[PATCH] git-am --one

[PATCH] git-am --one

[H. Peter Anvin]

[-- Attachment #1: Type: text/plain, Size: 344 bytes --]

This patch adds the --one (-o) option to git-am, to apply a single 
message in RFC 2822 format, as opposed to an mbox.  With some MUAs it's 
a lot easier to save individual messages than with mboxes, and either 
way the user may want to control the ordering if there are known 
interdependencies.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>

[-- Attachment #2: git-am-one --]
[-- Type: text/plain, Size: 2100 bytes --]

diff --git a/Documentation/git-am.txt b/Documentation/git-am.txt
index a415fe2..bec5c84 100644
--- a/Documentation/git-am.txt
+++ b/Documentation/git-am.txt
@@ -8,7 +8,7 @@ git-am - Apply a series of patches in a 
 
 SYNOPSIS
 --------
-'git-am' [--signoff] [--dotest=<dir>] [--utf8] [--binary] [--3way] <mbox>...
+'git-am' [--one] [--signoff] [--dotest=<dir>] [--utf8] [--binary] [--3way] <mbox>...
 'git-am' [--skip | --resolved]
 
 DESCRIPTION
@@ -19,6 +19,10 @@ current branch.
 
 OPTIONS
 -------
+--one::
+	Indicates that <mbox> contains a single message in RFC 2822
+	format, without mbox-style separators.
+
 --signoff::
 	Add `Signed-off-by:` line to the commit message, using
 	the committer identity of yourself.
diff --git a/git-am.sh b/git-am.sh
index 6ed527c..9628651 100755
--- a/git-am.sh
+++ b/git-am.sh
@@ -4,7 +4,7 @@
 . git-sh-setup
 
 usage () {
-    echo >&2 "usage: $0 [--signoff] [--dotest=<dir>] [--utf8] [--binary] [--3way] <mbox>"
+    echo >&2 "usage: $0 [--one] [--signoff] [--dotest=<dir>] [--utf8] [--binary] [--3way] <mbox>"
     echo >&2 "	or, when resuming"
     echo >&2 "	$0 [--skip | --resolved]"
     exit 1;
@@ -98,11 +98,13 @@ fall_back_3way () {
 }
 
 prec=4
-dotest=.dotest sign= utf8= keep= skip= interactive= resolved= binary=
+dotest=.dotest sign= utf8= keep= skip= interactive= resolved= binary= one=
 
 while case "$#" in 0) break;; esac
 do
 	case "$1" in
+	-o|--o|--on|--one)
+	one=t; shift;;
 	-d=*|--d=*|--do=*|--dot=*|--dote=*|--dotes=*|--dotest=*)
 	dotest=`expr "$1" : '-[^=]*=\(.*\)'`; shift ;;
 	-d|--d|--do|--dot|--dote|--dotes|--dotest)
@@ -165,8 +167,16 @@ else
 	mkdir -p "$dotest" || exit
 
 	# cat does the right thing for us, including '-' to mean
-	# standard input.
-	cat "$@" |
+	# standard input.  To prepend the header, we thus want a
+	# second cat, sigh...
+	if test "$one" = t
+	then
+		LANG=C date +'From - %a %b %d %T %Y' > "$dotest"/fakefrom
+		fakefrom="$dotest"/fakefrom
+	else
+		fakefrom=/dev/null
+	fi
+	cat "$@" | cat "$fakefrom" - |
 	git-mailsplit -d$prec "$dotest/" >"$dotest/last" || {
 		rm -fr "$dotest"
 		exit 1

Re: [PATCH] git-am --one

[Junio C Hamano]

"H. Peter Anvin" <hpa@zytor.com> writes:

> This patch adds the --one (-o) option to git-am, to apply a single 
> message in RFC 2822 format, as opposed to an mbox.  With some MUAs it's 
> a lot easier to save individual messages than with mboxes, and either 
> way the user may want to control the ordering if there are known 
> interdependencies.

> +	# standard input.  To prepend the header, we thus want a
> +	# second cat, sigh...
> +	if test "$one" = t
> +	then
> +		LANG=C date +'From - %a %b %d %T %Y' > "$dotest"/fakefrom
> +		fakefrom="$dotest"/fakefrom
> +	else
> +		fakefrom=/dev/null
> +	fi
> +	cat "$@" | cat "$fakefrom" - |
>  	git-mailsplit -d$prec "$dotest/" >"$dotest/last" || {
>  		rm -fr "$dotest"
>  		exit 1

I understand the motivation, but I suspect skipping mailsplit
might be simpler, like this untested code perhaps?

	case "$one" in
        '')
        	... original code ...
		;;
        *)
		one=`printf "%${prec}d" 1`
        	cat "$@" >"$dotest/$one"
                echo "$one" >"$dotest/last"
                ;;
	esac

Re: [PATCH] git-am --one

[H. Peter Anvin]

Junio C Hamano wrote:
> 
> I understand the motivation, but I suspect skipping mailsplit
> might be simpler, like this untested code perhaps?
> 
> 	case "$one" in
>         '')
>         	... original code ...
> 		;;
>         *)
> 		one=`printf "%${prec}d" 1`
>         	cat "$@" >"$dotest/$one"
>                 echo "$one" >"$dotest/last"
>                 ;;
> 	esac
> 

If that works, great.  I just implemented it in the "most obviously 
correct" way, meaning with as few changes as possible.

	-hpa

Re: [PATCH] git-am --one

[Junio C Hamano]

"H. Peter Anvin" <hpa@zytor.com> writes:

> Junio C Hamano wrote:
>> I understand the motivation, but I suspect skipping mailsplit
>> might be simpler, like this untested code perhaps?
>> 	case "$one" in
>>         '')
>>         	... original code ...
>> 		;;
>>         *)
>> 		one=`printf "%${prec}d" 1`
>>         	cat "$@" >"$dotest/$one"
>>                 echo "$one" >"$dotest/last"
>>                 ;;
>> 	esac
>>
>
> If that works, great.  I just implemented it in the "most obviously 
> correct" way, meaning with as few changes as possible.

Another thing you may probably want is to loop over "$@", so
that the flag is not --one anymore, but --2822 (or --bare as
opposed to mbox format) and do something like this:

	case "$series_of_2822_messages" in
        '') ... original code ... ;;
        *)
        	i=1
                for input
                do
			this=`printf "%${prec}d" $i
                        cp "$input" "$dotest/$this"
                        i=$(($i+1))
		done
                printf "%${prec}d" $# >"$dotest/last"
                ;;
	esac

Re: [PATCH] git-am --one

[H. Peter Anvin]

Junio C Hamano wrote:
> 
> Another thing you may probably want is to loop over "$@", so
> that the flag is not --one anymore, but --2822 (or --bare as
> opposed to mbox format) and do something like this:
> 
> 	case "$series_of_2822_messages" in
>         '') ... original code ... ;;
>         *)
>         	i=1
>                 for input
>                 do
> 			this=`printf "%${prec}d" $i
>                         cp "$input" "$dotest/$this"
>                         i=$(($i+1))
> 		done
>                 printf "%${prec}d" $# >"$dotest/last"
>                 ;;
> 	esac
> 

That is definitely a good idea.  I'll mess with it a bit.

	-hpa

Re: [PATCH] git-am --one

[Junio C Hamano]

"H. Peter Anvin" <hpa@zytor.com> writes:

> This is a multi-part message in MIME format.
> --------------080605010801030702080802
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
> Content-Transfer-Encoding: 7bit

BTW, this is totally offtopic, but I find it somewhat puzzling
to see somebody like you, one of Linus' trusted lieutenants, is
sending a patch as an attachment.