refact: rename stage to repository

This commit is contained in:
2024-08-07 11:39:33 +02:00
parent 7fb7bc6f5c
commit 2de0e5ef5c
11 changed files with 119 additions and 105 deletions

View File

@@ -0,0 +1,64 @@
from pathlib import Path
import pandas as pd
from .repository import AbstractRepository
class FSRepository(AbstractRepository):
def __init__(self, name, basepath, metadata_engine=None):
self.name = name
self.basepath = Path(basepath)
self._metadata_engine = metadata_engine
def ls(self, dir, only_files=False, only_directories=False, recursive=False) -> list[str]:
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:
if recursive:
return [str(f[0].relative_to(dirpath)) for f in dirpath.walk()]
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()]
def schemas(self, recursive=True) -> list[str]:
dirpath = self.basepath
return self.ls(dirpath, only_directories=True, recursive=True)
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
if schema == '.':
return table_path / table
return table_path / schema / table
def info(self, table:str, schema:str='.'):
table_path = self.build_table_path(table, schema)
pass
def read(self, table:str, schema:str='.', read_options={}):
table_path = self.build_table_path(table, schema)
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")
def write(self, table:str, content, schema:str='.'):
table_path = self.build_table_path(table, schema)
pass
def delete(self, table:str, schema:str='.'):
table_path = self.build_table_path(table, schema)
pass

View File

@@ -0,0 +1,5 @@
from abc import ABC
class AbstractMetadataEngine(ABC):
pass

View File

@@ -0,0 +1,36 @@
import abc
from .metadata import AbstractMetadataEngine
class AbstractRepository(abc.ABC):
metadata_engine = AbstractMetadataEngine
@abc.abstractmethod
def schemas():
""" List schemas """
raise NotImplementedError
@abc.abstractmethod
def tables(schema):
""" List table in schema"""
raise NotImplementedError
@abc.abstractmethod
def info(self, path):
""" Get infos about a file"""
raise NotImplementedError
@abc.abstractmethod
def read(self, path):
""" Get content of a file"""
raise NotImplementedError
@abc.abstractmethod
def write(self, path, content):
""" Write content into the file"""
raise NotImplementedError
@abc.abstractmethod
def delete(self, path):
""" Delete the file """
raise NotImplementedError