#!/bin/bash

# Bash script to download geoscrub_input table dump (created by AMK) from the

# vegbien database, and load it into the geoscrub database (i.e., the

# postgis database prepped with geonames.org data, GADM2 data, and

# associated mapping tables).

#

# Won't be necessary if we end up injecting all of the geoscrubbing and

# geovalidation functionality directly into vegbien itself. And if we

# end up implementing this stuff as a standalone service instead, we'd

# need to rethink (and generalize) how the input data is handled. But

# for now, this should at least serve as a placeholder that could be

# tweaked manually to load any arbitrary geoscrub input data table.

#

# Jim Regetz

# NCEAS

# Created Nov 2012

# Note, to force data to download from DATA_URL, ensure the DATAFILE is deleted

# before running this script.

DATA_URL="http://fs.vegpath.org/exports/geoscrub_input.no_header.cols=country,stateProvince,county,decimalLatitude,decimalLongitude.csv"

DATADIR="$(dirname $0)/input"

DATAFILE="${DATADIR}/geoscrub-corpus.csv"

if [[ ! -d "$DATADIR" ]]; then

    echo "making directory ${DATADIR}"

    mkdir -p "$DATADIR"

    if [[ $? != 0 ]]; then

        echo "Could not create directory ${DATADIR}"

        exit 1

    fi

fi

if [[ ! -r "$DATAFILE" ]]; then

    # download distinct records from vegbien

    wget -O "$DATAFILE" "$DATA_URL"

    if [[ $? != 0 ]]; then

        echo "Could not download input to ${DATAFILE}"

        exit 1

    fi

fi

# generate table

psql -c \

   'CREATE TABLE vegbien_geoscrub (

        country text,

        stateProvince text,

        county text,

        decimalLatitude double precision,

        decimalLongitude double precision

    )' geoscrub

# load

psql -c "COPY vegbien_geoscrub FROM '${DATAFILE}' WITH CSV" geoscrub