import { htmlAttributes } from './html-attributes';

export interface IndexedHtmlElement {
  tagName: string;
  start: number;
  bodyStart: number;
  selfClosing: boolean;
}

/** Import-local index over already protected/sanitized lexical markup. No DOM
 * or persistent cache: offsets belong only to this exact immutable string. */
export function indexHtmlElements(markup: string) {
  const byId = new Map<string, IndexedHtmlElement>();
  const namedAnchors = new Map<string, IndexedHtmlElement>();
  const definitions = new Map<string, { count: number; start: number }>();
  const pattern = /<([a-zA-Z][\w:-]*)\b([^>]*)>/g;
  let match: RegExpExecArray | null;
  while ((match = pattern.exec(markup))) {
    let id: string | undefined;
    let name: string | undefined;
    for (const attribute of htmlAttributes(match[2] ?? '')) {
      if (attribute.name === 'id' && id === undefined) id = attribute.value;
      if (attribute.name === 'name' && name === undefined) name = attribute.value;
    }
    if (!id && !name) continue;
    const element = { tagName: match[1], start: match.index, bodyStart: pattern.lastIndex, selfClosing: /\/\s*>$/.test(match[0]) };
    if (id && !byId.has(id)) byId.set(id, element);
    if (name && element.tagName.toLowerCase() === 'a' && !namedAnchors.has(name)) namedAnchors.set(name, element);
    for (const target of new Set([id, name])) {
      if (!target) continue;
      const previous = definitions.get(target);
      if (previous) previous.count++;
      else definitions.set(target, { count: 1, start: element.start });
    }
  }
  return { byId, namedAnchors, definitions };
}

export type HtmlElementIndex = ReturnType<typeof indexHtmlElements>;