feat(cli): download voice notes, files, and images from chat messages

src/cli.ts

@@ -19,7 +19,7 @@ import {
   searchPeople,
   sendMessage,
 } from "./api.js";
-import { getMyMri, login, logout } from "./auth.js";
+import { getMyMri, getToken, login, logout } from "./auth.js";
 import { getChatClient, getSearchClient, getStreamClient } from "./client.js";
 import {
   DATA_DIR,
@@ -29,13 +29,18 @@ import {
 import {
   convertTranscriptToGrouped,
   convertTranscriptToVtt,
+  extractFile,
+  extractImage,
+  extractMedia,
+  type FileInfo,
   type FormattedChat,
   type FormattedMessage,
   formatChatList,
   formatMember,
   formatMessage,
   formatPerson,
   getConversationDisplayName,
+  type MediaInfo,
   parseRecordings,
   stripHtml,
   transcriptFilename,
@@ -166,6 +171,118 @@ export function createProgram(): Command {
       printMessages(msgs.reverse());
     });
 
+  // media
+  program
+    .command("media <chat> [id]")
+    .description(
+      "Download a voice note / media / image message (default: newest media message).",
+    )
+    .option("-o, --out <file>", "Output file")
+    .action(async (chat: string, id: string | undefined, opts) => {
+      const convId = resolveConversationId(chat);
+      const client = await getChatClient();
+      const raw = await getMessages(client, convId, 60);
+      let target: { msg: Record<string, unknown>; media: MediaInfo } | null =
+        null;
+      for (const m of raw) {
+        const media = extractMedia(m) ?? extractImage(m);
+        if (!media) continue;
+        if (id && m.id !== id) continue;
+        target = { msg: m, media };
+        break;
+      }
+      if (!target || (id && target.msg.id !== id)) {
+        console.error(
+          id ? `No media message with id ${id}.` : "No media messages found.",
+        );
+        process.exit(1);
+      }
+      const token = await getToken("ic3");
+      if (!token) {
+        console.error(
+          "No valid Teams chat token found. Run `teams login` to authenticate.",
+        );
+        process.exit(1);
+      }
+      const resp = await fetch(target.media.url, {
+        headers: { Authorization: `Bearer ${token}` },
+      });
+      if (!resp.ok) {
+        console.error(`Download failed: HTTP ${resp.status}`);
+        process.exit(1);
+      }
+      const buf = Buffer.from(await resp.arrayBuffer());
+      const ct = resp.headers.get("content-type") ?? "";
+      const ext = ct.includes("webm")
+        ? "webm"
+        : ct.includes("jpeg") || ct.includes("jpg")
+          ? "jpg"
+          : ct.includes("png")
+            ? "png"
+            : "mp4";
+      const out = opts.out || `teams-media-${target.msg.id}.${ext}`;
+      writeFileSync(out, buf);
+      console.log(
+        `Saved ${out} (${buf.length} bytes, ${target.media.duration}).`,
+      );
+    });
+
+  // file
+  program
+    .command("file <chat> [match]")
+    .description(
+      "Download a file attachment (match = file-name substring or message id; default: newest file).",
+    )
+    .option("-o, --out <file>", "Output file")
+    .action(async (chat: string, match: string | undefined, opts) => {
+      const convId = resolveConversationId(chat);
+      const client = await getChatClient();
+      const raw = await getMessages(client, convId, 60);
+      let target: { msg: Record<string, unknown>; file: FileInfo } | null =
+        null;
+      for (const m of raw) {
+        const file = extractFile(m);
+        if (!file) continue;
+        if (
+          match &&
+          !(
+            m.id === match ||
+            String(file.fileName)
+              .toLowerCase()
+              .includes(String(match).toLowerCase())
+          )
+        )
+          continue;
+        target = { msg: m, file };
+        break;
+      }
+      if (!target) {
+        console.error(
+          match ? `No file matching '${match}'.` : "No file messages found.",
+        );
+        process.exit(1);
+      }
+      console.log(`File: ${target.file.fileName}`);
+      if (!target.file.url) {
+        console.error("No download URL on this message.");
+        process.exit(1);
+      }
+      // tempauth links are pre-authed; they expire shortly after the message
+      // is sent - a 403 means the sender has to re-share the file.
+      const resp = await fetch(target.file.url, { redirect: "follow" });
+      if (!resp.ok) {
+        console.error(
+          `Download failed: HTTP ${resp.status}${resp.status === 403 ? " (tempauth link likely EXPIRED - ask the sender to re-send the file)" : ""}`,
+        );
+        process.exit(1);
+      }
+      const buf = Buffer.from(await resp.arrayBuffer());
+      const safe = String(target.file.fileName).replace(/[/\\?%*:|"<>]/g, "_");
+      const out = opts.out || safe;
+      writeFileSync(out, buf);
+      console.log(`Saved ${out} (${buf.length} bytes).`);
+    });
+
   // send
   program
     .command("send <chat> <message>")

