data4impact/apps/data4impact-import-scripts/scripts/drug_bank_database/script.sh

53 lines
1.5 KiB
Bash

#!/bin/bash
excelFile="../../orig/drug_bank_database/Publication_citations_in_Drug_Bank_database.xlsx"
workdir=/tmp/drugbank
rm -rf "$workdir" && mkdir "$workdir"
echo
echo "Links from drugbank db Import:"
#--------------------------------
echo " - Generating csv file"
csv="$workdir/drugbank.csv"
xlsx2csv -c UTF-8 "$excelFile" > $csv
#--------------------------------
echo " - Recreating the drugbank database"
dbname=drugbank
dropdb $dbname --if-exists;
createdb $dbname;
psql $dbname -f schema.sql
if [[ -f "$csv" ]]; then
echo " - Importing data: $csv"
psql $dbname -c "COPY data(id,d_b_id,doi,pmcid,pmid,drug_substance,funder,section_of_drug_bank_entry_where_citation_occured) FROM '$csv' CSV HEADER;"
else
echo " - Invalid file: $csv"
fi
echo " - Fix funder names"
psql $dbname -c "UPDATE data SET funder='EC' WHERE funder = 'European Research Council'"
psql $dbname -c "UPDATE data SET funder='Austrian Science Fund FWF' WHERE funder = 'FWF'"
psql $dbname -c "UPDATE data SET funder='Swiss National Science Foundation SNSF' WHERE funder = 'Swiss National Science Foundation'"
#--------------------------------
echo " - Generating json files"
rm -f ../../jsonfiles/drug_bank_database/*.json
psql $dbname -f projects2json.sql | sed 's/\\\\/\\/g' > ../../jsonfiles/drug_bank_database/project.json
psql $dbname -f docOtherId2json.sql | sed 's/\\\\/\\/g' > ../../jsonfiles/drug_bank_database/docotherid.json
psql $dbname -f projDocOtherIds2json.sql | sed 's/\\\\/\\/g' > ../../jsonfiles/drug_bank_database/projectdocotherid.json
echo "Done."
echo