loudmouth-ai 0.1.0 → 0.2.0

package/extensions/msteams/src/attachments/shared.ts

@@ -1,235 +0,0 @@
-import type { MSTeamsAttachmentLike } from "./types.js";
-
-type InlineImageCandidate =
-  | {
-      kind: "data";
-      data: Buffer;
-      contentType?: string;
-      placeholder: string;
-    }
-  | {
-      kind: "url";
-      url: string;
-      contentType?: string;
-      fileHint?: string;
-      placeholder: string;
-    };
-
-export const IMAGE_EXT_RE = /\.(avif|bmp|gif|heic|heif|jpe?g|png|tiff?|webp)$/i;
-
-export const IMG_SRC_RE = /<img[^>]+src=["']([^"']+)["'][^>]*>/gi;
-export const ATTACHMENT_TAG_RE = /<attachment[^>]+id=["']([^"']+)["'][^>]*>/gi;
-
-export const DEFAULT_MEDIA_HOST_ALLOWLIST = [
-  "graph.microsoft.com",
-  "graph.microsoft.us",
-  "graph.microsoft.de",
-  "graph.microsoft.cn",
-  "sharepoint.com",
-  "sharepoint.us",
-  "sharepoint.de",
-  "sharepoint.cn",
-  "sharepoint-df.com",
-  "1drv.ms",
-  "onedrive.com",
-  "teams.microsoft.com",
-  "teams.cdn.office.net",
-  "statics.teams.cdn.office.net",
-  "office.com",
-  "office.net",
-  // Azure Media Services / Skype CDN for clipboard-pasted images
-  "asm.skype.com",
-  "ams.skype.com",
-  "media.ams.skype.com",
-  // Bot Framework attachment URLs
-  "trafficmanager.net",
-  "blob.core.windows.net",
-  "azureedge.net",
-  "microsoft.com",
-] as const;
-
-export const GRAPH_ROOT = "https://graph.microsoft.com/v1.0";
-
-export function isRecord(value: unknown): value is Record<string, unknown> {
-  return Boolean(value) && typeof value === "object" && !Array.isArray(value);
-}
-
-export function normalizeContentType(value: unknown): string | undefined {
-  if (typeof value !== "string") return undefined;
-  const trimmed = value.trim();
-  return trimmed ? trimmed : undefined;
-}
-
-export function inferPlaceholder(params: {
-  contentType?: string;
-  fileName?: string;
-  fileType?: string;
-}): string {
-  const mime = params.contentType?.toLowerCase() ?? "";
-  const name = params.fileName?.toLowerCase() ?? "";
-  const fileType = params.fileType?.toLowerCase() ?? "";
-
-  const looksLikeImage =
-    mime.startsWith("image/") || IMAGE_EXT_RE.test(name) || IMAGE_EXT_RE.test(`x.${fileType}`);
-
-  return looksLikeImage ? "<media:image>" : "<media:document>";
-}
-
-export function isLikelyImageAttachment(att: MSTeamsAttachmentLike): boolean {
-  const contentType = normalizeContentType(att.contentType) ?? "";
-  const name = typeof att.name === "string" ? att.name : "";
-  if (contentType.startsWith("image/")) return true;
-  if (IMAGE_EXT_RE.test(name)) return true;
-
-  if (
-    contentType === "application/vnd.microsoft.teams.file.download.info" &&
-    isRecord(att.content)
-  ) {
-    const fileType = typeof att.content.fileType === "string" ? att.content.fileType : "";
-    if (fileType && IMAGE_EXT_RE.test(`x.${fileType}`)) return true;
-    const fileName = typeof att.content.fileName === "string" ? att.content.fileName : "";
-    if (fileName && IMAGE_EXT_RE.test(fileName)) return true;
-  }
-
-  return false;
-}
-
-/**
- * Returns true if the attachment can be downloaded (any file type).
- * Used when downloading all files, not just images.
