Fix parse error of system default /usr/share/nano/*.nanorc

[snazy]

(Recent) nano packages in Ubuntu come with some .nanorc files preinstalled.

jline's NanorcParser sadly fails parsing a couple of the regular expressions.

This change translates the regular expressions to Java regular expressions.

The differences are described in org.jline.builtins.SyntaxHighlighter#posixToJavaRegex:

• The first ] in a bracket expression does not need to be escaped in Posix,translate to \].
• Same as above for a negating bracket expression like [^][], translate to [^\]\[].
• Any [ in a bracket expression does not need to be escaped in Posix, translate to \[.
• Any ] not in a bracket expression is valid in both Posix and Java, no translation.
• A backslash before the closing bracket like [.f\] is not an escape of the closing bracket, the backslash needs to be escaped for Java, translate to [.f\\].
• Do not perform the above translations within an escape via \, except for \< and \> to \b.
• Replace the Posix classes like [:word:] or [:digit:] to Java classes, inside and outside a bracket expression.

Test cases have been added.

There are however two regexes that still don't work, but those look invalid. To let jnano not trip over these, any PatternSyntaxException lets jnano just ignore the particular rule. A warning is logged in such cases.

Fixes #1156

[mattirn]

@snazy, I think we should do a more general fix rather than start to resolve patterns one by one.

In java regex characters [ and ] must be always escaped in a bracket expression. In POSIX regex [ character can be added in a bracket expression without escaping. Also ] character can be added without escaping if it is the first character of the bracket expression.

We can add a static method fixRegex(String regex) to Parse class that will escape [ and ] characters in a bracket expression.

[snazy]

Wonder whether this gets into parsing the regex itself. However, do you think its good enough to handle the cases that the Character classes chapter in Pattern's javadoc describes? I.e. looking for [ (unless it's not escaped as \[ and then handle the escaping [ ] ?

[mattirn]

Wonder whether this gets into parsing the regex itself. However, do you think its good enough to handle the cases that the Character classes chapter in Pattern's javadoc describes?

Regexes in nanorc-files follow POSIX standard. Intersections ([a-z&&[def]]), unions ([a-d[m-p]]) etc descriped in Pattern's javadoc are not valid POSIX regexes.

I.e. looking for [ (unless it's not escaped as \[ and then handle the escaping [ ] ?

In regex the first unescaped '[' starts a bracket expression. In POSIX a bracket expression is closed by unescaped ']' if it is not the next character after '[' which started the bracket expression. Every square bracket between the starting and closing brackets must be escaped in order to obtain valid java pattern.

[snazy]

Okay, I think I have it now. There are quite some more cases that needed to be handled. I've added real world test cases for those - all escapings yielded by posixToJavaRegex look correct to me. I didn't verify it in a "live jnano" though.

[snazy]

Updated the PR quite a bit in the 2nd commit.

Looking at the regexes and source rules printed by the processLocalNanorcFiles "test", the translations look reasonable.