import type { RichTextDocument } from "./ast.js";
import { DocBuilder, type ParseOptions } from "./builder.js";

// Single-pass, backtracking-safe token scan: URL | @mention | #hashtag.
const TOKENS =
  /(https?:\/\/[^\s<>"'`\u0000-\u001F]{1,2048})|(?<![\p{L}\p{N}_@])@([\p{L}\p{N}_.-]{1,64}(?:@[\p{L}\p{N}-]{1,63}(?:\.[\p{L}\p{N}-]{1,63}){0,8})?)|(?<![\p{L}\p{N}_#&])#([\p{L}\p{N}\p{M}_]{1,100})/gu;
const TRAILING_PUNCT = /[.,;:!?)\]}'"»”’]+$/;

function lineToNodes(builder: DocBuilder, line: string): void {
  let last = 0;
  for (const match of line.matchAll(TOKENS)) {
    const [whole, url, handle, tag] = match;
    let consumed = whole;
    if (match.index > last) builder.text(line.slice(last, match.index));
    if (url !== undefined) {
      const trimmed = url.replace(TRAILING_PUNCT, "");
      consumed = trimmed;
      builder.link(trimmed, builder.inlineText(trimmed.replace(/^https?:\/\//i, "")));
    } else if (handle !== undefined) {
      const clean = handle.replace(/[.-]+$/, "");
      consumed = "@" + clean;
      builder.mention(clean);
    } else if (tag !== undefined) {
      builder.hashtag(tag);
    }
    last = match.index + consumed.length;
  }
  if (last < line.length) builder.text(line.slice(last));
}

/** Plain provider text (Bluesky, Telegram, bios, …): newlines, URLs, mentions, hashtags. */
export function parsePlainText(input: unknown, options: ParseOptions = {}): RichTextDocument {
  const builder = new DocBuilder(options);
  if (typeof input !== "string") return builder.finish();
  const text = builder.clip(input).replace(/\r\n?/g, "\n");
  for (const paragraph of text.split(/\n{2,}/)) {
    paragraph.split("\n").forEach((line, i) => {
      if (i > 0) builder.lineBreak();
      lineToNodes(builder, line);
    });
    builder.endParagraph();
  }
  return builder.finish();
}