build_static.py

#!/usr/bin/env python3
"""
DataSignals Lab - static site generator.

Produces a fully static, buildless `public/` folder (pure HTML/CSS/JS + assets)
that can be drag-and-dropped into Netlify, or deployed with no build command.

Content comes from the existing Jekyll markdown (front-matter + body), the same
texts that are live on Apify and GitHub, wrapped in a premium template.

Run:  python3 build_static.py
"""
import os, re, glob, shutil, datetime, html as htmllib
import markdown

ROOT = os.path.dirname(os.path.abspath(__file__))
OUT  = os.path.join(ROOT, "public")
BASE = "https://datasignalslab.com"
# Apify affiliate referral id (FirstPromoter). Appended to every click-through
# apify.com link via add_referral() in render_page(); SEO/identity refs
# (JSON-LD "url"/sameAs, canonical) stay clean because they use no href=.
REFERRAL = "wnlxst"
APIFY_PROFILE = "https://apify.com/datasignalslab"
GITHUB_PROFILE = "https://github.com/datasignalslab"
# Sitemap-lastmod volgt de bouwdatum automatisch.
BUILD_DATE = __import__("datetime").date.today().isoformat()


def add_referral(html):
    """Append ?fpr=<REFERRAL> to every href pointing at apify.com (exact host).
    Leaves mcp.apify.com and other subdomains, JSON-LD and canonical untouched."""
    def repl(m):
        url = m.group(1)
        if "fpr=" in url:
            return m.group(0)
        frag = ""
        if "#" in url:
            url, frag = url.split("#", 1)
            frag = "#" + frag
        sep = "&" if "?" in url else "?"
        return f'href="{url}{sep}fpr={REFERRAL}{frag}"'
    return re.sub(r'href="(https://apify\.com[^"]*)"', repl, html)


def add_referral_text(text):
    """Same as add_referral but for bare apify.com URLs in plain-text/markdown
    (llms.txt / llms-full.txt). Skips mcp.apify.com and already-tagged URLs."""
    def repl(m):
        url = m.group(0)
        if "fpr=" in url:
            return url
        sep = "&" if "?" in url else "?"
        return f"{url}{sep}fpr={REFERRAL}"
    return re.sub(r'https://apify\.com[^\s)\]>"]*', repl, text)

# ---------------------------------------------------------------------------
# Product / page metadata (improved short copy for the homepage grid)
# ---------------------------------------------------------------------------
PRODUCTS = [
    # html, icon, short title, blurb, category
    ("sec-form4-insider-buying.html",      "icon-sec-form4.png",       "Insider Buying Clusters",   "When several insiders buy the same stock within days - scored 0-100 from SEC Form 4 filings.",                         "SEC & financial"),
    ("sec-form144-insider-selling.html",   "icon-form144-selling.png", "Insider Selling (Form 144)","Who is filing to sell, their role and the dollar value - the leading-indicator companion to Form 4.",                  "SEC & financial"),
    ("sec-13dg-activist-stakes.html",      "icon-13dg-activist.png",   "Activist Stakes (13D/G)",   "Who is taking a 5%+ stake, activist vs passive, with Item 4 intent, scored by impact.",                               "SEC & financial"),
    ("smart-money-13f.html",               "icon-smart-money-13f.png", "Smart Money 13F",           "Quarterly hedge-fund buys, exits and cross-fund consensus from 13F filings.",                                          "SEC & financial"),
    ("sec-8k-material-events.html",        "icon-sec-8k.png",          "8-K Material Events",       "Real-time 8-K events - M&A, executive changes, earnings - classified and materiality-scored.",                        "SEC & financial"),
    ("startup-funding-form-d.html",        "icon-formd-funding.png",   "Startup Funding (Form D)",  "New capital raises before the press release - fund noise filtered, amounts and executives parsed.",                    "SEC & financial"),
    ("biotech-catalyst-monitor.html",      "icon-biotech.png",         "Biotech Catalysts",         "Upcoming clinical-trial readouts and results from ClinicalTrials.gov, ranked by potential impact.",                   "Biotech & health"),
    ("fda-drug-approval-monitor.html",     "icon-fda-approval.png",    "FDA Drug Actions",          "Approvals, new indications and priority reviews from openFDA, scored as regulatory catalysts.",                       "Biotech & health"),
    ("nih-research-funding-monitor.html",  "icon-nih-research.png",    "NIH Research Funding",      "Grant momentum by organization or research topic from NIH RePORTER - an upstream R&D signal.",                        "Biotech & health"),
    ("gov-contract-awards-monitor.html",   "icon-gov-contracts.png",   "Gov Contract Awards",       "Recent federal contract wins per company from USAspending.gov, scored by size and recency.",                          "Government"),
    ("congress-trading-monitor.html",      "icon-congress-trading.png","Congress Trading",          "Recent stock trades by US House members from the official House Clerk disclosures, parsed and scored.",                "Government"),
    ("crypto-volume-momentum-scanner.html","icon-crypto-scanner.png",  "Crypto Volume & Momentum",  "A crypto screener that ranks coins by unusual turnover and momentum, stablecoins filtered.",                          "Crypto & app"),
    ("app-store-review-intelligence.html", "icon-app-review.png",      "App Review Intelligence",   "Aspect-based sentiment, emerging issues and competitor comparison from App Store reviews.",                            "Crypto & app"),
]
MCP_CARD = ("datasignals-mcp.html", "icon-datasignals-mcp.png", "MCP Server",
            "All 17 tools behind one Model Context Protocol endpoint for Claude, ChatGPT, Cursor and custom agents - free server, pay only per result.")

CATEGORY_ORDER = ["SEC & financial", "Biotech & health", "Government", "Crypto & app"]
ICON_BY_HTML = {p[0]: p[1] for p in PRODUCTS}
ICON_BY_HTML[MCP_CARD[0]] = MCP_CARD[1]

NAV = [
    ("Products", "/#products"),
    ("Reports", "/reports.html"),
    ("Pricing", "/pricing.html"),
    ("Performance", "/performance.html"),
    ("Confluence", "/confluence.html"),
    ("AI agents", "/datasignals-mcp.html"),
    ("Compare", "/compare.html"),
    ("Blog", "/blog.html"),
    ("About", "/about.html"),
]

md = markdown.Markdown(extensions=["tables", "fenced_code", "sane_lists", "attr_list", "md_in_html"])

# ---------------------------------------------------------------------------
# Custom line-art icon set (cohesive, data-tech, brand gradient stroke).
# 24x24 viewBox, stroke via shared <linearGradient id="dsg">.
# ---------------------------------------------------------------------------
# Icons carry their own gradient (unique id per call) so the diagonal gradient
# spans the whole icon, identical to the rendered Apify PNG tiles, in every browser.
GRAD_DEF = ""
_ICON_N = [0]

_ICON_PATHS = {
    # SEC & financial
    "sec-form4-insider-buying.html":
        '<polyline points="4,16 9,11 13,14 20,6"/><polyline points="15,6 20,6 20,11"/>'
        '<circle cx="9" cy="11" r="1.25"/><circle cx="13" cy="14" r="1.25"/>',
    "sec-form144-insider-selling.html":
        '<polyline points="4,8 9,13 13,10 20,18"/><polyline points="15,18 20,18 20,13"/>'
        '<circle cx="9" cy="13" r="1.25"/><circle cx="13" cy="10" r="1.25"/>',
    "sec-13dg-activist-stakes.html":
        '<line x1="6.5" y1="3" x2="6.5" y2="21"/><path d="M6.5 4 H18 L14.8 8 L18 12 H6.5"/>',
    "smart-money-13f.html":
        '<rect x="4" y="13" width="3.6" height="7" rx="1"/><rect x="10.2" y="9" width="3.6" height="11" rx="1"/>'
        '<rect x="16.4" y="4.5" width="3.6" height="15.5" rx="1"/>',
    "sec-8k-material-events.html":
        '<path d="M13 2.5 L5.5 13 H11 L10 21.5 L18.5 10.5 H12 Z"/>',
    "startup-funding-form-d.html":
        '<path d="M12 2.6c2.6 2.2 4 5.1 4 8.4 0 1.9-.8 3.5-1.7 4.6H9.7C8.8 14.5 8 12.9 8 11c0-3.3 1.4-6.2 4-8.4Z"/>'
        '<circle cx="12" cy="9.6" r="1.5"/><path d="M8.2 14 5.6 16.6v3l2.6-1.5M15.8 14l2.6 2.6v3l-2.6-1.5M12 16v4.5"/>',
    # Biotech & health
    "biotech-catalyst-monitor.html":
        '<path d="M8 3c0 4.5 8 6.5 8 9s-8 4.5-8 9"/><path d="M16 3c0 4.5-8 6.5-8 9s8 4.5 8 9"/>'
        '<line x1="9.3" y1="6.6" x2="14.7" y2="6.6"/><line x1="9.3" y1="17.4" x2="14.7" y2="17.4"/>',
    "fda-drug-approval-monitor.html":
        '<rect x="3" y="8.5" width="12.5" height="6" rx="3"/><line x1="9.25" y1="8.5" x2="9.25" y2="14.5"/>'
        '<path d="M14.5 17.5l2.2 2.2 4-4.2"/>',
    "nih-research-funding-monitor.html":
        '<path d="M9 3h6M10.2 3v6l-4.6 8.2A2 2 0 0 0 7.4 20.2h9.2a2 2 0 0 0 1.8-3L13.8 9V3"/>'
        '<line x1="8" y1="15" x2="16" y2="15"/>',
    # Government
    "gov-contract-awards-monitor.html":
        '<path d="M4 9 L12 4 L20 9"/><line x1="5.5" y1="9.5" x2="5.5" y2="18"/><line x1="9.8" y1="9.5" x2="9.8" y2="18"/>'
        '<line x1="14.2" y1="9.5" x2="14.2" y2="18"/><line x1="18.5" y1="9.5" x2="18.5" y2="18"/>'
        '<line x1="3.5" y1="20" x2="20.5" y2="20"/>',
    "congress-trading-monitor.html":
        '<path d="M7 11c0-3 2.2-5 5-5s5 2 5 5"/><line x1="12" y1="3" x2="12" y2="6"/><line x1="5" y1="11" x2="19" y2="11"/>'
        '<line x1="6.7" y1="11.5" x2="6.7" y2="19"/><line x1="10.2" y1="11.5" x2="10.2" y2="19"/>'
        '<line x1="13.8" y1="11.5" x2="13.8" y2="19"/><line x1="17.3" y1="11.5" x2="17.3" y2="19"/><line x1="4" y1="19.5" x2="20" y2="19.5"/>',
    # Crypto & app
    "crypto-volume-momentum-scanner.html":
        '<line x1="7" y1="3.5" x2="7" y2="20.5"/><rect x="5" y="7.5" width="4" height="7.5" rx="1"/>'
        '<line x1="16" y1="3.5" x2="16" y2="20.5"/><rect x="14" y="11" width="4" height="6" rx="1"/>',
    "app-store-review-intelligence.html":
        '<path d="M4 5.5a2 2 0 0 1 2-2h12a2 2 0 0 1 2 2v7a2 2 0 0 1-2 2H9.5l-3.5 3.5V14.5H6a2 2 0 0 1-2-2Z"/>'
        '<path d="M12 6.4l1 2 2.2.3-1.6 1.55.38 2.2L12 11.4l-1.98 1.05.38-2.2L8.8 8.7l2.2-.3Z"/>',
    # AI
    "datasignals-mcp.html":
        '<circle cx="12" cy="12" r="2.3" class="hub-core"/><circle cx="5" cy="6" r="1.7" class="hub-node n1"/>'
        '<circle cx="19" cy="6" r="1.7" class="hub-node n2"/><circle cx="5" cy="18" r="1.7" class="hub-node n3"/>'
        '<circle cx="19" cy="18" r="1.7" class="hub-node n4"/>'
        '<line x1="6.4" y1="7.1" x2="10.2" y2="10.6"/><line x1="17.6" y1="7.1" x2="13.8" y2="10.6"/>'
        '<line x1="6.4" y1="16.9" x2="10.2" y2="13.4"/><line x1="17.6" y1="16.9" x2="13.8" y2="13.4"/>',
}

def svg_icon(html, cls="dsg-ic"):
    paths = _ICON_PATHS.get(html, '<circle cx="12" cy="12" r="8"/>')
    extra = " hub" if html == "datasignals-mcp.html" else ""
    _ICON_N[0] += 1
    gid = f"dsg{_ICON_N[0]}"
    grad = (f'<defs><linearGradient id="{gid}" gradientUnits="userSpaceOnUse" x1="2" y1="2" x2="22" y2="22">'
            '<stop offset="0" stop-color="#14B8A6"/><stop offset=".55" stop-color="#22D3EE"/>'
            '<stop offset="1" stop-color="#7af0d8"/></linearGradient></defs>')
    return (f'<svg class="{cls}{extra}" viewBox="0 0 24 24" fill="none" stroke="url(#{gid})" '
            f'stroke-width="1.5" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">{grad}{paths}</svg>')