src/formatting.ts

@@ -131,21 +131,133 @@ export interface FormattedMessage {
   id: string;
 }
 
+export interface MediaInfo {
+  url: string;
+  duration: string;
+  kind: string;
+}
+
+// Voice notes and other card-based media: the playable URL and duration live
+// in the message's `properties.cards` JSON payload.
+export function extractMedia(msg: Record<string, unknown>): MediaInfo | null {
+  try {
+    const props = (msg.properties as Record<string, string>) ?? {};
+    const cards = props.cards ? JSON.parse(props.cards) : [];
+    for (const c of cards as Record<string, unknown>[]) {
+      const content = c?.content as Record<string, unknown> | undefined;
+      const media = content?.media as Record<string, string>[] | undefined;
+      if (Array.isArray(media) && media[0]?.url) {
+        const dm = String(content?.duration ?? "").match(/PT(?:(\d+)M)?(\d+)S/);
+        const dur = dm ? `${dm[1] ? `${dm[1]}m` : ""}${dm[2] ?? "?"}s` : "?";
+        const kind = String(content?.contentType ?? "").includes("audio")
+          ? "voice note"
+          : "media";
+        return { url: media[0].url, duration: dur, kind };
+      }
+    }
+  } catch {
+    // malformed cards JSON - not a media message
+  }
+  return null;
+}
+
+// Inline images: the downloadable object URL sits in the raw HTML `<img src>`
+// of the message content (e.g. .../v1/objects/<ref>/views/imgo).
+export function extractImage(msg: Record<string, unknown>): MediaInfo | null {
+  try {
+    const html = (msg.content as string) ?? "";
+    const m = html.match(/<img[^>]+src="(https:\/\/[^"]+\/objects\/[^"]+)"/i);
+    if (!m) return null;
+    const url = m[1].replace(/&amp;/g, "&");
+    const dim = html.match(/<img[^>]+width="(\d+)"/i);
+    return { url, duration: dim ? `${dim[1]}w` : "?", kind: "image" };
+  } catch {
+    // fall through
+  }
+  return null;
+}
+
+export interface FileInfo {
+  fileName: string;
+  objectUrl: string;
+  url: string;
+}
+
+// File attachments: `properties.files` carries the name plus a pre-authed
+// tempauth download URL (botFileProperties.url) with the SharePoint objectUrl
+// as fallback.
+export function extractFile(msg: Record<string, unknown>): FileInfo | null {
+  try {
+    const props = (msg.properties as Record<string, string>) ?? {};
+    const files = props.files ? JSON.parse(props.files) : [];
+    for (const f of files as Record<string, unknown>[]) {
+      if (f?.fileName) {
+        const bot = f?.botFileProperties as Record<string, string> | undefined;
+        const objectUrl = (f.objectUrl as string) ?? "";
+        return {
+          fileName: f.fileName as string,
+          objectUrl,
+          url: bot?.url ?? objectUrl,
+        };
+      }
+    }
+  } catch {
+    // malformed files JSON - not a file message
+  }
+  return null;
+}
+
 export function formatMessage(
   msg: Record<string, unknown>,
 ): FormattedMessage | null {
   const msgType = (msg.messagetype as string) ?? "";
   if (!["RichText/Html", "Text", "RichText"].includes(msgType)) return null;
 
   const sender = (msg.imdisplayname as string) ?? "Unknown";
+  const time = formatTimestamp(msg.originalarrivaltime as string);
+  const id = (msg.id as string) ?? "";
   const content = stripHtml((msg.content as string) ?? "");
-  if (!content) return null;
+  if (!content) {
+    // Media/file/image-only messages carry no text - render a marker row
+    // instead of dropping the message entirely.
+    const media = extractMedia(msg);
+    if (media) {
+      return {
+        time,
+        sender,
+        body: `[${media.kind} ${media.duration} — media id ${id}]`,
+        id,
+      };
+    }
+    const file = extractFile(msg);
+    if (file) {
+      return {
+        time,
+        sender,
+        body: `[file: ${file.fileName} — file id ${id}]`,
+        id,
+      };
+    }
+    const image = extractImage(msg);
+    if (image) {
+      return {
+        time,
+        sender,
+        body: `[${image.kind} ${image.duration} — img id ${id}]`,
+        id,
+      };
+    }
+    return null;
+  }
 
+  const img = extractImage(msg);
   return {
-    time: formatTimestamp(msg.originalarrivaltime as string),
+    time,
     sender,
-    body: content,
-    id: (msg.id as string) ?? "",
+    body: img
+      ? `${content} [${img.kind} ${img.duration} — img id ${id}]`
+      : content,
+    id,
   };
 }