fix(telegram): require callback chat context by houko · Pull Request #6871 · librefang/librefang

houko · 2026-08-09T23:57:07Z

Summary

drop callback-query events that have no Telegram message/chat context
stop emitting an empty channel ID that can create a synthetic mis-routed session
reuse the required message context for group and message-ID metadata

Verification

cargo test --manifest-path sdk/rust/librefang-sidecar-telegram/Cargo.toml callback_event_requires_chat_context
cargo test --manifest-path sdk/rust/librefang-sidecar-telegram/Cargo.toml --all-targets
cargo clippy --manifest-path sdk/rust/librefang-sidecar-telegram/Cargo.toml --all-targets -- -D warnings
git diff --check

The explanatory comment about populating the canonical top-level id was accidentally moved into message_event during the callback_event fix, where it was self-referential (message_event already set message_id unconditionally before this PR and needed no such note). Move it back next to the .message_id(message_id) call in callback_event where it actually applies, and rewrap it one sentence per line per the repo's prose convention.

houko · 2026-08-10T10:46:28Z

⚠️ Compliance flag (not a code-quality note): commit ca89f729 on this branch is authored as Claude <noreply@anthropic.com>. CLAUDE.md's commit-msg hook exists specifically to reject a commit whose author identity resolves to Claude/Anthropic, and the repo's git conventions explicitly disallow AI attribution — this slipped through because core.hooksPath isn't set by default in a fresh clone, so the version-controlled hook never ran for whatever session made this commit. I can't fix this myself: rewriting the offending commit needs a rebase + force-push, which I'm not authorized to do unilaterally. Flagging for a maintainer to rebase with --reset-author (or squash before merge, which will attribute the squashed commit to whoever GitHub picks — typically the PR author).

Generated by Claude Code

houko

Automated CLAUDE.md compliance check: one changelog fragment formatting nit below.

Generated by Claude Code

houko · 2026-08-10T10:47:02Z

@@ -0,0 +1 @@
+- Dropped Telegram callback events without chat context instead of routing them into an empty synthetic channel. (@houko)


changelog.d/README.md format rule: the fragment body must NOT start with a leading - — cargo xtask collect-fragments prepends its own - when assembling ## [Unreleased], so this will render as - - Dropped.... Please drop the leading - . Also missing the trailing (#6871) reference before (@houko) (not tooling-enforced, but prevents a duplicate generated line in the release notes).

Generated by Claude Code

houko

Automated correctness/CLAUDE.md review pass on top of the earlier round (Claude-attribution and changelog-format findings from the prior pass are already addressed/flagged — not repeating them). Verified the fail-closed guard is consistent with existing code, both accept/reject paths are tested, and inline-mode callbacks are not a regression. Left one comment on a cross-SDK parity gap for a maintainer call — see inline. cargo check, cargo test -p librefang-sidecar-telegram, and cargo clippy -p librefang-sidecar-telegram --all-targets -- -D warnings all pass clean on this branch. No mechanical fixes applied — nothing met the ≥90%-confidence bar for a direct commit.

Generated by Claude Code

houko · 2026-08-10T13:33:05Z

 pub fn callback_event(cq: &CallbackQuery) -> Option<Value> {
    let user = cq.from.as_ref()?;
+    let message = cq.message.as_ref()?;
+    if message.chat.id == 0 {


Verified: this mirrors the existing msg.chat.id == 0 guard already established in message_event (line 331, same file, pre-existing), so the fail-closed behavior here is consistent with the rest of the translator rather than a new invention.
Both directions are covered by the new test: missing message returns None, and message present with a degenerate/zero chat.id also returns None — while the pre-existing callback_event_emits_message_id_as_a_string_in_both_slots test covers the accept path with a real chat.id.
Inline-mode callbacks (Telegram sends inline_message_id instead of message for those) aren't a regression here: CallbackQuery (api/types.rs) never modeled inline_message_id in the first place, so those updates already had message: None and were being mis-routed to an empty synthetic channel before this fix, not correctly handled. Dropping them now is strictly safer than the prior behavior, not a new gap.

Cross-SDK parity gap worth a maintainer decision, not a bug in this PR (file untouched by this diff): sdk/python/librefang/sidecar/adapters/telegram.py's _callback_to_event (~line 1344-1349) only checks if not message: return None and if not isinstance(chat_id, int): return None — it has no equivalent to the chat.id == 0 guard this PR adds on the Rust side. 0 passes isinstance(chat_id, int), so a callback with a degenerate zero chat id would still route to channel_id="0" in Python, reproducing the exact synthetic-channel bug this PR fixes in Rust.
This asymmetry isn't new or specific to callbacks, either — Python's plain-message path (_update_to_event, ~line 1481-1483) only checks chat_id is None, never chat_id == 0, so the general message_event-equivalent guard (translator.rs line 331) also has no Python counterpart. Real Telegram traffic never sends chat id 0, so this only matters for malformed/spoofed webhook payloads, which is presumably why it wasn't caught by existing parity tests.
Should the same zero-id guard land in the Python adapter (this PR or a fast-follow), or is Python intentionally out of scope here? Leaving as a question rather than a diff since it's a different SDK/language and outside this PR's touched files.

Generated by Claude Code

houko · 2026-08-10T17:27:48Z

@@ -531,13 +525,10 @@ pub fn callback_event(cq: &CallbackQuery) -> Option<Value> {
        .channel_id(chat_id)
        .platform("telegram")


Reviewed the core logic independently and it holds up: callback_event correctly fails closed on both a missing message and a degenerate chat.id == 0, consistent with the pre-existing guard in message_event (line 331); confirmed update_to_event (line 607) propagates the None straight through rather than substituting a fallback channel. cargo test -p librefang-sidecar-telegram --bins (63 passed), cargo clippy --bins --all-targets -- -D warnings, and cargo fmt --check all clean on current HEAD.

