From 9eb4f8e1175dbe8cc35e7c2a04fc3a700f214863 Mon Sep 17 00:00:00 2001
From: Oleg Sheynin <osheynin@gmail.com>
Date: Sun, 18 May 2025 18:42:08 -0400
Subject: [PATCH] daily sim for docker.initial

---
 VERSION                                       |   2 +-
 .../sim_prepare}/prepare_crypto_simdata.sh    | 127 +++++-------------
 .../sim_prepare}/prepare_eqt_simdata.sh       |   4 +
 3 files changed, 41 insertions(+), 92 deletions(-)
 rename {research => scripts/mkt_data/sim_prepare}/prepare_crypto_simdata.sh (67%)
 rename {research => scripts/mkt_data/sim_prepare}/prepare_eqt_simdata.sh (99%)

diff --git a/VERSION b/VERSION
index 588b40e..849cb4f 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-2.1.3,remove cvttdata
+2.1.4,docker prepare sim - crypto 
diff --git a/research/prepare_crypto_simdata.sh b/scripts/mkt_data/sim_prepare/prepare_crypto_simdata.sh
similarity index 67%
rename from research/prepare_crypto_simdata.sh
rename to scripts/mkt_data/sim_prepare/prepare_crypto_simdata.sh
index ad47de8..4bc8ba8 100755
--- a/research/prepare_crypto_simdata.sh
+++ b/scripts/mkt_data/sim_prepare/prepare_crypto_simdata.sh
@@ -1,112 +1,54 @@
 #!/bin/bash
 
-# SQLite DDL for simulation
-# =========================
-
-
-# -- md_quotes
-
-# -- md_1min_bars
-
-
-usage() {
-  echo -n "Usage: $0"
-  echo -n " -d YYYMMDD Date>"
-  echo -n " [-O <output dir (./) >]"
-  echo -n " [-S <archive_source_host (cloud21)>]"
-  echo -n " [-D <orig_db_source> (cloud21)]"
-  echo
-  exit 1
-}
-
 # --------------------- Settings
-SourceHost=cloud21
-SourceRootDir=/opt/store/cvtt/md_archive/crypto
-DbSource=cloud21
+SOURCE_HOST=cloud21.cvtt.vpn
+SOURCE_ROOT_DIR=/opt/store/cvtt/md_archive/crypto
+DB_SOURCE=cloud28
+OUTPUT_DIR=/tmp
+DATE=20250516
 # --------------------- Settings
 
-while getopts ":d:O:S:D:" opt; do
-  case ${opt} in
-    d )
-      Date=$OPTARG
-      ;;
-    O )
-      OutputDir=$OPTARG
-      ;;
-    S )
-      SourceHost=$OPTARG
-      ;;
-    D )
-      DbSource=$OPTARG
-      ;;
-    \? )
-      echo "Invalid option: -$OPTARG" >&2
-      usage
-      ;;
-    : )
-      echo "Option -$OPTARG requires an argument." >&2
-      usage
-      ;;
-  esac
-done
-
-if [ -z ${Date}  ] ; then
-    echo "date is not specified"
-    usage
-fi
-if [ -z ${OutputDir}  ] ; then
-    OutputDir=.
+if [ -z ${DATE}  ] ; then
+    DATE=$(date -d 'yesterday' +'%Y%m%d')
 fi
 
-SourceHost+=".cvtt.vpn"
-mkdir -p ${OutputDir}
+if [ -z ${OUTPUT_DIR}  ] ; then
+    OUTPUT_DIR=.
+fi
 
-# --- Binance
-Instruments=( PAIR-ADA-USDT )
-Instruments+=( PAIR-BCH-USDT )
-Instruments+=( PAIR-BTC-USDT )
-Instruments+=( PAIR-DOT-USDT )
-Instruments+=( PAIR-ETH-USDT )
-Instruments+=( PAIR-LTC-USDT )
-Instruments+=( PAIR-SOL-USDT )
-Instruments+=( PAIR-USDC-USDT )
-Instruments+=( PAIR-XRP-USDT )
+echo "DATE=${DATE} SOURCE_HOST=${SOURCE_HOST}"
 
-# --- Coinbase
-Instruments+=( PAIR-ADA-USD )
-Instruments+=( PAIR-BCH-USD )
-Instruments+=( PAIR-BTC-USD )
-Instruments+=( PAIR-DOT-USD )
-Instruments+=( PAIR-ETH-USD )
-Instruments+=( PAIR-LTC-USD )
-Instruments+=( PAIR-SOL-USD )
-Instruments+=( PAIR-XRP-USD )
+mkdir -p ${OUTPUT_DIR}
 
