc_run ();
re_cache_gc ();
#endif /* ENABLED (JERRY_BUILTIN_REGEXP) */
} /* ecma_gc_run */
jerry-core/ecma/base/ecma-helpers-string.c
+2 -4
View File
@@ -2362,8 +2362,7 @@ ecma_string_trim_helper (const lit_utf8_byte_t **utf8_str_p, /**< [in, out] curr
{
read_size = lit_read_code_unit_from_utf8 (current_p, &ch);
if (!lit_char_is_white_space (ch)
&& !lit_char_is_line_terminator (ch))
if (!lit_char_is_white_space (ch))
{
nonws_start_p = current_p;
break;
@@ -2378,8 +2377,7 @@ ecma_string_trim_helper (const lit_utf8_byte_t **utf8_str_p, /**< [in, out] curr
{
read_size = lit_read_prev_code_unit_from_utf8 (current_p, &ch);
if (!lit_char_is_white_space (ch)
&& !lit_char_is_line_terminator (ch))
if (!lit_char_is_white_space (ch))
{
break;
}
jerry-core/ecma/builtin-objects/ecma-builtin-global.c
+10 -13
View File
@@ -223,13 +223,13 @@ ecma_builtin_global_object_decode_uri_helper (lit_utf8_byte_t *input_start_p, /*
continue;
}
ecma_char_t decoded_byte;
if (!lit_read_code_unit_from_hex (input_char_p + 1, 2, &decoded_byte))
uint32_t hex_value = lit_char_hex_lookup (input_char_p + 1, input_end_p, 2);
if (hex_value == UINT32_MAX)
{
return ecma_raise_uri_error (ECMA_ERR_MSG ("Invalid hexadecimal value."));
}
ecma_char_t decoded_byte = (ecma_char_t) hex_value;
input_char_p += URI_ENCODED_BYTE_SIZE;
if (decoded_byte <= LIT_UTF8_1_BYTE_CODE_POINT_MAX)
@@ -272,20 +272,18 @@ ecma_builtin_global_object_decode_uri_helper (lit_utf8_byte_t *input_start_p, /*
/* Input decode. */
if (*input_char_p != '%')
{
*output_char_p = *input_char_p;
output_char_p++;
input_char_p++;
*output_char_p++ = *input_char_p++;
continue;
}
ecma_char_t decoded_byte;
if (!lit_read_code_unit_from_hex (input_char_p + 1, 2, &decoded_byte))
uint32_t hex_value = lit_char_hex_lookup (input_char_p + 1, input_end_p, 2);
if (hex_value == UINT32_MAX)
{
ret_value = ecma_raise_uri_error (ECMA_ERR_MSG ("Invalid hexadecimal value."));
break;
}
ecma_char_t decoded_byte = (ecma_char_t) hex_value;
input_char_p += URI_ENCODED_BYTE_SIZE;
if (decoded_byte <= LIT_UTF8_1_BYTE_CODE_POINT_MAX)
@@ -337,17 +335,16 @@ ecma_builtin_global_object_decode_uri_helper (lit_utf8_byte_t *input_start_p, /*
}
else
{
ecma_char_t chr;
hex_value = lit_char_hex_lookup (input_char_p + 1, input_end_p, 2);
if (!lit_read_code_unit_from_hex (input_char_p + 1, 2, &chr)
|| ((chr & LIT_UTF8_EXTRA_BYTE_MASK) != LIT_UTF8_EXTRA_BYTE_MARKER))
if (hex_value == UINT32_MAX || (hex_value & LIT_UTF8_EXTRA_BYTE_MASK) != LIT_UTF8_EXTRA_BYTE_MARKER)
{
is_valid = false;
break;
}
octets[i] = (lit_utf8_byte_t) chr;
input_char_p += URI_ENCODED_BYTE_SIZE;
octets[i] = (lit_utf8_byte_t) hex_value;
