Create a Java rule to check for incorrect dates

[dpelle]

LanguageTool could have a Java rule to check for dates and report an error when the day of the week is inconsistent with the date.

For example, the following dates are all incorrect since the day (Tuesday) is inconsistent with the date (Apr 21st, 2014 is a Monday!)

• Tue Apr 21st, 2014
• Tuesday April 21st, 2014
• Tuesday 04-24-2014
• Tuesday 04/21/2014
• Tuesday 21/04/2014
• Tuesday 2014/04/21
• Tue 21/Apr/2014
• ...

We could also check for other incorrect dates such as:

• Feb 29th, 2014: it can't be correct since 2014 is not a leap year.
• Apr 31th, 2014: it can't be correct since April has only 30 days.

The rule could obviously be adapted to other languages.

[danielnaber]

I like this idea a lot. How could it be done? I see two approaches:

• Use SimpleDateParser, as described here http://stackoverflow.com/a/892204/1582948. SimpleDateParser works on plain text, not on our analyzed text.
• Use XML rules but do the actual date/day-of-week check in Java code. We'd have to change our XML to make calling Java possible,

Are there other approaches?

[dpelle]

Just seing that Grammalecte-0.4.1 French grammar checker used the idea of date checking, and implemented it. See discussion in French: http://www.dicollecte.org/thread.php?prj=fr&t=404

The discussion there says that apparently the French grammar checker Antidote also already does date checking, but MS-Word does not (I have neither of them).

[danielnaber]

This has now been implemented for German, see https://github.com/languagetool-org/languagetool/blob/master/languagetool-standalone/CHANGES.txt#L89. I'll try to add it for English, too.

[danielnaber]

Added for several other languages now, too.