Regex::Diagram.pm

Regex::Diagram.pm development page

I wrote a Perl Module which parses a regex and draws a diagram of it. I use this module to write a textbook for my tutorial at Perl/Ruby Conference in Kyoto.

Here's a (huge) example:
Regex used to parse regexes

Surprised? The module produces this chart AUTOMATICALLY from the regex! Actually, it generates a Tgif's object file. Thus you can edit, print, make eps, png, pdf... with Tgif.
The current status is just a three-day-hack. Anyway, here is the code:
Regex-Diagram-0.20.tar.gz

Please give me your comments.
Now, I want to ask native English speakers about the name. When I started the module, I named it "Regex::Diagram". But now I wonder if this name is better than others like "Regex::Chart", "Regex::FlowChart"... How do you think?

History and Acknowledgements

I learned drawing a flow chart for a regex in Jeffrey Friedl's tutorial at Perl Conference 1998. Jeffrey called it a "street map".

I gave a tutorial on regex at Open Source Matsuri '99. I wrote a small program which generates Tgif object file. But I had to write a strange language like this:

## -?(\d+(\.\d*)?|\.\d+)


#S,#m0,#ene,#v1,#anne,#m1,#g0,#ah11,#me,#g1,-#{lcurve1},#ge,#x-1,\
#m0,#ene,#v-2,#nne,#m1,#g0,#ese,#sse,#mlow, \
#g1,#ah2,#[\d],#m0,#ene,#annw,#m1,#g0,#h2,#me,#g1,#-,#h-3,#-,#{ureturn0},#ge,\
#{vfork1},#ah20,#mend,#g1,#ah1,#[\.],#h1, \
#h5,#m0,#ene,#v1,#annw,#m1,#g0,#h2,#me,#g1,#-,#[\d],#-,#wsw,#v-1,#asse,#ge,\
#ese,#v2,#asse,\
#glow,#ah12,#[\.],#h1, \
#[\d],#m0,#ene,#v1,#annw,#m1,#g0,#h2,#me,#g1,#-,#h-3,#-,#wsw,#v-1,#se,#ge,\
#h2,#en,#v-3,#ane,#gend,#E,

This means: "start symbol, mark here as 0, round curve east-north-east, vertical line 1 unit, curved arrow north-north-east, mark as 1, go back to point 0, ..." I wanted an automation from regex to this language, but had no time.

I found Jeff Pinyan's RegexParser module on the perl6-regex list. I thought now I have the parser to use! Unfortunately, his module doesn't provide a tree data structure. So I decided to write a parser myself (but stole his regex :P). The above regex is simplified from his so that it can run on Perl 5.005_03.
Having the parse-tree, I'm left with two problems: OO-based Tgif object in Perl, and tree to chart converter. I wrote them in three days.
Now I have enough tools to write my tutorial note.
You can look at my tutorial text here.

Future Works

These are only possibilites and not promises.

Decide a proper name for the module.
Write documentation in pod.
Either (a) merge Regex/Diagram/*.pm files into one Diagram.pm, or (b) separate into self-contained packages: Regex::Parser, Tgif::Object...
Post to CPAN.
Submit to TPJ? / Submit to TPC?
Attack on more regex: POSIX-char class, perl's other (?...) constructs (pssp... no! you can't parse (?{perl code}) with regex!).
Tell me your ideas! =)

Return

maeda@tokyo.pm.org
Last Modified: November 10, 2000.