upstream-tracker

#!/bin/bash
# Script to track mainline progress of patches in TI vendor tree
# Based on a script provided by Tomi Valkeinen <tomi.valkeinen@ti.com>
#
# License: GNU Public License, version 2.0

# paths we're interested in. speeds up processing quite a bit to limit the commits.
#paths="drivers/mmc/host"
count=$(perl -MPOSIX -le 'print LONG_MAX')
committers='\(nsekhar@ti.com\)\|\(kishon@ti.com\)'
devbranch=ti-linux/ti-linux-4.9.y
upstreambranch=linux-next/master
base=$(git describe $devbranch | cut -d "-" -f 1)
patchidweight=50
mentionweight=40
tmpfile1=`mktemp`
tmpfile2=`mktemp`
use_fuzz=0

function cleanup {
	rm -rf $tmpfile1
	rm -rf $tmpfile2
}
trap cleanup EXIT

# Generate list of commits with author name, headline, patch-id and commit-id
# Arg1: commit head to fetch commits from
# Arg2: list of committers to filter by
function commitlist {
	for c in $(git rev-list --max-count=$count --no-merges $1 ^$base --committer=$2 $paths)
	do
		git log -1 --pretty=format:"%an%x09%s%x09" $c
		patchid=`git show --pretty=fuller $c | git patch-id --stable | cut -d " " -f 1`
		printf "%s\t%s\n" $patchid $c
	done
}

# Update an existing file with new commits
# Arg1 : name of file to update
# Arg2 : commit head to fetch commits from
# Arg3 : list of committers to filter by
function updatefile {
	# Get base commit in existing file
	base=`head -1 $1 | awk -F'\t' '{print $4}'`
	# Get list of new commits from base
	commitlist $2 $3 > $tmpfile1
	# If there are new commits
	if [ -n "`cat $tmpfile1`" ]
	then
		# save existing file
		cp $1 $tmpfile2
		# update file with new + old commits
		cat $tmpfile1 $tmpfile2 > $1
		rm -f $tmpfile2
		echo "file $1 updated" 1>&2
	fi
	echo `wc -l $tmpfile1 | cut -d " " -f 1`
}

echo -n "development base: " 1>&2
git describe $devbranch 1>&2
echo -n "upstream base: " 1>&2
git describe $upstreambranch 1>&2
echo -n "patch base: " 1>&2
git describe $base 1>&2

# collect patch ids for commits in ti-linux
if [ -f ti-patch-id-list ]
then
	echo "using existing file ti-patch-id-list" 1>&2
	numcommits=`updatefile ti-patch-id-list $devbranch $committers`
	echo "updated ti-patch-id-list with $numcommits commits" 1>&2
else
	commitlist $devbranch $committers > ti-patch-id-list
	echo "done creating ti-patch-id-list" 1>&2
fi

# collect patch ids for commits in mainline
if [ -f upstream-patch-id-list ]
then
	echo "using existing file upstream-patch-id-list" 1>&2
	numcommits=`updatefile upstream-patch-id-list $upstreambranch '@'`
	echo "updated upstream-patch-id-list with $numcommits commits" 1>&2
else
	commitlist $upstreambranch '@' > upstream-patch-id-list
	echo "done creating upstream-patch-id-list" 1>&2
fi

# Find a headline with fuzz of 5 characters
# Echoes the line found, warns if multiple lines found, and returns the weight of the match (0-10)
function fuzzygrep {
	weight=5
	for i in `seq 0 $weight`
	do
		lines=`tre-agrep -k -E $i "$1" $2`
		if [ -z "$lines" ]
		then
			count=0
		else
			count=`echo "$lines" | wc -l`
		fi
		if [ $count -gt 1 ]
		then
			echo "warning: multiple matches found at fuzz" $i "for headline" \""$1"\" "(using only first match)" 1>&2
			echo "$lines" 1>&2
		fi
		if [ $count -gt 0 ]
		then
			weight=`expr $weight - $i`
			echo "$lines" | head -1
			return `expr $weight \* 2`
		fi
	done
}

# returns the shortened form of a commit
function shortcommit {
	echo `git log -1 --format="%h" $1`
}

# combine both lists into one
printf "Author\tHeadline\tTI Commit\tPatch-id match\tHeadline match\tPatch description mention\tScore\n"
while read l; do
	score=0
	author=`echo "$l" | awk -F'\t' '{print $1}'`
	headline=`echo "$l" | awk -F'\t' '{print $2}'`
	patchid=`echo "$l" | awk -F'\t' '{print $3}'`
	commitid=`echo "$l" | awk -F'\t' '{print $4}'`

	foundline=`grep -m 1 $patchid upstream-patch-id-list`

	upcommitid=`echo "$foundline" | awk -F'\t' '{print $4}'`
	if [ -z $upcommitid ]
	then
		upcommitid="NA"
	else
		upcommitid=$(shortcommit $upcommitid)
		score=`expr $score + $patchidweight`
	fi

	printf "%s\t%s\t%s\t%s\t" "$author" "$headline" $(shortcommit $commitid) $upcommitid

	if [ $use_fuzz -eq 1 ]
	then
		foundline="$(fuzzygrep "$headline" upstream-patch-id-list)"
		grepweight=$?
	else
		foundline=`grep "$headline" upstream-patch-id-list | head -1`
		grepweight=10
	fi

	upcommitid=`echo "$foundline" | awk -F'\t' '{print $4}'`
	if [ -z $upcommitid ]
	then
		upcommitid="NA"
	else
		upcommitid=$(shortcommit $upcommitid)
		score=`expr $score + $grepweight`
	fi

	printf "%s\t" $upcommitid

	upcommitid=`git show --pretty=fuller $commitid | grep -E "commit [[:alnum:]]{40} upstream" | cut -d " " -f 6`
	if [ -z $upcommitid ]
	then
		upcommitid="NA"
	else
		upcommitid=$(shortcommit $upcommitid)
		score=`expr $score + $mentionweight`
	fi

	printf "%s\t%s\n" $upcommitid $score

done < ti-patch-id-list