- */
-export function isDownloadableAttachment(att: MSTeamsAttachmentLike): boolean {
-  const contentType = normalizeContentType(att.contentType) ?? "";
-
-  // Teams file download info always has a downloadUrl
-  if (
-    contentType === "application/vnd.microsoft.teams.file.download.info" &&
-    isRecord(att.content) &&
-    typeof att.content.downloadUrl === "string"
-  ) {
-    return true;
-  }
-
-  // Any attachment with a contentUrl can be downloaded
-  if (typeof att.contentUrl === "string" && att.contentUrl.trim()) {
-    return true;
-  }
-
-  return false;
-}
-
-function isHtmlAttachment(att: MSTeamsAttachmentLike): boolean {
-  const contentType = normalizeContentType(att.contentType) ?? "";
-  return contentType.startsWith("text/html");
-}
-
-export function extractHtmlFromAttachment(att: MSTeamsAttachmentLike): string | undefined {
-  if (!isHtmlAttachment(att)) return undefined;
-  if (typeof att.content === "string") return att.content;
-  if (!isRecord(att.content)) return undefined;
-  const text =
-    typeof att.content.text === "string"
-      ? att.content.text
-      : typeof att.content.body === "string"
-        ? att.content.body
-        : typeof att.content.content === "string"
-          ? att.content.content
-          : undefined;
-  return text;
-}
-
-function decodeDataImage(src: string): InlineImageCandidate | null {
-  const match = /^data:(image\/[a-z0-9.+-]+)?(;base64)?,(.*)$/i.exec(src);
-  if (!match) return null;
-  const contentType = match[1]?.toLowerCase();
-  const isBase64 = Boolean(match[2]);
-  if (!isBase64) return null;
-  const payload = match[3] ?? "";
-  if (!payload) return null;
-  try {
-    const data = Buffer.from(payload, "base64");
-    return { kind: "data", data, contentType, placeholder: "<media:image>" };
-  } catch {
-    return null;
-  }
-}
-
-function fileHintFromUrl(src: string): string | undefined {
-  try {
-    const url = new URL(src);
-    const name = url.pathname.split("/").pop();
-    return name || undefined;
-  } catch {
-    return undefined;
-  }
-}
-
-export function extractInlineImageCandidates(
-  attachments: MSTeamsAttachmentLike[],
-): InlineImageCandidate[] {
-  const out: InlineImageCandidate[] = [];
-  for (const att of attachments) {
-    const html = extractHtmlFromAttachment(att);
-    if (!html) continue;
-    IMG_SRC_RE.lastIndex = 0;
-    let match: RegExpExecArray | null = IMG_SRC_RE.exec(html);
-    while (match) {
-      const src = match[1]?.trim();
-      if (src && !src.startsWith("cid:")) {
-        if (src.startsWith("data:")) {
-          const decoded = decodeDataImage(src);
-          if (decoded) out.push(decoded);
-        } else {
-          out.push({
-            kind: "url",
-            url: src,
-            fileHint: fileHintFromUrl(src),
-            placeholder: "<media:image>",
-          });
-        }
-      }
-      match = IMG_SRC_RE.exec(html);
-    }
-  }
-  return out;
-}
-
-export function safeHostForUrl(url: string): string {
-  try {
-    return new URL(url).hostname.toLowerCase();
-  } catch {
-    return "invalid-url";
-  }
-}
-
-function normalizeAllowHost(value: string): string {
-  const trimmed = value.trim().toLowerCase();
-  if (!trimmed) return "";
-  if (trimmed === "*") return "*";
-  return trimmed.replace(/^\*\.?/, "");
-}
-
-export function resolveAllowedHosts(input?: string[]): string[] {
-  if (!Array.isArray(input) || input.length === 0) {
-    return DEFAULT_MEDIA_HOST_ALLOWLIST.slice();
-  }
-  const normalized = input.map(normalizeAllowHost).filter(Boolean);
-  if (normalized.includes("*")) return ["*"];
-  return normalized;
-}
-
-function isHostAllowed(host: string, allowlist: string[]): boolean {
-  if (allowlist.includes("*")) return true;
-  const normalized = host.toLowerCase();
-  return allowlist.some((entry) => normalized === entry || normalized.endsWith(`.${entry}`));
-}
-
-export function isUrlAllowed(url: string, allowlist: string[]): boolean {
