⚡️Retornada a leitura usando pyarrow para performance

extracao/datasources/base.py

@@ -30,11 +30,11 @@ class Base:
 	folder: Union[str, Path] = Path(__file__).parent / 'arquivos' / 'saida'
 	read_cache: bool = False
 
-	def _read(self, stem: str) -> pd.DataFrame:
+	def _read(self, stem: str, backend: str = 'pyarrow') -> pd.DataFrame:
 		"""Lê o dataframe formado por self.folder / self.stem.parquet.gzip"""
 		file = Path(f'{self.folder}/{stem}.parquet.gzip')
 		try:
-			df = pd.read_parquet(file, dtype_backend='pyarrow')
+			df = pd.read_parquet(file, dtype_backend=backend)
 		except (ArrowInvalid, FileNotFoundError) as e:
 			raise ValueError(f'Error when reading {file}') from e
 		return df
@@ -43,7 +43,7 @@ def _save(self, df: pd.DataFrame, folder: Union[str, Path], stem: str) -> pd.Dat
 		"""Format, Save and return a dataframe"""
 		try:
 			file = Path(f'{folder}/{stem}.parquet.gzip')
-			df.to_parquet(file, compression='gzip', index=False, engine='pyarrow')
+			df.astype('string').to_parquet(file, compression='gzip', index=False, engine='pyarrow')
 		except (ArrowInvalid, ArrowTypeError) as e:
 			raise Exception(f'Não foi possível salvar o arquivo parquet') from e
 		return df
@@ -95,7 +95,7 @@ def register_log(
 		df['Log'] = df['Log'].astype('string', copy=False).fillna('[]')
 		df['Log'] = df['Log'].str.replace('^$', r'[]', regex=True)
 		log_function = partial(Base.format_log, processing=processing, column=column)
-		print(f'Logging {processing}...')
+		print(f'Logging: {processing}')
 		df.loc[row_filter, 'Log'] = df[row_filter].progress_apply(log_function, axis=1)
 
 	@staticmethod