Vorabpauschale gets parsed as Dividend

[egorlukomsky]

I just got the recent Vorabpauschale transactions and they were mapped to Dividends (expected type would be Taxes).
It uses the same eventType ssp_corporate_action_invoice_cash, but the Amount is negative this time.

Here's a full transaction json example:
vorabpauschale.json

[NiklasRosenstein]

Oof, that's a bit of a tricky one. I think we could have the Event._parse_type() function handle this:

diff --git a/pytr/event.py b/pytr/event.py
index ca44893..5857c98 100644
--- a/pytr/event.py
+++ b/pytr/event.py
@@ -99,19 +99,29 @@ class Event:
             Event: Event object
         """
         date: datetime = datetime.fromisoformat(event_dict["timestamp"][:19])
-        event_type: Optional[EventType] = cls._parse_type(event_dict)
-        title: str = event_dict["title"]
         value: Optional[float] = (
             v if (v := event_dict.get("amount", {}).get("value", None)) is not None and v != 0.0 else None
         )
+        event_type: Optional[EventType] = cls._parse_type(event_dict, value)
+        title: str = event_dict["title"]
         fees, isin, note, shares, taxes = cls._parse_type_dependent_params(event_type, event_dict)
         return cls(date, title, event_type, fees, isin, note, shares, taxes, value)
 
     @staticmethod
-    def _parse_type(event_dict: Dict[Any, Any]) -> Optional[EventType]:
-        event_type: Optional[EventType] = tr_event_type_mapping.get(event_dict.get("eventType", ""), None)
-        if event_dict.get("status", "").lower() == "canceled":
-            event_type = None
+    def _parse_type(event_dict: Dict[Any, Any], value: Optional[float]) -> Optional[EventType]:
+        raw_event_type = event_dict.get("eventType", "")
+
+        if raw_event_type == "ssp_corporate_action_invoice_cash":
+            # This is usually dividends, but with a negative value this is a "Vorabpauschale" (German tax
+            # that is applied to non-distributing ETFs before they are realized since 2018).
+            assert value is not None, "no value associated with `ssp_corporate_action_invoice_cash` event?"
+            event_type = PPEventType.TAXES if value < 0 else PPEventType.DIVIDEND
+
+        else:
+            event_type: Optional[EventType] = tr_event_type_mapping.get(raw_event_type, None)
+            if event_dict.get("status", "").lower() == "canceled":
+                event_type = None
+
         return event_type

But then the timeline has its own bit of mapping logic for the subfolder, which isn't bijective compared to the PPEventType mapping, and I'd want to avoid duplicating more of that logic.

pytr/pytr/timeline.py

Lines 141 to 148 in f1484b6

	subfolder = {
	"benefits_saveback_execution": "Saveback",
	"benefits_spare_change_execution": "RoundUp",
	"ssp_corporate_action_invoice_cash": "Dividende",
	"CREDIT": "Dividende",
	"INTEREST_PAYOUT_CREATED": "Zinsen",
	"SAVINGS_PLAN_EXECUTED": "Sparplan",
	}.get(event["eventType"])

pytr/pytr/event.py

Line 40 in f1484b6

tr_event_type_mapping = {

Medium-term I suppose we could consider taking on some of the work already discussed in #98 and using a set of rules not only to determine the output folders, but also to normalize event types and data fields.