Add Text.Parsing.Parser.Language and Token modules.

[cdepillabout]

This commit adds the Text.Parsing.Parser.Language and
Text.Parsing.Parser.Token modules. This is a straight port of modules
of the same name in Haskell's parsec library.

• Text.Parsec.Language
• Text.Parsec.Token

This commit also adds the (<??>) combinator, which acts like a flipped
(<?>). It is convenient to use with monadic parsers:

foo :: Parser String Unit
foo = "some message" <??> do
    string "foo"
    bar
    pure unit

[cdepillabout]

This commit adds a dependency on purescript-unicode.

If this dependency is too heavy, there are a few ways to get around it:

1. Release this PR as a separate library (just like what happened with purescript-unicode).

2. Functions from purescript-unicode are being used in the following places:

   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R759
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R762
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R765
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R768
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R771
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R774
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R777
   • https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R777

   It would be possible to reimplement these functions in the Text.Parsing.Parser.Token module with functions that just handle ASCII. The downside of this is that the parsers wouldn't be able to handle some unicode inputs.

If it is decided that the dependency on purescript-unicode is okay, then I will need to release purescript-unicode before this PR can be merged.

[cdepillabout]

One small problem I couldn't figure out was the functions near the bottom of the Text.Parsing.Parser.Token module:

https://github.com/purescript-contrib/purescript-parsing/pull/27/files#diff-13c443ec289a466b24a254814b927fa7R675

I tried to put them in the where clause of the makeTokenParser function, but I was running into the following two problems:

1. Depending on how I wrote the functions, sometimes the compiler was crashing with an error about the list passed to foldl1 being empty.
2. Sometimes I was getting a compiler error about not being able to have mutually recursive functions in a where clause.

I moved the functions in question to the top level and the problems disappered.

[michaelficarra]

You should publish this to bower. Never actually commit git dependencies, especially when it's not a SHA.

[cdepillabout]

@michaelficarra Thanks for taking a look at this PR.

I am planning to publish purescript-unicode to bower, but I am waiting for this PR to get reviewed.

I don't think I explained it well enough in the PR comments, but this PR shouldn't be merged until purescript-unicode is released. (Or if we decide not to use the purescript-unicode dependency, then it doesn't matter.)

[paf31]

What @michaelficarra said, plus a few code comments from me, plus:

• We should update the LICENSE file
• Could you please add some tests which mention haskellStyle or javaStyle specifically?

Thanks very much for working on this!

[cdepillabout]

@paf31, Thanks for taking the time to review the code. I have made the following changes:

• In commit 115989c, I changed the exports in Text.Parsing.Parser.Language and ext.Parsing.Parser.Token to be explicit.
• In commit 072ef28, I changed the documentation from Haddock-style to Markdown.
• In commit 930e594, I added Parsec's license to the LICENSE file.
• In commit 6c2745c, I added some simple tests for the haskellStyle and javaStyle LanguageDefs.
• In commit cc562ab, I fixed the purescript-unicode dependency in bower.json.

Now this PR should be ready for merging. (Or at least additional review.)

[paf31]

Sorry, I'm catching up on PRs. I'll get to this soon!

[cdepillabout]

No worries!

[paf31]

Thanks very much! And sorry again for the delay.