From 2ed436ef0e4041757143dcb732b577e699a56098 Mon Sep 17 00:00:00 2001
From: yiekheng <yiekheng@04080616.xyz>
Date: Sat, 9 May 2026 17:23:59 +0800
Subject: [PATCH] feat(bot): add Telegram media ingest into /data/media

---
 apps/bot/src/media/ingest.ts | 89 ++++++++++++++++++++++++++++++++++++
 1 file changed, 89 insertions(+)
 create mode 100644 apps/bot/src/media/ingest.ts

diff --git a/apps/bot/src/media/ingest.ts b/apps/bot/src/media/ingest.ts
new file mode 100644
index 0000000..e78b38d
--- /dev/null
+++ b/apps/bot/src/media/ingest.ts
@@ -0,0 +1,89 @@
+import { mkdir, writeFile } from "node:fs/promises";
+import { dirname } from "node:path";
+import { createHash } from "node:crypto";
+import { mediaFiles } from "@cmbot/db";
+import { newMediaPath, absoluteMediaPath } from "@cmbot/shared";
+import { db } from "../db.js";
+import { env } from "../env.js";
+import { logger } from "../logger.js";
+
+export type IngestInput = {
+  operatorId: string;
+  filenameOriginal: string;
+  mimeType: string;
+  buffer: Buffer;
+};
+
+export type IngestResult = {
+  mediaId: string;
+  storagePath: string;
+};
+
+export async function ingestMediaBuffer(input: IngestInput): Promise<IngestResult> {
+  const sha256 = createHash("sha256").update(input.buffer).digest("hex");
+  const storagePath = newMediaPath(input.filenameOriginal);
+  const absolute = absoluteMediaPath(storagePath, env.MEDIA_DIR);
+  await mkdir(dirname(absolute), { recursive: true });
+  await writeFile(absolute, input.buffer);
+
+  const [row] = await db
+    .insert(mediaFiles)
+    .values({
+      operatorId: input.operatorId,
+      filenameOriginal: input.filenameOriginal,
+      mimeType: input.mimeType,
+      sizeBytes: input.buffer.byteLength,
+      sha256,
+      storagePath,
+    })
+    .returning({ id: mediaFiles.id });
+
+  logger.info(
+    { mediaId: row!.id, sizeBytes: input.buffer.byteLength, sha256 },
+    "media: ingested",
+  );
+
+  return { mediaId: row!.id, storagePath };
+}
+
+/**
+ * Download a Telegram file by file_id and ingest it. Returns the new media row.
+ */
+export async function ingestTelegramFile(
+  operatorId: string,
+  apiBase: string,
+  botToken: string,
+  fileId: string,
+  defaultFilename: string,
+  mimeType: string,
+): Promise<IngestResult> {
+  // 1. getFile — Telegram returns a file_path
+  const getFileUrl = `${apiBase}/bot${botToken}/getFile?file_id=${encodeURIComponent(fileId)}`;
+  const getFileRes = await fetch(getFileUrl);
+  if (!getFileRes.ok) {
+    throw new Error(`Telegram getFile failed: ${getFileRes.status} ${getFileRes.statusText}`);
+  }
+  const getFileJson = (await getFileRes.json()) as {
+    ok: boolean;
+    result?: { file_path?: string };
+  };
+  if (!getFileJson.ok || !getFileJson.result?.file_path) {
+    throw new Error("Telegram getFile: missing file_path in response");
+  }
+  // 2. Download bytes
+  const downloadUrl = `${apiBase}/file/bot${botToken}/${getFileJson.result.file_path}`;
+  const dl = await fetch(downloadUrl);
+  if (!dl.ok) {
+    throw new Error(`Telegram file download failed: ${dl.status} ${dl.statusText}`);
+  }
+  const buffer = Buffer.from(await dl.arrayBuffer());
+
+  // The Telegram-side filename can be missing; fall back to defaultFilename.
+  const filename = getFileJson.result.file_path.split("/").pop() ?? defaultFilename;
+  return ingestMediaBuffer({
+    operatorId,
+    filenameOriginal: filename,
+    mimeType,
+    buffer,
+  });
+}