| 1 |
11442
|
psarando
|
#!/bin/bash
|
| 2 |
|
|
|
| 3 |
|
|
# Bash script to download geoscrub_input table dump (created by AMK) from the
|
| 4 |
10707
|
aaronmk
|
# vegbien database, and load it into the geoscrub database (i.e., the
|
| 5 |
|
|
# postgis database prepped with geonames.org data, GADM2 data, and
|
| 6 |
|
|
# associated mapping tables).
|
| 7 |
|
|
#
|
| 8 |
|
|
# Won't be necessary if we end up injecting all of the geoscrubbing and
|
| 9 |
|
|
# geovalidation functionality directly into vegbien itself. And if we
|
| 10 |
|
|
# end up implementing this stuff as a standalone service instead, we'd
|
| 11 |
|
|
# need to rethink (and generalize) how the input data is handled. But
|
| 12 |
|
|
# for now, this should at least serve as a placeholder that could be
|
| 13 |
|
|
# tweaked manually to load any arbitrary geoscrub input data table.
|
| 14 |
|
|
#
|
| 15 |
|
|
# Jim Regetz
|
| 16 |
|
|
# NCEAS
|
| 17 |
|
|
# Created Nov 2012
|
| 18 |
11479
|
psarando
|
#
|
| 19 |
|
|
# Paul Sarando
|
| 20 |
|
|
# iPlant Collaborative
|
| 21 |
|
|
# Updated Oct 2013
|
| 22 |
10707
|
aaronmk
|
|
| 23 |
11442
|
psarando
|
# Note, to force data to download from DATA_URL, ensure the DATAFILE is deleted
|
| 24 |
|
|
# before running this script.
|
| 25 |
11443
|
psarando
|
|
| 26 |
11479
|
psarando
|
DB_NAME="geoscrub"
|
| 27 |
11450
|
psarando
|
DB_USER="bien"
|
| 28 |
11558
|
psarando
|
DB_HOST_OPT=""
|
| 29 |
11493
|
psarando
|
SCRIPT_DIR="$(dirname $0)"
|
| 30 |
11587
|
aaronmk
|
DATA_URL="http://fs.vegpath.org/exports/geoscrub_input.csv"
|
| 31 |
11561
|
psarando
|
DATADIR="${HOME}/geoscrub_input"
|
| 32 |
11591
|
aaronmk
|
: "${DATAFILE:=geoscrub-corpus.csv}"
|
| 33 |
10707
|
aaronmk
|
|
| 34 |
11559
|
psarando
|
function usage {
|
| 35 |
|
|
echo "Usage: $0 [OPTIONS]" >&2
|
| 36 |
|
|
echo "Valid Options:" >&2
|
| 37 |
|
|
echo "-d, --dbname=DBNAME database name psql commands will connect to" >&2
|
| 38 |
|
|
echo "-h, --host=HOSTNAME database server host or socket directory" >&2
|
| 39 |
|
|
echo "-U, --username=USERNAME database user name" >&2
|
| 40 |
11561
|
psarando
|
echo "-i, --geoscrub-input Geoscrub input directory (default: ${HOME}/geoscrub_input)" >&2
|
| 41 |
|
|
echo " Delete this directory, or the input CSV in it," >&2
|
| 42 |
|
|
echo " to re-download the data." >&2
|
| 43 |
11559
|
psarando
|
exit 1;
|
| 44 |
|
|
}
|
| 45 |
|
|
|
| 46 |
|
|
while [[ $# -gt 0 ]]; do
|
| 47 |
|
|
case "$1" in
|
| 48 |
|
|
-\? | --help)
|
| 49 |
|
|
usage
|
| 50 |
|
|
;;
|
| 51 |
|
|
-h)
|
| 52 |
|
|
if [[ -z $2 ]]; then
|
| 53 |
|
|
echo "Option $1 requires an argument." >&2
|
| 54 |
|
|
usage
|
| 55 |
|
|
fi
|
| 56 |
|
|
DB_HOST_OPT="-h $2"
|
| 57 |
|
|
shift 2
|
| 58 |
|
|
;;
|
| 59 |
|
|
--host=*)
|
