You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
fix(httpv1): non-UTF-8 octets in HTTP/1 header values are re-encoded as UTF-8 on rebuild #1302
Describe the bug
HTTP/1 header values are decoded with a guessed charset, then re-encoded as UTF-8 when the message is rebuilt (httpv1.py around line 322). #1241 / #1301 fix the same problem for start lines, but not for header values.
To Reproduce HTTP(b'HTTP/1.1 200 OK\r\nX-Name: caf\xe9\r\n\r\n') rebuilds through from_data as …X-Name: caf\xc3\xa9….
Expected behavior
Header values round-trip byte-exactly. This needs a charset recorded for each header in the data model, or the raw bytes kept.
Activity
JarryShaw commentedon Oct 7, 2026
Blocked until #1301 merges, because both edit
ftp.py/httpv1.pyand their data models.JarryShaw commentedon Oct 7, 2026
Unblocked: #1301 merged at 19:09Z, so the file conflict is gone. Back to
pendinguntil a worker picks it up.fix(ftp,httpv1): keep the raw command case, HTTP/1 whitespace and fie…