big-rag

const HTML_EXTENSIONS = [".htm", ".html", ".xhtml"];
const XML_EXTENSIONS = [".xml"];
const FB2_EXTENSIONS = [".fb2"];
const BSL_EXTENSIONS = [".bsl"];
const MARKDOWN_EXTENSIONS = [".md", ".markdown", ".mdown", ".mdx", ".mkd", ".mkdn"];
const TEXT_EXTENSIONS = [".txt", ".text"];
const PDF_EXTENSIONS = [".pdf"];
const EPUB_EXTENSIONS = [".epub"];
const IMAGE_EXTENSIONS = [".bmp", ".jpg", ".jpeg", ".png"];
const ARCHIVE_EXTENSIONS = [".rar"];

const ALL_EXTENSION_GROUPS = [
  HTML_EXTENSIONS,
  XML_EXTENSIONS,
  FB2_EXTENSIONS,
  BSL_EXTENSIONS,
  MARKDOWN_EXTENSIONS,
  TEXT_EXTENSIONS,
  PDF_EXTENSIONS,
  EPUB_EXTENSIONS,
  IMAGE_EXTENSIONS,
  ARCHIVE_EXTENSIONS,
];

export const SUPPORTED_EXTENSIONS = new Set(
  ALL_EXTENSION_GROUPS.flatMap((group) => group.map((ext) => ext.toLowerCase())),
);

export const HTML_EXTENSION_SET = new Set(HTML_EXTENSIONS);
export const XML_EXTENSION_SET = new Set(XML_EXTENSIONS);
export const FB2_EXTENSION_SET = new Set(FB2_EXTENSIONS);
export const BSL_EXTENSION_SET = new Set(BSL_EXTENSIONS);
export const MARKDOWN_EXTENSION_SET = new Set(MARKDOWN_EXTENSIONS);
export const TEXT_EXTENSION_SET = new Set(TEXT_EXTENSIONS);
export const IMAGE_EXTENSION_SET = new Set(IMAGE_EXTENSIONS);

export function isHtmlExtension(ext: string): boolean {
  return HTML_EXTENSION_SET.has(ext.toLowerCase());
}

export function isXmlExtension(ext: string): boolean {
  return XML_EXTENSION_SET.has(ext.toLowerCase());
}

export function isFb2Extension(ext: string): boolean {
  return FB2_EXTENSION_SET.has(ext.toLowerCase());
}

export function isBslExtension(ext: string): boolean {
  return BSL_EXTENSION_SET.has(ext.toLowerCase());
}

export function isMarkdownExtension(ext: string): boolean {
  return MARKDOWN_EXTENSION_SET.has(ext.toLowerCase());
}

export function isPlainTextExtension(ext: string): boolean {
  return TEXT_EXTENSION_SET.has(ext.toLowerCase());
}

export function isTextualExtension(ext: string): boolean {
  return isMarkdownExtension(ext) || isPlainTextExtension(ext);
}

export function listSupportedExtensions(): string[] {
  return Array.from(SUPPORTED_EXTENSIONS.values()).sort();
}

/**
 * Parse user-provided comma/space/newline separated list of file extensions
 * Returns Set of extensions with dots, or null if input is empty (all types enabled)
 * Example: "pdf,epub,txt" -> Set { ".pdf", ".epub", ".txt" }
 */
export function parseEnabledExtensions(raw: string | null | undefined): Set<string> | null {
  if (!raw || raw.trim().length === 0) {
    return null; // null means "all supported types"
  }
  const set = new Set<string>();
  for (const token of raw.split(/[,;\s\n]+/)) {
    const trimmed = token.trim().toLowerCase();
    if (trimmed.length === 0) continue;
    // Add dot if missing
    const ext = trimmed.startsWith(".") ? trimmed : `.${trimmed}`;
    if (SUPPORTED_EXTENSIONS.has(ext)) {
      set.add(ext);
    } else {
      console.warn(`[Config] Unknown extension "${ext}" in enabledFileTypes, ignoring.`);
    }
  }
  return set.size > 0 ? set : null;
}