Files
duckdb/scripts/run.sh
2026-02-02 22:37:55 +01:00

130 lines
3.5 KiB
Bash

#!/bin/bash
set -e
DUCKDB_PORT="${DUCKDB_PORT:-8432}"
DUCKDB_DATABASE="${DUCKDB_DATABASE:-:memory:}"
DUCKDB_DATA_DIR="${DUCKDB_DATA_DIR:-/data/duckdb}"
DUCKDB_READ_ONLY="${DUCKDB_READ_ONLY:-false}"
# Create query handler script
create_handler() {
cat > /tmp/duckdb_handler.sh << 'HANDLER'
#!/bin/bash
DUCKDB_DATABASE="${DUCKDB_DATABASE:-:memory:}"
DUCKDB_READ_ONLY="${DUCKDB_READ_ONLY:-false}"
# Read HTTP request
read -r request_line
method=$(echo "$request_line" | cut -d' ' -f1)
path=$(echo "$request_line" | cut -d' ' -f2)
# Read headers
content_length=0
while read -r header; do
header=$(echo "$header" | tr -d '\r')
[ -z "$header" ] && break
if [[ "$header" =~ ^[Cc]ontent-[Ll]ength:\ *([0-9]+) ]]; then
content_length="${BASH_REMATCH[1]}"
fi
done
# Read body
body=""
if [ "$content_length" -gt 0 ]; then
body=$(head -c "$content_length")
fi
# Route request
send_response() {
local status="$1"
local content_type="$2"
local body="$3"
local body_length=${#body}
printf "HTTP/1.1 %s\r\n" "$status"
printf "Content-Type: %s\r\n" "$content_type"
printf "Content-Length: %d\r\n" "$body_length"
printf "Connection: close\r\n"
printf "\r\n"
printf "%s" "$body"
}
# Health check
if [ "$path" = "/health" ]; then
send_response "200 OK" "application/json" '{"status":"ok"}'
exit 0
fi
# Query endpoint
if [ "$path" = "/query" ] && [ "$method" = "POST" ]; then
# Extract SQL from JSON body
sql=$(echo "$body" | jq -r '.sql // empty')
if [ -z "$sql" ]; then
send_response "400 Bad Request" "application/json" '{"error":"Missing sql field"}'
exit 0
fi
# Build duckdb command
duckdb_args=()
if [ "$DUCKDB_READ_ONLY" = "true" ]; then
duckdb_args+=("-readonly")
fi
duckdb_args+=("-json")
duckdb_args+=("$DUCKDB_DATABASE")
# Execute query
start_time=$(date +%s%3N)
result=$(echo "$sql" | duckdb "${duckdb_args[@]}" 2>&1) || {
error_msg=$(echo "$result" | jq -Rs '.')
send_response "400 Bad Request" "application/json" "{\"success\":false,\"error\":$error_msg}"
exit 0
}
end_time=$(date +%s%3N)
time_ms=$((end_time - start_time))
# Parse result
if [ -z "$result" ] || [ "$result" = "[]" ]; then
send_response "200 OK" "application/json" "{\"success\":true,\"rows\":[],\"row_count\":0,\"time_ms\":$time_ms}"
else
row_count=$(echo "$result" | jq 'length')
columns=$(echo "$result" | jq -c '.[0] | keys')
rows=$(echo "$result" | jq -c '[.[] | [.[]]]')
response=$(jq -n \
--argjson columns "$columns" \
--argjson rows "$rows" \
--argjson row_count "$row_count" \
--argjson time_ms "$time_ms" \
'{success:true, columns:$columns, rows:$rows, row_count:$row_count, time_ms:$time_ms}')
send_response "200 OK" "application/json" "$response"
fi
exit 0
fi
# Not found
send_response "404 Not Found" "application/json" '{"error":"Not found. Use POST /query"}'
HANDLER
chmod +x /tmp/duckdb_handler.sh
# Export env vars for handler
export DUCKDB_DATABASE
export DUCKDB_READ_ONLY
}
# Serve HTTP API
serve_api() {
echo "Starting DuckDB HTTP API on port $DUCKDB_PORT..."
echo "Database: $DUCKDB_DATABASE"
echo "Data directory: $DUCKDB_DATA_DIR"
[ "$DUCKDB_READ_ONLY" = "true" ] && echo "Mode: read-only"
exec socat TCP-LISTEN:$DUCKDB_PORT,reuseaddr,fork EXEC:"/tmp/duckdb_handler.sh"
}
create_handler
serve_api