2024-07-27 15:39:09 +00:00
|
|
|
from pathlib import Path
|
2024-08-07 09:39:33 +00:00
|
|
|
|
2024-07-28 15:34:56 +00:00
|
|
|
import pandas as pd
|
2024-07-27 15:39:09 +00:00
|
|
|
|
2024-08-07 09:39:33 +00:00
|
|
|
from .repository import AbstractRepository
|
|
|
|
|
|
|
|
|
|
|
|
class FSRepository(AbstractRepository):
|
2024-07-27 17:19:36 +00:00
|
|
|
def __init__(self, name, basepath, metadata_engine=None):
|
|
|
|
self.name = name
|
|
|
|
|
2024-07-27 15:39:09 +00:00
|
|
|
self.basepath = Path(basepath)
|
|
|
|
self._metadata_engine = metadata_engine
|
|
|
|
|
2024-07-27 15:50:29 +00:00
|
|
|
def ls(self, dir, only_files=False, only_directories=False, recursive=False) -> list[str]:
|
2024-07-27 15:39:09 +00:00
|
|
|
dirpath = Path(dir)
|
|
|
|
|
|
|
|
if only_files:
|
|
|
|
return [str(f.relative_to(dirpath)) for f in dirpath.iterdir() if not f.is_dir()]
|
|
|
|
|
|
|
|
if only_directories:
|
2024-07-27 15:50:29 +00:00
|
|
|
if recursive:
|
|
|
|
return [str(f[0].relative_to(dirpath)) for f in dirpath.walk()]
|
|
|
|
|
2024-07-27 15:39:09 +00:00
|
|
|
return [str(f.relative_to(dirpath)) for f in dirpath.iterdir() if f.is_dir()]
|
|
|
|
|
|
|
|
return [str(f.relative_to(dirpath)) for f in dirpath.iterdir()]
|
|
|
|
|
2024-07-27 15:50:29 +00:00
|
|
|
def schemas(self, recursive=True) -> list[str]:
|
2024-07-27 15:39:09 +00:00
|
|
|
dirpath = self.basepath
|
2024-07-27 15:50:29 +00:00
|
|
|
return self.ls(dirpath, only_directories=True, recursive=True)
|
2024-07-27 15:39:09 +00:00
|
|
|
|
|
|
|
def tables(self, schema:str) -> list[str]:
|
|
|
|
dirpath = self.basepath / schema
|
|
|
|
return self.ls(dirpath, only_files=True)
|
|
|
|
|
|
|
|
def build_table_path(self, table:str, schema:str):
|
|
|
|
table_path = self.basepath
|
2024-07-27 15:50:29 +00:00
|
|
|
if schema == '.':
|
2024-07-27 15:39:09 +00:00
|
|
|
return table_path / table
|
|
|
|
return table_path / schema / table
|
|
|
|
|
2024-07-27 15:50:29 +00:00
|
|
|
def info(self, table:str, schema:str='.'):
|
2024-07-27 15:39:09 +00:00
|
|
|
table_path = self.build_table_path(table, schema)
|
|
|
|
pass
|
|
|
|
|
2024-07-28 15:34:56 +00:00
|
|
|
def read(self, table:str, schema:str='.', read_options={}):
|
2024-07-27 15:39:09 +00:00
|
|
|
table_path = self.build_table_path(table, schema)
|
2024-07-28 15:34:56 +00:00
|
|
|
extension = table_path.suffix
|
|
|
|
if extension == '.csv':
|
|
|
|
return pd.read_csv(table_path, **read_options)
|
|
|
|
|
|
|
|
if extension == '.xlsx':
|
|
|
|
return pd.read_excel(table_path, **read_options)
|
|
|
|
|
|
|
|
raise ValueError("Can't open the table")
|
2024-07-27 15:39:09 +00:00
|
|
|
|
2024-08-14 05:21:36 +00:00
|
|
|
def write(self, content, table:str, schema:str='.'):
|
2024-07-27 15:39:09 +00:00
|
|
|
table_path = self.build_table_path(table, schema)
|
|
|
|
pass
|
|
|
|
|
2024-07-27 15:50:29 +00:00
|
|
|
def delete(self, table:str, schema:str='.'):
|
2024-07-27 15:39:09 +00:00
|
|
|
table_path = self.build_table_path(table, schema)
|
|
|
|
pass
|