-echo "Date=${Date} Instruments=${Instruments[@]} OutputDir=${OutputDir}"
-echo Getting data from ${DataHost} ...
-
-year=$(date -d ${Date} +"%Y")
-month=$(date -d ${Date} +"%m")
-SourceDir="${SourceRootDir}/${DbSource}/${year}/${month}"
-SourceFile="${SourceDir}/${Date}.mktdata.db.gz"
+year=$(date -d ${DATE} +"%Y")
+month=$(date -d ${DATE} +"%m")
+SourceDir="${SOURCE_ROOT_DIR}/${DB_SOURCE}/${year}/${month}"
+SourceFile="${SourceDir}/${DATE}.mktdata.db.gz"
 
 Cmd="rsync -ahv"
-Cmd+=" ${SourceHost}:${SourceFile}"
-Cmd+=" $OutputDir/"
+Cmd+=" -e 'ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null'"
+Cmd+=" ${SOURCE_HOST}:${SourceFile}"
+Cmd+=" $OUTPUT_DIR/"
 echo ${Cmd}
 eval ${Cmd}
 
-Cmd="(cd ${OutputDir} && gunzip *.db.gz)"
+Cmd="(cd ${OUTPUT_DIR} && gunzip -f *.db.gz)"
 echo ${Cmd}
 eval ${Cmd}
 
-SourceDbFile="${OutputDir}/${Date}.mktdata.db"
-ResultDbFile="${OutputDir}/${Date}.crypto_sim_md.db"
+SourceDbFile="${OUTPUT_DIR}/${DATE}.mktdata.db"
+ResultDbFile="${OUTPUT_DIR}/${DATE}.crypto_sim_md.db"
 
+echo "SourceDbFile=${SourceDbFile}"
 echo "Creating Result Database File ${ResultDbFile}"
+cleanup() {
+    rm ${SourceDbFile}
+}
+trap cleanup EXIT
 
 echo "Creating table md_trades ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_trades (
     tstamp text,
     tstamp_ns integer,
@@ -119,12 +61,14 @@ CREATE TABLE  IF NOT EXISTS md_trades (
     condition text,
     tape text
 );
+
 CREATE UNIQUE INDEX  IF NOT EXISTS md_trades_uidx
 ON md_trades(tstamp_ns, exchange_id, instrument_id);
 EOF
 
 echo "Creating table md_quotes ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_quotes (
     tstamp text,
     tstamp_ns integer,
@@ -143,6 +87,7 @@ EOF
 
 echo "Creating table md_1min_bars ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_1min_bars (
 	tstamp text,
 	tstamp_ns integer,
@@ -162,6 +107,7 @@ EOF
 
 echo "Loading md_trades ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_trades SELECT
@@ -199,6 +145,7 @@ EOF
 
 echo "Loading md_quotes ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_quotes SELECT
@@ -237,6 +184,7 @@ EOF
 
 echo "Loading md_1min_bars ..."
 sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${SourceDbFile}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_1min_bars SELECT
@@ -273,16 +221,13 @@ COMMIT;
 DETACH source_db;
 EOF
 
-Cmd="rm ${SourceDbFile}"
-echo ${Cmd}
-eval ${Cmd}
-
+exit
 
 Cmd="gzip ${ResultDbFile}"
 echo ${Cmd}
 eval ${Cmd}
 
-Cmd="rsync -ahvv ${ResultDbFile}.gz cvtt@hs01.cvtt.vpn:/works/cvtt/md_archive/crypto/sim/"
+Cmd="rsync -ahvv ${ResultDbFile}.gz "
 echo ${Cmd}
 eval ${Cmd}
 
diff --git a/research/prepare_eqt_simdata.sh b/scripts/mkt_data/sim_prepare/prepare_eqt_simdata.sh
similarity index 99%
rename from research/prepare_eqt_simdata.sh
rename to scripts/mkt_data/sim_prepare/prepare_eqt_simdata.sh
index 47d93ba..8ee9735 100755
--- a/research/prepare_eqt_simdata.sh
+++ b/scripts/mkt_data/sim_prepare/prepare_eqt_simdata.sh
@@ -104,6 +104,7 @@ echo "Creating Result Database File ${ResultDbFile}"
 
 echo "Creating table md_trades ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_trades (
     tstamp text,
     tstamp_ns integer,
@@ -122,6 +123,7 @@ EOF
 
 echo "Creating table md_quotes ..."
 sqlite3 ${ResultDbFile} <<EOF
+.echo ON
 CREATE TABLE  IF NOT EXISTS md_quotes (
     tstamp text,
     tstamp_ns integer,
@@ -166,6 +168,7 @@ for sym in ${Symbols[@]}; do
 
     echo "Loading md_trades and md_quotes from ${src_qat_db} ..."
     sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${src_qat_db}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_trades SELECT
@@ -200,6 +203,7 @@ EOF
 
     echo "Loading md_1min_bars from ${src_hbar_db} ..."
     sqlite3  ${ResultDbFile} <<EOF
+.echo ON
 ATTACH '${src_hbar_db}' AS source_db;
 BEGIN;
 INSERT OR IGNORE INTO md_1min_bars SELECT