# ---------------------------------------------------------------------------
# Helpers
# ---------------------------------------------------------------------------
def congress_example_block():
    """Rendert een echt trade-record uit de live congres-data als voorbeeld op de
    productpagina (placeholder {{CONGRESS_EXAMPLE}}). Altijd vers uit de feed en
    toont de exacte JSON-vorm die de API en de MCP-tools teruggeven. Uitsluitend
    presentatie: leest alleen, wijzigt geen data of selectieregels."""
    import json as _j
    try:
        d = _j.load(open("/home/krtradingpro/apify/products/_congress_data.json"))
        trades = [t for t in (d.get("trades") or []) if t.get("ticker") and t.get("pdf_url")]
        if not trades:
            return "*(live example temporarily unavailable)*"
        # Sterkste, meest sprekende record: hoogste impact, dan grootste bedrag.
        t = sorted(trades, key=lambda x: (-(x.get("impact") or 0), -(x.get("amount_mid") or 0)))[0]
        rec = {
            "member": t.get("member", ""),
            "ticker": t.get("ticker", ""),
            "tx_type": t.get("tx_type", ""),
            "tx_date": t.get("tx_date", ""),
            "amount_low": t.get("amount_low", 0),
            "amount_high": t.get("amount_high", 0),
            "impact": t.get("impact", 0),
            "filing_date": t.get("filing_date", ""),
            "pdf_url": t.get("pdf_url", ""),
        }
        n = len([x for x in (d.get("trades") or []) if x.get("ticker")])
        js = _j.dumps(rec, indent=2, ensure_ascii=False)
        return (
            "One real record from the live feed, refreshed daily. This is the exact JSON the API "
            "and the MCP tools return for every trade:\n\n"
            "```json\n" + js + "\n```\n\n"
            "Every field is parsed straight from the official House Clerk filing linked in "
            "`pdf_url`. The `impact` score (0&ndash;100) ranks how market-moving a trade is, so an "
            "agent or a spreadsheet can sort it or set a threshold on it. All " + str(n) +
            " scored trades in the current window share this shape."
        )
    except Exception:
        return "*(live example temporarily unavailable)*"


def read_md(path):
    txt = open(path, encoding="utf-8").read()
    txt = txt.replace("{{VALUE_PROP}}", PRICING["value_prop"])
    if "{{CONGRESS_EXAMPLE}}" in txt:
        txt = txt.replace("{{CONGRESS_EXAMPLE}}", congress_example_block())
    meta = {}
    body = txt
    if txt.startswith("---"):
        end = txt.find("---", 3)
        block = txt[3:end]
        body = txt[end + 3:]
        for line in block.splitlines():
            m = re.match(r"(\w+):\s*(.*)", line)
            if m:
                meta[m.group(1)] = m.group(2).strip().strip('"')
    return meta, body

def first_h1(body):
    m = re.search(r"^#\s+(.+)$", body, re.M)
    return (m.group(1).strip() if m else "")

def strip_first_h1(body):
    return re.sub(r"^#\s+.+$", "", body, count=1, flags=re.M)

def to_html(body):
    md.reset()
    return md.convert(body)

def strip_keywords(body):
    """Remove visible keyword-stuffing lines (e.g. '*Keywords: a, b, c.*') and any
    horizontal-rule that directly precedes them. Keyword stuffing adds no ranking
    value and looks spammy to readers and to Google."""
    body = re.sub(r'(?im)^\s*-{3,}\s*\n+\s*\*?keywords:.*$', '', body)
    body = re.sub(r'(?im)^\s*\*?keywords:.*$', '', body)
    return body

def parse_faq(body):
    """Extract (question, answer) pairs from a '## FAQ' markdown section."""
    m = re.search(r'^##\s+FAQ\s*$(.*?)(?=^##\s|\Z)', body, re.M | re.S)
    if not m:
        return []
    out = []
    for q, a in re.findall(r'\*\*(.+?)\*\*\s*(.*?)(?=\n\*\*|\Z)', m.group(1), re.S):
        q = q.strip()
        a = re.sub(r'\s+', ' ', a).strip()
        if q and a:
            out.append((q, a))
    return out

def rewrite_links(h):
    # Jekyll-pretty internal links -> static files
    h = h.replace('href="/about"', 'href="/about.html"')
    h = h.replace('href="/"', 'href="/index.html"').replace('href="index.html"', 'href="/index.html"')
    return h

def esc(s):
    return htmllib.escape(s, quote=True)

# ---------------------------------------------------------------------------
# Shared chrome
# ---------------------------------------------------------------------------
def nav_html(active=""):
    links = ""
    for label, href in NAV:
        cls = "nav-link active" if label.lower() == active.lower() else "nav-link"
        links += f'<a class="{cls}" href="{href}">{label}</a>'
    return f"""
<header class="site-head" id="site-head">
  <div class="wrap head-inner">
    <a class="brand" href="/index.html" aria-label="DataSignals Lab home">
      <span class="brand-mark" aria-hidden="true"><span></span><span></span><span></span></span>
      <span class="brand-text">DataSignals<span class="brand-accent"> Lab</span></span>
    </a>
    <nav class="nav" id="nav">{links}</nav>
    <div class="head-cta">
      <a class="btn btn-solid" href="{APIFY_PROFILE}" target="_blank" rel="noopener">Browse on Apify</a>
    </div>
    <button class="nav-toggle" id="nav-toggle" aria-label="Menu" aria-expanded="false"><span></span><span></span><span></span></button>
  </div>
</header>"""

def footer_html():
    prod_links = "".join(
        f'<li><a href="/{p[0]}">{p[2]}</a></li>' for p in PRODUCTS
    )
    return f"""
<footer class="site-foot">
  <div class="wrap foot-grid">
    <div class="foot-brand">
      <a class="brand" href="/index.html">
        <span class="brand-mark" aria-hidden="true"><span></span><span></span><span></span></span>
        <span class="brand-text">DataSignals<span class="brand-accent"> Lab</span></span>
      </a>
      <p class="foot-tag">Official public data, turned into scored, decision-ready signals. Self-serve, pay per result, API and AI-agent ready.</p>
      <div class="foot-social">
        <a href="{APIFY_PROFILE}" target="_blank" rel="noopener">Apify Store</a>
        <a href="/datasignals-mcp.html">MCP server</a>
        <a href="/compare.html">Compare</a>
      </div>
    </div>
    <div class="foot-col">
      <h4>Data products</h4>
      <ul>{prod_links}</ul>
    </div>
    <div class="foot-col">
      <h4>Learn</h4>
      <ul>
        <li><a href="/compare.html">Compare</a></li>
        <li><a href="/performance.html">Performance</a></li>
        <li><a href="/proof.html">Proof</a></li>
        <li><a href="/confluence.html">Confluence</a></li>
        <li><a href="/congress/late-filings.html">STOCK Act late filings</a></li>
        <li><a href="/lobby-to-contracts.html">Lobby to contracts (beta)</a></li>
        <li><a href="/press.html">Press &amp; citations</a></li>
        <li><a href="/blog.html">Blog</a></li>
        <li><a href="/what-is-sec-form-144.html">What is Form 144?</a></li>
        <li><a href="/13d-vs-13g-explained.html">13D vs 13G</a></li>
        <li><a href="/about.html">About</a></li>
        <li><a href="/datasignals-mcp.html">MCP server</a></li>
        <li><a href="/privacy.html">Privacy</a></li>
        <li><a href="/terms.html">Terms</a></li>
      </ul>
    </div>
  </div>
  <div class="wrap foot-bottom">
    <p>&copy; <span id="yr">2026</span> DataSignals Lab. Data for research, screening and monitoring - not investment advice. Historical patterns do not guarantee future results.</p>
    <p class="foot-sources">Sources: SEC EDGAR &middot; openFDA &middot; ClinicalTrials.gov &middot; USAspending.gov &middot; US House Clerk &middot; NIH RePORTER &middot; CoinGecko &middot; Apple App Store</p>
    <p class="foot-contact">Questions or support: <a href="mailto:support@datasignalslab.com">support@datasignalslab.com</a></p>
  </div>
</footer>"""

PAGE_TMPL = """<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1">
<title>__TITLE__</title>
<meta name="description" content="__DESC__">
__VERIFY__<link rel="canonical" href="__CANON__">
<meta name="theme-color" content="#070d1c">
<meta property="og:type" content="__OGTYPE__">
<meta property="og:site_name" content="DataSignals Lab">
<meta property="og:title" content="__OGTITLE__">
<meta property="og:description" content="__DESC__">
<meta property="og:url" content="__CANON__">
<meta property="og:image" content="__BASE____OGIMG__">
<meta property="og:image:width" content="1200">
<meta property="og:image:height" content="630">
<meta property="og:image:alt" content="__OGTITLE__">
<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:title" content="__OGTITLE__">
<meta name="twitter:description" content="__DESC__">
<meta name="twitter:image" content="__BASE____OGIMG__">
<link rel="icon" href="/favicon.ico" sizes="any">
<link rel="icon" type="image/png" sizes="32x32" href="/assets/favicon-32x32.png">
<link rel="icon" type="image/png" sizes="16x16" href="/assets/favicon-16x16.png">
<link rel="icon" type="image/svg+xml" href="/assets/datasignalslab-profile.svg">
<link rel="apple-touch-icon" sizes="180x180" href="/assets/apple-touch-icon.png">
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800;900&family=Space+Grotesk:wght@500;600;700&family=JetBrains+Mono:wght@400;600&display=swap" rel="stylesheet">
<link rel="stylesheet" href="/assets/site.css?v=__ASSETV__">
<script type="application/ld+json">
{"@context":"https://schema.org","@type":"Organization","name":"DataSignals Lab","url":"https://datasignalslab.com","logo":"https://datasignalslab.com/assets/favicon-512.png","description":"Self-serve data products that turn official public sources (SEC EDGAR, openFDA, ClinicalTrials.gov, CoinGecko, Apple App Store) into scored, structured signals for traders, quants, fintech and AI agents.","sameAs":["https://apify.com/datasignalslab"]}
</script>
__JSONLD__
</head>
<body class="__BODYCLASS__">
__GRAD__
__NAV__
<main id="top">
__BODY__
</main>
__FOOT__
<script src="https://cdn.jsdelivr.net/npm/gsap@3.13.0/dist/gsap.min.js" defer></script>
<script src="https://cdn.jsdelivr.net/npm/gsap@3.13.0/dist/ScrollTrigger.min.js" defer></script>
<script src="/assets/site.js?v=__ASSETV__" defer></script>
</body>
</html>
"""

_ASSET_V = None
def asset_v():
    """Content-hash van site.css+site.js -> cache-busting querystring.
    Elke deploy met gewijzigde assets krijgt zo automatisch verse CSS/JS bij
    alle bezoekers (lost het stale-cache-probleem van 2 juli op)."""
    global _ASSET_V
    if _ASSET_V is None:
        import hashlib
        h = hashlib.md5()
        for f in ("assets/site.css", "assets/site.js"):
            h.update(open(os.path.join(ROOT, f), "rb").read())
        _ASSET_V = h.hexdigest()[:10]
    return _ASSET_V

def render_page(title, desc, canon_path, body, active="", extra_jsonld="", og_type="website", bodyclass="", og_title=None, og_image="/assets/og/og-default.png"):
    page = PAGE_TMPL
    canon = BASE + canon_path
    repl = {
        "__ASSETV__": asset_v(),
        "__TITLE__": esc(title),
        "__OGTITLE__": esc(og_title or title),
        "__DESC__": esc(desc),
        "__CANON__": canon,
        "__OGIMG__": og_image,
        "__BASE__": BASE,
        "__OGTYPE__": og_type,
        "__JSONLD__": extra_jsonld,
        "__GRAD__": GRAD_DEF,
        "__NAV__": nav_html(active),
        "__FOOT__": footer_html(),
        "__BODY__": body,
        "__BODYCLASS__": bodyclass,
        "__VERIFY__": verify_meta(),
    }
    for k, v in repl.items():
        page = page.replace(k, v)
    page = page.replace("</body>", analytics_snippet() + "</body>")
    return add_referral(page)


