#!/bin/bash

# Bash script to create a new postgis database and prep it with GADM

# data for geovalidation purposes.

#

# Basic workflow:

#  1. Load gadm2 data into database

#  2. Load geonames data into database

#  3. Update geonames to gadm2 data tables

# 

# Paul Sarando

# iPlant Collaborative

# Oct 2013

DB_NAME="geoscrub"

DB_USER="bien"

DB_HOST_OPT=""

SCRIPT_DIR="$(dirname $0)"

function usage {

    echo "Usage: $0 [OPTIONS]" >&2

    echo "Valid Options:" >&2

    echo "-d, --dbname=DBNAME      database name psql commands will connect to" >&2

    echo "-h, --host=HOSTNAME      database server host or socket directory" >&2

    echo "-U, --username=USERNAME  database user name" >&2

    echo ""

    echo "Update Options:" >&2

    echo "-G, --gadm-only             update only GADM data" >&2

    echo "-N, --geonames-only         update only geonames.org data" >&2

    echo "-M, --geonames-to-gadm-only update niether GADM nor geonames.org data," >&2

    echo "                            but geonames-to-gadm mappings are always updated." >&2

    echo ""

    echo "Input Data Options:" >&2

    echo "-g, --gadm-data          GADM data directory (default: ${HOME}/gadm_v2_shp)" >&2

    echo "                         Delete this directory, or the GADM data in it," >&2

    echo "                         to re-download the data." >&2

    echo "-m, --geonames-data      Geonames.org data directory (default: ${HOME}/geonames)" >&2

    echo "                         Delete this directory, or the geonames.org data in it," >&2

    echo "                         to re-download the data." >&2

    exit 1;

}

while [[ $# -gt 0  ]]; do

    case "$1" in

        -\? | --help)

            usage

            ;;

        -h)

            if [[ -z $2  ]];  then

                echo "Option $1 requires an argument." >&2

                usage

            fi

            DB_HOST_OPT="-h $2"

            shift 2

            ;;

        --host=*)

            DB_HOST_OPT="-h ${1#*=}"

            shift

            ;;

        -U)

            if [[ -z $2  ]];  then

                echo "Option $1 requires an argument." >&2

                usage

            fi

            DB_USER="$2"

            shift 2

            ;;

        --username=*)

            DB_USER="${1#*=}"

            shift

            ;;

        -d)

            if [[ -z $2  ]];  then

                echo "Option $1 requires an argument." >&2

                usage

            fi

            DB_NAME="$2"

            shift 2

            ;;

        --dbname=*)

            DB_NAME="${1#*=}"

            shift

            ;;

        -G | --gadm-only)

            NO_GEONAMES=TRUE

            shift

            ;;

        -N | --geonames-only)

            NO_GADM=TRUE

            shift

            ;;

        -M | --geonames-to-gadm-only)

            NO_GADM=TRUE

            NO_GEONAMES=TRUE

            shift

            ;;

        -g)

            if [[ -z $2  ]];  then

                echo "Option $1 requires an argument." >&2

                usage

            fi

            GADM_DATA_DIR_OPT="-g $2"

            shift 2

            ;;

        --gadm-data=*)

            GADM_DATA_DIR_OPT="$1"

            shift

            ;;

        -m)

            if [[ -z $2  ]];  then

                echo "Option $1 requires an argument." >&2

                usage

            fi

            GEONAMES_DATA_DIR_OPT="-m $2"

            shift 2

            ;;

        --geonames-data=*)

            GEONAMES_DATA_DIR_OPT="$1"

            shift

            ;;

        *)

            echo "Invalid option: $1" >&2

            usage

            ;;

    esac

done

function run_sql_script {

    local SCRIPT=$1

    psql -e -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" --set ON_ERROR_STOP=1 < "$SCRIPT"

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

        echo "Error while executing SQL script ${SCRIPT}"

        exit 1

    fi

}

echo "Updating geoscrub validation data."

if [[ -z $NO_GADM ]]; then

    "${SCRIPT_DIR}"/update_gadm_data.sh -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" $GADM_DATA_DIR_OPT

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

        echo "Could not update ${DB_NAME} database with GADM data."

        exit 1

    fi

fi

if [[ -z $NO_GEONAMES ]]; then

    "${SCRIPT_DIR}"/update_geonames_data.sh -U "$DB_USER" $DB_HOST_OPT -d "$DB_NAME" $GEONAMES_DATA_DIR_OPT

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

        echo "Could not update ${DB_NAME} database with geonames.org data."

        exit 1

    fi

fi

run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.country.sql"

run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.stateprovince.sql"

run_sql_script "${SCRIPT_DIR}/geonames-to-gadm.county.sql"

echo "Validation data successfully updated."