formail - mail (re)formatter
formail [+skip] [-total] [-bczfrktedqBY] [-p prefix]
[-D maxlen idcache]
[-x headerfield] [-X headerfield]
[-a headerfield] [-A headerfield]
[-i headerfield] [-I headerfield]
[-u headerfield] [-U headerfield]
[-R oldfield newfield]
[-n [maxprocs ]] [-m minfields] [-s [command [arg ...]]]
formail is a filter that can be used to force mail into mailbox format, perform `From '
escaping, generate auto-replying headers, do simple header munging/extracting or split up
a mailbox/digest/articles file. The mail/mailbox/article contents will be expected on
If formail is supposed to determine the sender of the mail, but is unable to find any, it
will substitute `foo@bar'.
If formail is started without any command line options, it will force any mail coming from
stdin into mailbox format and will escape all bogus `From ' lines with a `>'.
-v Formail will print its version number and exit.
-b Don't escape any bogus mailbox headers (i.e., lines starting with `From ').
Define a different quotation prefix. If unspecified it defaults to `>'.
-Y Assume traditional Berkeley mailbox format, ignoring any Content-Length: fields.
-c Concatenate continued fields in the header. Might be convenient when postprocessing
mail with standard (line oriented) text utilities.
-z Ensure a whitespace exists between field name and content. Zap fields which contain
only a single whitespace character. Zap leading and trailing whitespace on fields
extracted with -x.
-f Force formail to simply pass along any non-mailbox format (i.e., don't generate a
`From ' line as the first line).
-r Generate an auto-reply header. This will normally throw away all the existing fields
(except X-Loop:) in the original message, fields you wish to preserve need to be
named using the -i option. If you use this option in conjunction with -k, you can
prevent the body from being `escaped' by also specifying -b.
-k When generating the auto-reply header or when extracting fields, keep the body as
-t Trust the sender to have used a valid return address in his header. This causes for-
mail to select the header sender instead of the envelope sender for the reply. This
option should be used when generating auto-reply headers from news articles or when
the sender of the message is expecting a reply.
-s The input will be split up into separate mail messages, and piped into a program one
by one (a new program is started for every part). -s has to be the last option spec-
ified, the first argument following it is expected to be the name of a program, any
other arguments will be passed along to it. If you omit the program, then formail
will simply concatenate the split mails on stdout again. See FILENO.
Tell formail not to wait for every program to finish before starting the next (causes
splits to be processed in parallel). Maxprocs optionally specifies an upper limit on
the number of concurrently running processes.
-e Do not require empty lines to be preceding the header of a new message (i.e., the
messages could start on every line).
-d Tell formail that the messages it is supposed to split need not be in strict mailbox
format (i.e., allows you to split digests/articles or non-standard mailbox formats).
This disables recognition of the Content-Length: field.
Generate a log summary in the same style as procmail. This includes the entire "From
" line, the Subject: header field, the folder, and the size of the message in bytes.
The mailstat command can be used to summarize logs in this format.
-B Makes formail assume that it is splitting up a BABYL rmail file.
Allows you to specify the number of consecutive headerfields formail needs to find
before it decides it found the start of a new message, it defaults to 2.
-q Tells formail to (still detect but) be quiet about write errors, duplicate messages
and mismatched Content-Length: fields. This option is on by default, to make it dis-
play the messages use -q-.
-D maxlen idcache
Formail will detect if the Message-ID of the current message has already been seen
using an idcache file of approximately maxlen size. If not splitting, it will return
success if a duplicate has been found. If splitting, it will not output duplicate
messages. If used in conjunction with -r, formail will look at the mail address of
the envelope sender instead at the Message-ID.
Extract the contents of this headerfield from the header. Line continuations will be
left intact; if you want the value on a single line then you'll also need the -c
Same as -x, but also preserves/includes the field name.
Append a custom headerfield onto the header; but only if a similar field does not
exist yet. If you specify either one of the field names Message-ID: or Resent-Mes-
sage-ID: with no field contents, then formail will generate a unique message-ID for
Append a custom headerfield onto the header in any case.
Same as -A, except that any existing similar fields are renamed by prepending an
``Old-'' prefix. If headerfield consists only of a field-name, it will not be
Same as -i, except that any existing similar fields are simply removed. If header-
field consists only of a field-name, it effectively deletes the field.
Make the first occurrence of this field unique, and thus delete all subsequent occur-
rences of it.
Make the last occurrence of this field unique, and thus delete all preceding occur-
rences of it.
-R oldfield newfield
Renames all occurrences of the fieldname oldfield into newfield.
Skip the first skip messages while splitting.
Output at most total messages while splitting.
When renaming, removing, or extracting fields, partial fieldnames may be used to specify
all fields that start with the specified value.
By default, when generating an auto-reply header procmail selects the envelope sender from
the input message. This is correct for vacation messages and other automatic replies
regarding the routing or delivery of the original message. If the sender is expecting a
reply or the reply is being generated in response to the contents of the original message
then the -t option should be used.
RFC822, the original standard governing the format of Internet mail messages, did not
specify whether Resent header fields (those that begin with `Resent-', such as
`Resent-From:') should be considered when generating a reply. Since then, the recommended
usage of the Resent headers has evolved to consider them as purely informational and not
for use when generating a reply. This has been codified in RFC2822, the new Internet Mes-
sage Format standard, which states in part:
Resent fields are used to identify a message as having been reintroduced into the
transport system by a user. The purpose of using resent fields is to have the mes-
sage appear to the final recipient as if it were sent directly by the original
sender, with all of the original fields remaining the same....They MUST NOT be used
in the normal processing of replies or other such automatic actions on messages.
