fix(telegram): aggregate split inbound messages

This commit is contained in:
Peter Steinberger
2026-01-15 02:39:29 +00:00
parent e569f15631
commit aa74e28112
2 changed files with 196 additions and 0 deletions

View File

@@ -26,9 +26,23 @@ export const registerTelegramHandlers = ({
processMessage,
logger,
}) => {
const TELEGRAM_TEXT_FRAGMENT_START_THRESHOLD_CHARS = 4000;
const TELEGRAM_TEXT_FRAGMENT_MAX_GAP_MS = 1500;
const TELEGRAM_TEXT_FRAGMENT_MAX_ID_GAP = 1;
const TELEGRAM_TEXT_FRAGMENT_MAX_PARTS = 12;
const TELEGRAM_TEXT_FRAGMENT_MAX_TOTAL_CHARS = 50_000;
const mediaGroupBuffer = new Map<string, MediaGroupEntry>();
let mediaGroupProcessing: Promise<void> = Promise.resolve();
type TextFragmentEntry = {
key: string;
messages: Array<{ msg: TelegramMessage; ctx: unknown; receivedAtMs: number }>;
timer: ReturnType<typeof setTimeout>;
};
const textFragmentBuffer = new Map<string, TextFragmentEntry>();
let textFragmentProcessing: Promise<void> = Promise.resolve();
const processMediaGroup = async (entry: MediaGroupEntry) => {
try {
entry.messages.sort((a, b) => a.msg.message_id - b.msg.message_id);
@@ -51,6 +65,55 @@ export const registerTelegramHandlers = ({
}
};
const flushTextFragments = async (entry: TextFragmentEntry) => {
try {
entry.messages.sort((a, b) => a.msg.message_id - b.msg.message_id);
const first = entry.messages[0];
const last = entry.messages.at(-1);
if (!first || !last) return;
const combinedText = entry.messages.map((m) => m.msg.text ?? "").join("");
if (!combinedText.trim()) return;
const syntheticMessage: TelegramMessage = {
...first.msg,
text: combinedText,
caption: undefined,
caption_entities: undefined,
entities: undefined,
date: last.msg.date ?? first.msg.date,
};
const storeAllowFrom = await readTelegramAllowFromStore().catch(() => []);
const baseCtx = first.ctx as { me?: unknown; getFile?: unknown } & Record<string, unknown>;
const getFile =
typeof baseCtx.getFile === "function" ? baseCtx.getFile.bind(baseCtx) : async () => ({});
await processMessage(
{ message: syntheticMessage, me: baseCtx.me, getFile },
[],
storeAllowFrom,
{ messageIdOverride: String(last.msg.message_id) },
);
} catch (err) {
runtime.error?.(danger(`text fragment handler failed: ${String(err)}`));
}
};
const scheduleTextFragmentFlush = (entry: TextFragmentEntry) => {
clearTimeout(entry.timer);
entry.timer = setTimeout(async () => {
textFragmentBuffer.delete(entry.key);
textFragmentProcessing = textFragmentProcessing
.then(async () => {
await flushTextFragments(entry);
})
.catch(() => undefined);
await textFragmentProcessing;
}, TELEGRAM_TEXT_FRAGMENT_MAX_GAP_MS);
};
bot.on("callback_query", async (ctx) => {
const callback = ctx.callbackQuery;
if (!callback) return;
@@ -226,6 +289,68 @@ export const registerTelegramHandlers = ({
}
}
// Text fragment handling - Telegram splits long pastes into multiple inbound messages (~4096 chars).
// We buffer “near-limit” messages and append immediately-following parts.
const text = typeof msg.text === "string" ? msg.text : undefined;
const isCommandLike = (text ?? "").trim().startsWith("/");
if (text && !isCommandLike) {
const nowMs = Date.now();
const senderId = msg.from?.id != null ? String(msg.from.id) : "unknown";
const key = `text:${chatId}:${resolvedThreadId ?? "main"}:${senderId}`;
const existing = textFragmentBuffer.get(key);
if (existing) {
const last = existing.messages.at(-1);
const lastMsgId = last?.msg.message_id;
const lastReceivedAtMs = last?.receivedAtMs ?? nowMs;
const idGap = typeof lastMsgId === "number" ? msg.message_id - lastMsgId : Infinity;
const timeGapMs = nowMs - lastReceivedAtMs;
const canAppend =
idGap > 0 &&
idGap <= TELEGRAM_TEXT_FRAGMENT_MAX_ID_GAP &&
timeGapMs >= 0 &&
timeGapMs <= TELEGRAM_TEXT_FRAGMENT_MAX_GAP_MS;
if (canAppend) {
const currentTotalChars = existing.messages.reduce(
(sum, m) => sum + (m.msg.text?.length ?? 0),
0,
);
const nextTotalChars = currentTotalChars + text.length;
if (
existing.messages.length + 1 <= TELEGRAM_TEXT_FRAGMENT_MAX_PARTS &&
nextTotalChars <= TELEGRAM_TEXT_FRAGMENT_MAX_TOTAL_CHARS
) {
existing.messages.push({ msg, ctx, receivedAtMs: nowMs });
scheduleTextFragmentFlush(existing);
return;
}
}
// Not appendable (or limits exceeded): flush buffered entry first, then continue normally.
clearTimeout(existing.timer);
textFragmentBuffer.delete(key);
textFragmentProcessing = textFragmentProcessing
.then(async () => {
await flushTextFragments(existing);
})
.catch(() => undefined);
await textFragmentProcessing;
}
const shouldStart = text.length >= TELEGRAM_TEXT_FRAGMENT_START_THRESHOLD_CHARS;
if (shouldStart) {
const entry: TextFragmentEntry = {
key,
messages: [{ msg, ctx, receivedAtMs: nowMs }],
timer: setTimeout(() => {}, TELEGRAM_TEXT_FRAGMENT_MAX_GAP_MS),
};
textFragmentBuffer.set(key, entry);
scheduleTextFragmentFlush(entry);
return;
}
}
// Media group handling - buffer multi-image messages
const mediaGroupId = (msg as { media_group_id?: string }).media_group_id;
if (mediaGroupId) {

