#!/bin/bash

if [ ! $# -eq 4 ]
then
    echo "Incorrect number of arguments:" $# "/ 4"
    exit
fi

error=0

if [ ! -f $1 ]
then
    echo "Log file" $1 "not found"
    error=1
fi

if [ ! -f $2 ]
then
    echo "Eeg file" $2 "not found"
    error=1
fi

if [ ! -f $3 ]
then
    echo "Audio file" $3 "not found"
    error=1
fi

if [ $error -eq 1 ]
then
    echo "At least one file doesn't exist, goodbye"
    exit
fi

if [ ! -r $1 ]
then
    echo "Log file" $1 "not readable"
    error=1
fi

if [ ! -r $2 ]
then
    echo "Eeg file" $2 "not readable"
    error=1
fi

if [ ! -r $3 ]
then
    echo "Audio file" $3 "not readable"
    error=1
fi

if [ $error -eq 1 ]
then
    echo "At least one file is unreadable, goodbye"
    exit 0
fi


if [ ! -d $4 ]
then
    echo "Directory $4 doesnt exist, trying to create it"
    mkdir -p $4
    if [ $? -ne 0 ]
    then
        echo "Couldn't create directory $4, bye"
        exit
    fi
fi

if [ ! -w $4 ]
then
    echo "No write permissions for directory $4, bye"
    exit
fi

if [ "${EEG_TZ}" == "" ]
then
    EEG_TZ="UTC"
fi

eegDate=$(grep -E '^<StartRecordingDate>' "$2" | sed -E 's_.*>(.*)[.](.*)[.](.*)<.*_\3-\2-\1_')
eegFullHour=$(grep -E '^<StartRecordingTime>' "$2" | sed -E 's_.*>(.*)<.*_\1_')

if [ $eegDate = "" -o $eegFullHour = "" ]
then
    echo "Eeg start not found, bye"
fi

eegS=$(echo $eegDate $eegFullHour)
eegStart=$(date --date "$eegS" +%s.%N | sed -E 's_0+$__')
hz=$(grep -E '^<SamplingRate>' $2 | grep -oE '[0-9]+')

duplicates=$(grep -oE '^[^ ]+' timetable.log | sort | uniq -d)
# TODO check if not empty

wavStart=$(grep -E '^beep' $1 | grep -oE '[0-9.]+$')
#TODO check if eeg/wav start is empty

tickCount=$(grep -cE '^<tick>' $2)
firstTick=$(grep -nE '^<tick>' eeg.xml | head -n 1 | cut -d':' -f1)
echo "ticks = $tickCount"
eegDuration=$(echo "$tickCount / $hz" | bc -l)
eegEnd=$(echo "$eegStart + $eegDuration" | bc)

wavDuration=$(sox $3 -n stat 2>&1 | grep '^Length' | sed -E 's_.*[[:space:]]([.0-9]+$)_\1_')
wavEnd=$(echo "$wavStart + $wavDuration" | bc)

echo "[file] [start] [end] [duration]"
echo "eeg $eegStart $eegEnd $eegDuration"
echo "wav $wavStart $wavEnd $wavDuration"

st=$(cat $1)
IFS=$'\n'
for x in $st; do
    name=$(echo $x | grep -oE '^[^ ]+')

    # in case someone decides to put multiple spaces
    durationExpr=$(echo $x | sed -E 's_^.*[[:space:]]+([0-9.]+)[[:space:]]([0-9.]+)$_\2-\1_')

    # in unix time
    uStart=$(echo $durationExpr | cut -d'-' -f2)
    uEnd=$(echo $durationExpr | cut -d'-' -f1)
    duration=$(echo $durationExpr | bc | sed -E 's_^[.]_0._')

    pissOff="0"
    fail=$(echo "$duration < 0.2" | bc)
    if [ $fail -eq "1" ]
    then
        echo "Duration for $name is too short ($duration)"
        pissOff="1"
    fi

    fail=$(echo "$uStart < $eegStart" | bc)
    if [ $fail -eq "1" ]
    then
        echo "Start of $name is before eeg start"
        pissOff="1"
    fi

    fail=$(echo "$uStart < $wavStart" | bc)
    if [ $fail -eq "1" ]
    then
        echo "Start of $name is before wav start"
        pissOff="1"
    fi

    fail=$(echo "$uEnd > $eegEnd" | bc)
    if [ $fail -eq "1" ]
    then
        echo "End of $name is after eeg end"
        pissOff="1"
    fi

    fail=$(echo "$uEnd > $wavEnd" | bc)
    if [ $fail -eq "1" ]
    then
        echo "End of $name is after wav end"
        pissOff="1"
    fi

    if [ $pissOff -eq "1" ]
    then
        echo "At least one error encountered with $name, skipping"
        continue
    fi

    audioFileName="${4}/${name}_lar.wav"
    audioStart=$(echo "$uStart-$wavStart" | bc)
    #echo $audioFileName
    sox "$3" "$audioFileName" trim $audioStart $duration

    eegFileName="${4}/${name}_eeg.xml"
    startTick=$(echo "($uStart - $eegStart) * $hz" | bc | awk -F'.' '{print $1}')
    endTick=$(echo "($uEnd - $eegStart) * $hz" | bc | awk -F'.' '{print $1}')
    tickDiff=$(echo "$endTick - $startTick + 1" | bc)

    tailArg=$((endTick + firstTick))
    #echo "$startTick $endTick $tailArg"

    head -n $tailArg $2 | tail -n $tickDiff > $eegFileName
    echo "$name - done"
done