#!/usr/bin/bash

#

# Bash script used to download GHCN data and generate a simple log file

# recording basic info about what was obtained. The downloaded data and

# log file are stored in the directory specified by $DESTDIR below. Note

# that the subdirectory name is hardcoded to reflect the GHCN version in

# effect at the time this script was written and executed; it would

# need be modified if this script is used again in the future, either

# manually in the code or perhaps by parsing the ghcnd-version.txt file

# (if it can be assumed that the structure of this file is consistent

# over time).

#

# Written and executed by Jim Regetz on 29-May-2012, with post-hoc

# tweaking of filesystem metadata to preserve everything as owned by

# 'layers' and read-only to all project collaborators.

FTPDIR="ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily"

BASEDIR="/home/layers/data/climate/ghcn"

DESTDIR="$BASEDIR/v2.92-upd-2012052822"

# create destination for data

if [ -d $DESTDIR ]

then

    echo "target directory $DESTDIR already exists"

    exit 1

fi

mkdir $DESTDIR

# set up log file

LOGFILE="$DESTDIR/download-ghcn-daily.log"

if [ -f $LOGFILE ]

then

    echo "first remove existing log file '$LOGFILE'"

    exit 1

fi

echo "==========================================

 Download log for GHCN daily data

 Timestamp: `date`

 User: $USER

==========================================" >>$LOGFILE

echo -e "\n" >>$LOGFILE

# log snapshot of current FTP directory listing

echo "Remote directory listing:" >>$LOGFILE

curl "$FTPDIR/" >>$LOGFILE 2>/dev/null

echo -e "\n" >>$LOGFILE

# files we want to get...

FILES=(

  "readme.txt"

  "status.txt"

  "ghcnd-countries.txt"

  "ghcnd-inventory.txt"

  "ghcnd-states.txt"

  "ghcnd-stations.txt"

  "ghcnd-version.txt"

  "ghcnd_all.tar.gz"

)

# download each file and record some info to log

for file in ${FILES[@]}

do

    TARGET=$DESTDIR/$file

    if [ -f $TARGET ]

    then

        echo "$TARGET already exists -- skipping"

    else

        printf '%72s\n' | tr ' ' '-' >>$LOGFILE

        curl -o $TARGET "$FTPDIR/$file" >>$LOGFILE \

            -w "URL: %{url_effective}\nSize: %{size_download}\nTime: %{time_total}\n"

        echo "Location: $TARGET" >>$LOGFILE

        echo "MD5: `md5sum $TARGET | sed 's/ .*//'`" >>$LOGFILE

        echo -e "\n" >>$LOGFILE

    fi

done