428th-website-2026/scripts/build.py

#!/usr/bin/env python3
"""Build static localized HTML for 428th.com (ru / en / ua)."""

from __future__ import annotations

import json
import re
from pathlib import Path

ROOT = Path(__file__).resolve().parent.parent
SRC = ROOT / "src"
DIST = ROOT
SITE = "https://428th.com"

LOCALES = [
    {
        "id": "ru",
        "prefix": "",
        "html_lang": "ru",
        "hreflang": "ru",
        "dict": "ru",
        "og_locale": "ru_RU",
        "label": "RU",
    },
    {
        "id": "en",
        "prefix": "/en",
        "html_lang": "en",
        "hreflang": "en",
        "dict": "en",
        "og_locale": "en_US",
        "label": "EN",
    },
    {
        "id": "ua",
        "prefix": "/ua",
        "html_lang": "uk",
        "hreflang": "uk",
        "dict": "uk",
        "og_locale": "uk_UA",
        "label": "UA",
    },
]

PAGES = [
    {
        "id": "home",
        "template": "index.html",
        "title_key": "meta.title",
        "desc_key": "meta.description",
        "path_for": lambda prefix: f"{prefix}/" if prefix else "/",
        "out_for": lambda prefix: (
            DIST / "index.html"
            if not prefix
            else DIST / prefix.lstrip("/") / "index.html"
        ),
        "depth_for": lambda prefix: 0 if not prefix else 1,
    },
    {
        "id": "terminal",
        "template": "terminal.html",
        "title_key": "terminal.meta.title",
        "desc_key": "terminal.meta.description",
        "path_for": lambda prefix: f"{prefix}/terminal/" if prefix else "/terminal/",
        "out_for": lambda prefix: (
            DIST / "terminal" / "index.html"
            if not prefix
            else DIST / prefix.lstrip("/") / "terminal" / "index.html"
        ),
        "depth_for": lambda prefix: 1 if not prefix else 2,
    },
]


def asset_prefix(_depth: int) -> str:
    """Root-absolute assets — avoids FOUC on /terminal vs /terminal/ and deep locales."""
    return "/"


def t(translations: dict, dict_key: str, key: str) -> str:
    value = translations.get(dict_key, {}).get(key)
    if value is None:
        value = translations.get("ru", {}).get(key, "")
    return value


def localize_urls(html: str, prefix: str) -> str:
    if not prefix:
        return html

    html = re.sub(
        r'href="/terminal/?"',
        f'href="{prefix}/terminal/"',
        html,
    )
    html = re.sub(
        r'href="/#([^"]*)"',
        lambda m: f'href="{prefix}/#{m.group(1)}"',
        html,
    )
    return html


def fill_i18n(html: str, translations: dict, dict_key: str) -> str:
    def repl_el(match: re.Match) -> str:
        open_tag, key, close = match.group(1), match.group(2), match.group(4)
        open_tag = re.sub(r'\sdata-i18n="[^"]+"', "", open_tag)
        return f"{open_tag}{t(translations, dict_key, key)}{close}"

    html = re.sub(
        r'(<\w+\b[^>]*\sdata-i18n="([^"]+)"[^>]*>)(.*?)(</\w+>)',
        repl_el,
        html,
        flags=re.DOTALL,
    )

    def repl_placeholder(match: re.Match) -> str:
        before, key, after = match.group(1), match.group(2), match.group(3)
        value = esc_attr(t(translations, dict_key, key))
        return f'{before}placeholder="{value}"{after}'

    html = re.sub(
        r'(\s)data-i18n-placeholder="([^"]+)"([^>]*>)',
        repl_placeholder,
        html,
    )

    def repl_aria(match: re.Match) -> str:
        before, key, after = match.group(1), match.group(2), match.group(3)
        value = esc_attr(t(translations, dict_key, key))
        after = re.sub(r'\saria-label="[^"]*"', "", after)
        before = re.sub(r'\saria-label="[^"]*"', "", before)
        return f'{before} aria-label="{value}"{after}'

