formulation/scratch/test-fig-bars.mjs

import fs from "node:fs";

const doc = JSON.parse(fs.readFileSync("file.json", "utf8"));

function isProcedureHeader(text, label) {
  if (label !== "section_header") return false;
  const s = text.trim();
  if (/^chef's notes?:?$/i.test(s)) return false;
  if (/^table of contents|foreword|introduction|acknowledgements|how to use/i.test(s)) return false;
  return /procedure|assembly|variations?|baking|shaping|finishing|glaz|infusion/i.test(s) || s.endsWith(":");
}

function cleanProcedureHeader(raw) {
  const t = raw.trim().replace(/^['"]+|['":]+$/g, "").trim();
  if (/^procedure$/i.test(t)) return "";
  const cleaned = t.replace(/\s+procedure$/i, "").trim();
  return cleaned ? `${cleaned}:` : "";
}

function testPages(pages, title) {
  const texts = doc.texts.filter(t => pages.includes(t.prov?.[0]?.page_no));
  console.log(`\nTesting on ${title} (pages ${pages.join("-")}):`);

  let inProcedure = false;
  const steps = [];

  for (const t of texts) {
    const text = t.text.trim();
    if (/^chef's notes?:?$/i.test(text)) {
      inProcedure = false;
      continue;
    }
    if (isProcedureHeader(text, t.label)) {
      inProcedure = true;
      const heading = cleanProcedureHeader(text);
      if (heading) {
        steps.push({ id: `step_${steps.length + 1}`, order: steps.length + 1, instruction: heading });
      }
      continue;
    }
    if (inProcedure && (t.label === "list_item" || t.label === "text") && !/^\d+$/.test(text) && !/^yields?:/i.test(text)) {
      steps.push({ id: `step_${steps.length + 1}`, order: steps.length + 1, instruction: text });
    }
  }

  console.log("Extracted steps count:", steps.length);
  steps.forEach(s => console.log("  " + (s.instruction.endsWith(":") ? "📂 " : "   • ") + s.instruction));
}

testPages([312, 313], "Fig Bars");
testPages([26, 27], "Chocolate Puff Pastry");