-  try {
-    const parsed = new URL(url);
-    if (parsed.protocol !== "https:") return false;
-    return isHostAllowed(parsed.hostname, allowlist);
-  } catch {
-    return false;
-  }
-}

package/extensions/msteams/src/attachments/types.ts

@@ -1,37 +0,0 @@
-export type MSTeamsAttachmentLike = {
-  contentType?: string | null;
-  contentUrl?: string | null;
-  name?: string | null;
-  thumbnailUrl?: string | null;
-  content?: unknown;
-};
-
-export type MSTeamsAccessTokenProvider = {
-  getAccessToken: (scope: string) => Promise<string>;
-};
-
-export type MSTeamsInboundMedia = {
-  path: string;
-  contentType?: string;
-  placeholder: string;
-};
-
-export type MSTeamsHtmlAttachmentSummary = {
-  htmlAttachments: number;
-  imgTags: number;
-  dataImages: number;
-  cidImages: number;
-  srcHosts: string[];
-  attachmentTags: number;
-  attachmentIds: string[];
-};
-
-export type MSTeamsGraphMediaResult = {
-  media: MSTeamsInboundMedia[];
-  hostedCount?: number;
-  attachmentCount?: number;
-  hostedStatus?: number;
-  attachmentStatus?: number;
-  messageUrl?: string;
-  tokenError?: boolean;
-};

package/extensions/msteams/src/attachments.test.ts

@@ -1,424 +0,0 @@
-import { beforeEach, describe, expect, it, vi } from "vitest";
-
-import type { PluginRuntime } from "clawdbot/plugin-sdk";
-import { setMSTeamsRuntime } from "./runtime.js";
-
-const detectMimeMock = vi.fn(async () => "image/png");
-const saveMediaBufferMock = vi.fn(async () => ({
-  path: "/tmp/saved.png",
-  contentType: "image/png",
-}));
-
-const runtimeStub = {
-  media: {
-    detectMime: (...args: unknown[]) => detectMimeMock(...args),
-  },
-  channel: {
-    media: {
-      saveMediaBuffer: (...args: unknown[]) => saveMediaBufferMock(...args),
-    },
-  },
-} as unknown as PluginRuntime;
-
-describe("msteams attachments", () => {
-  const load = async () => {
-    return await import("./attachments.js");
-  };
-
-  beforeEach(() => {
-    detectMimeMock.mockClear();
-    saveMediaBufferMock.mockClear();
-    setMSTeamsRuntime(runtimeStub);
-  });
-
-  describe("buildMSTeamsAttachmentPlaceholder", () => {
-    it("returns empty string when no attachments", async () => {
-      const { buildMSTeamsAttachmentPlaceholder } = await load();
-      expect(buildMSTeamsAttachmentPlaceholder(undefined)).toBe("");
-      expect(buildMSTeamsAttachmentPlaceholder([])).toBe("");
-    });
-
-    it("returns image placeholder for image attachments", async () => {
-      const { buildMSTeamsAttachmentPlaceholder } = await load();
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          { contentType: "image/png", contentUrl: "https://x/img.png" },
-        ]),
-      ).toBe("<media:image>");
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          { contentType: "image/png", contentUrl: "https://x/1.png" },
-          { contentType: "image/jpeg", contentUrl: "https://x/2.jpg" },
-        ]),
-      ).toBe("<media:image> (2 images)");
-    });
-
-    it("treats Teams file.download.info image attachments as images", async () => {
-      const { buildMSTeamsAttachmentPlaceholder } = await load();
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          {
-            contentType: "application/vnd.microsoft.teams.file.download.info",
-            content: { downloadUrl: "https://x/dl", fileType: "png" },
-          },
-        ]),
-      ).toBe("<media:image>");
-    });
-
-    it("returns document placeholder for non-image attachments", async () => {
-      const { buildMSTeamsAttachmentPlaceholder } = await load();
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          { contentType: "application/pdf", contentUrl: "https://x/x.pdf" },
-        ]),
-      ).toBe("<media:document>");
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          { contentType: "application/pdf", contentUrl: "https://x/1.pdf" },
