_win_printf used to drop the buffer append and pad render for any window
in paged state, but only WIN_CHAT can recover lost messages via
chatwin_db_history on scroll-down. WIN_MUC, WIN_PRIVATE and WIN_AI have
no such fallback, so incoming and outgoing messages were silently lost
when the user was viewing history. Suppress the buffer/render only for
WIN_CHAT; for the rest, still bump unread_msg but let the message land
in the buffer so it becomes visible on scroll-down.
This also removes the manual paged/unread_msg reset before printing the
user message in cl_ev_send_ai_msg — it was a local workaround for the
same drop and is no longer needed.
The buffer-bottom reset in win_page_down (c167f487) is still required:
it is the only path that clears paged and unread_msg for non-chat
windows when the user scrolls back to the bottom (WIN_SCROLL_REACHED_BOTTOM
is set only on the is_chat DB branch).
Add an AI client module that integrates with OpenAI-compatible API
providers (OpenAI, Perplexity, and custom endpoints) to provide
AI-assisted chat within CProof. Users can start sessions with /ai start,
send prompts, receive responses in a dedicated AI window, switch between
providers and models, and manage API keys — all with tab-completion.
Providers are configured via /ai set commands with per-provider API keys,
endpoints, default models, and custom settings. Two default providers
(openai, perplexity) are seeded on first use. Provider state persists in
[ai/<name>] sections of the preferences keyfile with automatic migration
from the previous flat-key format.
The /ai command integrates into the existing command system with 8
subcommands covering provider management, session lifecycle, model
fetching, and conversation clearing. Autocomplete uses the standard
flat prefix-matching chain for reliable tab-completion at every nesting
level. A new ProfAiWin window type is added to the window system.
Architecture:
Async design: HTTP requests run on a background thread (pthread) to avoid blocking the ncurses UI loop; results are displayed on the main thread via direct function calls
Thread safety: AIProvider and AISession use atomic ref-counting and mutex-protected session state; the request thread snapshots all session data before making the HTTP call
Window validation: wins_ai_exists() prevents use-after-free when the user closes the AI window during an in-flight HTTP request (~60s)
Privacy: store:false is sent with every request to prevent providers from persisting conversations or using them for training
Response size limit: 10MB cap with immediate curl abort via CURL_WRITEFUNC_ERROR to prevent OOM
JSON parsing uses unified helpers for both chat responses and error
envelopes with consistent escape decoding. The response parser tries
Perplexity /v1/responses "text" field first, then falls back to OpenAI
"content". Error parsing extracts provider error.message from the
standard envelope format. Model parsing handles multiple API response
formats (OpenAI list, Perplexity, array) including edge cases.
Tests include 470+ lines of unit tests covering provider management,
session lifecycle, JSON parsing (multiple formats), autocomplete cycling,
and error handling, plus functional tests for /ai command dispatch.
A stub_ai.c module isolates unit tests from UI dependencies.
Implement /force-encryption command to configure forced encryption
settings with on|off and policy resend-to-confirm|block options.
The changes enhance user experience by clarifying encryption requirements
and providing actionable commands. Users can now press Enter again to
confirm unencrypted messages in resend-to-confirm mode.
Simple `ui_resize` is sufficient for rerendering all the elements.
Other calls create complexity, slow down the applications and even introduce mem leaks.
`ui_resize` technique is already used widely, including `cmd_editor` case.
Fixes#4
Instead of always returning a `strdup()`'ed version, maybe return NULL and
handle this in the calling code.
This is still not true for `plugins_pre_chat_message_display()`, where we
return the passed `messag` in case there are no plugins, since this would
require a bigger refactor of more parts.
This also
* merges the implementation of `sv_ev_incoming_private_message()` and
`sv_ev_delayed_private_message()`, since they differed only by a single
line.
* untangles the `#ifdef` mess in `cl_ev_send_muc_msg_corrected()`.
Signed-off-by: Steffen Jaeckel <s@jaeckel.eu>
In the documentation of the `muc_nick` function we can read:
> The nickname is owned by the chat room and should not be modified or freed
We should reflect that in the return type and `strdup` the value for the
plugin API.
Fixes: https://github.com/profanity-im/profanity/issues/2013
Feature discovery was marked as finished once we received a reply to the
initial request. The discovery mechanism allows to delegate the real
feature discovery to another service running on different domain and those
requests are created dynamically.
This was another instance causing the warning message described in #1940Fixes#1940 (once more)
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
Use a singleton `Jid` inside the connection instead of always re-creating
a `Jid` from the same string.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
There were multiple reports where after a reconnect the window of the
MUC that was last opened, was empty.
`muc_join()` creates an instance of a MUC, `presence_join_room()` works
with this instance. Therefore the instance has to exist before working on
it.
I'm not sure if this really fixes the issue, but at least it didn't
happen anymore after I applied this modification.
I can't remember how I stumbled over this, either while looking at debug
logs or while looking at Valgrind output while a reconnect happened, but
something went wrong. Then I came to the conclusion that this may fix
the issue and for now it did ... maybe it comes back, then my RCA was
wrong.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
Sometimes after a reconnect the current window would get cleared. This
was a deliberate change to fix the profanity window looking all garbled
up after providing the passphrase for a gpg key using pinentry-curses.
Fixes https://github.com/profanity-im/profanity/issues/1556
I think this PR already solves and improves the MAM situation a lot.
What's @MarcoPolo-PasTonMolo still wanted to do in this branch is:
* MAM for mucs
* Check if url and quotes autocompletion works fine
* Check if the api still works fine
* Resolve conflicts
Conflicts are solved with this commit.
MAM for mucs can be another feature PR.
The rest we can check while being on master. And more people can help
testing.
Only nicknames, photos, birthdays, addresses, telephone numbers, emails,
JIDs, titles, roles, notes, and URLs are supported
Due to the synopsis array not having enough space, `/vcard photo
open-self` and `/vcard photo save-self` are not documented properly in
the synopsis section of the `/vcard` command, but they are documented in
the arguments section
Fixed memory leak in vcard autocomplete (thanks to debXwoody)
The profanity-internal mechanism to allow connecting to a server isn't
easily portable to cURL. Therefor introduce a profanity-specific CAfile
which is managed individually and will be configured in libcurl calls.
Signed-off-by: Steffen Jaeckel <jaeckel-floss@eyet-services.de>
- Added JABBER_RAW_CONNECT[ING/ED] connection states
- Added cl_ev_connect_raw and session_connect_raw to conform to normal
connection functions
- Fixed SIGABRT during registration
- Added a check in cmd_register to ensure it's actually connected before
registering--but this will always fail atm
