Merge pull request #3 from elpablete/main

Add support for file 'fronteras'

src/asic/VERSION

@@ -1 +1 @@
-0.1.0
+0.2.0

src/asic/cli.py

@@ -5,8 +5,8 @@
 import pathlib
 from ssl import SSLZeroReturnError
 from typing import Optional
-import pydantic
 
+import pydantic
 import typer
 from rich.progress import track
 
@@ -92,7 +92,7 @@ def get_ftps(
     ftps.connect(
         port=ftps_port,
     )
-
+    logger.info(f"Login to FTP '{ftps_host}' as '{ftps_user}'")
     ftps.login(user=ftps_user, passwd=ftps_password.get_secret_value())
 
     ftps.prot_p()

src/asic/data/ASIC_FILE_CONFIG.jsonl

@@ -15,4 +15,5 @@
 {"code":"cliq", "visibility": "public","name_pattern":"(?P<code>cliq)(?P<name_month>[0-9]{2}).(?P<ext>[a-zA-Z0-9]+)","location_pattern":"/INFORMACION_XM/PUBLICOK/SIC/COMERCIA/{location_year:04}-{location_month:02}/","description":"Liquidación diaria de los agentes comercializadores"}
 {"code":"desbm", "visibility": "agent","name_pattern":"(?P<code>desbm)(?P<name_month>[0-9]{2})(?P<name_day>[0-9]{2}).(?P<ext>[a-zA-Z0-9]+)","location_pattern":"/INFORMACION_XM/USUARIOSK/{agent}/SIC/COMERCIA/{location_year:04}-{location_month:02}/","description":""}
 {"code":"desbmex", "visibility": "agent","name_pattern":"(?P<code>desbmex)(?P<name_month>[0-9]{2})(?P<name_day>[0-9]{2}).(?P<ext>[a-zA-Z0-9]+)","location_pattern":"/INFORMACION_XM/USUARIOSK/{agent}/SIC/COMERCIA/{location_year:04}-{location_month:02}/","description":""}
-{"code":"oefagnch", "visibility": "agent","name_pattern":"(?P<code>oefagnch)(?P<name_month>[0-9]{2})(?P<name_day>[0-9]{2}).(?P<ext>[a-zA-Z0-9]+)","location_pattern":"/INFORMACION_XM/USUARIOSK/{agent}/SIC/COMERCIA/{location_year:04}-{location_month:02}/","description":""}
+{"code":"oefagnch", "visibility": "agent","name_pattern":"(?P<code>oefagnch)(?P<name_month>[0-9]{2})(?P<name_day>[0-9]{2}).(?P<ext>[a-zA-Z0-9]+)","location_pattern":"/INFORMACION_XM/USUARIOSK/{agent}/SIC/COMERCIA/{location_year:04}-{location_month:02}/","description":""}
+{"code":"fronteras", "visibility": "agent","name_pattern":"(?P<agent>[a-zA-Z]{4})_(?P<code>FronterasComerciales)_(?P<name_day>[0-9]{2})-(?P<name_month>[0-9]{2})-(?P<name_year>[0-9]{4}).(?P<ext>xlsx)","location_pattern":"/INFORMACION_XM/USUARIOSK/{agent}/SIC/FRONTERAS/{location_year:04}-{location_month:02}/","description":"Formato con universo de fronteras de un agente, no parece tener un cronograma de publicación"}

src/asic/files/__init__.py

@@ -5,21 +5,22 @@
 from .adem import adem_preprocess
 from .aenc import aenc_preprocess
 from .afac import afac_preprocess
+from .balcttos import balcttos_preprocess
+from .cliq import cliq_preprocess
+from .desbm import desbm_preprocess
+from .desbmex import desbmex_preprocess
 from .dspcttos import dspcttos_preprocess
+from .fronteras import fronteras_preprocess
 from .ldcbmr import ldcbmr_preprocess
+from .oefagnch import oefagnch_preprocess
 from .pep import pep_preprocess
 from .pubfc import pubfc_preprocess
 from .pubfc_falla_hurto import pubfc_falla_hurto_preprocess
 from .sntie import sntie_preprocess
+from .tfroc import tfroc_preprocess
 from .tgrl import tgrl_preprocess
 from .trsd import trsd_preprocess
 from .trsm import trsm_preprocess
-from .tfroc import tfroc_preprocess
-from .cliq import cliq_preprocess
-from .balcttos import balcttos_preprocess
-from .desbm import desbm_preprocess
-from .desbmex import desbmex_preprocess
-from .oefagnch import oefagnch_preprocess
 
