[API] Add RegExp C API

[DanielBallaSZTE]

This patch supports creating a RegExp object through the C API.

JerryScript-DCO-1.0-Signed-off-by: Daniel Balla dballa@inf.u-szeged.hu

[LaszloLango]

Please fix the style issues.

[LaszloLango]

I was thinking on a slightly different approach. I'd avoid using the flag's enum. I'd simply create RegExp objects from strings in RegExp literal form, like:

jerry_value_t regexp = jerry_create_regexp("/a*/gim");

What do you guys think?

[zherczeg]

We would need syntax check then. That is complicated.

[LaszloLango]

@zherczeg it is not that complicated. The code is already there, just needs to be exposed. The only question is which API is better to use, which one is more intuitive.

[zherczeg]

Code is there? I think it requires a full parser environment to do this.

[LaszloLango]

Code is there? I think it requires a full parser environment to do this.

https://github.com/jerryscript-project/jerryscript/blob/master/jerry-core/parser/js/js-lexer.c#L2002

[zherczeg]

@LaszloLango First arg: parser_context_t *context_p :) So we would need to call a full parser with another flag to limit its functionality. Btw, if you want a regex this way, a jerry_eval do this for you.

[LaszloLango]

@zherczeg don't stop reading on the first line :) All you need from this function is using only the source code pointer (char *).

[zherczeg]

Regardless calling jerry_eval with a regex has the same effect, so it does not warrant a separate api function.

[DanielBallaSZTE]

@akosthekiss I've updated the PR, thank you for your suggestions, it simplified the patch a lot.
UPDATE: working on the travis failure

[akosthekiss]

LGTM

[DanielBallaSZTE]

@LaszloLango Do you have any other comments to the patch?

[LaszloLango]

I feel like my question is not really answered, but simply ignored. I am still questioning the form of this new API function. IMHO the literal form is easier to use and it is not difficult to implement.

[zherczeg]

@LaszloLango I would like to see your proposal as a patch (you can copy it into here if it is small). It is hard for me to decide which one looks better. However if your new API is the same as jerry_eval("/a*b/g") I would not support it since it is redundant.

[LaszloLango]

@zherczeg you are right, it would be the same. I can accept your argument. I am good with the current approach.

[zherczeg]

LGTM

[LaszloLango]

Only one minor thing before we merge the PR.

[LaszloLango]

LGTM