_VERIFY = None

def verify_meta():
    """Sectie 11: verificatie-metatags voor Google Search Console en Bing
    Webmaster. Injecteert pas zodra config/search_console.json echte tokens
    heeft; placeholders lekken nooit naar de output. De HTML-bestandsvariant
    gaat via config/verification/ (zie main), de DNS-TXT-variant staat als
    stappenplan in TODO.md."""
    global _VERIFY
    if _VERIFY is None:
        _VERIFY = ""
        try:
            cfg = _json_pricing.load(open(os.path.join(ROOT, "config", "search_console.json")))
            g = cfg.get("google_site_verification", "")
            b = cfg.get("bing_msvalidate_01", "")
            if g and not g.startswith("TODO"):
                _VERIFY += f'<meta name="google-site-verification" content="{g}">\n'
            if b and not b.startswith("TODO"):
                _VERIFY += f'<meta name="msvalidate.01" content="{b}">\n'
        except Exception:
            _VERIFY = ""
    return _VERIFY


_ANALYTICS = None

def analytics_snippet():
    """Sectie 16: GoatCounter (cookieloos, geen banner nodig). Injecteert pas
    zodra config/analytics.json een echte sitecode heeft; placeholders lekken
    nooit naar de output (check_consistency waakt daar ook over)."""
    global _ANALYTICS
    if _ANALYTICS is None:
        _ANALYTICS = ""
        try:
            cfg = _json_pricing.load(open(os.path.join(ROOT, "config", "analytics.json")))
            code = cfg.get("goatcounter_code", "")
            if code and not code.startswith("TODO"):
                _ANALYTICS = (f'<script data-goatcounter="https://{code}.goatcounter.com/count" '
                              f'async src="https://gc.zgo.at/count.js"></script>')
        except Exception:
            _ANALYTICS = ""
    return _ANALYTICS

def jsonld_block(d):
    import json
    return '<script type="application/ld+json">' + json.dumps(d, separators=(",", ":")) + "</script>"

def write(rel, content):
    path = os.path.join(OUT, rel)
    os.makedirs(os.path.dirname(path), exist_ok=True)
    open(path, "w", encoding="utf-8").write(content)

# ---------------------------------------------------------------------------
# Homepage
# ---------------------------------------------------------------------------
# ---------------------------------------------------------------------------
# Premium reports — gedeelde kaarten (homepage + /reports.html)
# ---------------------------------------------------------------------------
# Single source of truth: config/pricing.json (PROJECTINSTRUCTIE_v4.0.md sectie 3).
# Alle prijzen, limieten en checkout-URLs komen hieruit; check_consistency.py
# faalt de build bij elke afwijking in de output.
import json as _json_pricing
PRICING = _json_pricing.load(open(os.path.join(ROOT, "config", "pricing.json")))
_RP = PRICING["reports"]
CHECKOUT = {
    "formd_paid":    _RP["formd"]["checkout_snapshot"],
    "formd_free":    _RP["formd"]["checkout_preview"],
    "f13_paid":      _RP["13f"]["checkout_snapshot"],
    "f13_free":      _RP["13f"]["checkout_preview"],
    "congress_paid": _RP["congress"]["checkout_snapshot"],
    "congress_free": _RP["congress"]["checkout_preview"],
}
SNAPSHOT_PRICE = PRICING["tiers"]["snapshot"]["price_usd"]
PRO_MONTH = PRICING["tiers"]["pro"]["price_month_usd"]
PRO_YEAR = PRICING["tiers"]["pro"]["price_year_usd"]
MCP_FREE_CALLS = PRICING["tiers"]["free"]["mcp_free_calls_per_month"]
ACTOR_RESULT = f"${PRICING['actors']['price_per_result_usd']:.2f}"
# Kernzin (RONDE_FINAL punt 2): een bron, overal identiek, bewaakt door
# check_consistency.py. Markdown-bronnen gebruiken de placeholder {{VALUE_PROP}}.
VALUE_PROP = PRICING["value_prop"]
# Fondsenlijst 13F: single source config/funds13f.json (v4.0 sectie 7).
N_FUNDS_13F = len(_json_pricing.load(open(os.path.join(ROOT, "config", "funds13f.json")))["funds"])
# Stromenteller: single source config/streams.json (RONDE_FINAL punt 6.3);
# nieuwe stromen tellen automatisch mee op de homepage en in /api/streams.json.
N_STREAMS = len(_json_pricing.load(open(os.path.join(ROOT, "config", "streams.json")))["streams"])

def _fmt_usd(v):
    v = float(v)
    if v >= 1e9: return f"${v/1e9:.1f}B"
    if v >= 1e6: return f"${v/1e6:.0f}M"
    return f"${v/1e3:.0f}K"

def _tcase(s):
    return " ".join(w.capitalize() if not (w.isupper() and len(w) <= 4) else w
                    for w in str(s).split())

def _report_peeks():
    """Echte top-rijen uit de rapportdata: de 'live sample' in de kaarten.
    Ververst automatisch mee met elke deploy (dagelijkse cron). Faalt stil."""
    import json as _json
    formd, f13, congress = "", "", ""
    try:
        d = _json.load(open("/home/krtradingpro/apify/products/_formd_data.json"))
        rs = [r for r in d if r.get("type") == "capital_raise" and float(r.get("amount_sold") or 0) > 0]
        total = len([r for r in d if r.get("type") == "capital_raise"])
        rs.sort(key=lambda r: (-int(r.get("score", 0)), -float(r.get("amount_sold") or 0)))
        rows = "".join(
            f'<div class="pk-row"><span class="pk-name">{esc(_tcase(r["issuer"]))}</span>'
            f'<span class="pk-val">{_fmt_usd(r["amount_sold"])}</span>'
            f'<span class="pk-chip">{r.get("score","")}</span></div>'
            for r in rs[:3])
        if rows:
            formd = (f'<div class="report-peek mono"><div class="pk-head">Live sample &middot; latest filings</div>'
                     f'{rows}<div class="pk-more">+ {max(0, total-3)} more inside</div></div>')
    except Exception:
        pass
    try:
        d = _json.load(open("/home/krtradingpro/apify/products/_13f_data.json"))
        picks = next((x.get("picks") for x in d if x.get("type") == "consensus"), [])[:3]
        rows = "".join(
            f'<div class="pk-row"><span class="pk-name">{esc(_tcase(p["issuer"]))}</span>'
            f'<span class="pk-val">{p.get("funds_holding","?")} funds</span>'
            f'<span class="pk-chip">{p.get("buyers","?")} buying</span></div>'
            for p in picks)
        if rows:
            f13 = (f'<div class="report-peek mono"><div class="pk-head">Live sample &middot; consensus ranking</div>'
                   f'{rows}<div class="pk-more">50 ranked stocks inside</div></div>')
    except Exception:
        pass
    try:
        d = _json.load(open("/home/krtradingpro/apify/products/_congress_data.json"))
        ts = d.get("trades") or []
        rows = "".join(
            f'<div class="pk-row"><span class="pk-name">{esc(t["member"])} &middot; {esc(t.get("ticker") or "n/a")}</span>'
            f'<span class="pk-val">{_fmt_usd(t.get("amount_mid") or 0)}</span>'
            f'<span class="pk-chip">{t.get("impact","")}</span></div>'
            for t in ts[:3])
        if rows:
            congress = (f'<div class="report-peek mono"><div class="pk-head">Live sample &middot; latest disclosures</div>'
                        f'{rows}<div class="pk-more">+ {max(0, len(ts)-3)} more inside</div></div>')
    except Exception:
        pass
    return formd, f13, congress

def _cover_svg(title, source, slug):
    """Genereert een zelfstandige 16:9 SVG-cover in de huisstijl voor een rapport.
    Vervangt de ontbrekende cover-PNG's; geen externe assets, rendert altijd."""
    words = title.split()
    line1, line2 = title, ""
    if len(title) > 16 and len(words) > 1:
        # Split op ~helft, op woordgrens, zodat twee nette regels ontstaan.
        best, half = 0, len(title) / 2
        acc = 0
        for i, w in enumerate(words[:-1]):
            acc += len(w) + 1
            if abs(acc - half) < abs(best - half):
                best, split_at = acc, i + 1
        line1 = " ".join(words[:split_at])
        line2 = " ".join(words[split_at:])
    t1 = htmllib.escape(line1)
    t2 = htmllib.escape(line2)
    src = htmllib.escape(source.upper())
    y1, y2 = (430, 545) if t2 else (490, 0)
    line2_svg = (f'<text x="110" y="{y2}" fill="#ffffff" font-family="Space Grotesk, Inter, sans-serif" '
                 f'font-size="104" font-weight="700">{t2}</text>') if t2 else ""
    return (
        f'<svg class="report-cover-svg" viewBox="0 0 1600 900" width="100%" height="100%" '
        f'preserveAspectRatio="xMidYMid slice" role="img" '
        f'aria-label="{htmllib.escape(title)} &mdash; {src} report cover" '
        f'style="display:block">'
        f'<defs>'
        f'<linearGradient id="cbg{slug}" x1="0" y1="0" x2="1" y2="1">'
        f'<stop offset="0" stop-color="#0a1a26"/><stop offset="1" stop-color="#0c2130"/></linearGradient>'
        f'<linearGradient id="cac{slug}" x1="0" y1="0" x2="1" y2="0">'
        f'<stop offset="0" stop-color="#14B8A6"/><stop offset="1" stop-color="#22D3EE"/></linearGradient>'
        f'</defs>'
        f'<rect width="1600" height="900" fill="url(#cbg{slug})"/>'
        f'<rect x="0" y="0" width="1600" height="10" fill="url(#cac{slug})"/>'
        f'<path d="M110 690 L430 600 L720 660 L1010 470 L1300 540 L1520 360" fill="none" '
        f'stroke="url(#cac{slug})" stroke-width="7" opacity="0.35" stroke-linecap="round" stroke-linejoin="round"/>'
        f'<circle cx="1010" cy="470" r="16" fill="#22D3EE"/>'
        f'<text x="110" y="300" fill="#22D3EE" font-family="JetBrains Mono, monospace" '
        f'font-size="44" letter-spacing="8">{src}</text>'
        f'<text x="110" y="{y1}" fill="#ffffff" font-family="Space Grotesk, Inter, sans-serif" '
        f'font-size="104" font-weight="700">{t1}</text>'
        f'{line2_svg}'
        f'<text x="110" y="820" fill="#7fa6b8" font-family="Inter, sans-serif" font-size="38" '
        f'font-weight="600" letter-spacing="1">DataSignals Lab</text>'
        f'</svg>'
    )


