[Libguestfs] [PATCH 0/2] Add lightweight bindings for PCRE.

[Pino Toscano]

Hi,

(replying here since v2 of the series does not have this explanation.)

On Tuesday, 1 August 2017 16:00:15 CEST Richard W.M. Jones wrote:
> We'd like to use PCRE instead of the awful Str module.  However I
> don't necessarily want to pull in the extra dependency of ocaml-pcre,
> and in any case ocaml-pcre is rather difficult to use.
> 
> This introduces very simplified and lightweight bindings for PCRE.
> 
> They work rather like Str in that there is some global state (actually
> thread-local in this implementation) between the matching and the
> getting the substring, so you can write code like this:
> 
>   let re = PCRE.compile "(a+)b"
>   ...
> 
>   if PCRE.matches re "ccaaaabb" then (
>     let whole = PCRE.sub 0 in (* returns "aaaab" *)
>     let first = PCRE.sub 1 in (* returns "aaaa" *)
>     ...

Since we are providing a better module, with a different API (which
needs changes), what about removing the usage of a global state, in
favour of a match object holding the captures?  Something like
(starting from your example above):

  let re = PCRE.compile "(a+)b" in
  try
    let m = PCRE.match re "ccaaaabb" in
    let whole = PCRE.sub m 0 in (* returns "aaaab" *)
    let first = PCRE.sub m 1 in (* returns "aaaa" *)
  with Not_matched _ ->
    ...

This makes it possible to stop thinking about what was the last saved
state, and even keep the multiple results of matches at the same time.
Also the results are properly GC'ed once they get out of scope, and not
linger until the thread finish (or the program shutdown).
The drawback I see is that many of the Str usages are in chains of
"if ... else if ...", which could make the code slightly more complex.

Of course PCRE.matches ought to be left, but it would just return
whether the re matched, without changing any global state, and without
any result available.

WDYT?

-- 
Pino Toscano
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: This is a digitally signed message part.
URL: <http://listman.redhat.com/archives/libguestfs/attachments/20170802/20a60cf9/attachment.sig>