| 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 |
11479
|
psarando
|
DB_HOST="localhost"
|
| 29 |
11493
|
psarando
|
SCRIPT_DIR="$(dirname $0)"
|
| 30 |
11442
|
psarando
|
DATA_URL="http://fs.vegpath.org/exports/geoscrub_input.no_header.cols=country,stateProvince,county,decimalLatitude,decimalLongitude.csv"
|
| 31 |
11443
|
psarando
|
DATADIR="${SCRIPT_DIR}/input"
|
| 32 |
11442
|
psarando
|
DATAFILE="${DATADIR}/geoscrub-corpus.csv"
|
| 33 |
10707
|
aaronmk
|
|
| 34 |
11442
|
psarando
|
if [[ ! -d "$DATADIR" ]]; then
|
| 35 |
|
|
echo "making directory ${DATADIR}"
|
| 36 |
|
|
mkdir -p "$DATADIR"
|
| 37 |
|
|
|
| 38 |
|
|
if [[ $? != 0 ]]; then
|
| 39 |
|
|
echo "Could not create directory ${DATADIR}"
|
| 40 |
|
|
exit 1
|
| 41 |
|
|
fi
|
| 42 |
|
|
fi
|
| 43 |
|
|
|
| 44 |
|
|
if [[ ! -r "$DATAFILE" ]]; then
|
| 45 |
|
|
# download distinct records from vegbien
|
| 46 |
|
|
wget -O "$DATAFILE" "$DATA_URL"
|
| 47 |
|
|
|
| 48 |
|
|
if [[ $? != 0 ]]; then
|
| 49 |
|
|
echo "Could not download input to ${DATAFILE}"
|
| 50 |
|
|
exit 1
|
| 51 |
|
|
fi
|
| 52 |
|
|
fi
|
| 53 |
|
|
|
| 54 |
11443
|
psarando
|
echo "Loading vegbien data from ${DATAFILE}"
|
| 55 |
10707
|
aaronmk
|
|
| 56 |
11443
|
psarando
|
# clear previous data
|
| 57 |
11479
|
psarando
|
psql -e -U "$DB_USER" -h "$DB_HOST" -d "$DB_NAME" --set ON_ERROR_STOP=1 < "${SCRIPT_DIR}/truncate.vegbien_geoscrub.sql"
|
| 58 |
11443
|
psarando
|
if [[ $? != 0 ]]; then
|
| 59 |
|
|
echo "Could not clear data from vegbien_geoscrub tables."
|
| 60 |
|
|
exit 1
|
| 61 |
|
|
fi
|
| 62 |
|
|
|
| 63 |
11493
|
psarando
|
# load vegbien_geoscrub table with input data
|
| 64 |
11479
|
psarando
|
psql -U "$DB_USER" -h "$DB_HOST" -d "$DB_NAME" -c "\COPY vegbien_geoscrub FROM '${DATAFILE}' WITH CSV"
|
