When using org-protocol capture URLs on Windows, pages with non-ASCII titles or captured text results in a capture with incorrect values applied in the template. Despite such characters being properly URL-encoded in the capture URL, Emacs does not decode them back to valid Unicode characters. Despite this, the file to which the captures are sent is properly UTF-8 encoded, and doing a copy/paste of the text that gets mangled via the capture URL into the buffer works as expected.
A couple examples that I whipped up to demonstrate:
- Example 1
- URL: https://www.branah.com/japanese
- Capture URL:
org-protocol://capture?template=L&url=https%3A%2F%2Fwww.branah.com%2Fjapanese&title=Japanese%20Keyboard%20-%20%E6%97%A5%E6%9C%AC%E8%AA%9E%E3%81%AE%E3%82%AD%E3%83%BC%E3%83%9C%E3%83%BC%E3%83%89%20-%20Type%20Japanese%20Online&body= - Title: Japanese Keyboard - 日本語のキーボード - Type Japanese Online
- Captured as: Japanese Keyboard - ????????? - Type Japanese Online
- Example 2
- URL: https://www.washingtonpost.com/history/2021/05/03/dominos-pizza-noid-returns/
- Capture URL:
org-protocol://capture?template=L&url=https%3A%2F%2Fwww.washingtonpost.com%2Fhistory%2F2021%2F05%2F03%2Fdominos-pizza-noid-returns%2F&title=Domino%E2%80%99s%20Pizza%20Noid%3A%20Return%20of%20the%20mascot%20that%20drove%20a%20man%20to%20take%20hostages%20in%201989%20-%20The%20Washington%20Post&body= - Title: Domino’s Pizza Noid: Return of the mascot that drove a man to take hostages in 1989 - The Washington Post
- Captured as: Domino\222s Pizza Noid: Return of the mascot that drove a man to take hostages in 1989 - The Washington Post
I am running 27.2 on Windows 10, with Org 9.4 via ELPA.
Update: Calling emacsclientw or emacsclient directly with such org-protocol URLs seems to work just fine. Windows appears to be passing the decoded URL to emacsclientw.