| 60 |
|
|
DB_HOST_OPT="-h ${1#*=}"
|
| 61 |
|
|
shift
|
| 62 |
|
|
;;
|
| 63 |
|
|
-U)
|
| 64 |
|
|
if [[ -z $2 ]]; then
|
| 65 |
|
|
echo "Option $1 requires an argument." >&2
|
| 66 |
|
|
usage
|
| 67 |
|
|
fi
|
| 68 |
|
|
DB_USER="$2"
|
| 69 |
|
|
shift 2
|
| 70 |
|
|
;;
|
| 71 |
|
|
--username=*)
|
| 72 |
|
|
DB_USER="${1#*=}"
|
| 73 |
|
|
shift
|
| 74 |
|
|
;;
|
| 75 |
|
|
-d)
|
| 76 |
|
|
if [[ -z $2 ]]; then
|
| 77 |
|
|
echo "Option $1 requires an argument." >&2
|
| 78 |
|
|
usage
|
| 79 |
|
|
fi
|
| 80 |
|
|
DB_NAME="$2"
|
| 81 |
|
|
shift 2
|
| 82 |
|
|
;;
|
| 83 |
|
|
--dbname=*)
|
| 84 |
|
|
DB_NAME="${1#*=}"
|
| 85 |
|
|
shift
|
| 86 |
|
|
;;
|
| 87 |
11561
|
psarando
|
-i)
|
| 88 |
|
|
if [[ -z $2 ]]; then
|
| 89 |
|
|
echo "Option $1 requires an argument." >&2
|
| 90 |
|
|
usage
|
| 91 |
|
|
fi
|
| 92 |
|
|
DATADIR="$2"
|
| 93 |
|
|
shift 2
|
| 94 |
|
|
;;
|
| 95 |
|
|
--geoscrub-input=*)
|
| 96 |
|
|
DATADIR="${1#*=}"
|
| 97 |
|
|
shift
|
| 98 |
|
|
;;
|
| 99 |
11559
|
psarando
|
*)
|
| 100 |
|
|
echo "Invalid option: $1" >&2
|
| 101 |
|
|
usage
|
| 102 |
|
|
;;
|
| 103 |
|
|
esac
|
| 104 |
|
|
done
|
| 105 |
|
|
|
| 106 |
11442
|
psarando
|
if [[ ! -d "$DATADIR" ]]; then
|
| 107 |
|
|
echo "making directory ${DATADIR}"
|
| 108 |
|
|
mkdir -p "$DATADIR"
|
| 109 |
|
|
|
| 110 |
|
|
if [[ $? != 0 ]]; then
|
| 111 |
|
|
echo "Could not create directory ${DATADIR}"
|
| 112 |
|
|
exit 1
|
| 113 |
|
|
fi
|
| 114 |
|
|
fi
|
| 115 |
|
|
|
| 116 |
11561
|
psarando
|
if [[ ! -r "${DATADIR}/${DATAFILE}" ]]; then
|
| 117 |
11442
|
psarando
|
# download distinct records from vegbien
|
| 118 |
11561
|
psarando
|
wget -O "${DATADIR}/${DATAFILE}" "$DATA_URL"
|
| 119 |
11442
|
psarando
|
|
| 120 |
|
|
if [[ $? != 0 ]]; then
|
| 121 |
11561
|
psarando
|
echo "Could not download input to ${DATADIR}/${DATAFILE}"
|
| 122 |
11442
|
psarando
|
exit 1
|
| 123 |
|
|
fi
|
| 124 |
|
|
fi
|
| 125 |
|
|
|
| 126 |
11561
|
psarando
|
echo "Loading vegbien data from ${DATADIR}/${DATAFILE}"
|
| 127 |
10707
|
aaronmk
|
|
| 128 |
11443
|
psarando
|
# clear previous data
|
| 129 |
11558
|
psarando
|
psql -e -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" --set ON_ERROR_STOP=1 < "${SCRIPT_DIR}/truncate.vegbien_geoscrub.sql"
|
| 130 |
11443
|
psarando
|
if [[ $? != 0 ]]; then
|
| 131 |
|
|
echo "Could not clear data from vegbien_geoscrub tables."
|
| 132 |
|
|
exit 1
|
| 133 |
|
|
fi
|
| 134 |
|
|
|
| 135 |
11493
|
psarando
|
# load vegbien_geoscrub table with input data
|
| 136 |
11561
|
psarando
|
echo "Copying vegbien_geoscrub from ${DATADIR}/${DATAFILE}"
|
| 137 |
11586
|
aaronmk
|
psql -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" -c "\COPY vegbien_geoscrub FROM '${DATADIR}/${DATAFILE}' WITH CSV HEADER"
|
