added optimal position data

sysdata/parquet/parquet_optimal_positions.py

@@ -0,0 +1,60 @@
+
+from syscore.exceptions import missingData
+from sysdata.parquet.parquet_access import ParquetAccess
+from sysdata.production.optimal_positions import optimalPositionData
+from syslogging.logger import *
+
+from sysobjects.production.tradeable_object import (
+    instrumentStrategy,
+    listOfInstrumentStrategies,
+)
+
+import pandas as pd
+
+OPTIMAL_POSITION_COLLECTION = "optimal_positions"
+
+
+class parquetOptimalPositionData(optimalPositionData):
+    def __init__(self, parquet_access: ParquetAccess, log=get_logger("parquetOptimalPositionData")):
+
+        super().__init__(log=log)
+        self._parquet = parquet_access
+
+    def __repr__(self):
+        return "parquetOptimalPositionData"
+
+    @property
+    def parquet(self):
+        return self._parquet
+
+    def get_list_of_instrument_strategies_with_optimal_position(
+        self,
+    ) -> listOfInstrumentStrategies:
+
+        raw_list_of_instrument_strategies = self.parquet.get_all_identifiers_with_data_type(data_type=OPTIMAL_POSITION_COLLECTION)
+        list_of_instrument_strategies = [
+            instrumentStrategy.from_key(key)
+            for key in raw_list_of_instrument_strategies
+        ]
+
+        return listOfInstrumentStrategies(list_of_instrument_strategies)
+
+    def get_optimal_position_as_df_for_instrument_strategy(
+        self, instrument_strategy: instrumentStrategy
+    ) -> pd.DataFrame:
+
+        try:
+            ident = instrument_strategy.key
+            df_result = self.parquet.read_data_given_data_type_and_identifier(data_type=OPTIMAL_POSITION_COLLECTION, identifier=ident)
+        except:
+            raise missingData
+
+        return df_result
+
+    def write_optimal_position_as_df_for_instrument_strategy_without_checking(
+        self,
+        instrument_strategy: instrumentStrategy,
+        optimal_positions_as_df: pd.DataFrame,
+    ):
+        ident = instrument_strategy.key
+        self.parquet.write_data_given_data_type_and_identifier(data_type=OPTIMAL_POSITION_COLLECTION, identifier=ident, data_to_write=optimal_positions_as_df)

sysinit/transfer/backup_arctic_to_parquet.py

@@ -15,6 +15,7 @@
 from sysdata.parquet.parquet_multiple_prices import parquetFuturesMultiplePricesData
 from sysdata.parquet.parquet_spotfx_prices import parquetFxPricesData
 from sysdata.parquet.parquet_spreads import parquetSpreadsForInstrumentData
+from sysdata.parquet.parquet_optimal_positions import parquetOptimalPositionData
 
 from sysdata.csv.csv_futures_contracts import csvFuturesContractData
 from sysdata.csv.csv_contract_position_data import csvContractPositionData
@@ -85,7 +86,9 @@ def backup_arctic_to_parquet():
         do = true_if_answer_is_yes("Time series of spread costs?")
         if do:
             backup_spreads_to_parquet(backup_data)
-        #backup_optimal_positions(backup_data)
+        do = true_if_answer_is_yes("optimal positions?")
+        if do:
+            backup_optimal_positions(backup_data)
         #backup_roll_state_data(backup_data)
 
 
@@ -106,7 +109,7 @@ def get_data_blob(logname):
             parquetFuturesAdjustedPricesData,
             #csvFuturesContractData,
             parquetFxPricesData,
-            #csvOptimalPositionData,
+            parquetOptimalPositionData,
             #csvRollStateData,
             #csvSpreadCostData,
             parquetSpreadsForInstrumentData,
@@ -430,10 +433,25 @@ def backup_optimal_positions(data):
             )
         except missingData:
             continue
-        data.csv_optimal_position.write_optimal_position_as_df_for_instrument_strategy_without_checking(
+
+        try:
+            parquet_data = data.parquet_optimal_position.get_optimal_position_as_df_for_instrument_strategy(
+                instrument_strategy
+            )
+        except missingData:
+            parquet_data=[]
+
+        if len(parquet_data)>=len(arctic_data):
+            data.log.debug("skipping already written")
+
+        data.parquet_optimal_position.write_optimal_position_as_df_for_instrument_strategy_without_checking(
             instrument_strategy, arctic_data
         )
-        data.log.debug("Backed up %s  optimal position data" % str(instrument_strategy))
+        parquet_data = data.parquet_optimal_position.get_optimal_position_as_df_for_instrument_strategy(
+            instrument_strategy
+        )
+
+        data.log.debug("Backed up %s  optimal position data was %s now %s" % (str(instrument_strategy), str(arctic_data), str(parquet_data)))
 
 
 def backup_spread_cost_data(data):