-          { contentType: "application/pdf", contentUrl: "https://x/2.pdf" },
-        ]),
-      ).toBe("<media:document> (2 files)");
-    });
-
-    it("counts inline images in text/html attachments", async () => {
-      const { buildMSTeamsAttachmentPlaceholder } = await load();
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          {
-            contentType: "text/html",
-            content: '<p>hi</p><img src="https://x/a.png" />',
-          },
-        ]),
-      ).toBe("<media:image>");
-      expect(
-        buildMSTeamsAttachmentPlaceholder([
-          {
-            contentType: "text/html",
-            content: '<img src="https://x/a.png" /><img src="https://x/b.png" />',
-          },
-        ]),
-      ).toBe("<media:image> (2 images)");
-    });
-  });
-
-  describe("downloadMSTeamsAttachments", () => {
-    it("downloads and stores image contentUrl attachments", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn(async () => {
-        return new Response(Buffer.from("png"), {
-          status: 200,
-          headers: { "content-type": "image/png" },
-        });
-      });
-
-      const media = await downloadMSTeamsAttachments({
-        attachments: [{ contentType: "image/png", contentUrl: "https://x/img" }],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["x"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(fetchMock).toHaveBeenCalledWith("https://x/img");
-      expect(saveMediaBufferMock).toHaveBeenCalled();
-      expect(media).toHaveLength(1);
-      expect(media[0]?.path).toBe("/tmp/saved.png");
-    });
-
-    it("supports Teams file.download.info downloadUrl attachments", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn(async () => {
-        return new Response(Buffer.from("png"), {
-          status: 200,
-          headers: { "content-type": "image/png" },
-        });
-      });
-
-      const media = await downloadMSTeamsAttachments({
-        attachments: [
-          {
-            contentType: "application/vnd.microsoft.teams.file.download.info",
-            content: { downloadUrl: "https://x/dl", fileType: "png" },
-          },
-        ],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["x"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(fetchMock).toHaveBeenCalledWith("https://x/dl");
-      expect(media).toHaveLength(1);
-    });
-
-    it("downloads non-image file attachments (PDF)", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn(async () => {
-        return new Response(Buffer.from("pdf"), {
-          status: 200,
-          headers: { "content-type": "application/pdf" },
-        });
-      });
-      detectMimeMock.mockResolvedValueOnce("application/pdf");
-      saveMediaBufferMock.mockResolvedValueOnce({
-        path: "/tmp/saved.pdf",
-        contentType: "application/pdf",
-      });
-
-      const media = await downloadMSTeamsAttachments({
-        attachments: [{ contentType: "application/pdf", contentUrl: "https://x/doc.pdf" }],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["x"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(fetchMock).toHaveBeenCalledWith("https://x/doc.pdf");
-      expect(media).toHaveLength(1);
-      expect(media[0]?.path).toBe("/tmp/saved.pdf");
-      expect(media[0]?.placeholder).toBe("<media:document>");
-    });
-
-    it("downloads inline image URLs from html attachments", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn(async () => {
-        return new Response(Buffer.from("png"), {
-          status: 200,
-          headers: { "content-type": "image/png" },
-        });
-      });
-
-      const media = await downloadMSTeamsAttachments({
-        attachments: [
-          {
-            contentType: "text/html",
-            content: '<img src="https://x/inline.png" />',
-          },
-        ],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["x"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(media).toHaveLength(1);
