This repository was archived by the owner on Jan 23, 2023. It is now read-only.
Repository navigation
This repository was archived by the owner on Jan 23, 2023. It is now read-only.
Fix zwj counting #114
Description
When using ZWJ characters (like 🏳️🌈, which is made of 🏳️ + ZWJ + 🌈) Cawbird "miscounts" compared to Twitter. This causes tweets to become unpostable when they're actually valid.
| Character | Cawbird | Notes | |
|---|---|---|---|
| 🏳️ | 2 | 2/4 | Appears to be 4 because of VS-16 for "E0.7 white flag" but not in "white flag" |
| 🌈 | 2 | 2 | |
| 🏳️🌈 | 2 | 5/7 | 2/4 (flag) + 1 (ZWJ) + 2 (Rainbow) |
| 👨🏾 | 2 | 4 | 2 (face) + 2 (Fitzpatrick modifier) |
| 👨 | 2 | 2 |
This actually needs to be included in libtweetlength, but we bundle a copy of the library.
Activity
IBBoard commentedon Aug 29, 2020
This appears to be related to twitter/twitter-text#265 and the v3 parsing, which treats recognised emoji differently.
IBBoard commentedon Aug 29, 2020
For future testing, use the following tweet text and then paste each emoji:
The text should show 17 characters remaining at first (18 if it doesn't copy the trailing space that I put in), so you can see how many characters remain afterwards.
IBBoard commentedon Mar 14, 2021
The special case solution for this would be easy. But the handling of things like families and other ZWJ sequences is MUCH harder!
Fix handling of rainbow flag
Update to libtl e2a07a