# Rapporten waarvoor een cover-SVG wordt gegenereerd: (bestandsslug, titel, bron).
REPORT_COVERS = [
    ("formd-startup-raises", "Startup Capital Raises", "SEC Form D"),
    ("13f-consensus-q1-2026", "Smart-Money 13F Consensus", "SEC 13F"),
    ("congress-trading", "Congress Stock Trades", "House PTRs"),
]


def write_report_covers():
    """Schrijft de gegenereerde SVG-covers naar public/dl/ (de img-tags wijzen ernaar)."""
    os.makedirs(os.path.join(OUT, "dl"), exist_ok=True)
    for slug, title, source in REPORT_COVERS:
        cid = re.sub(r"[^a-z0-9]", "", slug)
        write(os.path.join("dl", f"cover-{slug}.svg"), _cover_svg(title, source, cid))


def report_cards_html():
    pk_formd, pk_f13, pk_congress = _report_peeks()
    return f"""<div class="report-grid">
        <article class="report-card reveal">
          <a class="report-cover" href="{CHECKOUT['formd_free']}" target="_blank" rel="noopener"><img src="/assets/cover-formd-startup-raises.png" alt="Startup Capital Raises, SEC Form D report cover" loading="lazy"></a>
          <div class="report-body">
            <span class="report-price">${SNAPSHOT_PRICE} snapshot &middot; or all reports in <a href="/pricing.html">Pro</a></span>
            <h3>Startup Capital Raises <span class="report-src">SEC Form D</span></h3>
            {pk_formd}
            <p>The biggest new private capital raises, days before the press. Pooled investment funds filtered out, ranked by a funding score. Refreshes daily.</p>
            <div class="report-actions">
              <a class="btn btn-solid" href="{CHECKOUT['formd_paid']}" target="_blank" rel="noopener">Get the report for ${SNAPSHOT_PRICE}</a>
              <a class="btn btn-ghost" href="{CHECKOUT['formd_free']}" target="_blank" rel="noopener">Free preview</a>
            </div>
          </div>
        </article>
        <article class="report-card reveal">
          <a class="report-cover" href="{CHECKOUT['f13_free']}" target="_blank" rel="noopener"><img src="/assets/cover-13f-consensus-q1-2026.png" alt="Smart-Money 13F Consensus report cover" loading="lazy"></a>
          <div class="report-body">
            <span class="report-price">${SNAPSHOT_PRICE} snapshot &middot; or all reports in <a href="/pricing.html">Pro</a></span>
            <h3>Smart-Money 13F Consensus <span class="report-src">SEC 13F</span></h3>
            {pk_f13}
            <p>What {N_FUNDS_13F} of the most-watched hedge funds bought, sold and crowded into last quarter, in one cross-fund consensus ranking. Refreshes each quarter.</p>
            <div class="report-actions">
              <a class="btn btn-solid" href="{CHECKOUT['f13_paid']}" target="_blank" rel="noopener">Get the report for ${SNAPSHOT_PRICE}</a>
              <a class="btn btn-ghost" href="{CHECKOUT['f13_free']}" target="_blank" rel="noopener">Free preview</a>
            </div>
          </div>
        </article>
        <article class="report-card reveal">
          <a class="report-cover" href="{CHECKOUT['congress_free']}" target="_blank" rel="noopener"><img src="/assets/cover-congress-trading.png" alt="Congress Stock Trades, House PTR report cover" loading="lazy"></a>
          <div class="report-body">
            <span class="report-price">${SNAPSHOT_PRICE} snapshot &middot; or all reports in <a href="/pricing.html">Pro</a></span>
            <h3>Congress Stock Trades <span class="report-src">House PTRs</span></h3>
            {pk_congress}
            <p>Every stock trade disclosed by U.S. House members, ranked by impact, with consensus stocks (tickers multiple members traded) and the most active traders. Refreshes weekly.</p>
            <div class="report-actions">
              <a class="btn btn-solid" href="{CHECKOUT['congress_paid']}" target="_blank" rel="noopener">Get the report for ${SNAPSHOT_PRICE}</a>
              <a class="btn btn-ghost" href="{CHECKOUT['congress_free']}" target="_blank" rel="noopener">Free preview</a>
            </div>
          </div>
        </article>
      </div>"""

def build_ticker():
    """Live ticker met echte Form D-raises + Congress-trades (koop groen / verkoop rood)
    uit de rapport-datamodellen. Wordt bij elke deploy opnieuw gegenereerd (dagelijkse
    formd-cron + wekelijkse congress-cron verversen 'm dus mee). Faalt stil per bron."""
    import json as _json

    def usd(v):
        v = float(v)
        if v >= 1e9: return f"${v/1e9:.1f}B"
        if v >= 1e6: return f"${v/1e6:.0f}M"
        return f"${v/1e3:.0f}K"

    def tc(s):
        return " ".join(w.capitalize() if not (w.isupper() and len(w) <= 4) else w
                        for w in str(s).split())

    formd_items, congress_items = [], []
    try:
        data = _json.load(open("/home/krtradingpro/apify/products/_formd_data.json"))
        raises = [r for r in data if r.get("type") == "capital_raise"
                  and float(r.get("amount_sold") or 0) >= 1_000_000]
        raises.sort(key=lambda r: -float(r.get("amount_sold") or 0))
        formd_items = [
            f'<span class="tk-item"><b>{esc(tc(r.get("issuer","")))}</b> raised {usd(r.get("amount_sold"))}</span>'
            for r in raises[:6]]
    except Exception:
        pass
    try:
        d = _json.load(open("/home/krtradingpro/apify/products/_congress_data.json"))
        ts = [t for t in (d.get("trades") or []) if t.get("ticker")
              and t.get("tx_type") in ("Purchase", "Sale")]
        congress_items = [
            (f'<span class="tk-item">'
             f'<span class="{"tk-buy" if t["tx_type"] == "Purchase" else "tk-sell"}">'
             f'{"&#9650;" if t["tx_type"] == "Purchase" else "&#9660;"} {esc(t["ticker"])}</span> '
             f'Rep. {esc(t["member"].split()[-1])} '
             f'{"bought" if t["tx_type"] == "Purchase" else "sold"} '
             f'${t.get("amount_low",0):,}&ndash;${t.get("amount_high",0):,}</span>')
            for t in ts[:6]]
    except Exception:
        pass

    # Afwisselen: raise, trade, raise, trade... (wat over is achteraan)
    items_list = []
    for a, b in zip(formd_items, congress_items):
        items_list += [a, b]
    items_list += formd_items[len(congress_items):] + congress_items[len(formd_items):]
    if len(items_list) < 4:
        return ""
    items = "".join(i + '<span class="tk-sep">&middot;</span>' for i in items_list)
    lead = ('<span class="tk-live"><span class="tk-dot"></span>Live &middot; SEC &amp; Congress filings</span>'
            '<span class="tk-sep">&middot;</span>')
    half = lead + items
    return (f'<a class="ticker" href="/#reports" aria-label="Latest startup capital raises and Congress trades">'
            f'<div class="tk-track">{half}{half}</div></a>')

def build_home():
    # product grid grouped by category
    sections = ""
    for cat in CATEGORY_ORDER:
        cards = ""
        for hpage, icon, title, blurb, c in PRODUCTS:
            if c != cat:
                continue
            cards += f"""
        <a class="card reveal" href="/{hpage}">
          <span class="card-ic">{svg_icon(hpage)}</span>
          <h3>{esc(title)}</h3>
          <p>{esc(blurb)}</p>
          <span class="card-go">View product &rarr;</span>
        </a>"""
        sections += f"""
      <div class="cat-row">
        <h3 class="cat-label">{esc(cat)}</h3>
        <div class="card-grid">{cards}</div>
      </div>"""

    stats = f"""
    <div class="stat-strip wrap reveal">
      <div class="stat"><b>{len(PRODUCTS) + 1}</b><span>data products</span></div>
      <div class="stat"><b>{len(PRODUCTS) + 2}</b><span>tools over MCP</span></div>
      <div class="stat"><b>3</b><span>premium reports</span></div>
      <div class="stat"><b>{N_STREAMS}</b><span>signal streams, hash-chained daily</span></div>
    </div>"""

    mcp = f"""
    <section class="agents wrap reveal" id="agents">
      <div class="agents-card">
        <div class="agents-text">
          <span class="eyebrow">For AI agents</span>
          <h2>One MCP endpoint. 15 live financial-data tools.</h2>
          <p>Connect Claude, ChatGPT, Cursor or any MCP-capable agent once and it can pull real-time SEC, FDA, clinical-trial, government and crypto data on demand - parsed and scored, never raw HTML. One free call to the flagship confluence_signals ranks the strongest cross-source signals right now. The server is free and you only pay the underlying tools' standard {ACTOR_RESULT} per result.</p>
          <div class="btn-row">
            <a class="btn btn-solid" href="/datasignals-mcp.html">Explore the MCP server</a>
            <a class="btn btn-ghost" href="{APIFY_PROFILE}/datasignals-mcp" target="_blank" rel="noopener">Get it on Apify</a>
          </div>
          <div class="agents-logos">
            <span>LangChain</span><span>LlamaIndex</span><span>Flowise</span><span>Zapier</span><span>Make</span><span>Webhooks</span>
          </div>
        </div>
        <div class="agents-visual" aria-hidden="true">
          <span class="agents-ic">{svg_icon(MCP_CARD[0], cls="dsg-ic")}</span>
          <code class="code-pop">agent.call("hedge_fund_13f", ticker="NVDA")</code>
        </div>
      </div>
    </section>"""

    why = """
    <section class="why wrap reveal">
      <h2 class="sec-title">Why DataSignals Lab</h2>
      <div class="why-grid">
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><circle cx="12" cy="12" r="8"/><circle cx="12" cy="12" r="3"/></svg></div><h4>Official sources only</h4><p>SEC EDGAR, openFDA, ClinicalTrials.gov, USAspending, CoinGecko, Apple App Store. Free, legal, no personal data.</p></div>
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><rect x="4" y="4" width="16" height="16" rx="3"/><path d="M8 13l3 3 5-7"/></svg></div><h4>Transparent scoring</h4><p>Every 0-100 score is explained, so you can verify and reproduce it - no black boxes.</p></div>
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><path d="M6 18 18 6"/><path d="M9 6h9v9"/></svg></div><h4>Verifiable</h4><p>Every result links back to the original filing or source. Trust, then check.</p></div>
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><path d="M20 12a8 8 0 1 1-2.34-5.66"/><path d="M20 4v4h-4"/></svg></div><h4>Reliable</h4><p>Each product is monitored daily and built on stable, official APIs.</p></div>
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><path d="M13 2.5 5.5 13H11L10 21.5 18.5 10.5H12Z"/></svg></div><h4>Self-serve</h4><p>Pay per result on the data tools, or one <a href="/pricing.html">Pro plan</a> for all three reports. API and AI-agent ready.</p></div>
        <div class="why-item"><div class="why-ic"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round"><circle cx="12" cy="12" r="3"/><path d="M12 4v2M12 18v2M4 12h2M18 12h2M6.3 6.3l1.4 1.4M16.3 16.3l1.4 1.4M17.7 6.3l-1.4 1.4M7.7 16.3l-1.4 1.4"/></svg></div><h4>Drops into your stack</h4><p>JSON output, webhooks, schedules, Zapier and Make - or call it as an agent tool.</p></div>
      </div>
    </section>"""

    hero = f"""
    <section class="hero">
      <div class="hero-bg" aria-hidden="true"><span class="blob b1"></span><span class="blob b2"></span><span class="grid-overlay"></span><canvas id="sig" class="sig"></canvas></div>
      <div class="wrap hero-inner">
        <span class="eyebrow reveal">SEC &middot; FDA &middot; clinical-trial &middot; gov &middot; crypto &middot; app data</span>
        <h1 class="reveal" id="hero-h1">Official public data, turned into <span class="grad">scored signals</span>.</h1>
        <p class="hero-sub reveal">{esc(VALUE_PROP)}</p>
        <p class="hero-sub reveal">DataSignals Lab takes official datasets - SEC EDGAR, openFDA, ClinicalTrials.gov, USAspending, the App Store - and adds a transparent analysis layer. You get an answer, not a pile of rows. Pay per result, call it from an API, or wire it straight into an AI agent.</p>
        <div class="btn-row reveal">
          <a class="btn btn-solid btn-lg" href="#products">Explore the products</a>
          <a class="btn btn-ghost btn-lg" href="/datasignals-mcp.html">Free MCP for agents</a>
        </div>
        <p class="hero-meta reveal">Verifiable sources &middot; free previews &middot; built for traders, quants, fintech and researchers</p>
      </div>
      <div class="scroll-cue" aria-hidden="true"><span></span></div>
    </section>"""

