Minor changes to allow for subclassing

[ghost]

This there are a few minor changes to lib/regexify.rb to allow for easier use of the Regexify class as a base class. There is a new file called examples/example.rb to demonstrate. examples/example.rb contains three different subclasses of Regexify of varying complexity. The monkey-patch of Regexify is just for the convenience of the test cases at the bottom.

I hope you will consider merging this into the master branch of your repository.

Thanks for letting me contribute.

-- Steve Witten

[jusleg]

Hey Steve, is there a reason why you closed this PR? It looks really promising 😄

[ghost]

Probably because this is the first one I've ever done. I'll re-open it. Sorry for the newbie mistake.

…

-- sw

On Sun, Oct 14, 2018 at 7:20 PM Justin Léger ***@***.***> wrote: Hey Steve, is there a reason why you closed this PR? — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACPr9fdP1FgSvXJmrLYzkrJ9GAQw64T9ks5uk_DngaJpZM4Xba9l> .

[jusleg]

@stevewi All good. I like what I see. Perhaps we could introduce your variations as options in the main gem. What do you think?

[ghost]

There are very few changes in your code itself:

Logic changes

• Saving the constant PATTERNS into an instance variable and using the instance variable instead of the constant in the extract_regex method.
• You might consider splitting PATTERNS into a module of it's own with methods to manage it's contents (if you did this, you wouldn't need to store PATTERNS in an instance variable so subclasses could manage it...I considered doing this but opted for the quick/dirty method I used for this).

Visibility changes

• Making the instance variables protected instead of private so subclasses can access them without methods.
• Making the Error class public so users and subclasses of Regexify can throw it.
• All other methods retain their original visibility.

The real action is in the example (examples/example.rb)...a demonstration of the power that these changes can provide (the monkey patch addition of to_s was purely for convenience since it will output something different than Regexp.to_s):

• RegexifyWithOptions allows you to initialize/change the options of the Regexp (see http://www.rubyguides.com/2015/06/ruby-regex/ ... down at the bottom) to change how the Regexp works.
• RegexifyWithAnchors adds methods to insert anchors (see http://www.rexegg.com/regex-anchors.html) to the generated Regexp.
• RegexifyANSI causes the generated Regexp to use ANSI notation for various common patterns: [[:upper:]] for [A-Z], [[:digit:]] for [0-9] etc. It also demonstrates how a subclass can add new named patterns; e.g., :hexdigit for [a-fA-F0-9] or [[:xdigit:]].

I have an application that scans in-/out-bound mail for certain sets of regexps and sets a tag header in the message depending on what it finds. I'm certainly no expert at writing regexps and most of the ones I have are fairly simple from a guru's point of view (they do, however, use the \b anchor and the imx options a lot 😉).

I'm looking for a way to build/edit regexps more easily (and in symbols that my pea-brain understands) -- perhaps using Regexify's simple grammar for a rudimentary file representation using YAML or JSON such that reading/parsing a file would result in an array of Regexify objects.

Anyway, that's my motivation...

-- sw

P.S. I would also like any file format to include a minimal amount of reuse of expressions using Boolean logic (similar to SpamAssassin rules: https://wiki.apache.org/spamassassin/WritingRules) and some rudimentary amount of forward chaining on substrings using Regexp capture groups (see https://stackoverflow.com/questions/9303984/ruby-regexp-group-matching-assign-variables-on-1-line#9304049).

[ghost]

It's your gem... I only want to use it without having to support it 😉 I don't know what you mean by 'options in the main gem'... Can you please explain? I'm afraid I'm not very familiar with gem packaging either. Be aware that the 'RegexifyANSI' class may generate nonsensical Regexps -- and that the class is an example of what is possible...not what is possible and correct (I think it needs a little more thought). I posted a note about the changes and my motivation for them on the PR. I don't know if you've read it but, if not, I'd encourage you to. It sort of lays out a vision of what I want to accomplish with Regexify. Regards, Steve

…

On Mon, Oct 15, 2018 at 9:12 AM Justin Léger ***@***.***> wrote: @stevewi <https://github.com/stevewi> All good. I like what I see. Perhaps we could introduced your variations as options in the main gem. What do you think? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ACPr9aIrjI_FdQZJ6Ezsr0dBRWnuqv4Vks5ulLP_gaJpZM4Xba9l> .

[ghost]

On Mon, Oct 15, 2018 at 11:04 AM Steve Witten <caponecicero@gmail.com> wrote: <snip />

I don't know what you mean by 'options in the main gem'... Can you please explain? I'm afraid I'm not very familiar with gem packaging either.

P.S. What I've submitted is totally backwards-compatible from a functionality perspective with your gem. I'm always leary of introducing scads of incomprehensible (and possibly incompatible) options. -- sw

…

[ghost]

Removed a case...when construct & replaced with 3 equivalent <statement> if <condition> constructs...less code and possibly faster...and possible solves one of the codebeat issues.