progress - intermediate for md checklists

scripts/mkt_data/alpaca_hist_md/alpaca_assets.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+CredsFile=${1}
+CredKey=${2}
+
+if [ "${CredsFile}" == "" ] ; then
+    CredsFile="${HOME}/.creds"
+fi
+
+if [ "${CredKey}" == "" ] ; then
+    CredKey="ALPACA_SANDBOX"
+fi
+
+key=$(jq -r ".[\"${CredKey}\"] | .api_key" ${CredsFile})
+secret=$(jq -r ".[\"${CredKey}\"] | .secret_key" ${CredsFile})
+
+Cmd="curl -s --request GET --url 'https://paper-api.alpaca.markets/v2/assets?status=active'"
+Cmd="${Cmd} --header 'APCA-API-KEY-ID: ${key}'"
+Cmd="${Cmd} --header 'APCA-API-SECRET-KEY: ${secret}'"
+Cmd="${Cmd} --header 'accept: application/json'"
+
+eval ${Cmd}
+
+# split string into array
+Instruments=()
+for Inst in $(eval ${Cmd})
+do
+    Instruments+=("$Inst")
+done

scripts/mkt_data/alpaca_hist_md/alpaca_md.sh

@@ -0,0 +1,114 @@
+#!/bin/bash
+
+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  "${open_time}" ]; then
+        return 0
+    else
+        return 1
+    fi
+
+}
+export -f is_business_day
+
+
+if [ "${Start}" == "" ] ; then
+    Start=$(date -d "yesterday" "+%Y-%m-%d")
+fi
+
+if ! is_business_day ${Start}; then
+    echo "${Start} is not business day in US"
+    exit
+fi
+
+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}"
+
+# export PyScript=/home/cvtt/prod/cvttpy/exchanges/alpaca/hist_mkt_data.py
+
+export PYTHONPATH=/home/cvtt/prod
+export Python=/home/cvtt/.pyenv/python3.10-venv/bin/python3
+export Config=http://cloud16.cvtt.vpn:6789/apps/minimal_md
+export PyScript=/home/cvtt/prod/cvttpy/exchanges/alpaca/hist_md/hist_md_bars.py
+
+export OutputDir=/home/cvtt/prod/archive/md_archive/equity/alpaca_md # Local
+export LogDir=/home/cvtt/prod/logs/alpaca_md
+
+mkdir -p ${LogDir}
+
+run_proc() {
+    Inst=${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 ${Cmd}
+}
+
+export -f run_proc
+
+if [ "$InstListFile" == "" ]; then
+
+    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-/'"
+else
+    Cmd="cat ${InstListFile}"
+fi
+
+# 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

scripts/mkt_data/alpaca_hist_md/load_rel_liq.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+export PYTHONPATH=${HOME}/prod
+
+Python=${HOME}/.pyenv/python3.10-venv/bin/python3
+Script=${HOME}/prod/cvttpy/exchanges/alpaca/hist_md/rl_calc_loader.py
+DbFile=${HOME}/prod/data/rel_liquidity.db
+Config=http://cloud23.cvtt.vpn:6789/apps/minimal_md_eqt
+
+Cmd="${Python}"
+Cmd="${Cmd} ${Script}"
+Cmd="${Cmd} --config=${Config}"
+Cmd="${Cmd} --db_file=${DbFile}"
+
+echo ${Cmd}
+eval ${Cmd}
+
+echo Done ${0} ${*}

scripts/mkt_data/alpaca_hist_md/run_alpaca_hist_md.sh

@@ -0,0 +1,50 @@
+#!/bin/bash
+
+is_business_day() {
+    dt=${1}
+
+    CalendarURL=http://cloud16.cvtt.vpn:8000/api/v1/markets/hours?mic=XNYS
+    open_time=$(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
+
+# Dates are in descending order
+FinishedDateFile=${HOME}/prod/run/alpaca_md_last_date
+Script=${HOME}/prod/run/alpaca_md.sh
+
+last_date=$(cat ${FinishedDateFile})
+
+if [ "${last_date}" == "" ]; then
+    echo "process is currently on"
+    exit
+fi
+
+while true; do
+    new_date=$(date -d "${last_date} - 1 day" "+%Y-%m-%d")
+    if is_business_day ${new_date}; then
+        break
+    else
+        last_date=${new_date}
+    fi
+done
+
+
+echo Collecting data for ${new_date}
+
+Cmd="${Script} ${new_date}"
+
+echo "${0} Starting at $(date)"
+
+echo > ${FinishedDateFile}
+echo ${Cmd} && eval ${Cmd}
+echo $new_date > ${FinishedDateFile}
+
+echo "${0} Done at $(date)"
+
+