    sigviz = """
    <section class="sigviz wrap" id="how-signals">
      <div class="sec-head reveal">
        <h2 class="sec-title">From raw filing to <span class="grad">scored signal</span></h2>
        <p class="sec-lead">Every product runs the same pipeline: pull the official source, parse the filing, score it 0&ndash;100, return clean JSON. Scroll to watch a signal take shape.</p>
      </div>
      <div class="viz-panel reveal">
        <div class="viz-head mono"><span class="tk-dot"></span>SEC EDGAR &rarr; PARSE &rarr; SCORE &rarr; JSON</div>
        <div class="viz-stage">
        <svg id="vizsvg" viewBox="0 0 800 300" preserveAspectRatio="none" role="img" aria-label="Example: share price line with scored signals from official filings">
          <defs>
            <linearGradient id="vzstroke" x1="0" y1="0" x2="1" y2="0">
              <stop offset="0" stop-color="#14B8A6"/><stop offset="1" stop-color="#22D3EE"/>
            </linearGradient>
            <linearGradient id="vzfill" x1="0" y1="0" x2="0" y2="1">
              <stop offset="0" stop-color="rgba(34,211,238,.16)"/><stop offset="1" stop-color="rgba(34,211,238,0)"/>
            </linearGradient>
          </defs>
          <line class="vz-grid" x1="0" y1="80" x2="800" y2="80"/>
          <line class="vz-grid" x1="0" y1="160" x2="800" y2="160"/>
          <line class="vz-grid" x1="0" y1="240" x2="800" y2="240"/>
          <path id="vzarea" fill="url(#vzfill)" stroke="none"
            d="M0,232 C60,224 95,246 145,238 C195,230 215,196 265,200 C315,204 335,162 385,156 C435,150 455,182 505,162 C555,142 575,104 625,98 C675,92 705,124 755,108 L800,102 L800,300 L0,300 Z"/>
          <path id="vzpath" fill="none" stroke="url(#vzstroke)" stroke-width="2.5" stroke-linecap="round"
            d="M0,232 C60,224 95,246 145,238 C195,230 215,196 265,200 C315,204 335,162 385,156 C435,150 455,182 505,162 C555,142 575,104 625,98 C675,92 705,124 755,108 L800,102"/>
        </svg>
        <div class="vz-overlay" aria-hidden="true">
          <div class="vz-mk down" style="left:33.1%;top:66.7%"><span class="vz-pt"></span><span class="vz-tag2 mono">FORM 4 &middot; CLUSTER BUY &middot; 84</span></div>
          <div class="vz-mk up" style="left:63.1%;top:54%"><span class="vz-pt"></span><span class="vz-tag2 mono">8-K &middot; MATERIAL EVENT &middot; 71</span></div>
          <div class="vz-mk down last" style="left:94.4%;top:36%"><span class="vz-pt"></span><span class="vz-tag2 mono">13F &middot; CONSENSUS ADD &middot; 92</span></div>
        </div>
        </div>
        <div class="viz-legend mono"><span class="lg-line"></span>share price (example)<span class="lg-gap"></span><span class="lg-dot"></span>scored signal from an official filing</div>
        <div class="viz-foot mono">score = size &times; freshness &times; conviction &middot; every result links to the official filing</div>
      </div>
    </section>"""

    products_sec = f"""
    <section class="products wrap" id="products">
      <div class="sec-head reveal">
        <h2 class="sec-title">The data products</h2>
        <p class="sec-lead">Thirteen self-serve data feeds plus one free MCP server, fourteen products in total. Each feed takes a clean official source and returns structured, scored, decision-ready output - with a link back to the original filing. The MCP server bundles the thirteen feeds for AI agents and adds the free confluence_signals flagship, a multi-signal company scan, a name-to-ticker resolver and a usage meter, seventeen tools in one endpoint.</p>
      </div>
      {sections}
      <div class="cat-row">
        <h3 class="cat-label">For AI agents</h3>
        <div class="card-grid">
          <a class="card card-mcp reveal" href="/{MCP_CARD[0]}">
            <span class="card-ic">{svg_icon(MCP_CARD[0])}</span>
            <h3>{esc(MCP_CARD[2])} <span class="badge">free</span></h3>
            <p>{esc(MCP_CARD[3])}</p>
            <span class="card-go">View MCP server &rarr;</span>
          </a>
        </div>
      </div>
    </section>"""

    cta = """
    <section class="endcta reveal">
      <div class="wrap endcta-inner">
        <h2>Start with a single run.</h2>
        <p>No account games, no demos to sit through. Pick a feed, run it, pay only for the results you get back.</p>
        <div class="btn-row">
          <a class="btn btn-solid btn-lg" href="#products">Browse all products</a>
          <a class="btn btn-ghost btn-lg" href="/compare.html">How we compare</a>
        </div>
      </div>
    </section>"""

    reports_sec = f"""
    <section class="reports wrap" id="reports">
      <div class="sec-head reveal">
        <h2 class="sec-title">Premium reports</h2>
        <p class="sec-lead">The same official sources, distilled into one clean, ranked document you can read in minutes. Try a free preview, buy a ${SNAPSHOT_PRICE} snapshot, or get all three refreshed monthly with <a href="/pricing.html">DataSignals Pro</a>.</p>
      </div>
      {report_cards_html()}
      <p class="report-note">Free previews collect your email and unlock the full report &middot; secure checkout by Lemon Squeezy &middot; instant access</p>
    </section>"""

    trust = """
    <div class="trust-bar wrap reveal" aria-label="Listed on">
      <span class="tb-label mono">Listed on</span>
      <a class="tb-chip" href="https://apify.com/datasignalslab" target="_blank" rel="noopener"><span class="tb-ck">&#10003;</span>Apify Store</a>
      <a class="tb-chip" href="/datasignals-mcp.html"><span class="tb-ck">&#10003;</span>Official MCP Registry</a>
      <a class="tb-chip" href="https://smithery.ai/servers/datasignalslab/datasignals-lab" target="_blank" rel="noopener"><span class="tb-ck">&#10003;</span>Smithery</a>
      <span class="tb-chip tb-static">Secure checkout &middot; Lemon Squeezy &middot; Merchant of Record</span>
    </div>"""

    body = hero + build_ticker() + trust + stats + sigviz + reports_sec + products_sec + mcp + why + cta
    title = "SEC, FDA & Crypto Data as Scored Signals | DataSignals Lab"
    desc = "Decision-ready data feeds from official public sources: SEC insider, 13F, 8-K, Form D, FDA, clinical-trial, congress, crypto and app-review signals. Pay per result."
    website = jsonld_block({
        "@context": "https://schema.org", "@type": "WebSite",
        "name": "DataSignals Lab", "url": BASE,
        "description": desc,
        "publisher": {"@type": "Organization", "name": "DataSignals Lab", "url": BASE},
    })
    write("index.html", render_page(title, desc, "/", body, active="Products",
                                    bodyclass="home", extra_jsonld=website,
                                    og_image="/assets/og/og-home.png"))

# ---------------------------------------------------------------------------
# Article-style pages (products, guides, compare, about)
# ---------------------------------------------------------------------------
def build_reports_page():
    """Eigen SEO-pagina voor de premium rapporten (naast de homepage-sectie):
    rankt op 'form d report' / '13f consensus report', met Product-JSON-LD."""
    import json as _json
    cards = report_cards_html()
    body = f"""
    <section class="reports wrap" id="reports" style="padding-top:2.4rem">
      <nav class="crumb"><a href="/index.html">Home</a> <span>/</span> <span class="crumb-cur">Reports</span></nav>
      <div class="sec-head reveal">
        <h2 class="sec-title" style="font-size:clamp(2rem,5vw,3rem)">Premium <span class="grad">SEC data reports</span></h2>
        <p class="sec-lead">The same official sources behind our live data products, distilled into one clean, ranked document you can read in minutes. Start with a free preview, buy one report as a ${SNAPSHOT_PRICE} snapshot, or get all three refreshed monthly with <a href="/pricing.html">DataSignals Pro</a>.</p>
      </div>
      {cards}
      <p class="report-note">Free previews collect your email and unlock the full report &middot; secure checkout by Lemon Squeezy (Merchant of Record) &middot; instant access</p>

      <div class="sec-head reveal" style="margin-top:3.2rem">
        <h2 class="sec-title" style="font-size:1.6rem">What's inside</h2>
      </div>
      <div class="why-grid">
        <div class="why-item reveal">
          <h4>Startup Capital Raises, SEC Form D &middot; ${SNAPSHOT_PRICE}</h4>
          <p>Every qualifying raise ($1M+) ranked by a funding score (size &times; freshness &times; type). Amounts raised and offered, industry, securities type, state and filing date. The biggest raises in detail with disclosed executives, and a direct link to the official SEC filing for every row. Pooled investment funds are filtered out, so you see the real operating-company startup signal, days before (and often without) any press.</p>
        </div>
        <div class="why-item reveal">
          <h4>Smart-Money 13F Consensus, SEC 13F &middot; ${SNAPSHOT_PRICE}</h4>
          <p>The latest 13F filings of {N_FUNDS_13F} of the most-watched managers, from Berkshire Hathaway and Citadel to Duquesne and Point72, distilled into a cross-fund consensus table: 50 stocks ranked by conviction (how many funds hold + buying activity), plus each fund's biggest new positions, adds, exits and trims. Every figure traceable to the official EDGAR filing.</p>
        </div>
        <div class="why-item reveal">
          <h4>Congress Stock Trades, House PTRs &middot; ${SNAPSHOT_PRICE}</h4>
          <p>Every stock trade disclosed by U.S. House members this period, parsed from the official House Clerk filings and ranked by an impact score (size &times; freshness). Consensus stocks (the tickers multiple members traded, with buy/sell lean), the most active members, and the biggest trades in detail. Every trade links to the official PTR filing.</p>
        </div>
      </div>

      <div class="sec-head reveal" style="margin-top:3.2rem">
        <h2 class="sec-title" style="font-size:1.6rem">FAQ</h2>
      </div>
      <div class="faq reveal">
        <div class="faq-item"><p class="faq-q">Snapshot or Pro, what is the difference?</p>
        <p class="faq-a">A snapshot is a one-off ${SNAPSHOT_PRICE} purchase of one report: the current edition plus 30 days of updates. DataSignals Pro is ${PRO_MONTH} a month (or ${PRO_YEAR} a year) for all three reports, refreshed monthly, plus MCP access for AI agents and email alerts. See the <a href="/pricing.html">pricing page</a>.</p></div>
        <div class="faq-item"><p class="faq-q">Where does the data come from?</p>
        <p class="faq-a">Directly from official public sources: SEC EDGAR Form D and Form 13F filings, and the U.S. House Clerk financial-disclosure feed. Every row in a report links back to the original filing so you can verify it yourself.</p></div>
        <div class="faq-item"><p class="faq-q">Is this investment advice?</p>
        <p class="faq-a">No. The reports are data and research, not investment advice and not an offer or solicitation. Amounts in Form D filings are self-reported by the issuer; historical 13F positioning does not guarantee future results.</p></div>
        <div class="faq-item"><p class="faq-q">Need this live or on demand instead?</p>
        <p class="faq-a">Yes. The same signals are available as self-serve <a href="/#products">data products</a> on Apify, and AI agents can call them through our free MCP server. The reports are the curated layer on top: cross-filing analysis, ranking and context in one finished document.</p></div>
      </div>

