Merge branch 'developer' into 'developer-ev'

# Conflicts:
#   dags/dag_conf.yml
#   dags/dag_transformacion_bcom.py

dags/components/DatabaseOperation/DatabaseLoad.py

 import pandas as pd
+from enums.DatabaseTypeEnum import DatabaseTypeEnum
+from sqlalchemy.types import VARCHAR
 
 import logging
 
@@ -8,9 +10,28 @@ logger = logging.getLogger()
 def save_from_dataframe(df: pd.DataFrame, tablename: str, connection) -> bool:
     save = True
     try:
+        chunksize = 2000
+        # db_type = connection.db_type
+        connection = connection.engine
+        # print(df["CREACION_PRODUCTO"].value_counts())
         with connection.connect() as conn:
-            df.to_sql(tablename, conn, if_exists='append', index=False, chunksize=500)
+            # if db_type == DatabaseTypeEnum.ORACLE.value:
+            #     df.info()
+            #     aux = df.columns[df.dtypes == 'object'].tolist()
+            #     print(aux)
+            #     dtyp = {}
+            #     for col in aux:
+            #         print(col)
+            #         print(df[col].dtype)
+            #         df[col] = df[col].astype(str)
+            #         dtyp.update({col: VARCHAR(df[col].str.len().max())})
+            #     # dtyp = {c: VARCHAR(df[c].str.len().max()) for c in aux}
+            #     print(dtyp)
+            #     df.to_sql(tablename, conn, if_exists='append', dtype=dtyp, index=False, chunksize=chunksize)
+            # else:
+            df.to_sql(tablename, conn, if_exists='append', index=False, chunksize=chunksize)
     except Exception as e:
         logger.error(f"Error guardando resultados desde dataframe. {e}")
+        raise AssertionError(f"Error guardando resultados desde dataframe. {e}")
     finally:
         return save

dags/components/Databases/Oracle.py

@@ -101,7 +101,8 @@ class Oracle:
     def get_all_tablenames(self) -> List[str]:
         tablenames = []
         try:
-            command = f"SELECT table_name FROM all_tables WHERE OWNER='{self.user}'"
+            user = self.user.upper()
+            command = f"SELECT table_name FROM all_tables WHERE OWNER='{user}'"
             with self.engine.connect() as conn:
                 tablenames = conn.execute(command).all()
         except Exception as e:

dags/components/Extractor.py

 from typing import Any, Dict
 import json
+import time
 
 import numpy as np
 import pandas as pd
@@ -169,12 +170,14 @@ def extract_from_source(command, source_conn, intern_conn, chunksize: int, **kwa
             # Traemos el iterator
             iterator = get_iterator(command, chunksize, source_engine)
             logger.info(f"Número de pasos para migrar datos: {steps}")
+            start_time = time.time()
             for step in range(steps):
                 dataframe = next(iterator)
                 dataframe = dataframe.fillna(value=np.nan)
-                save = save_from_dataframe(dataframe, tablename, intern_conn.engine)
+                save = save_from_dataframe(dataframe, tablename, intern_conn)
                 if save:
                     logger.info(f"Guardado correctamente dataframe en el paso {step+1}")
+            logger.info(f"Tiempo del Task de descarga de scripts: {round(time.time() - start_time, 3)} segundos")
     except Exception as e:
         delete = delete_table(tablename, intern_conn.engine)
         if delete:

dags/dag_conf.yml

@@ -23,10 +23,10 @@ app:
       service:
       schema: intern_db
   chunksize: 4000
-  label_multiple_select: TABLE
-  source_mask: selectDA  # Sufijo (S)
-  procedure_mask: procedureDA  # S
-  transformation_mask: transformDA  # S
+  label_multiple_select: TABLENAME
+  source_mask: select  # Sufijo (S)
+  procedure_mask: procedure  # S
+  transformation_mask: transform  # S
   prefix_order_delimiter: .
   cloud_provider: aws
   scripts:

dags/dag_inform_process.py

@@ -20,7 +20,7 @@ logger = logging.getLogger()
 DAG_NAME = "INFORM_PROCESS"
 
