Can ECMAScript be fully compatible with PCRE?

[reinaldorauch]

I was trying to write a regex that uses a long pattern and tryied to reutilize it with the (?n) syntax but I saw that ECMAScript is not fully compatible with PCRE. There is a reason for that? If not, I would suggest that the language become fully compatible with PCRE, so we could use the full power of them.

[leobalter]

I'm not sure if this is right place to discuss this, but I think it's a long step forward to make it fully compatible with PCRE. Add the fact this might be not retrocompatible and eventually break the web.

I think the best approach here would be gradually adding RegExp features approximating the compatibility, with that, we would need to describe them in spec text. Would you be willing to help on that?

[leobalter]

I'm not sure if this is right place to discuss this, but I think it's a long step forward to make it fully compatible with PCRE. Add the fact this might be not retrocompatible and eventually break the web.

I think the best approach here would be gradually adding RegExp features approximating the compatibility, with that, we would need to describe them in spec text. Would you be willing to help on that?

[reinaldorauch]

I fully agree with that. I think now that I've rushed when posting because after I searched the mailing list and found discussions about and learned the magnitude of the job.

Of course I can help but I'm not sure that I am knowledgeable enough to endure this task.
I just wanted to open (now reopen) the discution around that.

Thanks for the reply @leobalter

[reinaldorauch]

I fully agree with that. I think now that I've rushed when posting because after I searched the mailing list and found discussions about and learned the magnitude of the job.

Of course I can help but I'm not sure that I am knowledgeable enough to endure this task.
I just wanted to open (now reopen) the discution around that.

Thanks for the reply @leobalter

[littledan]

@reinaldorauch Which feature do you want to use from PCRE which is missing in JavaScript?

[littledan]

@reinaldorauch Which feature do you want to use from PCRE which is missing in JavaScript?

[reinaldorauch]

@littledan back reference for pattern groups, like instead of doing this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?:[a-zA-Z\u00C0-\u017F]+ )*(?:[a-zA-Z\u00C0-\u017F]+)$/

I would do this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?1)*(?1)$/;

or something approximate.

[reinaldorauch]

@littledan back reference for pattern groups, like instead of doing this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?:[a-zA-Z\u00C0-\u017F]+ )*(?:[a-zA-Z\u00C0-\u017F]+)$/

I would do this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?1)*(?1)$/;

or something approximate.

[claudepache]

@littledan back reference for pattern groups, like instead of doing this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?:[a-zA-Z\u00C0-\u017F]+ )*(?:[a-zA-Z\u00C0-\u017F]+)$/

I would do this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?1)*(?1)$/;

or something approximate.

@reinaldorauch There is an obvious bug in your example: (?1) refers to the subpattern defined by the first capturing group, and there is no capturing group in your example.

---

The feature of subpatterns is described here:

http://www.pcre.org/current/doc/html/pcre2pattern.html#subpatternsassubroutines

But the feature is not only about to “not repeating oneself”, for which use case there exists already another technique: let subpattern = '(?:[a-zA-Z\u00C0-\u017F]+)'; const r = new RegExp(`^${subpattern} ${subpattern}*${subpattern}$`);. It is more useful in the context of recursive patterns:

http://www.pcre.org/current/doc/html/pcre2pattern.html#recursion

[claudepache]

@littledan back reference for pattern groups, like instead of doing this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?:[a-zA-Z\u00C0-\u017F]+ )*(?:[a-zA-Z\u00C0-\u017F]+)$/

I would do this:

const r = /^(?:[a-zA-Z\u00C0-\u017F]+) (?1)*(?1)$/;

or something approximate.

@reinaldorauch There is an obvious bug in your example: (?1) refers to the subpattern defined by the first capturing group, and there is no capturing group in your example.

---

The feature of subpatterns is described here:

http://www.pcre.org/current/doc/html/pcre2pattern.html#subpatternsassubroutines

But the feature is not only about to “not repeating oneself”, for which use case there exists already another technique: let subpattern = '(?:[a-zA-Z\u00C0-\u017F]+)'; const r = new RegExp(`^${subpattern} ${subpattern}*${subpattern}$`);. It is more useful in the context of recursive patterns:

http://www.pcre.org/current/doc/html/pcre2pattern.html#recursion