ArchiveTeam
<a href>'s unescaped differently dependent on locale
echo '<html><body><a href="%2525CCfile.html" /></body></html>' > file.html
python3 -m http.server 8888 &
env -i ../wget-lua/src/wget -nv --recursive http://127.0.0.1:8888/file.html
env -i LANG=en_US.UTF-8 ../wget-lua/src/wget -nv --recursive http://127.0.0.1:8888/file.html
The outputs are respectively
...
2024-12-13 16:48:29 ERROR 404: File not found.
127.0.0.1 - - [13/Dec/2024 16:48:29] "GET /%25CCfile.html HTTP/1.1" 404 -
http://127.0.0.1:8888/%25CCfile.html:
2024-12-13 16:48:29 ERROR 404: File not found.
...
and
...
2024-12-13 16:48:52 ERROR 404: File not found.
127.0.0.1 - - [13/Dec/2024 16:48:52] code 404, message File not found
127.0.0.1 - - [13/Dec/2024 16:48:52] "GET /%2525CCfile.html HTTP/1.1" 404 -
http://127.0.0.1:8888/%2525CCfile.html:
2024-12-13 16:48:52 ERROR 404: File not found.
...
Notice the difference in the filenames it tries to GET. The behavior of the second is the one I wanted (I don't know if it's correct according to standards, but I hope that it is), but the first is what we get in warrior projects because something in Seesaw clears the environment.
My dev environment is not the official Docker image and there's a small chance that a difference between my installation and that is causing this; but this did come to my attention due to a crash in someone running a production Cohost script so that seems unlikely.
May investigate this myself but Cohost seems liable to go down soon so that can wait.
(This is not an issue with displaying the URLs in wget's output; the same thing happens if you print them in download_child_p or have the link go to a "server" on a port with a listening netcat)