ai-chatbot/lib/openclaw-chat/media-attachments.ts

export const OPENCLAW_CHAT_VIDEO_UPLOAD_PURPOSE = "openclaw-chat-video";
export const OPENCLAW_CHAT_FOLDER_UPLOAD_PURPOSE = "openclaw-chat-folder";
export const OPENCLAW_CHAT_FILE_UPLOAD_PURPOSE = "openclaw-chat-file";
export const OPENCLAW_CHAT_FILE_MAX_BYTES = 2 * 1024 * 1024 * 1024;
export const OPENCLAW_CHAT_VIDEO_MAX_FILES = 2;
export const OPENCLAW_CHAT_VIDEO_MAX_BYTES = OPENCLAW_CHAT_FILE_MAX_BYTES;
export const OPENCLAW_CHAT_FILE_MAX_FILES = 10;
export const OPENCLAW_CHAT_FOLDER_MAX_ARCHIVES = 5;
export const OPENCLAW_CHAT_FOLDER_MAX_FILES = 2000;
export const OPENCLAW_CHAT_FOLDER_MAX_TOTAL_BYTES = OPENCLAW_CHAT_FILE_MAX_BYTES;
export const OPENCLAW_CHAT_MEDIA_MAX_ATTACHMENTS = 16;
export const OPENCLAW_CHAT_VIDEO_ACCEPT =
  "video/mp4,video/quicktime,video/webm,video/x-m4v,video/x-matroska,.mp4,.mov,.webm,.m4v,.mkv";
export const OPENCLAW_CHAT_FILE_ACCEPT =
  ".zip,.rar,.7z,.tar,.gz,.tgz,.pdf,.txt,.md,.doc,.docx,.ppt,.pptx,.xls,.xlsx,.csv,.tsv,.json,.xml,.mp3,.wav,.m4a,.aac,.flac,.ogg,.psd,.ai,.dxf,.dwg,.step,.stp,application/zip,application/x-zip-compressed,application/vnd.rar,application/x-rar-compressed,application/x-7z-compressed,application/x-tar,application/gzip,application/pdf,application/msword,application/vnd.openxmlformats-officedocument.wordprocessingml.document,application/vnd.ms-powerpoint,application/vnd.openxmlformats-officedocument.presentationml.presentation,application/vnd.ms-excel,application/vnd.openxmlformats-officedocument.spreadsheetml.sheet,text/plain,text/csv,text/tab-separated-values,audio/mpeg,audio/wav,audio/mp4,audio/aac,audio/flac,audio/ogg";

const VIDEO_CONTENT_TYPES = new Set([
  "video/mp4",
  "video/quicktime",
  "video/webm",
  "video/x-m4v",
  "video/x-matroska",
]);

const VIDEO_CONTENT_TYPE_BY_EXTENSION: Record<string, string> = {
  m4v: "video/x-m4v",
  mkv: "video/x-matroska",
  mov: "video/quicktime",
  mp4: "video/mp4",
  webm: "video/webm",
};

const VIDEO_EXTENSION_PATTERN = /\.(mp4|mov|webm|m4v|mkv)(?:$|[?#])/i;
const ARCHIVE_EXTENSION_PATTERN = /\.zip(?:$|[?#])/i;
const OPENCLAW_FILE_EXTENSION_PATTERN =
  /\.(zip|rar|7z|tar|gz|tgz|pdf|txt|md|docx?|pptx?|xlsx?|csv|tsv|json|xml|mp3|wav|m4a|aac|flac|ogg|psd|ai|dxf|dwg|step|stp|folder\.json)(?:$|[?#])/i;
const BLOCKED_INLINE_FILE_EXTENSION_PATTERN =
  /\.(html?|xhtml|svg|js|mjs|cjs|exe|dll|msi|bat|cmd|com|scr|ps1|sh|jar|apk|dmg|pkg|deb|rpm)(?:$|[?#])/i;
const BLOCKED_INLINE_FILE_CONTENT_TYPES = new Set([
  "application/java-archive",
  "application/javascript",
  "application/vnd.android.package-archive",
  "application/x-apple-diskimage",
  "application/x-dosexec",
  "application/x-msdownload",
  "application/x-msi",
  "application/x-sh",
  "application/xhtml+xml",
  "image/svg+xml",
  "text/html",
  "text/javascript",
]);

const FILE_CONTENT_TYPE_BY_EXTENSION: Record<string, string> = {
  "7z": "application/x-7z-compressed",
  csv: "text/csv",
  gz: "application/gzip",
  rar: "application/vnd.rar",
  tar: "application/x-tar",
  tgz: "application/gzip",
  tsv: "text/tab-separated-values",
  xlsx: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
  zip: "application/zip",
};

