| 1 |
11496
|
psarando
|
#!/bin/bash
|
| 2 |
|
|
|
| 3 |
|
|
# Bash script to create a new postgis database and prep it with GADM
|
| 4 |
|
|
# data for geovalidation purposes.
|
| 5 |
|
|
#
|
| 6 |
|
|
# Basic workflow:
|
| 7 |
|
|
# 1. Load gadm2 data into database
|
| 8 |
|
|
# 2. Load geonames data into database
|
| 9 |
|
|
# 3. Update geonames to gadm2 data tables
|
| 10 |
|
|
#
|
| 11 |
|
|
# Paul Sarando
|
| 12 |
|
|
# iPlant Collaborative
|
| 13 |
|
|
# Oct 2013
|
| 14 |
|
|
|
| 15 |
|
|
DB_NAME="geoscrub"
|
| 16 |
|
|
DB_USER="bien"
|
| 17 |
11558
|
psarando
|
DB_HOST_OPT=""
|
| 18 |
11496
|
psarando
|
SCRIPT_DIR="$(dirname $0)"
|
| 19 |
|
|
|
| 20 |
11559
|
psarando
|
function usage {
|
| 21 |
|
|
echo "Usage: $0 [OPTIONS]" >&2
|
| 22 |
|
|
echo "Valid Options:" >&2
|
| 23 |
|
|
echo "-d, --dbname=DBNAME database name psql commands will connect to" >&2
|
| 24 |
|
|
echo "-h, --host=HOSTNAME database server host or socket directory" >&2
|
| 25 |
|
|
echo "-U, --username=USERNAME database user name" >&2
|
| 26 |
11561
|
psarando
|
echo ""
|
| 27 |
11560
|
psarando
|
echo "Update Options:" >&2
|
| 28 |
|
|
echo "-G, --gadm-only update only GADM data" >&2
|
| 29 |
|
|
echo "-N, --geonames-only update only geonames.org data" >&2
|
| 30 |
|
|
echo "-M, --geonames-to-gadm-only update niether GADM nor geonames.org data," >&2
|
| 31 |
|
|
echo " but geonames-to-gadm mappings are always updated." >&2
|
| 32 |
11561
|
psarando
|
echo ""
|
| 33 |
|
|
echo "Input Data Options:" >&2
|
| 34 |
|
|
echo "-g, --gadm-data GADM data directory (default: ${HOME}/gadm_v2_shp)" >&2
|
| 35 |
|
|
echo " Delete this directory, or the GADM data in it," >&2
|
| 36 |
|
|
echo " to re-download the data." >&2
|
| 37 |
|
|
echo "-m, --geonames-data Geonames.org data directory (default: ${HOME}/geonames)" >&2
|
| 38 |
|
|
echo " Delete this directory, or the geonames.org data in it," >&2
|
| 39 |
|
|
echo " to re-download the data." >&2
|
| 40 |
11559
|
psarando
|
exit 1;
|
| 41 |
|
|
}
|
| 42 |
|
|
|
| 43 |
|
|
while [[ $# -gt 0 ]]; do
|
| 44 |
|
|
case "$1" in
|
| 45 |
|
|
-\? | --help)
|
| 46 |
|
|
usage
|
| 47 |
|
|
;;
|
| 48 |
|
|
-h)
|
| 49 |
|
|
if [[ -z $2 ]]; then
|
| 50 |
|
|
echo "Option $1 requires an argument." >&2
|
| 51 |
|
|
usage
|
| 52 |
|
|
fi
|
| 53 |
|
|
DB_HOST_OPT="-h $2"
|
| 54 |
|
|
shift 2
|
| 55 |
|
|
;;
|
| 56 |
|
|
--host=*)
|
| 57 |
|
|
DB_HOST_OPT="-h ${1#*=}"
|
| 58 |
|
|
shift
|
| 59 |
|
|
;;
|
| 60 |
|
|
-U)
|
| 61 |
|
|
if [[ -z $2 ]]; then
|
| 62 |
|
|
echo "Option $1 requires an argument." >&2
|
| 63 |
|
|
usage
|
| 64 |
|
|
fi
|
| 65 |
|
|
DB_USER="$2"
|
| 66 |
|
|
shift 2
|
| 67 |
|
|
;;
|
| 68 |
|
|
--username=*)
|
| 69 |
|
|
DB_USER="${1#*=}"
|
| 70 |
|
|
shift
|
| 71 |
|
|
;;
|
| 72 |
|
|
-d)
|
| 73 |
|
|
if [[ -z $2 ]]; then
|
| 74 |
|
|
echo "Option $1 requires an argument." >&2
|
| 75 |
|
|
usage
|
| 76 |
|
|
fi
|
| 77 |
|
|
DB_NAME="$2"
|
| 78 |
|
|
shift 2
|
