'authors-prog' for git-svn when importing subversion repositories from SourceForge

sf2email.sh

#!/bin/bash

# sf2email v0.2 (2011-05-11)
# By Robert Quattlebaum <darco@deepdarc.com>
#
# PUBLIC DOMAIN
#
# This shell script was originally designed to be used as the 'authors-prog'
# for git-svn. It takes the sourceforge username as an argument, and outputs
# the name and email address of the individual to stdout.
#
# The results are cached locally to speed up future checks.
#
# This script uses cURL to grab the user's sourceforge page and then scrapes
# their real name from the HTML.
#
# Up-to-date versions of this script can be found here:
#
#     <https://gist.github.com/gists/935424>
#
# Warning: Currently only tested on MacOS X.
#
# Feedback and improvements are welcome!
#

[ $# = 1 ] || {
	[ $# > 1 ] && echo "$0: error: Too many arguments."
	echo "sf2email - Converts a sourceforge username to an name+email line"
	echo "Usage: $0 " '<sf-username>'
	exit 1
} 2>&1

which -s curl || {
	echo "$0: error: 'curl' command not found."
	exit 2
} 1>&2

SF_USERNAME=$1

# Handle "(no author)" cases. 
[ "$SF_USERNAME" = "(no author)" ] && {
	echo "(no author) <nobody@localhost>"
	exit 0
}

SF_USERNAME_LC=`echo $1 | tr '[:upper:]' '[:lower:]' | tr -C -d '[:alnum:].'`

[ "$SF_USERNAME_LC" = "." -o "$SF_USERNAME_LC" = ".." -o "$SF_USERNAME_LC" = "" ] && {
	echo "$0: error: Invalid username \"$SF_USERNAME\""
	exit 2
} 1>&2

CACHE_DIR="$TMPDIR/sf2email.cache"

# If the cache dir doesn't exist, make it.
[ -d $CACHE_DIR ] || mkdir -p $CACHE_DIR

if [ -r "$CACHE_DIR/$SF_USERNAME_LC" ]
then # Cache hit
	SF_EMAIL=`cat "$CACHE_DIR/$SF_USERNAME_LC"`
else # Cache miss
	
	# Try to scrape their real name from their user page
	SF_REALNAME="$(
		curl -s "http://sourceforge.net/users/$SF_USERNAME" |
			grep '<label>Public Name:</label>' |
			sed -E '
				s/<[^>]*>//g;
				s/Public Name://g;
				s/^[[:space:]]*([^[:space:]])/\1/;
				s/([^[:space:]])[[:space:]]*$/\1/;
			'
	)"
	
	# If the above command didn't work, we can just try using the username as-is
	[ "$SF_REALNAME" = "" ] && {
		SF_REALNAME=$SF_USERNAME
		echo "$0: warning: Unable to scrape real name for \"$SF_USERNAME\"." 1>&2
	}

	SF_EMAIL="$SF_REALNAME <$SF_USERNAME_LC@users.sourceforge.net>"
	
	# Cache the result to avoid future processing.
	echo $SF_EMAIL > "$CACHE_DIR/$SF_USERNAME"
fi

echo $SF_EMAIL