#!/bin/bash # --- Settings export PYTHONPATH=/ export Python=python3.10 export Config=http://cloud16.cvtt.vpn:6789/apps/minimal_md export PyScript=/cvttpy/exchanges/alpaca/hist_md/hist_md_bars.py export OutputDir=/app/data/alpaca_md # Local export LogDir=/logs mkdir -p ${OutputDir} mkdir -p ${LogDir} usage() { echo "Usage: $0 [ (30}) ] [] " exit 1 } echo "CommandLine: ${*}" Start=${1} NumJobs=${2} # InstListFile=${3} export CalendarURL=http://cloud16.cvtt.vpn:8000/api/v1/markets/hours?mic=XNYS is_business_day() { dt=${1} open_time=$(curl -s "${CalendarURL}&start=${dt}&end=${dt}" | jq '.[] | .open_time') if [ -n "${open_time}" ]; then return 0 else return 1 fi } export -f is_business_day if [ -z ${Start} ] ; then echo "start is not specified, getting yesterday..." Start=$(date -d "yesterday" "+%Y-%m-%d") echo "Start=${Start}" fi while true; do if is_business_day ${Start}; then break fi echo "${Start} is not business day in US" Start=$(date -d "${Start} - 1 day" "+%Y-%m-%d") done End=$(date -d "${Start} + 1 day" "+%Y-%m-%d") # Exclude today if [ "${End}" == $(date '+%Y-%m-%d') ] ; then End=$(date -d 'yesterday' ''+%Y-%m-%d'') fi if [ "${NumJobs}" == "" ] ; then NumJobs=30 fi echo "Start=${Start} End=${End} NumJobs=${NumJobs}" run_proc() { Inst=${1} Start=${2} End=${3} echo "Running for $Inst" Cmd="${Python} ${PyScript}" Cmd="${Cmd} --config=${Config}" Cmd="${Cmd} --output_dir=${OutputDir}/${Start:0:4}" Cmd="${Cmd} --instruments=ALPACA:${Inst}" Cmd="${Cmd} --start=${Start}" if [ "${End}" != "" ]; then Cmd="${Cmd} --end=${End}" fi Cmd="${Cmd} --log_file=${LogDir}/${Inst}.log" Cmd="${Cmd} --log_level=WARNING" echo ${Cmd} eval ${Cmd} } export -f run_proc key=$(jq -r '.["ALPACA_SANDBOX"] | .api_key' ~/.creds) secret=$(jq -r '.["ALPACA_SANDBOX"] | .secret_key' ~/.creds) Cmd="curl -s --request GET --url 'https://paper-api.alpaca.markets/v2/assets?status=active'" Cmd="${Cmd} --header 'APCA-API-KEY-ID: ${key}'" Cmd="${Cmd} --header 'APCA-API-SECRET-KEY: ${secret}'" Cmd="${Cmd} --header 'accept: application/json'" Cmd="${Cmd} | jq '.[] | select(.class == \"us_equity\" and .exchange != \"OTC\") | .symbol'" Cmd="${Cmd} | sed 's/\"//g'" Cmd="${Cmd} | sed 's/^/STOCK-/'" # split string into array Instruments=() for Inst in $(eval ${Cmd}) do Instruments+=("$Inst") done slice_size=500 # 10K symbols parallel cannot handle for ((ii=0; ii <${#Instruments[@]}; ii+=slice_size)); do InstSlice=("${Instruments[@]:ii:slice_size}") parallel -j ${NumJobs} run_proc {} ${Start} ${End} ::: "${InstSlice[@]}" done echo "Compressing" for file in $(find ${OutputDir} -type f -name '*db' -print ) do echo "Compressing ${file}" gzip ${file} done Source=/app/data/ Targets= Targets="${Targets} oleg@homestore.cvtt.vpn:/works/cvtt/md_archive/equity" Targets="${Targets} cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/equity" for tgt in ${Targets} do Cmd="/usr/bin/rsync -ahv ${Source} ${tgt}" echo $Cmd eval $Cmd done # DbHost="" # Date="" # Schemas=coinbase,bnbspot,bnbfut # while getopts ":h:d:s:" opt; do # case ${opt} in # h ) # DbHost=$OPTARG # ;; # d ) # Date=$OPTARG # ;; # s ) # Schemas=$OPTARG # ;; # \? ) # echo "Invalid option: -$OPTARG" >&2 # usage # ;; # : ) # echo "Option -$OPTARG requires an argument." >&2 # usage # ;; # esac # done # if [ -z "${DbHost}" ] || [ -z "$Date" ]; then # echo "DbHost=${DbHost} Date=${Date}. Cmdline: ${*}" # usage # fi # CredKey="" # if [ "${DbHost}" == "cvttdata" ] ; then # CredKey=TSDB_MD_CVTTDATA_RO # elif [ "${DbHost}" == "cloud21" ] ; then # CredKey=TSDB_MD_CLD21_RO # else # echo "host ${DbHost} is not valid source" # usage # fi # export PYTHONPATH=/ # Cmd="python3.10" # Cmd="${Cmd} cvttpy/research/utils/archive_ts_md.py" # Cmd="${Cmd} --config=http://cloud16.cvtt.vpn:6789/apps/md_recorder" # Cmd="${Cmd} --db_credentials_key=${CredKey}" # Cmd="${Cmd} --credentials_file=/.creds" # Cmd="${Cmd} --date=${Date}" # Cmd="${Cmd} --schemas=${Schemas}" # Cmd="${Cmd} --root_dir=/app/data/${DbHost}" # Cmd="${Cmd} --format=SQLite" # Cmd="${Cmd} --compress" # # Cmd="${Cmd} --log_file=/logs/%T.crypto_md_day.log" # echo ${Cmd} # eval ${Cmd} # echo "Data Sync" # Source=/app/data/ # Targets= # Targets="${Targets} oleg@homestore.cvtt.vpn:/works/cvtt/md_archive/crypto" # Targets="${Targets} cvtt@cloud21.cvtt.vpn:/opt/store/cvtt/md_archive/crypto" # for tgt in ${Targets} # do # Cmd="/usr/bin/rsync -ahv ${Source} ${tgt}" # echo $Cmd # eval $Cmd # done