run

#! /bin/bash
#

CONTAINER="[ stanfordcni/pfile-mr-classifier ]"


##############################################################################
# Configure paths

FLYWHEEL_BASE=/flywheel/v0
OUTPUT_DIR=$FLYWHEEL_BASE/output
INPUT_DIR=$FLYWHEEL_BASE/input/pfile


##############################################################################
# Parse configuration

function parse_config {

  CONFIG_FILE=$FLYWHEEL_BASE/config.json
  MANIFEST_FILE=$FLYWHEEL_BASE/manifest.json

  if [[ -f $CONFIG_FILE ]]; then
    echo "$(cat $CONFIG_FILE | jq -r '.config.'$1)"
  else
    CONFIG_FILE=$MANIFEST_FILE
    echo "$(cat $MANIFEST_FILE | jq -r '.config.'$1'.default')"
  fi
}


##############################################################################
# Set Time Zone

TZ="$(parse_config 'timezone')"
echo "${CONTAINER}  Setting time zone to: $TZ"
echo "$TZ" > /etc/timezone && ln -snf /usr/share/zoneinfo/"$TZ" /etc/localtime
dpkg-reconfigure -f noninteractive tzdata


##############################################################################
# Check I/O directories and Generate metadata

# Check that /output directory is empty
if [ "-d" "$OUTPUT_DIR" ]
    then
        if [ "$(ls -A $OUTPUT_DIR)" ]; then
            echo -e "$CONTAINER  Warning $OUTPUT_DIR is not Empty! Results may be overwritten."
        fi
    else
        echo -e "$CONTAINER  $OUTPUT_DIR not found. It will be created."
        mkdir $OUTPUT_DIR
fi


# Check for input
input_file=`find $INPUT_DIR -type f -name "*.7*" | head -1`

if [[ -f $input_file ]]; then

  # Generate the truncated header from the zip file, or other, and export to csv
  echo -e "${CONTAINER}  Extracting header from $input_file"
  header_file=/tmp/pfile_header
  if echo "$input_file" | grep -q ".7.zip"; then
    echo -e "${CONTAINER}  Found pfile zip archive - unzipping, extracting, and generating truncated header..."
    unzip -p "$input_file" "*/*.7.gz" | head -c 1000000 | gunzip > $header_file 2>/dev/null
  elif echo "$input_file" | grep -q ".7.gz"; then
    echo -e "${CONTAINER}  Found gzipped pfile - extracting and generating truncated header..."
    cat $input_file | head -c 1000000 | gunzip > $header_file 2>/dev/null
  elif echo "$input_file" | grep -q ".7"; then
    echo -e "${CONTAINER}  Found pfile - generating truncated header..."
    cat $input_file | head -c 1000000 > $header_file 2>/dev/null
  fi

  # Generate header file
  /opt/pfile-tools/pfile_tools/scripts/dump_pfile_header.py $header_file --separator=, > $header_file.csv

  if [[ $? != 0 ]]; then
    echo -e "${CONTAINER}  Error generating header!"
    exit 11
  fi

  # Generate the metadata
  echo -e "${CONTAINER}  Generating metadata..."
  PYHONPATH=$PYTHONPATH:/flywheel/v0/ python $FLYWHEEL_BASE/pfile-mr-classifier.py \
    "${input_file}" \
    "${header_file}" \
    "${header_file}.csv" \
    "$OUTPUT_DIR"
    if [[ $? != 0 ]]; then
      echo -e "${CONTAINER}  Error generating metadata!"
      exit 1
    fi
else
  echo -e "${CONTAINER}  No inputs were provided and $INPUT_DIR has no valid input files!"
  exit 1
fi


##############################################################################
# Check for outputs and exit

outputs=`find $OUTPUT_DIR -type f -name ".metadata.json"`

# If outputs exist, then go on...
if [[ -z $outputs ]]
    then
        echo -e "$CONTAINER  No results found in output directory... Exiting"
        exit 1
    else
        chmod -R 777 $OUTPUT_DIR
        echo -e "$CONTAINER  Success!"
fi

exit 0