Implement ECMA262 RegExp builtin object and RegExp engine #169
Conversation
LaszloLango
added
ecma core
|
The RegExp engine is not feature complete, but the core features are working. The patch is huge, so we can start the review process now. The RegExp engine is blocking a few string builtin function, so we need this patch with the core features and we can add the unimplemented things in a separate PR. |
|
@egavrin, please review |
| ECMA_TRY_CATCH (obj_this, ecma_op_to_object (this_arg), ret_value); | ||
|
|
||
| ecma_object_t *obj_p = ecma_get_object_from_value (obj_this); | ||
| ecma_property_t *bytecode_prop = ecma_get_internal_property (obj_p, ECMA_INTERNAL_PROPERTY_REGEXP_BYTECODE); |
There was a problem hiding this comment.
What happens if there is no bytecode on the this object?
For example:
var re = new RegExp("a")
re.exec.call({}, "a")
|
I've updated the PR. |
| string_desc_p->is_stack_var = false; | ||
| string_desc_p->container = ECMA_STRING_CONTAINER_HEAP_CHUNKS; | ||
| string_desc_p->hash = ecma_chars_buffer_calc_hash_last_chars (string_p, length); | ||
|
|
| for (uint32_t i = 0; i < re_ctx.num_of_captures; i += 2) | ||
| { | ||
| ecma_string_t *index_str_p = ecma_new_ecma_string_from_uint32 (i / 2); | ||
| if (re_ctx.saved_p[i] && re_ctx.saved_p[i + 1] && re_ctx.saved_p[i + 1] >= re_ctx.saved_p[i]) |
|
@LaszloLango we're leaving too many things for "unicode integration". General coding style, including comments provided by reviewers can be fixed within 3-4 hours. And the whole PR would be really OK. I clearly understand the fact that not all components of the engine follow the common style, but we definitely should fix this. So, I have no intention to increase technical debt of the project. Please, fix all issues in the pull request carefully. |
I wouldn't mind creating a single patch from the work, but creating 7 requires too much work. At least another week. This is just slowing us down unnecessarily, especially with such a tight deadlines. Regex is required for full conformance, and the success of Jerry project. Working on this forever would jeopardize our aims! |
LaszloLango
force-pushed
the
regexp_dev
branch
2 times, most recently
from
727b54f to
a59cd65
Compare
|
@egavrin, I've updated the PR. Please check |
|
@egavrin, I've updated the PR. Please check |
| * @return unsigned integer | ||
| */ | ||
| uint32_t | ||
| ecma_char_hex_to_int (char hex) |
There was a problem hiding this comment.
Did you removed this function from lexer.cpp?
JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
…ber of characters, instead of zero-terminated string. JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
…ion, ECMA-262 v5, 15.10.2.6); move hex_to_int from lexer to jerry-core/ecma/base/ecma-helpers-char.cpp, renaming it to ecma_char_hex_to_int. JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
…, RegExp and RegExp.prototype built-in objects. JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
- add regular expressions support to JS parser and interpreter; - add tests for regular expressions. JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected] JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]
JerryScript-DCO-1.0-Signed-off-by: Szilard Ledan [email protected]
JerryScript-DCO-1.0-Signed-off-by: László Langó [email protected]