}
}
jerry-core/ecma/builtin-objects/ecma-builtin-json.c
+3 -8
View File
@@ -174,18 +174,13 @@ ecma_builtin_json_parse_string (ecma_json_token_t *token_p) /**< token argument
}
case LIT_CHAR_LOWERCASE_U:
{
if ((end_p - current_p <= ECMA_JSON_HEX_ESCAPE_SEQUENCE_LENGTH))
uint32_t hex_value = lit_char_hex_lookup (current_p + 1, end_p, ECMA_JSON_HEX_ESCAPE_SEQUENCE_LENGTH);
if (hex_value == UINT32_MAX)
{
goto invalid_string;
}
ecma_char_t code_unit;
if (!(lit_read_code_unit_from_hex (current_p + 1, ECMA_JSON_HEX_ESCAPE_SEQUENCE_LENGTH, &code_unit)))
{
goto invalid_string;
}
ecma_stringbuilder_append_char (&result_builder, code_unit);
ecma_stringbuilder_append_char (&result_builder, (ecma_char_t) hex_value);
current_p += ECMA_JSON_HEX_ESCAPE_SEQUENCE_LENGTH + 1;
break;
}
jerry-core/ecma/builtin-objects/ecma-builtins.c
+3 -5
View File
@@ -505,12 +505,10 @@ ecma_instantiate_builtin (ecma_builtin_id_t obj_builtin_id) /**< built-in id */
ext_object_p->u.class_prop.class_id = LIT_MAGIC_STRING_REGEXP_UL;
const re_compiled_code_t *bc_p = NULL;
ecma_value_t ret_value = re_compile_bytecode (&bc_p,
ecma_get_magic_string (LIT_MAGIC_STRING_EMPTY_NON_CAPTURE_GROUP),
RE_FLAG_EMPTY);
re_compiled_code_t *bc_p = re_compile_bytecode (ecma_get_magic_string (LIT_MAGIC_STRING_EMPTY_NON_CAPTURE_GROUP),
RE_FLAG_EMPTY);
JERRY_ASSERT (ecma_is_value_empty (ret_value));
JERRY_ASSERT (bc_p != NULL);
ECMA_SET_INTERNAL_VALUE_POINTER (ext_object_p->u.class_prop.u.value, bc_p);
jerry-core/ecma/operations/ecma-regexp-object.c
+1080 -780
View File
File diff suppressed because it is too large Load Diff
jerry-core/ecma/operations/ecma-regexp-object.h
+71 -17
View File
@@ -44,30 +44,73 @@ typedef enum
} ecma_regexp_flags_t;
/**
* Structure for storing capturing group results
* Class escapes
*/
typedef enum
{
RE_ESCAPE__START, /**< escapes start */
RE_ESCAPE_DIGIT = RE_ESCAPE__START, /**< digit */
RE_ESCAPE_NOT_DIGIT, /**< not digit */
RE_ESCAPE_WORD_CHAR, /**< word char */
RE_ESCAPE_NOT_WORD_CHAR, /**< not word char */
RE_ESCAPE_WHITESPACE, /**< whitespace */
RE_ESCAPE_NOT_WHITESPACE, /**< not whitespace */
RE_ESCAPE__COUNT, /**< escape count */
} ecma_class_escape_t;
/**
* Character class flags escape count mask size.
*/
#define RE_CLASS_ESCAPE_COUNT_MASK_SIZE (3u)
/**
* Character class flags escape count mask.
*/
#define RE_CLASS_ESCAPE_COUNT_MASK ((1 << RE_CLASS_ESCAPE_COUNT_MASK_SIZE) - 1u)
/**
* Character class flags that are present in the upper bits of the class flags byte, while the 3 least significant bits
* hold a value that contains the number of class escapes present in the character class.