-      expect(fetchMock).toHaveBeenCalledWith("https://x/inline.png");
-    });
-
-    it("stores inline data:image base64 payloads", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const base64 = Buffer.from("png").toString("base64");
-      const media = await downloadMSTeamsAttachments({
-        attachments: [
-          {
-            contentType: "text/html",
-            content: `<img src="data:image/png;base64,${base64}" />`,
-          },
-        ],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["x"],
-      });
-
-      expect(media).toHaveLength(1);
-      expect(saveMediaBufferMock).toHaveBeenCalled();
-    });
-
-    it("retries with auth when the first request is unauthorized", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn(async (_url: string, opts?: RequestInit) => {
-        const hasAuth = Boolean(
-          opts &&
-          typeof opts === "object" &&
-          "headers" in opts &&
-          (opts.headers as Record<string, string>)?.Authorization,
-        );
-        if (!hasAuth) {
-          return new Response("unauthorized", { status: 401 });
-        }
-        return new Response(Buffer.from("png"), {
-          status: 200,
-          headers: { "content-type": "image/png" },
-        });
-      });
-
-      const media = await downloadMSTeamsAttachments({
-        attachments: [{ contentType: "image/png", contentUrl: "https://x/img" }],
-        maxBytes: 1024 * 1024,
-        tokenProvider: { getAccessToken: vi.fn(async () => "token") },
-        allowHosts: ["x"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(fetchMock).toHaveBeenCalled();
-      expect(media).toHaveLength(1);
-      expect(fetchMock).toHaveBeenCalledTimes(2);
-    });
-
-    it("skips urls outside the allowlist", async () => {
-      const { downloadMSTeamsAttachments } = await load();
-      const fetchMock = vi.fn();
-      const media = await downloadMSTeamsAttachments({
-        attachments: [{ contentType: "image/png", contentUrl: "https://evil.test/img" }],
-        maxBytes: 1024 * 1024,
-        allowHosts: ["graph.microsoft.com"],
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(media).toHaveLength(0);
-      expect(fetchMock).not.toHaveBeenCalled();
-    });
-  });
-
-  describe("buildMSTeamsGraphMessageUrls", () => {
-    it("builds channel message urls", async () => {
-      const { buildMSTeamsGraphMessageUrls } = await load();
-      const urls = buildMSTeamsGraphMessageUrls({
-        conversationType: "channel",
-        conversationId: "19:thread@thread.tacv2",
-        messageId: "123",
-        channelData: { team: { id: "team-id" }, channel: { id: "chan-id" } },
-      });
-      expect(urls[0]).toContain("/teams/team-id/channels/chan-id/messages/123");
-    });
-
-    it("builds channel reply urls when replyToId is present", async () => {
-      const { buildMSTeamsGraphMessageUrls } = await load();
-      const urls = buildMSTeamsGraphMessageUrls({
-        conversationType: "channel",
-        messageId: "reply-id",
-        replyToId: "root-id",
-        channelData: { team: { id: "team-id" }, channel: { id: "chan-id" } },
-      });
-      expect(urls[0]).toContain(
-        "/teams/team-id/channels/chan-id/messages/root-id/replies/reply-id",
-      );
-    });
-
-    it("builds chat message urls", async () => {
-      const { buildMSTeamsGraphMessageUrls } = await load();
-      const urls = buildMSTeamsGraphMessageUrls({
-        conversationType: "groupChat",
-        conversationId: "19:chat@thread.v2",
-        messageId: "456",
-      });
-      expect(urls[0]).toContain("/chats/19%3Achat%40thread.v2/messages/456");
-    });
-  });
-
-  describe("downloadMSTeamsGraphMedia", () => {
-    it("downloads hostedContents images", async () => {
-      const { downloadMSTeamsGraphMedia } = await load();
-      const base64 = Buffer.from("png").toString("base64");
-      const fetchMock = vi.fn(async (url: string) => {
-        if (url.endsWith("/hostedContents")) {