const GENERIC_FILE_CONTENT_TYPES = new Set([
  "application/gzip",
  "application/octet-stream",
  "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
  "application/vnd.rar",
  "application/vnd.yedir.folder-manifest+json",
  "application/x-7z-compressed",
  "application/x-rar-compressed",
  "application/x-tar",
  "application/x-zip-compressed",
  "application/zip",
  "text/csv",
  "text/tab-separated-values",
]);

export type OpenClawChatMediaAttachment = {
  type: "image" | "video" | "file";
  url: string;
  name?: string;
  content_type?: string;
};

type OpenClawChatAttachmentLike = {
  type?: unknown;
  url?: unknown;
  name?: unknown;
  contentType?: unknown;
  content_type?: unknown;
};

type OpenClawChatVideoFileLike = {
  name: string;
  type: string;
};

function normalizeContentType(value: unknown) {
  return String(value || "")
    .split(";", 1)[0]
    .trim()
    .toLowerCase();
}

function getExtension(value: string) {
  const cleanValue = String(value || "").split(/[?#]/, 1)[0];
  return cleanValue.split(".").pop()?.trim().toLowerCase() || "";
}

function isHttpUrl(value: string) {
  try {
    const parsed = new URL(value);
    return parsed.protocol === "http:" || parsed.protocol === "https:";
  } catch {
    return false;
  }
}

export function getOpenClawChatVideoContentType(
  file: OpenClawChatVideoFileLike
) {
  const contentType = normalizeContentType(file.type);
  if (VIDEO_CONTENT_TYPES.has(contentType)) {
    return contentType;
  }
  return VIDEO_CONTENT_TYPE_BY_EXTENSION[getExtension(file.name)] || "";
}

export function isOpenClawChatVideoFile(file: OpenClawChatVideoFileLike) {
  return Boolean(getOpenClawChatVideoContentType(file));
}

export function isOpenClawChatVideoAttachment(
  url: string,
  contentType?: string | null
) {
  return (
    normalizeContentType(contentType).startsWith("video/") ||
    VIDEO_EXTENSION_PATTERN.test(String(url || ""))
  );
}

export function isOpenClawChatArchiveAttachment(
  url: string,
  contentType?: string | null
) {
  const normalizedContentType = normalizeContentType(contentType);
  return (
    normalizedContentType === "application/zip" ||
    normalizedContentType === "application/x-zip-compressed" ||
    ARCHIVE_EXTENSION_PATTERN.test(String(url || ""))
  );
}

export function getOpenClawChatFileContentType(file: {
  name: string;
  type?: string | null;
}) {
  const normalized = normalizeContentType(file.type);
  if (
    normalized.startsWith("image/") ||
    normalized.startsWith("video/") ||
    BLOCKED_INLINE_FILE_CONTENT_TYPES.has(normalized) ||
    BLOCKED_INLINE_FILE_EXTENSION_PATTERN.test(file.name)
  ) {
    return "";
  }
  if (
    normalized &&
    normalized !== "application/octet-stream" &&
    !BLOCKED_INLINE_FILE_CONTENT_TYPES.has(normalized)
  ) {
    return normalized;
  }
  if (file.name.toLowerCase().endsWith(".folder.json")) {
    return "application/vnd.yedir.folder-manifest+json";
  }
  return (
    FILE_CONTENT_TYPE_BY_EXTENSION[getExtension(file.name)] ||
    (OPENCLAW_FILE_EXTENSION_PATTERN.test(file.name)
      ? "application/octet-stream"
      : "")
  );
}

export function isOpenClawChatFile(file: {
  name: string;
  type?: string | null;
}) {
  return Boolean(getOpenClawChatFileContentType(file));
}

export function isOpenClawChatFolderFileAllowed(file: {
  name: string;
  type?: string | null;
}) {
  const normalized = normalizeContentType(file.type);
  return !(
    BLOCKED_INLINE_FILE_CONTENT_TYPES.has(normalized) ||
    BLOCKED_INLINE_FILE_EXTENSION_PATTERN.test(file.name)
  );
}

export function isOpenClawChatFileAttachment(
  url: string,
  contentType?: string | null
) {
  const normalizedContentType = normalizeContentType(contentType);
  return (
    GENERIC_FILE_CONTENT_TYPES.has(normalizedContentType) ||
    (Boolean(normalizedContentType) &&
      !normalizedContentType.startsWith("image/") &&
      !normalizedContentType.startsWith("video/") &&
      !BLOCKED_INLINE_FILE_CONTENT_TYPES.has(normalizedContentType)) ||
    OPENCLAW_FILE_EXTENSION_PATTERN.test(String(url || ""))
  );
}

export function normalizeOpenClawChatMediaAttachments(
  value: unknown
): OpenClawChatMediaAttachment[] {
  if (!Array.isArray(value)) {
    return [];
  }

  const result: OpenClawChatMediaAttachment[] = [];
  const seenUrls = new Set<string>();
  let videoCount = 0;
  let archiveCount = 0;

