feat(dateParser): Add support for HH, H, mm, m, ss, s formats

[wesleycho]

• Add support for HH, H, mm, m, ss, s formats from Angular's dateFilter
• Add support for : character in format expression

Working Plunker

This addresses #3159.

[chrisirhc]

I think tests will fail if a DATETIME_FORMAT from $locale containing : is used.
Should probably add some of those formats in the dateparser.spec below (e.g. a format of short).

[wesleycho]

Should I do a replace on the next line instead then?

[chrisirhc]

I think a better fix might be to escape all special characters in the regex before doing new RegExp on the createParser function.

d3.requote has a nice regexp to do this. See: https://github.com/mbostock/d3/blob/master/src/format/requote.js

[chrisirhc]

Did a review. The main issue I see is the fix for : may not be adequate.

[wesleycho]

Made the change requested - can you re-review this?

[chrisirhc]

Should instead do this right before the new RegExp() call inside createParser as that shows why this character replacement is needed (it's to be fed into the RegExp constructor).

There's also less weird characters to process in the createParser's formatCodeToRegex loop.

[wesleycho]

Can you explain the less weird characters part? Just moving this inside the createParser function isn't enough, there seems to be some specificity required as to when the parsing happens since the regex gets some additional characters through the iteration over formatCodeToRegex. The loop adds an open and closed set of parentheses to the regex. This means that it would be too late to modify the format string at that stage.

[chrisirhc]

Hm yea, I'm referring to replacing the characters right before the parentheses are added.

With the less weird characters, I'm referring to that if given a format of hh:mm, the createParser will now receive hh\:mmas the format when you do this here. It doesn't need that extra \ when looking for hh or mm (in the angular.forEach(formatCodeToRegex… block), it only needs the extra \ right before creating the regex property when feeding it into the RegExp constructor with the parentheses.

[wesleycho]

As per discussion between @chrisirhc and I on Slack, this turns out to be the best spot for modifying the format given how the iteration in formatCodeToRegex is implemented.

[chrisirhc]

Did another review

[wesleycho]

Linking related issues for reference #2508, #2509, #3159