feat: diversas melhorias no processamento de dados

Author: Yulle Ulguim

.idea/.gitignore

@@ -5,7 +5,7 @@
 import pandas as pd
 
 
-def run_daily(tickers: list):
+def run_daily(tickers: list, start: datetime, end: datetime):
     print(tickers)
 
     client = MongoClient("mongodb://192.168.31.188:27017/")
@@ -17,17 +17,24 @@ def run_daily(tickers: list):
     mt5.initialize()
 
     for ticker in tickers:
+        if collection.count_documents({'ticker': ticker}) > 0:
+            continue
+
         print(ticker)
 
         data = mt5.copy_rates_range(
             ticker,
             mt5.TIMEFRAME_D1,
-            datetime(2021, 12, 1),
-            datetime(2023, 12, 31),
+            start,
+            end,
         )
 
         data = pd.DataFrame(data)
 
+        if 'time' not in data.columns:
+            print('{} no time', ticker)
+            continue
+
         # Converter data
         data['time'] = pd.to_datetime(data['time'], unit='s')
 

src/daily.py

@@ -19,11 +19,15 @@ def run_hourly(tickers: list):
             ticker,
             mt5.TIMEFRAME_H1,
             datetime(2023, 12, 1),
-            datetime(2023, 12, 19),
+            datetime.now(),
         )
 
         data = pd.DataFrame(data)
 
+        if 'time' not in data.columns:
+            print('{} no time', ticker)
+            continue
+
         print(data)
 
         data['time'] = pd.to_datetime(data['time'], unit='s')

src/hourly.py

@@ -9,19 +9,21 @@
 
 client = MongoClient("mongodb://192.168.31.188:27017/")
 
-client.drop_database('b3')
+# client.drop_database('b3')
 
-# tickers_list = open('resources/tickers.txt', 'r')
-# tickers_list = tickers_list.readlines()
-#
+tickers_list = open('resources/tickers.txt', 'r')
+tickers_list = list(map(lambda x: x.strip(), tickers_list.readlines()))
+
+# tickers_list = ['FRAS3', 'GGPS3']
 # count = 0
 # for line in tickers_list:
 #     count += 1
 #     print("Line{}: {}".format(count, line.strip()))
 
-tickers_list = ['FRAS3', 'GGPS3']
+start = datetime(2022, 1, 1)
+end = datetime.now()
 
-run_daily(tickers_list)
+run_daily(tickers_list, start, end)
 run_hourly(tickers_list)