      <div class="btn-row reveal" style="margin-top:2.4rem">
        <a class="btn btn-solid btn-lg" href="{CHECKOUT['formd_free']}" target="_blank" rel="noopener">Try the free Form D preview</a>
        <a class="btn btn-ghost btn-lg" href="{CHECKOUT['f13_free']}" target="_blank" rel="noopener">Try the free 13F preview</a>
        <a class="btn btn-ghost btn-lg" href="{CHECKOUT['congress_free']}" target="_blank" rel="noopener">Try the free Congress preview</a>
      </div>
    </section>"""

    prods = ""
    for name, desc, img, url in (
        ("Startup Capital Raises, SEC Form D Report",
         "The biggest new private capital raises from SEC Form D filings, ranked by a funding score, days before the press. Refreshes daily.",
         "/assets/cover-formd-startup-raises.png", CHECKOUT["formd_paid"]),
        ("Smart-Money 13F Consensus Report",
         f"What {N_FUNDS_13F} of the most-watched hedge funds bought, sold and crowded into last quarter, one cross-fund consensus ranking. Refreshes quarterly.",
         "/assets/cover-13f-consensus-q1-2026.png", CHECKOUT["f13_paid"]),
        ("Congress Stock Trades Report, House PTRs",
         "Every stock trade disclosed by U.S. House members, ranked by impact, with consensus stocks and the most active members. Refreshes weekly.",
         "/assets/cover-congress-trading.png", CHECKOUT["congress_paid"]),
    ):
        prods += jsonld_block({
            "@context": "https://schema.org", "@type": "Product",
            "name": name, "description": desc, "image": BASE + img,
            "brand": {"@type": "Brand", "name": "DataSignals Lab"},
            "offers": {"@type": "Offer", "price": f"{SNAPSHOT_PRICE:.2f}", "priceCurrency": "USD",
                       "url": url, "availability": "https://schema.org/InStock",
                       # Digitale download: geen fysieke verzending, geen retour
                       # (silences GSC merchant-listing warnings, eerlijk gemarkeerd)
                       "shippingDetails": {"@type": "OfferShippingDetails",
                                           "doesNotShip": True},
                       "hasMerchantReturnPolicy": {
                           "@type": "MerchantReturnPolicy",
                           "applicableCountry": ["US", "GB", "NL", "DE", "FR", "CA", "AU"],
                           "returnPolicyCategory": "https://schema.org/MerchantReturnNotPermitted"}},
        })

    write("reports.html", render_page(
        f"Premium SEC & Congress Data Reports (${SNAPSHOT_PRICE}) | DataSignals Lab",
        f"Reports from official filings: startup capital raises (Form D), 13F smart-money consensus, Congress stock trades. Snapshot ${SNAPSHOT_PRICE} or Pro ${PRO_MONTH}/month.",
        "/reports.html", body, active="Reports", extra_jsonld=prods,
        og_image="/assets/og/og-default.png"))

def build_article(src, out, active=""):
    meta, body = read_md(src)
    h1 = first_h1(body) or meta.get("product", "")
    desc = meta.get("description", "")
    apify = meta.get("apify", "")
    product = meta.get("product", "")
    body_wo = strip_first_h1(body)
    body_wo = strip_keywords(body_wo)
    article = rewrite_links(to_html(body_wo))

    # Render the FAQ as clear, separated question/answer blocks instead of one
    # run-together paragraph. Parse from the keyword-stripped body so a trailing
    # keyword line cannot bleed into the last answer.
    faq = parse_faq(body_wo)
    if faq:
        faq_html = ('<div class="faq">' + ''.join(
            f'<div class="faq-item"><p class="faq-q">{esc(q)}</p>'
            f'<p class="faq-a">{esc(a)}</p></div>' for q, a in faq) + '</div>')
        article = re.sub(r'(<h2[^>]*>FAQ</h2>\s*)<p>.*?</p>',
                         lambda m: m.group(1) + faq_html, article, count=1, flags=re.S)

    has_icon = out in _ICON_PATHS
    canon = "/" + out

    # hero
    crumb_mid = f'<a href="/#products">Products</a>' if apify else ''
    if not apify and out in ("compare.html", "about.html", "blog.html"):
        crumb_mid = ""
    crumb = f'<nav class="crumb"><a href="/index.html">Home</a>'
    if apify:
        crumb += ' <span>/</span> <a href="/#products">Products</a>'
    crumb += f' <span>/</span> <span class="crumb-cur">{esc(h1[:48])}{"&hellip;" if len(h1)>48 else ""}</span></nav>'

    cta_html = ""
    if apify:
        cta_html = f"""
        <div class="hero-actions">
          <a class="btn btn-solid btn-lg" href="{apify}" target="_blank" rel="noopener">Run on Apify - {ACTOR_RESULT} / result</a>
          <a class="btn btn-ghost btn-lg" href="/datasignals-mcp.html">Use via MCP (free)</a>
        </div>"""

    icon_html = f'<span class="art-hero-ic">{svg_icon(out)}</span>' if has_icon else ""

    hero = f"""
    <section class="art-hero">
      <div class="hero-bg" aria-hidden="true"><span class="blob b1"></span><span class="grid-overlay"></span><canvas class="sig"></canvas></div>
      <div class="wrap art-hero-inner">
        {crumb}
        <div class="art-hero-top">
          {icon_html}
          <div>
            <h1>{esc(h1)}</h1>
            <p class="art-lead">{esc(desc)}</p>
          </div>
        </div>
        {cta_html}
      </div>
    </section>"""

    aside = ""
    if apify:
        # related products quick links (exclude self)
        rel = "".join(
            f'<li><a href="/{p[0]}">{esc(p[2])}</a></li>'
            for p in PRODUCTS if p[0] != out
        )
        aside = f"""
      <aside class="art-aside">
        <div class="aside-card">
          <span class="aside-price">{ACTOR_RESULT}<small>/ result</small></span>
          <a class="btn btn-solid btn-block" href="{apify}" target="_blank" rel="noopener">Run on Apify</a>
          <a class="btn btn-ghost btn-block" href="/datasignals-mcp.html">Call it from an agent</a>
          <ul class="aside-meta">
            <li>Pay per result billing</li>
            <li>JSON output &middot; API &middot; schedule</li>
            <li>Monitored daily &middot; sources linked</li>
          </ul>
        </div>
        <div class="aside-card aside-more">
          <h4>More feeds</h4>
          <ul class="aside-links">{rel}</ul>
        </div>
      </aside>"""

    trust_html = ""
    if apify:
        trust_html = """
    <div class="wrap"><div class="trust-strip">
      <span><b>Official public source</b></span>
      <span><b>Updated &amp; monitored daily</b></span>
      <span><b>Every result links to its filing</b></span>
      <span><b>Pay per result &middot; free previews</b></span>
    </div></div>"""

    layout_cls = "art-layout has-aside" if apify else "art-layout"
    body_html = f"""
    {hero}
    {trust_html}
    <div class="wrap {layout_cls}">
      <article class="article">{article}</article>
      {aside}
    </div>"""

    extra = ""
    if apify and product:
        sw = jsonld_block({
            "@context": "https://schema.org", "@type": "SoftwareApplication",
            "name": product, "applicationCategory": "BusinessApplication",
            "operatingSystem": "Web, API", "description": desc, "url": apify,
            "offers": {"@type": "Offer", "price": "0.20", "priceCurrency": "USD"},
            "provider": {"@type": "Organization", "name": "DataSignals Lab", "url": BASE},
        })
        bc = jsonld_block({
            "@context": "https://schema.org", "@type": "BreadcrumbList",
            "itemListElement": [
                {"@type": "ListItem", "position": 1, "name": "Home", "item": BASE + "/"},
                {"@type": "ListItem", "position": 2, "name": product, "item": BASE + canon},
            ],
        })
        extra = sw + bc

    if faq:
        extra += jsonld_block({
            "@context": "https://schema.org", "@type": "FAQPage",
            "mainEntity": [
                {"@type": "Question", "name": q,
                 "acceptedAnswer": {"@type": "Answer", "text": a}} for q, a in faq
            ],
        })

    title_tag = (product + " | DataSignals Lab") if product else (h1 + " | DataSignals Lab")
    og = f"/assets/og/{out[:-5]}.png" if apify else "/assets/og/og-default.png"
    write(out, render_page(title_tag, desc, canon, body_html, active=active,
                           extra_jsonld=extra, og_type="article", bodyclass="article-page",
                           og_title=h1, og_image=og))

# ---------------------------------------------------------------------------
# Blog
# ---------------------------------------------------------------------------
MONTHS = ["", "January", "February", "March", "April", "May", "June", "July",
          "August", "September", "October", "November", "December"]

def post_slug(fname):
    base = os.path.basename(fname)[:-3]
    return base[11:]  # strip YYYY-MM-DD-

def post_date(fname):
    base = os.path.basename(fname)
    y, m, d = int(base[0:4]), int(base[5:7]), int(base[8:10])
    return datetime.date(y, m, d)

def build_posts():
    posts = sorted(glob.glob(os.path.join(ROOT, "_posts", "*.md")), reverse=True)
    cards = []
    for p in posts:
        meta, body = read_md(p)
        slug = post_slug(p)
        dt = post_date(p)
        datestr = f"{MONTHS[dt.month]} {dt.day}, {dt.year}"
        title = meta.get("title", first_h1(body))
        desc = meta.get("description", "")
        article = rewrite_links(to_html(strip_keywords(body)))
        url = f"/blog/{slug}/"
        # post page
        hero = f"""
    <section class="art-hero post-hero">
      <div class="hero-bg" aria-hidden="true"><span class="blob b1"></span><span class="grid-overlay"></span><canvas class="sig"></canvas></div>
      <div class="wrap art-hero-inner">
        <nav class="crumb"><a href="/index.html">Home</a> <span>/</span> <a href="/blog.html">Blog</a></nav>
        <span class="post-date">{datestr}</span>
        <h1>{esc(title)}</h1>
        <p class="art-lead">{esc(desc)}</p>
      </div>
    </section>"""
        # CTA naar de gratis preview van het rapport dat bij het onderwerp past
        # (sectie 10); valt terug op de rapportenpagina bij overige onderwerpen.
        s = slug.lower()
        if "congress" in s or "pelosi" in s or "senator" in s or "stock-act" in s or "45-day-rule" in s or "transaction-report" in s:
            cta = (f'Want the signal instead of the raw filings? '
                   f'<a href="{CHECKOUT["congress_free"]}" target="_blank" rel="noopener" data-ev="preview_submit">'
                   f'Get the free Congress Trades preview</a>.')
        elif "13f" in s or "hedge-fund" in s or "berkshire" in s:
            cta = (f'Want the signal instead of the raw filings? '
                   f'<a href="{CHECKOUT["f13_free"]}" target="_blank" rel="noopener" data-ev="preview_submit">'
                   f'Get the free 13F Consensus preview</a>.')
        elif "form-d" in s or "formd" in s or "raises" in s or "startup" in s or "regulation-d" in s:
            cta = (f'Want the signal instead of the raw filings? '
                   f'<a href="{CHECKOUT["formd_free"]}" target="_blank" rel="noopener" data-ev="preview_submit">'
                   f'Get the free Form D preview</a>.')
        else:
            cta = ('Want the signal instead of the raw filings? '
                   '<a href="/reports.html">Get a free report preview</a>.')
        body_html = f"""
    {hero}
    <div class="wrap art-layout">
      <article class="article">{article}
        <hr>
        <p class="post-foot">{cta} Prefer the tool to the write-up? Browse <a href="/index.html">all data feeds</a> or connect the <a href="/datasignals-mcp.html">free MCP server</a>.</p>
      </article>
    </div>"""
        bc = jsonld_block({
            "@context": "https://schema.org", "@type": "BlogPosting",
            "headline": title, "description": desc,
            "datePublished": dt.isoformat(), "url": BASE + url,
            "author": {"@type": "Organization", "name": "DataSignals Lab"},
            "publisher": {"@type": "Organization", "name": "DataSignals Lab",
                          "logo": {"@type": "ImageObject", "url": BASE + "/assets/favicon-512.png"}},
            "image": BASE + "/assets/og-banner.png",
        })
        write(f"blog/{slug}/index.html",
              render_page(title + " | DataSignals Lab", desc, url, body_html,
                          active="Blog", extra_jsonld=bc, og_type="article",
                          bodyclass="article-page", og_title=title))
        cards.append((dt, datestr, title, desc, url))

    # blog index
    meta, body = read_md(os.path.join(ROOT, "blog.md"))
    intro = body.split("{% for", 1)[0]
    intro = strip_first_h1(intro)
    intro_html = rewrite_links(to_html(intro))

    def card(dt, datestr, title, desc, url):
        return f"""
        <a class="post-card reveal" href="{url}">
          <span class="post-card-date">{datestr}</span>
          <h3>{esc(title)}</h3>
          <p>{esc(desc)}</p>
          <span class="card-go">Read post &rarr;</span>
        </a>"""