View File

@@ -409,3 +409,74 @@ describe("telegram media groups", () => {
MEDIA_GROUP_TEST_TIMEOUT_MS,
);
});
describe("telegram text fragments", () => {
beforeEach(() => {
// These tests rely on real setTimeout aggregation; guard against leaked fake timers.
vi.useRealTimers();
});
const TEXT_FRAGMENT_POLL_TIMEOUT_MS = process.platform === "win32" ? 30_000 : 15_000;
const TEXT_FRAGMENT_TEST_TIMEOUT_MS = process.platform === "win32" ? 45_000 : 20_000;
const waitForFragmentProcessing = async (
replySpy: ReturnType<typeof vi.fn>,
expectedCalls: number,
) => {
await expect
.poll(() => replySpy.mock.calls.length, { timeout: TEXT_FRAGMENT_POLL_TIMEOUT_MS })
.toBe(expectedCalls);
};
it(
"buffers near-limit text and processes sequential parts as one message",
async () => {
const { createTelegramBot } = await import("./bot.js");
const replyModule = await import("../auto-reply/reply.js");
const replySpy = replyModule.__replySpy as unknown as ReturnType<typeof vi.fn>;
onSpy.mockReset();
replySpy.mockReset();
createTelegramBot({ token: "tok" });
const handler = onSpy.mock.calls.find((call) => call[0] === "message")?.[1] as (
ctx: Record<string, unknown>,
) => Promise<void>;
expect(handler).toBeDefined();
const part1 = "A".repeat(4050);
const part2 = "B".repeat(50);
await handler({
message: {
chat: { id: 42, type: "private" },
message_id: 10,
date: 1736380800,
text: part1,
},
me: { username: "clawdbot_bot" },
getFile: async () => ({}),
});
await handler({
message: {
chat: { id: 42, type: "private" },
message_id: 11,
date: 1736380801,
text: part2,
},
me: { username: "clawdbot_bot" },
getFile: async () => ({}),
});
expect(replySpy).not.toHaveBeenCalled();
await waitForFragmentProcessing(replySpy, 1);
expect(replySpy).toHaveBeenCalledTimes(1);
const payload = replySpy.mock.calls[0][0] as { RawBody?: string; Body?: string };
expect(payload.RawBody).toContain(part1.slice(0, 32));
expect(payload.RawBody).toContain(part2.slice(0, 32));
},
TEXT_FRAGMENT_TEST_TIMEOUT_MS,
);
});