fix: tighten telegram html chunking

src/telegram/format.test.ts

@@ -113,11 +113,19 @@ describe("markdownToTelegramHtml", () => {
     expect(res).toContain("trailing ||");
   });
 
-  it("splits long html text without breaking balanced tags", () => {
-    const chunks = splitTelegramHtmlChunks(`<b>${"A".repeat(5000)}</b>`, 4000);
+  it("splits long multiline html text without breaking balanced tags", () => {
+    const chunks = splitTelegramHtmlChunks(`<b>${"A\n".repeat(2500)}</b>`, 4000);
     expect(chunks.length).toBeGreaterThan(1);
     expect(chunks.every((chunk) => chunk.length <= 4000)).toBe(true);
-    expect(chunks[0]).toMatch(/^<b>.*<\/b>$/);
-    expect(chunks[1]).toMatch(/^<b>.*<\/b>$/);
+    expect(chunks[0]).toMatch(/^<b>[\s\S]*<\/b>$/);
+    expect(chunks[1]).toMatch(/^<b>[\s\S]*<\/b>$/);
+  });
+
+  it("fails loudly when a leading entity cannot fit inside a chunk", () => {
+    expect(() => splitTelegramHtmlChunks(`A&amp;${"B".repeat(20)}`, 4)).toThrow(/leading entity/i);
+  });
+
+  it("fails loudly when tag overhead leaves no room for text", () => {
+    expect(() => splitTelegramHtmlChunks("<b><i><u>x</u></i></b>", 10)).toThrow(/tag overhead/i);
   });
 });

src/telegram/format.ts

@@ -270,13 +270,15 @@ function findTelegramHtmlSafeSplitIndex(text: string, maxLength: number): number
     return text.length;
   }
   const normalizedMaxLength = Math.max(1, Math.floor(maxLength));
-  let splitAt = normalizedMaxLength;
   const lastAmpersand = text.lastIndexOf("&", normalizedMaxLength - 1);
-  const lastSemicolon = text.lastIndexOf(";", normalizedMaxLength - 1);
-  if (lastAmpersand > lastSemicolon) {
-    splitAt = lastAmpersand;
+  if (lastAmpersand === -1) {
+    return normalizedMaxLength;
   }
-  return splitAt > 0 ? splitAt : normalizedMaxLength;
+  const lastSemicolon = text.lastIndexOf(";", normalizedMaxLength - 1);
+  if (lastAmpersand < lastSemicolon) {
+    return normalizedMaxLength;
+  }
+  return lastAmpersand;
 }
 
 function popTelegramHtmlTag(tags: TelegramHtmlTag[], name: string): void {
@@ -300,15 +302,15 @@ export function splitTelegramHtmlChunks(html: string, limit: number): string[] {
   const chunks: string[] = [];
   const openTags: TelegramHtmlTag[] = [];
   let current = "";
-  let chunkHasContent = false;
+  let chunkHasPayload = false;
 
   const resetCurrent = () => {
     current = buildTelegramHtmlOpenPrefix(openTags);
-    chunkHasContent = false;
+    chunkHasPayload = false;
   };
 
   const flushCurrent = () => {
-    if (!chunkHasContent) {
+    if (!chunkHasPayload) {
       return;
     }
     chunks.push(`${current}${buildTelegramHtmlCloseSuffix(openTags)}`);
@@ -321,24 +323,31 @@ export function splitTelegramHtmlChunks(html: string, limit: number): string[] {
       const available =
         normalizedLimit - current.length - buildTelegramHtmlCloseSuffixLength(openTags);
       if (available <= 0) {
-        const prefix = buildTelegramHtmlOpenPrefix(openTags);
-        if (!chunkHasContent && current === prefix) {
-          current += remaining;
-          chunkHasContent = true;
-          remaining = "";
-          break;
+        if (!chunkHasPayload) {
+          throw new Error(
+            `Telegram HTML chunk limit exceeded by tag overhead (limit=${normalizedLimit})`,
+          );
         }
         flushCurrent();
         continue;
       }
       if (remaining.length <= available) {
         current += remaining;
-        chunkHasContent = true;
+        chunkHasPayload = true;
         break;
       }
       const splitAt = findTelegramHtmlSafeSplitIndex(remaining, available);
+      if (splitAt <= 0) {
+        if (!chunkHasPayload) {
+          throw new Error(
+            `Telegram HTML chunk limit exceeded by leading entity (limit=${normalizedLimit})`,
+          );
+        }
+        flushCurrent();
+        continue;
+      }
       current += remaining.slice(0, splitAt);
-      chunkHasContent = true;
+      chunkHasPayload = true;
       remaining = remaining.slice(splitAt);
       flushCurrent();
     }
@@ -363,7 +372,7 @@ export function splitTelegramHtmlChunks(html: string, limit: number): string[] {
     if (!isClosing) {
       const nextCloseLength = isSelfClosing ? 0 : `</${tagName}>`.length;
       if (
-        chunkHasContent &&
+        chunkHasPayload &&
         current.length +
           rawTag.length +
           buildTelegramHtmlCloseSuffixLength(openTags) +
@@ -375,7 +384,9 @@ export function splitTelegramHtmlChunks(html: string, limit: number): string[] {
     }
 
