#!/bin/bash

[ -f ~/.d-i_manual_daily ] || exit 1
. ~/.d-i_manual_daily

[ -f "$ROOT/bin/langlist" ] || exit 1

# Order of languages is order in list.
languages=$(egrep -v "^[[:space:]]*(|#.*)$" $ROOT/bin/langlist | cut -f 1)

[ -e "$source" ] && cd $source || exit 1

[ -f "$ROOT/do_not_build" ] && exit 1

## Create directories
[ -e "$logdir" ] || mkdir -p "$logdir"

# Ignore the following type of lines:
# - empty
# - containing only reference variable (e.g. to .xml file)
# - containing an encoding setting
# - containing a (one-line) comment
# - containing only tags
grepexp="^[[:space:]]*(|&[-_.[:alnum:]]+;|<\?.+\?>|<!--.*-->|<[^<>]+>([[:space:]]*(<[^<>]+>))*)[[:space:]]*$"

lines_tot=0
files_tot=0
for lang in $languages; do
    lines_en=0
    lines_tr=0
    lines_not_tr=0
    lines_missing=0
    files_tr=0
    files_not_tr=0
    files_missing=0

    # Skip languages that do not have XML files
    [ -d $source/$lang/ ] || continue

    if [ $lang = "en" ] ; then
        for file_en in $(find en/ -name "*.xml"); do
            egrep -v "$grepexp" $file_en >/tmp/file_en.$$
            lines_en=$(wc -l /tmp/file_en.$$ | cut -d " " -f 1)
            [ $lines_en -eq 0 ] && continue

            lines_tot=$((lines_tot + $lines_en))
            files_tot=$((files_tot + 1))
        done
        echo -e "$lang\t\t$lines_tot\t$files_tot"
        continue
    fi

    for file_en in $(find en/ -name "*.xml"); do
        egrep -v "$grepexp" $file_en >/tmp/file_en.$$
        lines_en=$(wc -l /tmp/file_en.$$ | cut -d " " -f 1)
        [ $lines_en -eq 0 ] && continue

        file_tr=$(echo "$file_en" | sed "s/en\//$lang\//")
        if [ -f $file_tr ] ; then
            egrep -v "$grepexp" $file_tr >/tmp/file_tr.$$
            if diff -w -q /tmp/file_en.$$ /tmp/file_tr.$$ >/dev/null || \
               egrep -q "^<\!--[[:space:]]*original version.*untranslated[[:space:]]*-->" $file_tr ; then
                lines_not_tr=$(($lines_not_tr + $lines_en))
                files_not_tr=$(($files_not_tr + 1))
		# echo $file_en
            else
                lines_tr=$(($lines_tr + $lines_en))
                files_tr=$(($files_tr + 1))
            fi
        else
            lines_missing=$(($lines_missing + $lines_en))
            files_missing=$(($files_missing + 1))
            # echo "$lang missing: $file_en"
        fi

    done
    pct_trans=$(($lines_tr * 100 / $lines_tot))

    files_need_update=$(./scripts/doc-check $lang 2>/dev/null | \
        egrep "^en/.*[[:digit:]]* -> [[:digit:]]*" | wc -l)

    echo -e "$lang\t$pct_trans\t$lines_tr\t$files_tr\t$lines_not_tr\t$files_not_tr\t$lines_missing\t$files_missing\t$files_need_update"
done

rm /tmp/*.$$
exit 0