Use date from Received header if Date header is badly malformed

While importing some previously exported mails (which piler had filed as being sent 1970/01/01) I noticed these mails have really bad Date: headers, sometimes only mentioning the month and day and not even the year. Let’s not start about time zones…

Date: 20091023171906
Date: Tue, 26 Aug   XXXXXXXX XXXXXX
Date: Mon, 03 Nov   XXXXXXXXXXXX
Date: Sat, 30 Dec 1899 00:00:00 +0300
Date: Sun, 24 Aug   XXXX (XWXXXX)
Date: Fri, 04 Jul   Xrodkowoeuropejski czas stand.
Date: $DATE
Date: 2/25/2015 12:32 PM
Date: Wed, 17 Sep   Jerusalem Standard Time
Date: Donnerstag 26-Dez-2002 01:31:03
Date: 20100122133637

I would propose to also use the earliest date from the Received headers and use it in place of the date header if they differ to much. That could look roughly like this (in parser.c / parse_line, after testing for “Received:”):

if (state->message_state == MSG_RECEIVED && state->is_1st_header == 1)
  split buf at ';'
  store last split component in new property state->received_date

when parsing the Date, compare it to state->received_date instead of sdata->now. If the difference 
is more than a few days, use the received date as Sent date.

Please excuse the pseudocode, my C is rather rusty, so I didn’t really try.

I think that would give a much better approximation of the message date. After all, the servers along the way adhere to the format standards much better than some random mail client…

Comments (4)