import { inflateSync } from 'node:zlib';

/**
 * The text a pdfkit document prints, for assertions in tests: inflates the
 * content streams and joins the strings of every text-showing operator
 * (hex strings of the standard fonts, one line per operator). Not a PDF
 * parser - it reads what pdfkit writes with the built-in Helvetica.
 */
export function pdfText(pdf: Buffer): string {
  const source = pdf.toString('latin1');
  const lines: string[] = [];
  const streams = source.matchAll(/stream\r?\n([\s\S]*?)\r?\nendstream/g);
  for (const [, raw] of streams) {
    let content: string;
    try {
      content = inflateSync(Buffer.from(raw, 'latin1')).toString('latin1');
    } catch {
      continue;
    }
    for (const operator of content.matchAll(
      /\[([^\]]*)\]\s*TJ|<([0-9a-fA-F]+)>\s*Tj/g,
    )) {
      const parts = operator[1] ?? `<${operator[2]}>`;
      const text = [...parts.matchAll(/<([0-9a-fA-F]*)>/g)]
        .map(([, hex]) => Buffer.from(hex, 'hex').toString('latin1'))
        .join('');
      lines.push(text);
    }
  }
  return lines.join('\n');
}