local_build_db_all

#!/bin/bash

# Exit script on any error
set -e

# Check for required arguments
if [[ -z $1 || -z $2 ]]
then 
	echo "Please run with via the following: $0 agr.local.alliancemine.postgres.server postgres"
	exit 1
fi

echo "Starting Alliancemine DB Build: Host: $1 User: $2"
date

echo "Checking if the database 'alliancemine' exists and dropping it if it does..."
psql -U $2 -h $1 -c "DROP DATABASE IF EXISTS alliancemine;"

echo "Checking if the database 'items-alliancemine' exists and dropping it if it does..."
psql -U $2 -h $1 -c 'DROP DATABASE IF EXISTS "items-alliancemine";'

# Database and integration sources setup
fasta_sources=(alliance-wb-fasta alliance-fb-fasta alliance-zfin-fasta alliance-mgi-fasta alliance-rgd-fasta)
ontology_sources=(do go eco mmo emapa zfa wbbt fbbt so psi-mi-ontology go-slim)
sgd_sources=(sgd sgd-gff sgd-gff-utr sgd-db-utr fungi-homologs cgob-homologs cglabrata-homologs pombe-homologs homolog-genes sgd-complementation-db sgd-complexes sgd-protein-properties sgd-protein-ntermini)
alliance_sources=(alliance-genes go-annotation alliance-disease alliance-orthologs alliance-alleles alliance-expression entrez-organism)
postprocess_sources=(create-references do-sources transfer-sequences create-attribute-indexes summarise-objectstore)
solr_indexes=(create-autocomplete-index create-search-index)

# Create databases
echo createdb -U $2 -h $1 alliancemine || true
createdb -U $2 -h $1 alliancemine || true
echo createdb -U $2 -h $1 items-alliancemine || true
createdb -U $2 -h $1 items-alliancemine || true

# Pull Secrets file
echo "Pulling secrets file from AWS."
aws secretsmanager get-secret-value --region=us-east-1 --secret-id IntermineLocalPropertiesFile --query SecretString --output text > alliancemine.properties

# Build database
echo "./gradlew --stacktrace --no-daemon builddb"
./gradlew --no-daemon --stacktrace  builddb
date

# Integrate various sources
for i in "${fasta_sources[@]}"
do
	echo "./gradlew --stacktrace --no-daemon integrate -Psource=$i"
	./gradlew --no-daemon --stacktrace integrate -Psource=$i
	date
done

for i in "${ontology_sources[@]}"
do
	echo "./gradlew --stacktrace --no-daemon integrate -Psource=$i"
	./gradlew --no-daemon --stacktrace integrate -Psource=$i
	date
done

for i in "${sgd_sources[@]}"
do
	echo "./gradlew --no-daemon integrate -Psource=$i"
	./gradlew --no-daemon --stacktrace integrate -Psource=$i
	date
done

for i in "${alliance_sources[@]}"
do
	echo "./gradlew --stacktrace --no-daemon integrate -Psource=$i"
	./gradlew --no-daemon --stacktrace integrate -Psource=$i
	date
done

for i in "${postprocess_sources[@]}"
do
	echo "./gradlew --stacktrace --no-daemon postprocess -Pprocess=$i"
	./gradlew --no-daemon --stacktrace postprocess -Pprocess=$i
	date
done

# Dump the database
echo pg_dump -F c -U $2 -h $1 -f /root/data/alliancemine.dump alliancemine
pg_dump -F c -U $2 -h $1 -f /root/data/alliancemine.dump alliancemine
date

# Backup DB File to S3
backup_file_name=$(date "+%Y-%m-%d_%H:%M:%S").dump
echo "Backup DB File to S3: aws s3 cp /root/data/alliancemine.dump s3://agr-db-backups/alliancemine/local/$backup_file_name"
aws s3 cp /root/data/alliancemine.dump s3://agr-db-backups/alliancemine/local/$backup_file_name

# Keep only the last 5 backups
echo "Checking and maintaining only the last 5 backups..."
backup_files=$(aws s3 ls s3://agr-db-backups/alliancemine/local/ | awk '{print $4}' | sort)
backup_count=$(echo "$backup_files" | wc -l)

if [ "$backup_count" -gt 5 ]; then
    echo "More than 5 backups exist. Removing the oldest ones..."
    backups_to_remove=$(echo "$backup_files" | head -n -$((5)))
    for file in $backups_to_remove; do
        aws s3 rm s3://agr-db-backups/alliancemine/local/$file
        echo "Removed $file"
    done
fi

echo "Finished Alliancemine DB Build"
date