#!/bin/bash
# Copyright (c) 2004-2006 by Dennis Stampfer <seppy@debian.org>
# Copyright (c) 2006 by Frans Pop <fjp@debian.org>
# Licensed under the GNU General Public License
#
# Fetches sources for the installer packages and generates
# some translation-related statistics.
#
# As I am not a good bash-programmer, please imagine someone else 
# to be the author when reading this source! :)
#
# Files:
#   o packages-list
#     Names of the packages to download
#   o languages 
#     what languages to be checked?
#   o paths
#     where are the po-files located in this package?
#     (sometimes more than one, like shadow)
#
#   o packages (folder)
#     The packages get downloaded into $TEMPFILE, cleaned up
#     and moved to packages/.
# 
#

SCRIPTNAME="Installation Guide"
ROOT="/srv/d-i_manual.daily"
WD="$ROOT/bin/po_stats"
TEMP_DIR="$WD/temp"
TARGET_DIR="$ROOT/build/po_stats"
PO_DIR="$ROOT/manual/po"

# We want the manual files group writable
umask 0002

cd $WD
if [ -f manual.lock ]; then
	echo "A lockfile manual.lock exists. Exiting...";
	exit
fi
touch manual.lock

mkdir -p $TARGET_DIR

export GETTEXT="/usr/bin"
APT_OPT=""
export LANG_NAMES="../langlist"
TEMPFILE="/tmp/.gen-stats-temp-$RANDOM"
LANG=C

#some functions:
get_stat () { # how many strings are translated/untranslated/fuzzy
	$GETTEXT/msgfmt --statistics $1 2>&1 | tr "," "\n" | \
		sed -nr "/ $2/ {s/ *([0-9]+) $2.*/\1/; p}"
}

translator () { # who is last-translator?
	sed -rn "/Last-Translator: / {s/[^:]*: +(.*)\\\n.*/\1/; p}" $1
}

print_shit () { # print standard-text
	echo -n "Files fetched and report generated on: " >> $TARGET_DIR/$lang.txt
	date -u  >> $TARGET_DIR/$lang.txt
}

#########################################################################
echo -n "$SCRIPTNAME script start (by user $USER, logname $LOGNAME): "
date
echo "Working directory is $WD"


rm -rf $TARGET_DIR/*
rm -rf $TEMP_DIR
mkdir $TEMP_DIR

#check how many strings at all
echo "Counting strings..."
TOTALU=0
LANG=C

POT_LIST=`ls -1 $PO_DIR/pot/`
for file in $POT_LIST; do
	U=`get_stat $PO_DIR/pot/$file untranslated`
	echo "$file has $U untranslated. ($TOTALU)"

	TOTALU=`expr $TOTALU + $U || echo 0`
	if [ "$TOTALU" = "0" -o -z "$TOTALU" ]; then
		echo "Error while counting untranslated strings in templates."
	fi
done
echo "Total strings: $TOTALU=100%"
echo

#now check:
RANK=""

#following 3 variables for generating the graph. they include all
#strings of all languages of all packages
GR_TRANSLATED=0
GR_FUZZY=0
GR_UNTRANSLATED=0
rm -f $TEMP_DIR/rank.txt
rm -rf $TEMPFILE
for lang in `ls -1 $PO_DIR | grep -v "pot"`; do #run for every language
	[ -d "$PO_DIR/$lang" ] || continue

	NAME=$(grep "^$lang[[:space:]]" "$LANG_NAMES" | cut -f 7)
	if [ "$NAME" ]; then
		echo "$NAME ($lang)..."
	else
		echo "Skipping '$lang': not defined in '$LANG_NAMES'"
		continue
	fi

	cat <<EOF > $TARGET_DIR/$lang.txt
Statistics of Installation Guide translation for $NAME
See: http://d-i.alioth.debian.org/manual/

Outdated:
EOF
	MISSING=""

	#for total strings of a language
	TU=0
	TT=0
	TF=0
	for f in $POT_LIST; do #for every POT file
		file=`echo $f | sed "s/t$//g"`;
		fileprt="$file"
		file="$PO_DIR/$lang/$file";

		echo  "   $file..."
		if [ -e $file ]; then #if the file does exist
			LASTT=`translator $file`

			#strings of this package:
			U=`get_stat $file untranslated`
			F=`get_stat $file fuzzy`
			T=`get_stat $file translated`
			test -z "$U" && U=0
			test -z "$F" && F=0
			test -z "$T" && T=0

			#get infos:
			TU=`expr $TU + $U`
			TT=`expr $TT + $T`
			TF=`expr $TF + $F`

			#for the graph:
			GR_TRANSLATED=`expr $GR_TRANSLATED + $TT`
			GR_UNTRANSLATED=`expr $GR_UNTRANSLATED + $TU`
			GR_FUZZY=`expr $GR_FUZZY + $TF`
      			
			#don't show completely translated files:
			if [ "$U" = "0" -a "$F" = "0" ]; then
				OK="  $fileprt: $T"t" [$LASTT]\n$OK"
				echo -n
			else
				echo "  $fileprt: $T"t"$F"f"$U"u" [$LASTT]" >> $TARGET_DIR/$lang.txt
			fi

		else #=if not existing
			MISSING_U=`get_stat $PO_DIR/pot/$f untranslated`
			test -z $MISSING_U && MISSING_U=0
			MISSING="$MISSING  $fileprt: "$MISSING_U"u\n"
			#echo "  !!$path/$lang.po"
		fi
	done # POT files
 
	echo -e "\nTranslated:" >> $TARGET_DIR/$lang.txt
	echo -e "$OK" >> $TARGET_DIR/$lang.txt
	OK=""

  	echo -e "Missing files:\n$MISSING\n" >>  $TARGET_DIR/$lang.txt
	P=`echo "$TT * 100 / $TOTALU" | bc`
  	STR=`echo "$TT + $TU + $TF" | bc`
  	echo -e "Global statistics: "$TT"t"$TF"f"$TU"u ($P%)     (total strings: $TOTALU=100%)\n(From all $lang.po files)" >> $TARGET_DIR/$lang.txt
  	if [ "$STR" -gt "$TOTALU" ]; then
		echo -e "  You have more strings translated/fuzzy/untranslated than in\n  the pot-templates. Please check!" >> $TARGET_DIR/$lang.txt
	fi
	echo >> $TARGET_DIR/$lang.txt

	print_shit

	LEAD="000";
	if [ ${#LEAD} -gt ${#P} ]; then
		P="${LEAD:0:${#LEAD}-${#P}}$P"
	else
		P="$P"
	fi
	echo -e "$P%\t $lang" >> $TEMP_DIR/rank.txt

	echo "$lang done. ($P%)"
	echo
done # languages

#generate ranking:
$WD/sort-ranktxt < $TEMP_DIR/rank.txt > ../index.podata

#move files to their destination
cp HEADER.txt $TARGET_DIR/

#cleanup
echo "Cleanup..."
rm -rf $TEMP_DIR
rm -rf $TEMPFILE
rm -rf messages messages.mo

echo -n "Script end: "
date

rm $WD/manual.lock