Final code for MUCN EBSCO upload.

[sitka/sitka-tools.git] / marc_export_custom / mucn-ebsco-full-export-parallel.sh

#!/bin/bash

DIR=$(readlink -f $(dirname $0))
#cd /srv/openils/bin

PGUSER=evergreen
PGHOST=db1.sitka.bclibraries.ca
PSQL="psql -A -t -U $PGUSER -h $PGHOST"
DATABASE=evergreen
MUCN_QUERY="select distinct record FROM asset.call_number where owning_lib in (select distinct id from actor.org_unit_descendants(1400)) and not deleted and label != '##URI##';"
MARC_PARAMS="--config /srv/openils/conf/opensrf_core.xml --items --location MUCN --collapse_to_depth 2 --timeout 300 --force901 --encoding UTF-8 --onlyholdings --quiet"
MARC_EXPORT_BIN="${DIR}/marc_export_custom"
MARC_EXPORT="${MARC_EXPORT_BIN} ${MARC_PARAMS}"

[ -z "${T}" ] && T=$(date -u +%Y%m%dT%H%M%SZ)
[ -z "${OUTDIR}" ] && OUTDIR=$(mktemp --tmpdir -d outlook_export.XXXXXXX)

process_batch() {
	local f="$1" sql="$2" split_count="$3"
	if [ ! -f ${OUTDIR}/${f}-${T}.id ]; then
		echo "Getting IDs"
		$PSQL -c "$sql" $DATABASE | sort | uniq >${OUTDIR}/${f}-${T}.id
	fi
	if [ ! -f ${OUTDIR}/${f}.id.targets ]; then
		echo "Splitting work"
		split_suffix ${OUTDIR}/${f}-${T}.id ${OUTDIR}/${f}-${T}-split. .id $split_count
		echo "Making targets"
		find ${OUTDIR} -name "${f}-${T}-split.[0-9]*[0-9].id" \
		| sed 's,.id$,.marc,g' \
		>${OUTDIR}/${f}.id.targets
	fi

	make ${MAKEOPTS} -f Makefile-marc_export \
	MARC_EXPORT_BIN="${MARC_EXPORT_BIN}" \
	MARC_PARAMS="$MARC_PARAMS --exclusion_ini ${f}.ini" \
	$(<${OUTDIR}/${f}.id.targets)

	find ${OUTDIR} -name "${f}-${T}-split*.marc" \
	-exec cat \{} \; \
	>${OUTDIR}/${f}-${T}.marc
}

split_suffix() {
	local input="$1" output_prefix="$2" output_suffix="$3" split_count="$4"
	split -a 7 -d -C ${split_count} ${input} ${output_prefix}
	find $(dirname ${output_prefix}) -name "$(basename ${output_prefix})*" -exec mv -f \{\} \{\}${output_suffix} \;
}

# media for BC ELN post-secondary libraries
export SPLIT_COUNT=10000
F=mucn_ebsco SQL="$MUCN_QUERY"
process_batch "$F" "${SQL}" $SPLIT_COUNT