    # Nieuwste zes bovenaan als eigen blok (RONDE_2 punt 3: Learn toont de
    # zes nieuwste), daaronder het volledige archief. cards staat al op
    # datum aflopend (bestandsnaam begint met YYYY-MM-DD).
    latest_html = "".join(card(*c) for c in cards[:6])
    archive_html = "".join(card(*c) for c in cards[6:])
    grids = f"""
      <h2 class="sec-title" style="margin:2.5rem 0 0">Newest <span class="grad">guides</span></h2>
      <p class="sec-lead">The six most recently published guides.</p>
      <div class="post-grid" style="padding:1.4rem 0 1rem">{latest_html}</div>"""
    if archive_html:
        grids += f"""
      <h2 class="sec-title" style="margin:1.6rem 0 0">All guides</h2>
      <div class="post-grid" style="padding:1.4rem 0 5rem">{archive_html}</div>"""
    hero = f"""
    <section class="art-hero">
      <div class="hero-bg" aria-hidden="true"><span class="blob b1"></span><span class="grid-overlay"></span><canvas class="sig"></canvas></div>
      <div class="wrap art-hero-inner">
        <nav class="crumb"><a href="/index.html">Home</a> <span>/</span> <span class="crumb-cur">Blog</span></nav>
        <h1>Blog</h1>
        <div class="art-lead">{intro_html}</div>
      </div>
    </section>"""
    body_html = f"""
    {hero}
    <div class="wrap">{grids}
    </div>"""
    write("blog.html", render_page("Blog | DataSignals Lab", meta.get("description", ""),
                                   "/blog.html", body_html, active="Blog", bodyclass="article-page"))

# ---------------------------------------------------------------------------
# 404
# ---------------------------------------------------------------------------
def build_404():
    body = """
    <section class="art-hero">
      <div class="hero-bg" aria-hidden="true"><span class="blob b1"></span><span class="grid-overlay"></span><canvas class="sig"></canvas></div>
      <div class="wrap art-hero-inner" style="text-align:center">
        <h1 style="font-size:clamp(3rem,12vw,7rem);margin:0">404</h1>
        <p class="art-lead" style="margin-inline:auto">That page wandered off. Let's get you back to the data.</p>
        <div class="btn-row" style="justify-content:center">
          <a class="btn btn-solid btn-lg" href="/index.html">All products</a>
          <a class="btn btn-ghost btn-lg" href="/blog.html">Read the blog</a>
        </div>
      </div>
    </section>"""
    write("404.html", render_page("Page not found | DataSignals Lab",
          "The page you were looking for could not be found.", "/404.html", body))

# ---------------------------------------------------------------------------
# llms.txt - curated map for AI agents / LLMs (llmstxt.org standard)
# ---------------------------------------------------------------------------
def build_llms_txt():
    L = []
    L.append("# DataSignals Lab\n")
    L.append("> Self-serve data products that turn official public sources (SEC EDGAR, "
             "openFDA, ClinicalTrials.gov, NIH RePORTER, USAspending.gov, US House Clerk, "
             "CoinGecko, Apple App Store) into scored, structured signals for traders, quants, "
             f"fintech and AI agents. Each product runs as an Apify Actor at about {ACTOR_RESULT} per "
             "result; all of them are also callable by AI agents through one free MCP server.\n")
    L.append("## For AI agents (MCP)\n")
    L.append(f"- [DataSignals Lab MCP server]({BASE}/datasignals-mcp.html): one free MCP "
             "endpoint exposing all 17 tools to AI agents (Claude, ChatGPT, Cursor). "
             f"The server is free - you only pay each underlying tool's standard {ACTOR_RESULT} per "
             "result. Listed in the official MCP registry.")
    L.append("- MCP endpoint (streamable HTTP): `https://datasignalslab--datasignals-mcp.apify.actor/mcp` "
             "- point any MCP client here and pass your own Apify API token as a Bearer header. "
             "Free tier: 50 MCP calls per month; billable results cost "
             f"{ACTOR_RESULT} each, charged only on a non-empty result (resolve_company and usage are free).")
    L.append("- [Machine-readable server description](https://github.com/datasignalslab/datasignals-mcp/blob/main/server.json): "
             "name, categories, tags, every tool with the question it answers, the free tier, the price and the "
             "accepted payment rail - the format the official MCP registry reads.\n")
    L.append("## Machine-readable\n")
    L.append(f"- [OpenAPI specification]({BASE}/api/openapi.json): all public endpoints, "
             "the MCP endpoint and its 17 tools, in one machine-readable spec.")
    L.append(f"- [Stream registry]({BASE}/api/streams.json): every data stream with source, "
             "cadence, schema and changelog.")
    L.append(f"- [Confluence cases]({BASE}/api/confluence-public.json): tickers hit by multiple "
             "independent streams within 10 days, public with a 7 day delay.")
    L.append(f"- [Proof chain status]({BASE}/api/proof.json): the daily hash chain over the "
             f"published signal set; download the full chain at {BASE}/proof/chain.jsonl and each day's "
             f"fingerprint at {BASE}/proof/days/, independently timestamped in the Bitcoin blockchain via OpenTimestamps.")
    L.append(f"- [Reasoning]({BASE}/api/reasoning.json): confluence candidates ranked by conviction "
             "with contributing signals and source links, plus a per-signal track record.\n")
    L.append("## Products\n")
    for hpage, _icon, title, blurb, _cat in PRODUCTS:
        L.append(f"- [{title}]({BASE}/{hpage}): {blurb}")
    L.append("")
    L.append("## Guides\n")
    for name in ("what-is-sec-form-144", "13d-vs-13g-explained"):
        meta, body = read_md(os.path.join(ROOT, name + ".md"))
        title = meta.get("title", first_h1(body))
        desc = meta.get("description", "")
        L.append(f"- [{title}]({BASE}/{name}.html): {desc}")
    L.append("")
    L.append("## Blog\n")
    for p in sorted(glob.glob(os.path.join(ROOT, "_posts", "*.md")), reverse=True):
        meta, body = read_md(p)
        title = meta.get("title", first_h1(body))
        desc = meta.get("description", "")
        L.append(f"- [{title}]({BASE}/blog/{post_slug(p)}/): {desc}")
    L.append("")
    L.append("## More\n")
    L.append(f"- [Run on Apify]({APIFY_PROFILE}?fpr={REFERRAL}): browse and run every product "
             "on the Apify marketplace (pay per result, no subscription).")
    L.append(f"- [How we compare]({BASE}/compare.html): DataSignals Lab vs QuiverQuant, "
             "WhaleWisdom, Fintel and Appbot.")
    L.append(f"- [Performance]({BASE}/performance.html): hypothetical returns of the top signals "
             "versus the S&P 500, under rules fixed in advance.")
    L.append(f"- [Proof]({BASE}/proof.html): how the daily hash chain makes the track record "
             "tamper-evident, and how to verify it yourself.")
    L.append(f"- [Confluence]({BASE}/confluence.html): tickers hit by multiple independent "
             "streams within 10 days.")
    L.append(f"- [Ask the data]({BASE}/reasoning.html): a ranked, evidence-based answer to which "
             "ticker has the strongest confluence case right now, and the track record of signals "
             "that already reached a measurement horizon.")
    L.append(f"- [About]({BASE}/about.html): what DataSignals Lab is and how the data is sourced.")
    L.append("")
    write("llms.txt", add_referral_text("\n".join(L)))


def build_llms_full():
    """llms-full.txt: the complete readable text of every page in one file, so an
    agent can ingest the whole catalogue in a single fetch (llmstxt.org pattern)."""
    def clean(body):
        body = strip_keywords(body)
        # drop any Liquid/Jekyll tags that may linger in source markdown
        body = re.sub(r"\{%.*?%\}", "", body, flags=re.S)
        return body.strip()
    parts = ["# DataSignals Lab - full content for AI agents\n",
             "> Complete text of every product page, guide and blog post on "
             f"datasignalslab.com. Each product runs as an Apify Actor (~{ACTOR_RESULT}/result) and "
             "is also callable through one free MCP server. See /llms.txt for the curated "
             "index and /datasignals-mcp.html for the MCP endpoint.\n"]
    # compare.md bestaat niet meer: /compare.html komt uit funnel.py en
    # config/competitors.json (de md-versie was verouderd en week af).
    page_order = [p[0][:-5] for p in PRODUCTS] + [
        "datasignals-mcp", "what-is-sec-form-144", "13d-vs-13g-explained", "about"]
    for name in page_order:
        meta, body = read_md(os.path.join(ROOT, name + ".md"))
        title = meta.get("title", first_h1(body))
        parts.append(f"\n\n---\n\n# {title}\nURL: {BASE}/{name}.html\n\n{clean(body)}")
    for p in sorted(glob.glob(os.path.join(ROOT, "_posts", "*.md")), reverse=True):
        meta, body = read_md(p)
        title = meta.get("title", first_h1(body))
        parts.append(f"\n\n---\n\n# {title}\nURL: {BASE}/blog/{post_slug(p)}/\n\n{clean(body)}")
    write("llms-full.txt", add_referral_text("\n".join(parts)) + "\n")

# ---------------------------------------------------------------------------
# Machineleesbare laag (RONDE_FINAL punt 5): /api/ met OpenAPI-specificatie,
# stromenregister (schema + changelog per stroom), publieke confluence met
# vertraging en de proof-ketenstatus. Bron: config/streams.json e.a.
# ---------------------------------------------------------------------------
def build_api():
    import datetime as _dt
    import json as _json
    today = _dt.date.today()
    streams_cfg = _json.load(open(os.path.join(ROOT, "config", "streams.json")))["streams"]
    conf_cfg = _json.load(open(os.path.join(ROOT, "config", "confluence.json")))

    def jwrite(rel, obj):
        write(rel, _json.dumps(obj, indent=1, ensure_ascii=False) + "\n")

    # 1. Stromenregister: schema en changelog per stroom.
    jwrite("api/streams.json", {
        "generated": today.isoformat(), "site": BASE,
        "count": len(streams_cfg), "streams": streams_cfg,
        "note": "Live stream data is part of DataSignals Pro (daily alert mail and Pro area) "
                "and callable per result through the MCP server. This endpoint documents "
                "every stream: source, cadence, schema and changelog.",
        "mcp_endpoint": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
    })

    # 2. Publieke confluence-gevallen, met de vaste vertraging (zelfde regel als de pagina).
    delay_cutoff = (today - _dt.timedelta(days=conf_cfg["free_delay_days"])).isoformat()
    history_cutoff = (today - _dt.timedelta(days=90)).isoformat()
    visible, seen_t = [], set()
    try:
        archived = [_json.loads(line) for line in
                    open(os.path.join(ROOT, "maintenance", "confluence", "archive.jsonl"))]
    except OSError:
        archived = []
    for c in reversed(archived):
        if c["archived"] <= delay_cutoff and c["archived"] >= history_cutoff and c["ticker"] not in seen_t:
            seen_t.add(c["ticker"])
            visible.append(c)
    jwrite("api/confluence-public.json", {
        "generated": today.isoformat(), "definition": conf_cfg["copy"]["definition"],
        "delay_days": conf_cfg["free_delay_days"],
        "note": conf_cfg["copy"]["free_note"], "count": len(visible), "cases": visible,
    })

    # 3. Proof-ketenstatus (volledige keten staat op de publieke mirror).
    chain = []
    try:
        chain = [_json.loads(line) for line in
                 open(os.path.join(ROOT, "maintenance", "proof", "chain.jsonl")) if line.strip()]
    except OSError:
        pass
    jwrite("api/proof.json", {
        "generated": today.isoformat(), "links": len(chain),
        "first_date": chain[0]["date"] if chain else None,
        "last_date": chain[-1]["date"] if chain else None,
        "last_hash": chain[-1]["hash"] if chain else None,
        "chain_url": BASE + "/proof/chain.jsonl",
        "day_files": BASE + "/proof/days/",
        "timestamp_anchor": "Bitcoin blockchain via OpenTimestamps (per-day .json.ots)",
        "how_to_verify": BASE + "/proof.html",
    })

