gsheet feeder + db WIP

src/databases/__init__.py

@@ -0,0 +1,2 @@
+from .database import Database
+from .gsheet_db import GsheetsDb

src/databases/database.py

@@ -1,9 +1,11 @@
 from __future__ import annotations
 from dataclasses import dataclass
 from abc import abstractmethod, ABC
+from typing import Union
 from metadata import Metadata
 from steps.step import Step
 
+
 @dataclass
 class Database(Step, ABC):
     name = "database"
@@ -11,11 +13,30 @@ class Database(Step, ABC):
     def __init__(self, config: dict) -> None:
         # without this STEP.__init__ is not called
         super().__init__(config)
-        
 
-    # only for typing...
     def init(name: str, config: dict) -> Database:
+        # only for typing...
         return Step.init(name, config, Database)
 
     @abstractmethod
-    def enrich(self, item: Metadata) -> Metadata: pass
+    def started(self, item: Metadata) -> None:
+        """signals the DB that the given item archival has started"""
+        pass
+
+    def failed(self, item: Metadata) -> None:
+        """update DB accordingly for failure"""
+        pass
+
+    def aborted(self, item: Metadata) -> None:
+        """abort notification if user cancelled after start"""
+        pass
+
+    # @abstractmethod
+    def fetch(self, item: Metadata) -> Union[Metadata, bool]:
+        """check if the given item has been archived already"""
+        return False
+
+    @abstractmethod
+    def done(self, item: Metadata) -> None:
+        """archival result ready - should be saved to DB"""
+        pass

src/databases/gsheet_db.py

@@ -0,0 +1,64 @@
+from typing import Union, Tuple
+import gspread
+
+# from metadata import Metadata
+from loguru import logger
+
+# from . import Enricher
+from databases import Database
+from metadata import Metadata
+from steps.gsheet import Gsheets
+from utils import GWorksheet
+
+
+class GsheetsDb(Database):
+    """
+        NB: only works if GsheetFeeder is used. 
+        could be updated in the future to support non-GsheetFeeder metadata 
+    """
+    name = "gsheet_db"
+
+    def __init__(self, config: dict) -> None:
+        # without this STEP.__init__ is not called
+        super().__init__(config)
+
+    @staticmethod
+    def configs() -> dict:
+        return {}
+
+    def started(self, item: Metadata) -> None:
+        logger.warning(f"STARTED {item}")
+        gw, row = self._retrieve_gsheet(item)
+        gw.set_cell(row, 'status', 'Archive in progress')
+
+    def failed(self, item: Metadata) -> None:
+        logger.error(f"FAILED {item}")
+        self._safe_status_update(item, 'Archive failed')
+
+    def aborted(self, item: Metadata) -> None:
+        logger.warning(f"ABORTED {item}")
+        self._safe_status_update(item, '')
+
+    def fetch(self, item: Metadata) -> Union[Metadata, bool]:
+        """check if the given item has been archived already"""
+        # TODO: this should not be done at the feeder stage then!
+        return False
+
+    def done(self, item: Metadata) -> None:
+        """archival result ready - should be saved to DB"""
+        logger.success(f"DONE {item}")
+        gw, row = self._retrieve_gsheet(item)
+        self._safe_status_update(item, 'done')
+        pass
+
+    def _safe_status_update(self, item: Metadata, new_status: str) -> None:
+        try:
+            gw, row = self._retrieve_gsheet(item)
+            gw.set_cell(row, 'status', new_status)
+        except Exception as e:
+            logger.debug(f"Unable to update sheet: {e}")
+
+    def _retrieve_gsheet(self, item: Metadata) -> Tuple[GWorksheet, int]:
+        gw: GWorksheet = item.get("gsheet").get("worksheet")
+        row: int = item.get("gsheet").get("row")
+        return gw, row