user-docs

export type SourceKind = "file" | "https" | "github" | "huggingface" | "conversation" | "confluence";

export type RawContentType = "markdown" | "html" | "text";

export interface ImageRef {
  url: string;
  altText?: string;
  title?: string;
  sourceId: string;
  sourceKind: SourceKind;
  baseUrl?: string;
}

export interface SourceDocument {
  sourceId: string;
  sourceKind: SourceKind;
  canonicalUrl?: string;
  title?: string;
  rawContent: string;
  rawContentType: RawContentType;
  baseUrl?: string;
  fetchedAt: string;
  version?: string;
  metadata?: Record<string, unknown>;
}

export interface IndexedDocument {
  sourceId: string;
  sourceKind: SourceKind;
  canonicalUrl?: string;
  title?: string;
  content: string;
  contentHash: string;
  baseUrl?: string;
  version?: string;
  imageRefs: ImageRef[];
  metadata: Record<string, unknown>;
}

export interface SourceAdapterContext {
  fetchTimeoutMs: number;
  maxBytes: number;
  maxPages: number;
  githubToken?: string;
  huggingFaceToken?: string;
}

export interface SourceAdapter {
  canHandle(source: string): boolean;
  load(source: string, context: SourceAdapterContext): Promise<SourceDocument[]>;
}