1
|
#!/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 geoscrub input data into database
|
8
|
# 2. Scrub geoscrub input data with the geonames.sql script
|
9
|
# 3. Scrub geoscrub input data with the geovalidate.sql script
|
10
|
#
|
11
|
# Paul Sarando
|
12
|
# iPlant Collaborative
|
13
|
# Oct 2013
|
14
|
|
15
|
DB_NAME="geoscrub"
|
16
|
DB_USER="bien"
|
17
|
DB_HOST_OPT=""
|
18
|
SCRIPT_DIR="$(dirname $0)"
|
19
|
|
20
|
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
|
echo "Input Data Options:" >&2
|
27
|
echo "-i, --geoscrub-input Geoscrub input directory (default: ${HOME}/geoscrub_input)" >&2
|
28
|
echo " Delete this directory, or the input CSV in it," >&2
|
29
|
echo " to re-download the data." >&2
|
30
|
exit 1;
|
31
|
}
|
32
|
|
33
|
while [[ $# -gt 0 ]]; do
|
34
|
case "$1" in
|
35
|
-\? | --help)
|
36
|
usage
|
37
|
;;
|
38
|
-h)
|
39
|
if [[ -z $2 ]]; then
|
40
|
echo "Option $1 requires an argument." >&2
|
41
|
usage
|
42
|
fi
|
43
|
DB_HOST_OPT="-h $2"
|
44
|
shift 2
|
45
|
;;
|
46
|
--host=*)
|
47
|
DB_HOST_OPT="-h ${1#*=}"
|
48
|
shift
|
49
|
;;
|
50
|
-U)
|
51
|
if [[ -z $2 ]]; then
|
52
|
echo "Option $1 requires an argument." >&2
|
53
|
usage
|
54
|
fi
|
55
|
DB_USER="$2"
|
56
|
shift 2
|
57
|
;;
|
58
|
--username=*)
|
59
|
DB_USER="${1#*=}"
|
60
|
shift
|
61
|
;;
|
62
|
-d)
|
63
|
if [[ -z $2 ]]; then
|
64
|
echo "Option $1 requires an argument." >&2
|
65
|
usage
|
66
|
fi
|
67
|
DB_NAME="$2"
|
68
|
shift 2
|
69
|
;;
|
70
|
--dbname=*)
|
71
|
DB_NAME="${1#*=}"
|
72
|
shift
|
73
|
;;
|
74
|
-i)
|
75
|
if [[ -z $2 ]]; then
|
76
|
echo "Option $1 requires an argument." >&2
|
77
|
usage
|
78
|
fi
|
79
|
GEOSCRUB_INPUT_OPT="-i $2"
|
80
|
shift 2
|
81
|
;;
|
82
|
--geoscrub-input=*)
|
83
|
GEOSCRUB_INPUT_OPT="$1"
|
84
|
shift
|
85
|
;;
|
86
|
*)
|
87
|
echo "Invalid option: $1" >&2
|
88
|
usage
|
89
|
;;
|
90
|
esac
|
91
|
done
|
92
|
|
93
|
function run_sql_script {
|
94
|
local SCRIPT=$1
|
95
|
|
96
|
psql -e -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" --set ON_ERROR_STOP=1 < "$SCRIPT"
|
97
|
if [[ $? != 0 ]]; then
|
98
|
echo "Error while executing SQL script ${SCRIPT}"
|
99
|
exit 1
|
100
|
fi
|
101
|
}
|
102
|
|
103
|
"${SCRIPT_DIR}"/load-geoscrub-input.sh -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" $GEOSCRUB_INPUT_OPT
|
104
|
if [[ $? != 0 ]]; then
|
105
|
echo "Could not load ${DB_NAME} database with geonames.org data."
|
106
|
exit 1
|
107
|
fi
|
108
|
|
109
|
echo "Scrubbing input with geonames data..."
|
110
|
run_sql_script "${SCRIPT_DIR}/geonames.sql"
|
111
|
|
112
|
echo "Scrubbing input with geovalidate data..."
|
113
|
run_sql_script "${SCRIPT_DIR}/geovalidate.sql"
|
114
|
|
115
|
echo "Input successfully scrubbed."
|
116
|
echo "Scrubbed input available in the geoscrub table of the ${DB_NAME} database."
|
117
|
|