disclosure-bureau/web/lib/doc-summary.ts

/**
 * Extract a short summary from a wiki/documents/<id>.md body.
 *
 * Priority:
 *   1. `enthusiast_pitch_pt_br` / `enthusiast_pitch_en` in frontmatter (Johnny Harris-style
 *      generated by scripts/34-generate-doc-pitches.py) — preferred when present
 *   2. `## Sumário Executivo (PT-BR)` (5 docs have this, synthesized by Sonnet 4.6)
 *   3. `## Executive Summary (EN)` (same 5 docs)
 *   4. First substantial paragraph in the body (skipping headings, blockquotes, callouts)
 *
 * Strips markdown formatting (asterisks, backticks, wiki-links) and returns
 * a plain-text snippet capped at ~280 chars (≈3 lines of card width).
 *
 * pickPitch(): returns the Johnny Harris-style pitch directly (preserves markdown
 * for rich rendering in cards/wikis).
 */
const MAX_CHARS = 280;

function stripMd(s: string): string {
  return s
    // wiki-links → display text
    .replace(/\[\[([^\]|]+?)(?:\|([^\]]+))?\]\]/g, (_full, target: string, alias?: string) =>
      (alias ?? target).trim(),
    )
    // markdown links [text](url) → text
    .replace(/\[([^\]]+?)\]\([^)]*?\)/g, "$1")
    // emphasis & code
    .replace(/[*_`~]+/g, "")
    // heading hash
    .replace(/^#+\s*/gm, "")
    // bullets
    .replace(/^\s*[-*+]\s+/gm, "")
    // ordered list markers
    .replace(/^\s*\d+\.\s+/gm, "")
    // blockquote markers
    .replace(/^>\s*/gm, "")
    // collapse whitespace
    .replace(/\s+/g, " ")
    .trim();
}

function findSection(body: string, headingRe: RegExp): string | null {
  const lines = body.split("\n");
  let inSection = false;
  const captured: string[] = [];
  for (const line of lines) {
    if (headingRe.test(line)) {
      inSection = true;
      continue;
    }
    if (inSection) {
      // next ## or ### heading ends this section
      if (/^#{1,3}\s/.test(line)) break;
      captured.push(line);
    }
  }
  const text = stripMd(captured.join("\n"));
  return text.length >= 40 ? text : null;
}

function firstParagraph(body: string): string {
  // Skip leading H1, callouts (>), bare headings; pick first paragraph ≥ 80 chars
  const lines = body.split("\n");
  const buffer: string[] = [];
  for (const line of lines) {
    const t = line.trim();
    if (!t) {
      if (buffer.length > 0) {
        const text = stripMd(buffer.join(" "));
        if (text.length >= 80) return text;
        buffer.length = 0;
      }
      continue;
    }
    if (/^#{1,6}\s/.test(t)) continue;
    if (t.startsWith(">")) continue;
    if (t.startsWith("|")) continue; // skip tables
    if (/^[-*+]\s/.test(t)) continue; // skip bullet starts
    buffer.push(t);
  }
  // trailing buffer
  if (buffer.length > 0) {
    const text = stripMd(buffer.join(" "));
    if (text.length >= 40) return text;
  }
  return "";
}

/** Return the Johnny Harris pitch from frontmatter if present, else null. */
export function pickPitch(
  fm: Record<string, unknown> | undefined,
  lang: "pt" | "en" = "pt",
): string | null {
  if (!fm) return null;
  const key = lang === "en" ? "enthusiast_pitch_en" : "enthusiast_pitch_pt_br";
  const v = fm[key];
  if (typeof v === "string" && v.trim().length > 20) return v.trim();
  // Cross-fall: prefer ANY pitch (pt or en) over heuristic summary
  const other = lang === "en" ? "enthusiast_pitch_pt_br" : "enthusiast_pitch_en";
  const v2 = fm[other];
  if (typeof v2 === "string" && v2.trim().length > 20) return v2.trim();
  return null;
}

export function summarize(body: string, lang: "pt" | "en" = "pt"): string {
  if (!body) return "";
  const ptSection = findSection(body, /^##\s+Sum[áa]rio Executivo\s*\(PT-BR\)/i);
  const enSection = findSection(body, /^##\s+Executive Summary\s*\(EN\)/i);
  const narrativeSection = findSection(body, /^##\s+Narrative Arc\s*\(EN\)/i);

  let chosen: string;
  if (lang === "pt") {
    chosen = ptSection ?? enSection ?? narrativeSection ?? firstParagraph(body);
  } else {
    chosen = enSection ?? narrativeSection ?? ptSection ?? firstParagraph(body);
  }
  if (!chosen) return "";
  if (chosen.length <= MAX_CHARS) return chosen;
  // truncate at last word boundary before MAX_CHARS
  const cut = chosen.slice(0, MAX_CHARS);
  const lastSpace = cut.lastIndexOf(" ");
  return (lastSpace > 200 ? cut.slice(0, lastSpace) : cut) + "…";
}