    # 3b. Redeneerlaag (Werkpakket G): gerangschikte confluence-kandidaten op
    # conviction, plus de trackrecord-samenvatting van signalen die al een
    # meethorizon hebben doorlopen. Bron: scripts/reasoning.py en
    # scripts/trackrecord.py; geen nieuwe selectieregels (config/confluence.json
    # en config/performance_rules.json blijven ongewijzigd, alleen gelezen).
    reasoning = {}
    try:
        reasoning = _json.load(open(os.path.join(ROOT, "maintenance", "reasoning", "reasoning.json")))
    except OSError:
        pass
    track = {}
    try:
        track = _json.load(open(os.path.join(ROOT, "maintenance", "reasoning", "trackrecord.json")))
    except OSError:
        pass
    jwrite("api/reasoning.json", {
        "generated": reasoning.get("generated", today.isoformat()),
        "question": reasoning.get("question"),
        "definition": reasoning.get("definition"),
        "window_days": reasoning.get("window_days"),
        "min_streams": reasoning.get("min_streams"),
        "summary": reasoning.get("summary"),
        "count": reasoning.get("count", 0),
        "count_confluence_cases": reasoning.get("count_confluence_cases", 0),
        "candidates": reasoning.get("candidates", []),
        "track_record": {
            "definition": track.get("definition"),
            "n_signals_archived": track.get("n_signals_archived", 0),
            "n_horizon_results": track.get("n_horizon_results", 0),
            "n_hit": track.get("n_hit", 0),
            "n_beat_benchmark": track.get("n_beat_benchmark", 0),
            "results": track.get("results", []),
        },
        "note": "Ranked, evidence-based answer to a canonical question over the existing confluence data "
                "(config/confluence.json), plus a per-signal track record of archived signals whose measurement "
                "horizon has already elapsed (config/performance_rules.json). See /reasoning.html and "
                "/performance.html.",
    })

    # 4. OpenAPI 3.1-specificatie van de publieke endpoints + MCP-verwijzing.
    def get_op(summary, desc):
        return {"get": {"summary": summary, "description": desc,
                        "responses": {"200": {"description": "OK"}}}}

    mcp_tools = ["confluence_signals", "company_intelligence", "resolve_company",
                 "insider_trading_clusters", "insider_selling_form144", "hedge_fund_13f",
                 "sec_8k_events", "activist_stakes_13dg", "biotech_catalysts", "fda_drug_actions",
                 "government_contracts", "congress_trading", "nih_research_funding",
                 "startup_funding_form_d", "crypto_momentum_scanner", "app_review_intelligence",
                 "usage"]
    jwrite("api/openapi.json", {
        "openapi": "3.1.0",
        "info": {
            "title": "DataSignals Lab public surface",
            "version": today.isoformat(),
            "description": PRICING["value_prop"] + " This spec covers the free public JSON "
                           "endpoints. Live signal data is available through the MCP server "
                           "(per result) or DataSignals Pro (subscription).",
            "contact": {"email": "support@datasignalslab.com", "url": BASE},
        },
        "servers": [{"url": BASE}],
        "paths": {
            "/api/streams.json": get_op("Stream registry",
                                        "Every data stream with source, cadence, machine-readable schema and changelog."),
            "/api/confluence-public.json": get_op("Confluence cases (delayed)",
                                                  f"Confluence cases, public with a {conf_cfg['free_delay_days']} day delay. Live for Pro."),
            "/api/proof.json": get_op("Proof chain status",
                                      "Daily hash chain over the published signal set; full chain at /proof/chain.jsonl, each day independently timestamped in Bitcoin via OpenTimestamps."),
            "/api/reasoning.json": get_op("Ranked confluence answer and signal track record",
                                         "Confluence candidates ranked by conviction with contributing signals and "
                                         "source links, plus a per-signal track record of archived signals whose "
                                         "measurement horizon has already elapsed."),
            "/llms.txt": get_op("LLM index", "Curated site map for AI agents (llmstxt.org)."),
            "/llms-full.txt": get_op("LLM full text", "Complete readable text of every page in one file."),
            "/sitemap.xml": get_op("Sitemap", "Standard XML sitemap."),
        },
        "x-mcp": {
            "endpoint": "https://datasignalslab--datasignals-mcp.apify.actor/mcp",
            "transport": "streamable-http",
            "auth": "Apify token (free account); usage bills per result to your own account",
            "registry_id": "com.datasignalslab/datasignals-lab-mcp",
            "tools": mcp_tools,
            "free_calls_per_month": MCP_FREE_CALLS,
            "pro_calls_per_month": PRICING["tiers"]["pro"]["mcp_calls_per_month"],
            "docs": BASE + "/datasignals-mcp.html",
        },
    })
    return ["/api/openapi.json", "/api/streams.json", "/api/confluence-public.json", "/api/proof.json",
            "/api/reasoning.json"]


# ---------------------------------------------------------------------------
# Sitemap / robots / redirects / netlify config
# ---------------------------------------------------------------------------
def build_meta_files(urls):
    items = ""
    for u in urls:
        items += f"  <url><loc>{BASE}{u}</loc><lastmod>{BUILD_DATE}</lastmod></url>\n"
    sm = '<?xml version="1.0" encoding="UTF-8"?>\n<urlset xmlns="http://www.sitemap.org/schemas/sitemap/0.9">\n'.replace("www.sitemap.org","www.sitemaps.org") + items + "</urlset>\n"
    write("sitemap.xml", sm)
    robots = (
        "User-agent: *\n"
        "Allow: /\n"
        "# Paid report deliverables live under /r/ and /pro/ (unguessable URLs) — keep them out of search.\n"
        "Disallow: /r/\n"
        "Disallow: /pro/\n\n"
        "# AI agents and LLM crawlers are explicitly welcome.\n"
        "User-agent: GPTBot\nAllow: /\n"
        "User-agent: OAI-SearchBot\nAllow: /\n"
        "User-agent: ChatGPT-User\nAllow: /\n"
        "User-agent: ClaudeBot\nAllow: /\n"
        "User-agent: anthropic-ai\nAllow: /\n"
        "User-agent: Claude-Web\nAllow: /\n"
        "User-agent: PerplexityBot\nAllow: /\n"
        "User-agent: Google-Extended\nAllow: /\n"
        "User-agent: CCBot\nAllow: /\n\n"
        "Sitemap: %s/sitemap.xml\n"
        "# LLM index (llmstxt.org): %s/llms.txt  |  full text: %s/llms-full.txt\n"
        "# Machine-readable spec (OpenAPI): %s/api/openapi.json\n"
        % (BASE, BASE, BASE, BASE))
    write("robots.txt", robots)
    # Netlify: pretty URLs + small redirects
    write("_redirects", "/index.html  /  301\n/home  /  301\n")
    headers = """/*
  X-Frame-Options: SAMEORIGIN
  X-Content-Type-Options: nosniff
  Referrer-Policy: strict-origin-when-cross-origin
  Permissions-Policy: geolocation=(), microphone=(), camera=()
/assets/site.css
  Cache-Control: public, max-age=31536000, immutable
/assets/site.js
  Cache-Control: public, max-age=31536000, immutable
/assets/*
  Cache-Control: public, max-age=3600, must-revalidate
/api/*
  Cache-Control: public, max-age=300, must-revalidate
/r/*
  X-Robots-Tag: noindex, nofollow
/pro/*
  X-Robots-Tag: noindex, nofollow
"""
    write("_headers", headers)

# ---------------------------------------------------------------------------
# Proof artifacts (display-only publish)
# ---------------------------------------------------------------------------
def publish_proof():
    """Copy the proof chain and each day's fingerprint + Bitcoin OpenTimestamps
    anchor verbatim into public/proof/ so the proof page can link to a live,
    self-hosted per-day file and anyone can recompute the chain and verify the
    Bitcoin timestamp without depending on any third-party host.

    This only *copies* files that the daily pipeline (scripts/proof_chain.py)
    already produced under maintenance/proof/. It never regenerates, reorders or
    alters the chain, the day hashes or the tags - display only. The .bak working
    files are internal and are not published.
    """
    src_days = os.path.join(ROOT, "maintenance", "proof", "days")
    dst_proof = os.path.join(OUT, "proof")
    dst_days = os.path.join(dst_proof, "days")
    os.makedirs(dst_days, exist_ok=True)
    chain = os.path.join(ROOT, "maintenance", "proof", "chain.jsonl")
    if os.path.isfile(chain):
        shutil.copy2(chain, os.path.join(dst_proof, "chain.jsonl"))
    if os.path.isdir(src_days):
        for fn in sorted(os.listdir(src_days)):
            # the day fingerprint (.json) and its Bitcoin OTS proof (.json.ots) only
            if fn.endswith(".json") or fn.endswith(".json.ots"):
                shutil.copy2(os.path.join(src_days, fn), os.path.join(dst_days, fn))


# ---------------------------------------------------------------------------
# Main
# ---------------------------------------------------------------------------
def main():
    if os.path.isdir(OUT):
        shutil.rmtree(OUT)
    os.makedirs(OUT)

    # assets
    shutil.copytree(os.path.join(ROOT, "assets"), os.path.join(OUT, "assets"),
                    ignore=shutil.ignore_patterns("css"))
    shutil.copy(os.path.join(ROOT, "favicon.ico"), os.path.join(OUT, "favicon.ico"))
    publish_proof()

    build_home()

    article_pages = [
        "sec-form4-insider-buying", "sec-form144-insider-selling", "sec-13dg-activist-stakes",
        "smart-money-13f", "sec-8k-material-events", "startup-funding-form-d",
        "biotech-catalyst-monitor", "fda-drug-approval-monitor", "nih-research-funding-monitor",
        "gov-contract-awards-monitor", "congress-trading-monitor",
        "crypto-volume-momentum-scanner", "app-store-review-intelligence", "datasignals-mcp",
    ]
    for name in article_pages:
        build_article(os.path.join(ROOT, name + ".md"), name + ".html", active="Products")

    # compare.html wordt door funnel.build_compare() gegenereerd uit config/competitors.json
    build_article(os.path.join(ROOT, "about.md"), "about.html", active="About")
    build_article(os.path.join(ROOT, "what-is-sec-form-144.md"), "what-is-sec-form-144.html", active="Guides")
    build_article(os.path.join(ROOT, "13d-vs-13g-explained.md"), "13d-vs-13g-explained.html", active="Guides")
    build_article(os.path.join(ROOT, "privacy.md"), "privacy.html", active="")
    build_article(os.path.join(ROOT, "terms.md"), "terms.html", active="")

    build_posts()
    build_404()
    api_urls = build_api()
    build_llms_txt()
    build_llms_full()

    build_reports_page()

    # Funnelpaginas (PROJECTINSTRUCTIE_v4.0.md secties 3-6, 10): pricing,
    # landingspaginas, compare, performance, programmatische SEO.
    import funnel
    funnel_urls = funnel.build_all()

    # collect urls for sitemap
    urls = ["/"]
    for name in article_pages + ["reports", "compare", "about", "what-is-sec-form-144", "13d-vs-13g-explained", "blog", "privacy", "terms"]:
        urls.append("/" + name + ".html")
    for p in sorted(glob.glob(os.path.join(ROOT, "_posts", "*.md"))):
        urls.append(f"/blog/{post_slug(p)}/")
    urls += funnel_urls
    # api_urls bewust niet in de sitemap (JSON, geen indexeerbare pagina's);
    # vindbaar via llms.txt, robots.txt en de agents-pagina.
    del api_urls
    build_meta_files(urls)

    # Sectie 11: verificatiebestanden voor Search Console / Bing Webmaster.
    # Alles wat in config/verification/ ligt (google*.html, BingSiteAuth.xml)
    # komt in de siteroot terecht.
    verif_dir = os.path.join(ROOT, "config", "verification")
    if os.path.isdir(verif_dir):
        for fn in os.listdir(verif_dir):
            shutil.copy2(os.path.join(verif_dir, fn), os.path.join(OUT, fn))
            print("Verificatiebestand gekopieerd:", fn)

    n = sum(len(f) for _, _, f in os.walk(OUT))
    print("Built", n, "files into", OUT)

    # Kwaliteitspoorten (sectie 16.4 en 17 laag 1.4): kapotte interne links en
    # prijsafwijkingen zijn faalredenen. De build slaagt pas als beide schoon zijn.
    broken = funnel.check_internal_links(OUT)
    if broken:
        print("KAPOTTE INTERNE LINKS (%d):" % len(broken))
        for page, href in broken[:50]:
            print(f"  - {page} -> {href}")
        raise SystemExit(1)
    print("Linkcheck geslaagd: geen kapotte interne links.")

    import check_consistency
    if check_consistency.main() != 0:
        raise SystemExit(1)

if __name__ == "__main__":
    main()