*/
typedef enum
{
RE_CLASS_HAS_CHARS = (1 << 5), /**< contains individual characters */
RE_CLASS_HAS_RANGES = (1 << 6), /**< contains character ranges */
RE_CLASS_INVERT = (1 << 7), /**< inverted */
} ecma_char_class_flags_t;
/**
* Structure for matching capturing groups and storing their result
*/
typedef struct
{
const lit_utf8_byte_t *begin_p; /**< capture start pointer */
const lit_utf8_byte_t *end_p; /**< capture end pointer */
const uint8_t *bc_p; /**< group bytecode pointer */
uint32_t iterator; /**< iteration counter */
uint32_t subcapture_count; /**< number of nested capturing groups */
} ecma_regexp_capture_t;
/**
* Structure for matching non-capturing groups
*/
typedef struct
{
const lit_utf8_byte_t *begin_p; /**< substring start pointer */
const lit_utf8_byte_t *end_p; /**< substring end pointer */
} ecma_regexp_capture_t;
const uint8_t *bc_p; /**< group bytecode pointer */
uint32_t iterator; /**< iteration counter */
uint32_t subcapture_start; /**< first nested capturing group index */
uint32_t subcapture_count; /**< number of nested capturing groups */
} ecma_regexp_non_capture_t;
/**
* Check if an ecma_regexp_capture_t contains a defined capture
*/
#define ECMA_RE_IS_CAPTURE_DEFINED(c) ((c)->begin_p != NULL && (c)->end_p >= (c)->begin_p)
#define ECMA_RE_IS_CAPTURE_DEFINED(c) ((c)->begin_p != NULL)
ecma_value_t
ecma_regexp_get_capture_value (const ecma_regexp_capture_t *const capture_p);
/**
* Structure for storing non-capturing group results
*/
typedef struct
{
const lit_utf8_byte_t *str_p; /**< string pointer */
} ecma_regexp_non_capture_t;
#if (JERRY_STACK_LIMIT != 0)
/**
* Value used ase result when stack limit is reached
@@ -82,27 +125,38 @@ typedef struct
#define ECMA_RE_STACK_LIMIT_REACHED(p) (false)
#endif /* JERRY_STACK_LIMIT != 0 */
/**
* Offset applied to qmax when encoded into the bytecode.
*
* It's common for qmax to be Infinity, which is represented a UINT32_MAX. By applying the offset we are able to store
* it in a single byte az zero.
*/
#define RE_QMAX_OFFSET 1
/**
* RegExp executor context
*/
typedef struct
{
const lit_utf8_byte_t *input_end_p; /**< end of input string */
const lit_utf8_byte_t *input_start_p; /**< start of input string */
const lit_utf8_byte_t *input_end_p; /**< end of input string */
uint32_t captures_count; /**< number of capture groups */
ecma_regexp_capture_t *captures_p; /**< capturing groups */
uint32_t non_captures_count; /**< number of non-capture groups */
ecma_regexp_capture_t *captures_p; /**< capturing groups */
ecma_regexp_non_capture_t *non_captures_p; /**< non-capturing groups */
uint32_t *iterations_p; /**< number of iterations */
uint16_t flags; /**< RegExp flags */
uint8_t char_size; /**< size of encoded characters */
} ecma_regexp_ctx_t;
#if ENABLED (JERRY_ES2015)
lit_code_point_t ecma_regexp_unicode_advance (const lit_utf8_byte_t **str_p, const lit_utf8_byte_t *end_p);
#endif /* ENABLED (JERRY_ES2015) */
ecma_object_t *ecma_op_regexp_alloc (ecma_object_t *new_target_obj_p);
ecma_value_t ecma_regexp_exec_helper (ecma_object_t *regexp_object_p,
ecma_string_t *input_string_p);
ecma_string_t *ecma_regexp_read_pattern_str_helper (ecma_value_t pattern_arg);
lit_code_point_t ecma_regexp_canonicalize (lit_code_point_t ch, bool is_ignorecase);
lit_code_point_t ecma_regexp_canonicalize_char (lit_code_point_t ch);
lit_code_point_t ecma_regexp_canonicalize_char (lit_code_point_t ch, bool unicode);
ecma_value_t ecma_regexp_parse_flags (ecma_string_t *flags_str_p, uint16_t *flags_p);
void ecma_regexp_create_and_initialize_props (ecma_object_t *re_object_p,
ecma_string_t *source_p,