In `Timing.format_timestamp`, if `self.timezone_offset` is `None` (which is explicitly allowed by the `Optional[int]` type hint), it instantiates `TimeZone(None)`.
When `datetime.fromtimestamp()` calls the `utcoffset()` method on this timezone object, it executes `timedelta(hours=self.offset)`, which evaluates to `timedelta(hours=None)`. This raises a `TypeError: unsupported type for timedelta hours component: NoneType`, causing the application to crash during export.
Affected files: data_model.py
Signed-off-by: Tang Vu <vuminhtang2212@gmail.com>
I realized the `jid_map` table might be missing after reviewing @lifnej's work in ee7db80. This fix adds use the preflight check result for the table before querying it.
I plan to apply this same pattern to other sections where `jid_map` is used.
WhatsApp doesn't show when a reaction was made, and I don't want to mess with a popup in the HTML yet. Let’s just fetch the data for now. It might come in handy later.
Credit to @tlcameron3 from #79
Refactored the brute-force offset search in `_decrypt_crypt14` to use `ProcessPoolExecutor` for better parallelism and performance. Improved progress reporting and clean shutdown on success or interruption.
Implements the latest LID mapping changes. This should fully addresses #188 and likely resolves#144 (validation required). Note: A successful fix for #144 deprecates the pending workaround in #168. Additionally, resolved a bug where chat filters were not working for newly created chat rooms.
Moved the "Processing reactions..." log entry to occur after the `message_add_on` table check. This prevents the log from appearing on the old WhatsApp schema
This change replaces the inefficient N+1 SQL query pattern with a pre-computed hash map. By fetching `ZSTANZAID` and `ZTEXT` pairs globally before processing, the exporter can resolve quoted message content in O(1) time.
Crucially, this maintains parity with the Android exporter by ensuring that replies to messages outside the current date or chat filters are still correctly rendered, providing full conversational context without the performance penalty of repeated database hits.