| 79 |
|
|
;;
|
| 80 |
|
|
--dbname=*)
|
| 81 |
|
|
DB_NAME="${1#*=}"
|
| 82 |
|
|
shift
|
| 83 |
|
|
;;
|
| 84 |
11560
|
psarando
|
-G | --gadm-only)
|
| 85 |
|
|
NO_GEONAMES=TRUE
|
| 86 |
|
|
shift
|
| 87 |
|
|
;;
|
| 88 |
|
|
-N | --geonames-only)
|
| 89 |
|
|
NO_GADM=TRUE
|
| 90 |
|
|
shift
|
| 91 |
|
|
;;
|
| 92 |
|
|
-M | --geonames-to-gadm-only)
|
| 93 |
|
|
NO_GADM=TRUE
|
| 94 |
|
|
NO_GEONAMES=TRUE
|
| 95 |
|
|
shift
|
| 96 |
|
|
;;
|
| 97 |
11561
|
psarando
|
-g)
|
| 98 |
|
|
if [[ -z $2 ]]; then
|
| 99 |
|
|
echo "Option $1 requires an argument." >&2
|
| 100 |
|
|
usage
|
| 101 |
|
|
fi
|
| 102 |
|
|
GADM_DATA_DIR_OPT="-g $2"
|
| 103 |
|
|
shift 2
|
| 104 |
|
|
;;
|
| 105 |
|
|
--gadm-data=*)
|
| 106 |
|
|
GADM_DATA_DIR_OPT="$1"
|
| 107 |
|
|
shift
|
| 108 |
|
|
;;
|
| 109 |
|
|
-m)
|
| 110 |
|
|
if [[ -z $2 ]]; then
|
| 111 |
|
|
echo "Option $1 requires an argument." >&2
|
| 112 |
|
|
usage
|
| 113 |
|
|
fi
|
| 114 |
|
|
GEONAMES_DATA_DIR_OPT="-m $2"
|
| 115 |
|
|
shift 2
|
| 116 |
|
|
;;
|
| 117 |
|
|
--geonames-data=*)
|
| 118 |
|
|
GEONAMES_DATA_DIR_OPT="$1"
|
| 119 |
|
|
shift
|
| 120 |
|
|
;;
|
| 121 |
11559
|
psarando
|
*)
|
| 122 |
|
|
echo "Invalid option: $1" >&2
|
| 123 |
|
|
usage
|
| 124 |
|
|
;;
|
| 125 |
|
|
esac
|
| 126 |
|
|
done
|
| 127 |
|
|
|
| 128 |
11496
|
psarando
|
function run_sql_script {
|
| 129 |
|
|
local SCRIPT=$1
|
| 130 |
|
|
|
| 131 |
11558
|
psarando
|
psql -e -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" --set ON_ERROR_STOP=1 < "$SCRIPT"
|
| 132 |
11496
|
psarando
|
if [[ $? != 0 ]]; then
|
| 133 |
|
|
echo "Error while executing SQL script ${SCRIPT}"
|
| 134 |
|
|
exit 1
|
| 135 |
|
|
fi
|
| 136 |
|
|
}
|
| 137 |
|
|
|
| 138 |
|
|
echo "Updating geoscrub validation data."
|
| 139 |
|
|
|
| 140 |
11560
|
psarando
|
if [[ -z $NO_GADM ]]; then
|
| 141 |
11561
|
psarando
|
"${SCRIPT_DIR}"/update_gadm_data.sh -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" $GADM_DATA_DIR_OPT
|
| 142 |
11560
|
psarando
|
if [[ $? != 0 ]]; then
|
| 143 |
|
|
echo "Could not update ${DB_NAME} database with GADM data."
|
| 144 |
|
|
exit 1
|
| 145 |
|
|
fi
|
| 146 |
11496
|
psarando
|
fi
|
| 147 |
|
|
|
| 148 |
11560
|
psarando
|
if [[ -z $NO_GEONAMES ]]; then
|
| 149 |
11561
|
psarando
|
"${SCRIPT_DIR}"/update_geonames_data.sh -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" $GEONAMES_DATA_DIR_OPT
|
| 150 |
11560
|
psarando
|
if [[ $? != 0 ]]; then
|
| 151 |
|
|
echo "Could not update ${DB_NAME} database with geonames.org data."
|
| 152 |
|
|
exit 1
|
| 153 |
|
|
fi
|
| 154 |
11496
|
psarando
|
fi
|
| 155 |
|
|
|
| 156 |
|
|
run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.country.sql"
|
| 157 |
|
|
run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.stateprovince.sql"
|
| 158 |
|
|
run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.county.sql"
|
| 159 |
|
|
|
| 160 |
|
|
echo "Validation data successfully updated."
|
