Bug #13661: DateTime.parse parses strings with a month abbreviation after a word boundary as a valid date

• Status: Rejected
• Assignee: -
• Target version: -
• ruby -v: 2.4.1
• Backport: 2.2: UNKNOWN, 2.3: UNKNOWN, 2.4: UNKNOWN

[jbcden (Jacob Chae)]

When parsing a string like "[e-mail]" it parses it into a DateTime i.e.:

require 'date'
DateTime.parse('[e-mail]')
#=> #<DateTime: 2017-03-01T00:00:00+00:00 ((2457814j,0s,0n),+0s,2299161j)>

It seems like https://github.com/ruby/ruby/blob/9c9c46e5a30fb47068337932eefde759b2763551/ext/date/date_parse.c#L1612 is the source of the issue. I think this might just require a slight tweak of the regex, it looks like things like "[e-mail]" work but not "[e-mail]".

[jeremyevans0 (Jeremy Evans)]

• Status changed from Open to Rejected

{Date,DateTime}.parse use a loose parsing. They assume that somewhere in the string, there is something representing a date or time, and do their best to find it. Only if they can find nothing related to a date or time do they raise an exception.

In this case, [e-mail] is parsed as the month March:

DateTime._parse('[e-mail]')
# => {:mon=>3}

As no year or day is given, March 1 of the current year at midnight is used.

While this loose parsing is a fairly common compliant with the methods, it isn't a bug. As the documentation states: This method does not function as a validator. You should not pass strings to {Date,DateTime}.parse unless you think they contain a date.