While formail now ignores Resent headers when generating header replies, versions of for-
mail prior to 3.14 gave such headers a high precedence. If the old behavior is needed for
established applications it can be specified by calling formail with the option `-a
Resent-' in addition to the -r and -t options. This usage is deprecated and should not be
used in new applications.
While splitting, formail assigns the message number currently being output to this
variable. By presetting FILENO, you can change the initial message number being used
and the width of the zero-padded output. If FILENO is unset it will default to 000.
If FILENO is non-empty and does not contain a number, FILENO generation is disabled.
To split up a digest one usually uses:
formail +1 -ds >>the_mailbox_of_your_choice
formail +1 -ds procmail
To remove all Received: fields from the header:
formail -I Received:
To remove all fields except From: and Subject: from the header:
formail -k -X From: -X Subject:
To supersede the Reply-To: field in a header you could use:
formail -i "Reply-To: foo@bar"
To convert a non-standard mailbox file into a standard mailbox file you can use:
formail -ds <old_mailbox >>new_mailbox
Or, if you have a very tolerant mailer:
formail -a Date: -ds <old_mailbox >>new_mailbox
To extract the header from a message:
formail -X ""
sed -e '/^$/ q'
To extract the body from a message:
formail -I ""
sed -e '1,/^$/ d'
mail(1), binmail(1), sendmail(8), procmail(1), sed(1), sh(1), RFC822, RFC2822, RFC1123
Can't fork Too many processes on this machine.
Content-Length: field exceeds actual length by nnn bytes
The Content-Length: field in the header specified a length that was
longer than the actual body. This causes this message to absorb a
number of subsequent messages following it in the same mailbox.
Couldn't write to stdout
The program that formail was trying to pipe into didn't accept all
the data formail sent to it; this diagnostic can be suppressed by
the -q option.
Duplicate key found: x The Message-ID or sender x in this message was found in the id-
cache; this diagnostic can be suppressed by the -q option.
Failed to execute "x" Program not in path, or not executable.
File table full Too many open files on this machine.
Invalid field-name: "x"
The specified field-name "x" contains control characters, or cannot
be a partial field-name for this option.
You can save yourself and others a lot of grief if you try to avoid using this autoreply
feature on mails coming through mailinglists. Depending on the format of the incoming
mail (which in turn depends on both the original sender's mail agent and the mailinglist
setup) formail could decide to generate an autoreply header that replies to the list.
In the tradition of UN*X utilities, formail will do exactly what you ask it to, even if it
results in a non-RFC822 compliant message. In particular, formail will let you generate
header fields whose name ends in a space instead of a colon. While this is correct for
the leading `From ' line, that line is not a header field so much as the message separator
for the mbox mailbox format. Multiple occurrences of such a line or any other colonless
header field will be considered by many mail programs, including formail itself, as the
beginning of a new message. Others will consider the message to be corrupt. Because of
this, you should not use the -i option with the `From ' line as the resulting renamed
line, `Old-From ', will probably not do what you want it to. If you want to save the
original `From ' line, rename it with the -R option to a legal header field such as `X-
When formail has to generate a leading `From ' line it normally will contain the current
date. If formail is given the option `-a Date:', it will use the date from the `Date:'
field in the header (if present). However, since formail copies it verbatim, the format
will differ from that expected by most mail readers.
If formail is instructed to delete or rename the leading `From ' line, it will not auto-
matically regenerate it as usual. To force formail to regenerate it in this case, include
-a 'From '.
If formail is not called as the first program in a pipe and it is told to split up the in-
put in several messages, then formail will not terminate until the program it receives the
input from closes its output or terminates itself.
If formail is instructed to generate an autoreply mail, it will never put more than one
address in the `To:' field.
Formail is eight-bit clean.
When formail has to determine the sender's address, every RFC822 conforming mail address
is allowed. Formail will always strip down the address to its minimal form (deleting ex-
cessive comments and whitespace).
The regular expression that is used to find `real' postmarks is:
"\n\nFrom [\t ]*[^\t\n ]+[\t ]+[^\n\t ]"
If a Content-Length: field is found in a header, formail will copy the number of specified
bytes in the body verbatim before resuming the regular scanning for message boundaries
(except when splitting digests or Berkeley mailbox format is assumed).
Any header lines immediately following the leading `From ' line that start with `>From '
are considered to be a continuation of the `From ' line. If instructed to rename the
`From ' line, formail will change each leading `>' into a space, thereby transforming
those lines into normal RFC822 continuations.
Calling up formail with the -h or -? options will cause it to display a command-line help
This program is part of the procmail mail-processing-package (v3.22) available at
http://www.procmail.org/ or ftp.procmail.org in pub/procmail/.
There exists a mailinglist for questions relating to any program in the procmail package:
for submitting questions/answers.
for subscription requests.
If you would like to stay informed about new versions and official patches send a sub-
scription request to
(this is a readonly list).
Stephen R. van den Berg
Philip A. Guenther
BuGless 2001/08/04 FORMAIL(1)