[leonlowitzki]

Two orders of the same instrument executed within the same minute end up as one PDF: the later download overwrites the earlier one.

Cause

dl_doc already handles this case: if the computed path is already queued, it appends the document id (dl.py:455-462). But the check never fires:

self.filepaths: List[str] = []           # holds str
...
if filepath in self.filepaths:            # filepath is a Path → never equal to a str
...
self.filepaths.append(str(filepath))

Path("a.pdf") in ["a.pdf"] is False, so both documents are queued under the same path and the second write wins.

Real-world case

Two limit buys of the same stock, executed 27 seconds apart within one minute. Both render to Trades/<date> <HH:MM> Abrechnung - <name> - Limit-Buy-Order.pdf; only the later settlement survived. Same for Auftragsbestätigung and Kosteninformation.

Fix

Compare as strings (two lines). The second document now lands at … - Limit-Buy-Order (<doc id>).pdf, which is what the existing code intended.

Test

test_dl_paths_same_minute_same_title passes the Abrechnung from buy_new.json and a copy with another id through dl_doc with the same title and minute. It fails on master (both get the same path) and passes with the fix. Full suite: 285 passed; ruff check and ruff format --check clean.

Scope

This fixes collisions within one run only. If the first document was downloaded in an earlier run, the second still hits filepath.is_file() and is skipped. #398 (recognise documents by id, for #391) addresses the cross-run side; the two changes don't overlap.