-          return new Response(
-            JSON.stringify({
-              value: [
-                {
-                  id: "1",
-                  contentType: "image/png",
-                  contentBytes: base64,
-                },
-              ],
-            }),
-            { status: 200 },
-          );
-        }
-        if (url.endsWith("/attachments")) {
-          return new Response(JSON.stringify({ value: [] }), { status: 200 });
-        }
-        return new Response("not found", { status: 404 });
-      });
-
-      const media = await downloadMSTeamsGraphMedia({
-        messageUrl: "https://graph.microsoft.com/v1.0/chats/19%3Achat/messages/123",
-        tokenProvider: { getAccessToken: vi.fn(async () => "token") },
-        maxBytes: 1024 * 1024,
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(media.media).toHaveLength(1);
-      expect(fetchMock).toHaveBeenCalled();
-      expect(saveMediaBufferMock).toHaveBeenCalled();
-    });
-
-    it("merges SharePoint reference attachments with hosted content", async () => {
-      const { downloadMSTeamsGraphMedia } = await load();
-      const hostedBase64 = Buffer.from("png").toString("base64");
-      const shareUrl = "https://contoso.sharepoint.com/site/file";
-      const fetchMock = vi.fn(async (url: string) => {
-        if (url.endsWith("/hostedContents")) {
-          return new Response(
-            JSON.stringify({
-              value: [
-                {
-                  id: "hosted-1",
-                  contentType: "image/png",
-                  contentBytes: hostedBase64,
-                },
-              ],
-            }),
-            { status: 200 },
-          );
-        }
-        if (url.endsWith("/attachments")) {
-          return new Response(
-            JSON.stringify({
-              value: [
-                {
-                  id: "ref-1",
-                  contentType: "reference",
-                  contentUrl: shareUrl,
-                  name: "report.pdf",
-                },
-              ],
-            }),
-            { status: 200 },
-          );
-        }
-        if (url.startsWith("https://graph.microsoft.com/v1.0/shares/")) {
-          return new Response(Buffer.from("pdf"), {
-            status: 200,
-            headers: { "content-type": "application/pdf" },
-          });
-        }
-        if (url.endsWith("/messages/123")) {
-          return new Response(
-            JSON.stringify({
-              attachments: [
-                {
-                  id: "ref-1",
-                  contentType: "reference",
-                  contentUrl: shareUrl,
-                  name: "report.pdf",
-                },
-              ],
-            }),
-            { status: 200 },
-          );
-        }
-        return new Response("not found", { status: 404 });
-      });
-
-      const media = await downloadMSTeamsGraphMedia({
-        messageUrl: "https://graph.microsoft.com/v1.0/chats/19%3Achat/messages/123",
-        tokenProvider: { getAccessToken: vi.fn(async () => "token") },
-        maxBytes: 1024 * 1024,
-        fetchFn: fetchMock as unknown as typeof fetch,
-      });
-
-      expect(media.media).toHaveLength(2);
-    });
-  });
-
-  describe("buildMSTeamsMediaPayload", () => {
-    it("returns single and multi-file fields", async () => {
-      const { buildMSTeamsMediaPayload } = await load();
-      const payload = buildMSTeamsMediaPayload([
-        { path: "/tmp/a.png", contentType: "image/png" },
-        { path: "/tmp/b.png", contentType: "image/png" },
-      ]);
-      expect(payload.MediaPath).toBe("/tmp/a.png");
-      expect(payload.MediaUrl).toBe("/tmp/a.png");
-      expect(payload.MediaPaths).toEqual(["/tmp/a.png", "/tmp/b.png"]);
-      expect(payload.MediaUrls).toEqual(["/tmp/a.png", "/tmp/b.png"]);
-      expect(payload.MediaTypes).toEqual(["image/png", "image/png"]);
-    });
-  });
-});