Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:73262
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain hoa-project.net from 95.130.10.56 cause and error)
Message-ID: <53280D52.5040209@hoa-project.net>
Date: Tue, 18 Mar 2014 10:09:38 +0100
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:29.0) Gecko/20100101 Thunderbird/29.0a2
MIME-Version: 1.0
To: internals@lists.php.net
References: <CAEKnhAEFCALpqA8_Gsr-TH_4GUmscrGhLRD70VcAFFV3ij68PQ@mail.gmail.com>
In-Reply-To: <CAEKnhAEFCALpqA8_Gsr-TH_4GUmscrGhLRD70VcAFFV3ij68PQ@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] New JSON parser
From: ivan.enderlin@hoa-project.net ("Ivan Enderlin @ Hoa")

On 13/03/2014 20:48, Jakub Zelenka wrote:
> Hi,
Hi Jakub,


> I have create a new JSON parser using conditional re2c and pure pull Bi=
son
> parser. It's a native UTF-8 parser licensed under PHP license (it can b=
e
> used for Evil though :) ). The extension is available at
>
> https://github.com/bukka/php-jsond
Very nice work, thanks!


> [snip]
>
> I need to do more testing before creating RFC for replacing the current=

> parser. There is still space for further improvements. If anyone has an=
y
> ideas, please let me know. Or if you could test it, that would be great=

> too! ;)
For my PhD thesis in the automatic testing domain, I have created some=20
grammar-based testing algorithms, based on our dedicated LL(k) compiler=20
compiler (with its dedicated grammar description language called PP).=20
Please, see the article [1] (along with the presentation [2] and all the =

details [3] about the article and the conference) and also the tool [4]=20
(called Hoa\Compiler). In this article, my experimentation consisted to=20
generate a lot of JSON strings (based on the JSON grammar [5] written in =

PP) and compared them against the JSON parsers of Gecko and PHP. Now I=20
re-play this experimentation but I compare all the generated data with=20
ext/json and ext/jsond to see if there is no potential regression. Also, =

I test it with a bounded exhaustive algorithm: it means we generate all=20
possible JSON strings up to a given size (the unit is the number of=20
tokens in a sequence, so `{`, `true` or `foo` are tokens). Note that we=20
have two ther algorithms: uniform random generation and coverage-based=20
generation.

I have created a little repository to share my work [6]. I have=20
generated all sequences up to 15 tokens, which represents 356'327 data=20
and no one has failed. Congrats!
Just for the record, a good test is a test that fails. Here, I have=20
detected no regeression, and because I have previously compared the=20
ext/json with the JSON parser from Gecko, we can consider your=20
implementation as =E2=80=9Csafe=E2=80=9D.

This is my little contribution of the morning :-). You can use this work =

to generate data in a static file and use them to compare the memory and =

CPU usage between ext/json and ext/jsond also.


Best regards.

[1] http://hoa-project.net/Literature/Research/Amost12.html
[2] http://keynote.hoa-project.net/Amost12/EDGB12.pdf
[3] http://hoa-project.net/Event/Amost12.html
[4] https://github.com/hoaproject/Compiler
[5] https://github.com/hoaproject/Json/blob/master/Grammar.pp
[6] https://github.com/Hywan/jsond-test

--=20
Ivan Enderlin
Developer of Hoa
http://hoa-project.net/

PhD. student at DISC/Femto-ST (Vesontio) and INRIA (Cassis)
http://disc.univ-fcomte.fr/ and http://www.inria.fr/

Member of HTML and WebApps Working Group of W3C
http://w3.org/