     current += rawTag;
-    chunkHasContent = true;
+    if (isSelfClosing) {
+      chunkHasPayload = true;
+    }
     if (isClosing) {
       popTelegramHtmlTag(openTags, tagName);
     } else if (!isSelfClosing) {

src/telegram/send.test.ts

@@ -1288,6 +1288,35 @@ describe("sendMessageTelegram", () => {
     });
     expect(res.messageId).toBe("91");
   });
+
+  it("preserves caller plain-text fallback across chunked html parse retries", async () => {
+    const chatId = "123";
+    const htmlText = `<b>${"A".repeat(5000)}</b>`;
+    const plainText = `${"P".repeat(2500)}${"Q".repeat(2500)}`;
+    const parseErr = new Error(
+      "400: Bad Request: can't parse entities: Can't find end of the entity starting at byte offset 9",
+    );
+    const sendMessage = vi
+      .fn()
+      .mockRejectedValueOnce(parseErr)
+      .mockResolvedValueOnce({ message_id: 90, chat: { id: chatId } })
+      .mockRejectedValueOnce(parseErr)
+      .mockResolvedValueOnce({ message_id: 91, chat: { id: chatId } });
+    const api = { sendMessage } as unknown as { sendMessage: typeof sendMessage };
+
+    const res = await sendMessageTelegram(chatId, htmlText, {
+      token: "tok",
+      api,
+      textMode: "html",
+      plainText,
+    });
+
+    expect(sendMessage).toHaveBeenCalledTimes(4);
+    const plainFallbackCalls = [sendMessage.mock.calls[1], sendMessage.mock.calls[3]];
+    expect(plainFallbackCalls.map((call) => String(call?.[1] ?? "")).join("")).toBe(plainText);
+    expect(plainFallbackCalls.every((call) => !String(call?.[1] ?? "").includes("<"))).toBe(true);
+    expect(res.messageId).toBe("91");
+  });
 });
 
 describe("reactMessageTelegram", () => {

src/telegram/send.ts

@@ -108,6 +108,36 @@ function resolveTelegramMessageIdOrThrow(
   throw new Error(`Telegram ${context} returned no message_id`);
 }
 
+function splitTelegramPlainTextFallback(text: string, chunkCount: number, limit: number): string[] {
+  if (!text) {
+    return [];
+  }
+  const normalizedLimit = Math.max(1, Math.floor(limit));
+  if (chunkCount <= 1 || text.length <= normalizedLimit) {
+    return [text];
+  }
+  if (text.length > chunkCount * normalizedLimit) {
+    const chunks: string[] = [];
+    for (let start = 0; start < text.length; start += normalizedLimit) {
+      chunks.push(text.slice(start, start + normalizedLimit));
+    }
+    return chunks;
+  }
+  const chunks: string[] = [];
+  let offset = 0;
+  for (let index = 0; index < chunkCount && offset < text.length; index += 1) {
+    const remainingChars = text.length - offset;
+    const remainingChunks = chunkCount - index;
+    const nextChunkLength =
+      remainingChunks === 1
+        ? remainingChars
+        : Math.min(normalizedLimit, Math.ceil(remainingChars / remainingChunks));
+    chunks.push(text.slice(offset, offset + nextChunkLength));
+    offset += nextChunkLength;
+  }
+  return chunks;
+}
+
 const PARSE_ERR_RE = /can't parse entities|parse entities|find end of the entity/i;
 const THREAD_NOT_FOUND_RE = /400:\s*Bad Request:\s*message thread not found/i;
 const MESSAGE_NOT_MODIFIED_RE =
@@ -660,10 +690,14 @@ export async function sendMessageTelegram(
     rawText: string,
     context: string,
   ): Promise<{ messageId: string; chatId: string }> => {
-    const chunks = splitTelegramHtmlChunks(rawText, 4000).map((chunk) => ({
+    const htmlChunks = splitTelegramHtmlChunks(rawText, 4000);
+    const plainTextChunks = opts.plainText
+      ? splitTelegramPlainTextFallback(opts.plainText, htmlChunks.length, 4000)
+      : [];
+    const chunks = htmlChunks.map((chunk, index) => ({
       rawText: chunk,
       htmlText: chunk,
-      plainText: chunk,
+      plainText: plainTextChunks[index],
     }));
 
     let lastMessageId = "";