Separate from the logic: commit ca89f729c ("fix: relocate misplaced message_id builder comment to callback_event") on this branch has author identity Claude <noreply@anthropic.com> — same class of issue already flagged on PR #6868 (commit 8568134b4) and #6870 (commit aa6170be7). The commit's content is fine (a correct, harmless comment relocation), but the identity itself violates the repo's no-AI-attribution rule and will need the same rebase/re-author treatment before merge. Flagging rather than fixing since correcting it requires a history rewrite this session isn't able to force-push.

Generated by Claude Code

houko

Automated review pass: found an AI-authored commit (author identity) in this branch's history, flagged inline for maintainer judgment since fixing it needs a history rewrite I won't force-push. The functional change itself looks correct and well-tested.

Generated by Claude Code

houko · 2026-08-10T18:56:18Z

Flagging for maintainer judgment rather than editing directly (fixing requires rewriting history, which I won't force-push):

Commit ca89f729 ("fix: relocate misplaced message_id builder comment to callback_event") on this branch has author identity Claude <noreply@anthropic.com>. CLAUDE.md's "No AI / Claude attribution" section is explicit that a commit whose author identity resolves to Claude/Anthropic is disallowed even when the message text itself is clean — this is exactly what the repo's own commit-msg hook checks via GIT_AUTHOR_IDENT. I can't fix this without rewriting the branch's history and force-pushing, which I'm not permitted to do. Recommend rebasing to correct the author identity on this commit before merge.

The substantive change itself (drop callback events with no chat context, require message.chat.id != 0) looks correct and is covered by callback_event_requires_chat_context. No other findings.

Generated by Claude Code

houko

Automated pass on head d01434da1e1f7ebaea787d6750bb5789e0819aa6.

Logic: callback_event fails closed before any dispatch — cq.message.as_ref()? then chat.id == 0 guard both run before content/metadata are built, matching the pre-existing message_event pattern (no prior panic path existed here; the bug being fixed was silent misrouting to an empty synthetic channel, not a crash). update_to_event's unconditional answer_callback_query spawn only needs cq.id and doesn't bypass the guard. callback_event_requires_chat_context covers both the realistic case (no message, e.g. inline-mode callbacks) and the defensive chat.id == 0 case.

Changelog: changelog.d/fixed/6871-telegram-callback-channel-id.md format is already corrected (no leading - , trailing (#6871) (@houko)).

One outstanding item, consolidated below rather than repeated in full.

Generated by Claude Code

houko · 2026-08-10T23:59:07Z

Status (previously flagged 3x on this thread, consolidating rather than repeating): commit ca89f729c is still authored as Claude <noreply@anthropic.com>, unchanged since the last pass. Still blocking on a maintainer rebase (--reset-author or squash-on-merge) before this can land; no new detail to add.

Generated by Claude Code

fix(telegram): require callback chat context

c213331

houko force-pushed the fix/telegram-callback-channel-id branch from b82bd1f to c213331 Compare August 10, 2026 00:02

github-actions Bot added area/sdk JavaScript and Python SDKs size/M 50-249 lines changed labels Aug 10, 2026

houko and others added 3 commits August 10, 2026 13:21

Merge branch 'main' into fix/telegram-callback-channel-id

99e147a

Merge branch 'main' into fix/telegram-callback-channel-id

05ff732

houko commented Aug 10, 2026

View reviewed changes

docs: fix changelog fragment format for telegram callback fix

d01434d

houko commented Aug 10, 2026

View reviewed changes

This was referenced Aug 10, 2026

fix(telegram): reject malformed media groups #6870

Merged

fix(telegram): scale multipart timeouts #6868

Merged

fix(telegram): honor retry-after headers #6866

Merged

houko commented Aug 10, 2026

View reviewed changes

houko enabled auto-merge (squash) August 12, 2026 00:40

houko merged commit 547483f into main Aug 12, 2026
39 checks passed

houko deleted the fix/telegram-callback-channel-id branch August 12, 2026 00:40

houko mentioned this pull request Aug 18, 2026

release: v2026.8.19 #7710

Merged

Navigation Menu

Repository navigation

fix(telegram): require callback chat context - #6871

fix(telegram): require callback chat context#6871
houko merged 5 commits into
mainfrom
fix/telegram-callback-channel-id

houko commented Aug 9, 2026

Uh oh!

houko commented Aug 10, 2026

Uh oh!

houko left a comment

Uh oh!

houko Aug 10, 2026

Uh oh!

houko left a comment

Uh oh!

houko Aug 10, 2026

Uh oh!

houko Aug 10, 2026

Uh oh!

houko left a comment

Uh oh!

houko Aug 10, 2026

Uh oh!

houko left a comment

Uh oh!

houko Aug 10, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		@@ -0,0 +1 @@
		- Dropped Telegram callback events without chat context instead of routing them into an empty synthetic channel. (@houko)

		@@ -531,13 +525,10 @@ pub fn callback_event(cq: &CallbackQuery) -> Option<Value> {
		.channel_id(chat_id)
		.platform("telegram")

Conversation

houko commented Aug 9, 2026

Summary

Verification

Uh oh!

houko commented Aug 10, 2026

Uh oh!

houko left a comment

Choose a reason for hiding this comment

Uh oh!

houko Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

houko left a comment

Choose a reason for hiding this comment

Uh oh!

houko Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

houko Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

houko left a comment

Choose a reason for hiding this comment

Uh oh!

houko Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

houko left a comment

Choose a reason for hiding this comment

Uh oh!

houko Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants