https://gitlab.synchro.net/main/sbbs/-/commit/804e26960745a6d1d29be32f
Modified Files:
src/sbbs3/getmsg.cpp sbbs.h
Log Message:
Fix CP437 header templates printed as UTF-8 for UTF-8-flagged messages

msghdr_text() converts a CP437 text.dat template (MsgSubj, MsgTo,
MsgFrom, the message list formats) to UTF-8 into a 128-byte buffer
when a message's header fields are flagged UTF-8. The stock MsgSubj
string is 72 CP437 bytes with 30 box-drawing characters: exactly 128
bytes as UTF-8. Since 99b0b63767 (2019-08-03) that conversion only
succeeded by writing the NUL-terminator one byte past the buffer.

bb5ff19c18 (2026-07-29) made cp437_to_utf8_str() reserve the
terminator inside the passed size, so the MsgSubj conversion now fails
and msghdr_text() falls back to the raw CP437 template, which the
caller still prints in P_UTF8 mode, i.e. untranslated. On a UTF-8
terminal the first header line of every message with the HFIELDS_UTF8
aux attribute (e.g. all posts from a system that flags web-posted
headers as UTF-8) came out as CP437 mojibake. Messages without the
flag never take this path and were unaffected.

Size the buffer for any text.dat item (MAX_TEXTDAT_ITEM_LEN CP437
characters of at most three UTF-8 bytes each). And since text.ini and bbs.replace_text() can install longer strings, make the failure path
coherent: msghdr_text() logs a warning and returns NULL when the
template can't be converted, and show_msghdr() (via the new
show_msghdr_line()) and list_msg() then display that line entirely in
CP437, header fields included, instead of mixing charsets. A template
with nothing to convert is returned as-is rather than being treated as
a failure.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

---
þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net