 # from .undefined import undefined_preprocess
 
@@ -43,6 +44,7 @@ class SupportedFiles(str, enum.Enum):
     DESBM = "desbm", desbm_preprocess
     DESBMEX = "desbmex", desbmex_preprocess
     OEFAGNCH = "oefagnch", oefagnch_preprocess
+    FRONTERAS = "fronteras", fronteras_preprocess
 
     def __new__(cls, code, preprocessor):
         obj = str.__new__(cls, [code])

src/asic/files/fronteras.py

@@ -0,0 +1,104 @@
+import logging
+import pathlib
+
+# Third party imports
+import pandas as pd
+
+# Local application imports
+from asic.reader import FileReader
+
+from ..metadata import FileItemInfo
+
+logger = logging.getLogger(__name__)
+
+fronteras_format = {
+    "type": "xlsx",
+    "sheet_name": "Fronteras Comerciales",
+    "encoding": "cp1252",
+    "dt_fields": {
+        "Registrada por primera vez el": {"format": "%d-%m-%Y"},
+        "Inicio Representación Frontera": {"format": "%d-%m-%Y"},
+        "Último Cambio en Medidor Ppal o Transformadores": {"format": "%d-%m-%Y"},
+        "Fecha de Calibración Med. Ppal": {"format": "%d-%m-%Y"},
+        "Actualizada por última vez el": {"format": "%Y-%m-%d %H:%M:%S"},
+        "Último Cambio en Medidor Resp.": {"format": "%d-%m-%Y"},
+        "Fecha de Calibración Med. Resp.": {"format": "%d-%m-%Y"},
+    },
+    "dtype": {
+        "Código SIC": str,
+        "Nombre de la Frontera": str,
+        "Registrada en el Mercado por": str,
+        "Registrada por primera vez el": str,
+        "Factor de Pérdidas": float,
+        "NIU": str,
+        "NIT": str,
+        "Voltaje (kV)": str,
+        "Nivel de Tensión": str,
+        "Tipo Punto Medicion": str,
+        "Representante Frontera": str,
+        "Inicio Representación Frontera": str,
+        "Representante Anterior": str,
+        "Operador de Red": str,
+        "Operador de Red de la Zona": str,
+        "Agente Exportador": str,
+        "Agente Importador": str,
+        "Tipo de Frontera": str,
+        "Nombre del CGM": str,
+        "Código SIC DDV": str,
+        "Predio ID": str,
+        "Nombre del Predio": str,
+        "Representante de la DDV": str,
+        "Número de Serie Med. Ppal": str,
+        "Marca Med. Ppal": str,
+        "Clase del Medidor": str,
+        "Clase del CT": str,
+        "Clase del PT": str,
+        "Último Cambio en Medidor Ppal o Transformadores": str,
+        "Entidad Calibradora Med. Ppal.": str,
+        "Fecha de Calibración Med. Ppal": str,
+        "Actualizada por última vez el": str,
+        "Número de Serie Med. Resp.": str,
+        "Marca Med. Resp.": str,
+        "Último Cambio en Medidor Resp.": str,
+        "Entidad Calibradora Med. Resp.": str,
+        "Fecha de Calibración Med. Resp.": str,
+        "Actualizada por última vez el.1": str,  # Added the ".1" since the column name is repeated
+        "Es Agrupadora": str,
+        "Factor PSF (Para agrupadoras)": float,
+        "Agrupada bajo la": str,
+        "Es Principal de Modelo Embebido": str,
+        "Embebida bajo la": str,
+        "Factor Acordado": float,
+        "Factor Ajuste": float,
+        "Factor de Pérdidas Frontera Principal (para embebidas)": float,
+        "Código CIIU": str,
+        "Clasificación Industral General": str,
+        "Clasificación Industrial Específica": str,
+        "NIT.1": str,  # Added the ".1" since the column name is repeated
+        "Departamento": str,
+        "Municipio": str,
+        "Centro Poblado": str,
+        "Dirección": str,
+        "Latitud (°)": float,
+        "Longitud (°)": float,
+        "Altitud (msnm)": int,
+        "Código SIC Submercado": str,
+    },
+}
+
+
+class Fronteras(FileReader):
+    def __init__(self):
+        return super().__init__(fronteras_format.copy())
+
+
+def fronteras_preprocess(filepath: pathlib.Path, item: FileItemInfo) -> pd.DataFrame:
+    """
+    fronteras: se publica un archivo por dia.
+    versiones: No aplica
+    """
+    fronteras_reader = Fronteras()
+    total = fronteras_reader.read(filepath)
+
+    return_col = ["Código SIC", "Nombre de la Frontera"]
+    return total[return_col]