    html = re.sub(
        r'(<\w+\b[^>]*)\sdata-i18n-aria="([^"]+)"([^>]*>)',
        repl_aria,
        html,
    )
    return html


def page_url(locale: dict, page: dict) -> str:
    return SITE + page["path_for"](locale["prefix"])


def alternate_links(page: dict) -> str:
    lines = []
    for loc in LOCALES:
        lines.append(
            f'<link rel="alternate" hreflang="{loc["hreflang"]}" href="{page_url(loc, page)}">'
        )
    lines.append(
        f'<link rel="alternate" hreflang="x-default" href="{page_url(LOCALES[0], page)}">'
    )
    return "\n".join(lines)


def lang_switch(locale: dict, page: dict) -> str:
    parts = []
    order = ["en", "ua", "ru"]
    by_id = {loc["id"]: loc for loc in LOCALES}
    for i, lid in enumerate(order):
        loc = by_id[lid]
        href = page["path_for"](loc["prefix"])
        active = ' class="active"' if loc["id"] == locale["id"] else ""
        aria = ' aria-current="page"' if loc["id"] == locale["id"] else ""
        parts.append(
            f'<a href="{href}" data-set-lang="{loc["id"]}"{active}{aria}>{loc["label"]}</a>'
        )
        if i < len(order) - 1:
            parts.append(" //")
    return "\n      ".join(parts)


def esc_attr(value: str) -> str:
    return (
        value.replace("&", "&amp;")
        .replace('"', "&quot;")
        .replace("<", "&lt;")
        .replace(">", "&gt;")
    )


def head_meta(locale: dict, page: dict, translations: dict) -> str:
    title = t(translations, locale["dict"], page["title_key"])
    desc = t(translations, locale["dict"], page["desc_key"])
    url = page_url(locale, page)
    return f"""<title>{title}</title>
<meta name="description" content="{esc_attr(desc)}">
<meta name="robots" content="index, follow">
<link rel="canonical" href="{url}">
{alternate_links(page)}

<meta property="og:type" content="website">
<meta property="og:site_name" content="428th">
<meta property="og:locale" content="{locale["og_locale"]}">
<meta property="og:url" content="{url}">
<meta property="og:title" content="{esc_attr(title)}">
<meta property="og:description" content="{esc_attr(desc)}">
<meta property="og:image" content="{SITE}/og.png">
<meta property="og:image:width" content="1200">
<meta property="og:image:height" content="630">

<meta name="twitter:card" content="summary_large_image">
<meta name="twitter:title" content="{esc_attr(title)}">
<meta name="twitter:description" content="{esc_attr(desc)}">
<meta name="twitter:image" content="{SITE}/og.png">

<meta name="theme-color" content="#C4622A">"""


LANG_REDIRECT_SCRIPT = r"""<script>
(function () {
  var pageLang = document.documentElement.getAttribute('data-page-lang');
  var saved = localStorage.getItem('lang');
  if (saved === 'uk') { saved = 'ua'; localStorage.setItem('lang', 'ua'); }

  function normalizePath(path) {
    if (!path || path === '/') return '/';
    return path.replace(/\/+$/, '') || '/';
  }

  function pathFor(lang) {
    var path = location.pathname || '/';
    var rest = path.replace(/^\/(en|ua)(?=\/|$)/, '') || '/';
    if (rest === '/index.html') rest = '/';
    rest = normalizePath(rest);

    if (lang === 'ru') {
      if (rest === '/') return '/';
      if (rest === '/terminal') return '/terminal/';
      return rest;
    }
    if (rest === '/') return '/' + lang + '/';
    if (rest === '/terminal') return '/' + lang + '/terminal/';
    return '/' + lang + rest;
  }

  function needsRedirect(target) {
    return normalizePath(location.pathname) !== normalizePath(target);
  }

  if (saved === 'ru' || saved === 'en' || saved === 'ua') {
    if (saved !== pageLang) {
      var target = pathFor(saved);
      if (needsRedirect(target)) location.replace(target);
    }
    return;
  }

  if (pageLang !== 'ru') {
    localStorage.setItem('lang', pageLang);
    return;
  }

  var browser = String(navigator.language || '').slice(0, 2).toLowerCase();
  var preferred = 'ru';
  if (browser === 'en') preferred = 'en';
  else if (browser === 'uk' || browser === 'ua') preferred = 'ua';

  if (preferred !== 'ru') {
    var preferredTarget = pathFor(preferred);
    if (needsRedirect(preferredTarget)) location.replace(preferredTarget);
  } else {
    localStorage.setItem('lang', 'ru');
  }
})();
</script>"""