  for (const item of value) {
    if (!item || typeof item !== "object") {
      continue;
    }

    const raw = item as OpenClawChatAttachmentLike;
    const url = String(raw.url || "").trim();
    if (!url || !isHttpUrl(url) || seenUrls.has(url)) {
      continue;
    }

    const requestedType = String(raw.type || "").trim().toLowerCase();
    const rawContentType = normalizeContentType(
      raw.content_type || raw.contentType
    );
    const isVideo =
      requestedType === "video" ||
      isOpenClawChatVideoAttachment(url, rawContentType);
    const isFile =
      !isVideo &&
      (requestedType === "file" ||
        requestedType === "archive" ||
        isOpenClawChatFileAttachment(url, rawContentType));

    if (isVideo && videoCount >= OPENCLAW_CHAT_VIDEO_MAX_FILES) {
      continue;
    }
    if (isFile && archiveCount >= OPENCLAW_CHAT_FILE_MAX_FILES) {
      continue;
    }
    if (result.length >= OPENCLAW_CHAT_MEDIA_MAX_ATTACHMENTS) {
      break;
    }

    const name = String(raw.name || "")
      .trim()
      .slice(0, 200);
    const type = isVideo ? "video" : isFile ? "file" : "image";
    const fallbackContentType = isVideo
      ? VIDEO_CONTENT_TYPE_BY_EXTENSION[getExtension(url)] || "video/mp4"
      : isFile
        ? getOpenClawChatFileContentType({ name: name || url, type: rawContentType }) ||
          "application/octet-stream"
        : "image/jpeg";
    const contentType = isFile
      ? rawContentType || fallbackContentType
      : rawContentType.startsWith(`${type}/`)
        ? rawContentType
        : fallbackContentType;
    result.push({
      type,
      url,
      ...(name ? { name } : {}),
      content_type: contentType,
    });
    seenUrls.add(url);
    if (isVideo) {
      videoCount += 1;
    }
    if (isFile) {
      archiveCount += 1;
    }
  }

  return result;
}

export function buildOpenClawVideoDisplayText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const videos = attachments.filter((attachment) => attachment.type === "video");
  if (videos.length === 0) {
    return "";
  }

  const lines = videos.map(
    (attachment, index) => `- ${attachment.name || `视频${index + 1}`}`
  );
  return `已上传视频：\n${lines.join("\n")}`;
}

export function buildOpenClawVideoReferenceText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const videos = attachments.filter((attachment) => attachment.type === "video");
  if (videos.length === 0) {
    return "";
  }

  const lines = videos.map(
    (attachment, index) =>
      `- ${attachment.name || `视频${index + 1}`}: ${attachment.url}`
  );
  return [
    "用户上传了以下视频，请调用视频理解能力并按视频文件地址处理：",
    lines.join("\n"),
  ].join("\n");
}

export function buildOpenClawArchiveDisplayText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const archives = attachments.filter(
    (attachment) => attachment.type === "file"
  );
  if (archives.length === 0) {
    return "";
  }

  const lines = archives.map(
    (attachment, index) =>
      `- ${attachment.name || `文件夹压缩包${index + 1}.zip`}`
  );
  return `已上传文件夹：\n${lines.join("\n")}`;
}

export function buildOpenClawFileDisplayText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const files = attachments.filter((attachment) => attachment.type === "file");
  if (files.length === 0) {
    return "";
  }

  const lines = files.map(
    (attachment, index) => `- ${attachment.name || `文件附件${index + 1}`}`
  );
  return `已上传文件：\n${lines.join("\n")}`;
}

export function buildOpenClawArchiveReferenceText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const archives = attachments.filter(
    (attachment) => attachment.type === "file"
  );
  if (archives.length === 0) {
    return "";
  }

  const lines = archives.map(
    (attachment, index) =>
      `- ${attachment.name || `文件夹压缩包${index + 1}.zip`}: ${attachment.url}`
  );
  return [
    "用户上传了以下文件夹压缩包，压缩包保留原目录结构，请下载并解压后处理：",
    lines.join("\n"),
  ].join("\n");
}

export function buildOpenClawFileReferenceText(
  attachments: OpenClawChatMediaAttachment[]
) {
  const files = attachments.filter((attachment) => attachment.type === "file");
  if (files.length === 0) {
    return "";
  }

  const lines = files.map(
    (attachment, index) =>
      `- ${attachment.name || `文件附件${index + 1}`}: ${attachment.url} (${attachment.content_type || "application/octet-stream"})`
  );
  return [
    "用户上传了以下文件。表格请下载后读取，压缩包请下载并解压；application/vnd.yedir.folder-manifest+json 是保留目录结构和各文件下载地址的文件夹清单：",
    lines.join("\n"),
  ].join("\n");
}