 # Change this path if is deployed in prod or dev
-MAIN_PATH = "/opt/airflow/dags/"
+MAIN_PATH = "/root/airflow/dags/"
 
 DEFAULT_ARGS = {
     'owner': 'BCOM',

dags/dag_transformacion_tacomventas_promoresidencial.py

@@ -234,7 +234,7 @@ def set_dag_1():
     from yaml.loader import SafeLoader
     # Cambiar conf_path dependiendo del ambiente, en prod usando k8 y contenedores usar /opt/airflow/dags/app_conf.yml
     # En desarrollo, cualquiera que apunte a su carpeta dags
-    conf_path = "/root/airflow/dags/app_conf.yml"
+    conf_path = "/opt/airflow/dags/app_conf.yml"
     with open(conf_path) as f:
         data = yaml.load(f, Loader=SafeLoader)
         general_cnf = data["general"]

dags/procedure_definition.json

@@ -29,8 +29,8 @@
       },
       {
         "name": "NU_ADDON",
-        "datatype": "NUMBER",
-        "decimal_precision": 0
+        "datatype": "TEXT",
+        "maxLength": 5
       },
       {
         "name": "NB_PAQUETE",
@@ -53,11 +53,11 @@
       },
       {
         "name": "FH_ACTIVACION",
-        "datatype": "DATE"
+        "datatype": "DATETIME"
       },
       {
         "name": "FH_OPERACION",
-        "datatype": "DATE"
+        "datatype": "DATETIME"
       },
       {
         "name": "TP_SERVICIO",
@@ -81,7 +81,7 @@
       },
       {
         "name": "FH_CARGA",
-        "datatype": "DATE"
+        "datatype": "DATETIME"
       },
       {
         "name": "NU_ANIO",
@@ -141,7 +141,8 @@
     ],
     "indexes": [
       "CD_PAQUETE", "NU_ADDON", "CD_CLIENTE"
-    ]
+    ],
+    "save_output": true
   },
   {
   "identifier": "PROMOCIONES_RESIDENCIAL",
@@ -240,7 +241,7 @@
     "indexes": ["CD_PAQUETE"]
 },
   {
-  "identifier": "CATALOGO_PROMOCIONES",
+  "identifier": "CATALOGO_PROMOCION",
   "fields": [
     {
       "name": "NOMBRE_PRODUCTO",
@@ -255,7 +256,7 @@
   ]
 },
   {
-  "identifier": "RELACION_PROMOCION_3PA2P",
+  "identifier": "TEMP_PROMO",
   "fields": [
     {
       "name": "TRESP",
@@ -270,7 +271,7 @@
   ]
 },
   {
-  "identifier": "RELACION_POIDPAQUETE",
+  "identifier": "RELACION_POID_PAQ",
   "fields": [
     {
       "name": "POID_PRODUCT",
@@ -300,38 +301,13 @@
   ]
 },
     {
-  "identifier": "PAQUETES_NOPROMOCION",
-  "fields": [
-    {
-      "name": "CD_PAQUETE",
-      "datatype": "TEXT",
-      "maxLength": 50
-    }
-  ]
-},
-    {
-  "identifier": "PROCEDURE_1",
+  "identifier": "ADDONS_UNICO",
   "fields": [
-    {
-      "name": "CD_FOLIO",
-      "datatype": "TEXT",
-      "maxLength": 100
-    },
-    {
-      "name": "CD_CUENTA",
-      "datatype": "TEXT",
-      "maxLength": 100
-    },
     {
       "name": "CD_PAQUETE",
       "datatype": "TEXT",
       "maxLength": 50
-    },
-    {
-      "name": "NB_PAQUETE",
-      "datatype": "TEXT",
-      "maxLength": 200
     }
   ]
 }
-]
+]
\ No newline at end of file