THEME_SCRIPT = """<script>
(function () {
  var saved = localStorage.getItem('theme');
  var theme = (saved === 'light' || saved === 'dark')
    ? saved
    : (window.matchMedia('(prefers-color-scheme: dark)').matches ? 'dark' : 'light');
  document.documentElement.setAttribute('data-theme', theme);
})();
</script>"""


def home_href(locale: dict) -> str:
    return f"{locale['prefix']}/" if locale["prefix"] else "/"


def terminal_href(locale: dict) -> str:
    return f"{locale['prefix']}/terminal/" if locale["prefix"] else "/terminal/"


def build_page(locale: dict, page: dict, translations: dict, template: str) -> str:
    prefix = locale["prefix"]
    depth = page["depth_for"](prefix)
    assets = asset_prefix(depth)

    html = template
    html = html.replace("{{HTML_LANG}}", locale["html_lang"])
    html = html.replace("{{PAGE_LANG}}", locale["id"])
    html = html.replace("{{HEAD_META}}", head_meta(locale, page, translations))
    html = html.replace("{{THEME_SCRIPT}}", THEME_SCRIPT)
    html = html.replace("{{LANG_REDIRECT_SCRIPT}}", LANG_REDIRECT_SCRIPT)
    html = html.replace("{{HOME_HREF}}", home_href(locale))
    html = html.replace("{{TERMINAL_HREF}}", terminal_href(locale))
    html = html.replace("{{ASSET_PREFIX}}", assets)

    html = html.replace(
        "{{FORM_STATUS_LOADING}}",
        esc_attr(t(translations, locale["dict"], "form.statusLoading")),
    )
    html = html.replace(
        "{{FORM_STATUS_SUCCESS}}",
        esc_attr(t(translations, locale["dict"], "form.statusSuccess")),
    )

    html = fill_i18n(html, translations, locale["dict"])
    html = localize_urls(html, prefix)
    # After localize_urls so RU /terminal/ in the switch is not rewritten to /en/terminal/
    html = html.replace("{{LANG_SWITCH}}", lang_switch(locale, page))
    return html


def write_sitemap() -> None:
    urls = []
    for page in PAGES:
        for loc in LOCALES:
            loc_url = page_url(loc, page)
            if page["id"] == "home" and loc["id"] == "ru":
                priority = "1.0"
            elif page["id"] == "home" or loc["id"] == "ru":
                priority = "0.9"
            else:
                priority = "0.8"
            urls.append(
                f"""  <url>
    <loc>{loc_url}</loc>
    <changefreq>weekly</changefreq>
    <priority>{priority}</priority>
  </url>"""
            )
    content = (
        '<?xml version="1.0" encoding="UTF-8"?>\n'
        '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n'
        + "\n".join(urls)
        + "\n</urlset>\n"
    )
    (DIST / "sitemap.xml").write_text(content, encoding="utf-8")


def main() -> None:
    translations = json.loads((SRC / "translations.json").read_text(encoding="utf-8"))
    templates = {
        "index.html": (SRC / "index.html").read_text(encoding="utf-8"),
        "terminal.html": (SRC / "terminal.html").read_text(encoding="utf-8"),
    }

    for page in PAGES:
        for loc in LOCALES:
            out = page["out_for"](loc["prefix"])
            out.parent.mkdir(parents=True, exist_ok=True)
            html = build_page(loc, page, translations, templates[page["template"]])
            out.write_text(html, encoding="utf-8")
            print(f"wrote {out.relative_to(DIST)}")

    write_sitemap()
    print("wrote sitemap.xml")
    print("build ok")


if __name__ == "__main__":
    main()