pw/scripts/postbuild.py

#!/usr/bin/env python3
"""Efterbehandling af Zolas output i public/:

- overskrift-id'er som på den gamle side (bevarer æøå, så gamle #ankre stadig virker)
- sitemap-index.xml + sitemap-0.xml med lastmod fra git og frontmatter
- sw.js med precache-liste
- PW_PREVIEW=1: hele sitet flyttes ind under /udkast, uden sitemap og service worker
"""
import hashlib
import html
import json
import os
import pathlib
import re
import subprocess

OUT = pathlib.Path("public")
SITE = "https://asger.weirsoe.dk"
PREVIEW = os.environ.get("PW_PREVIEW") == "1"
BASE = "/udkast"


def pages():
    return sorted(OUT.rglob("*.html"))


def slug(text):
    """Samme regler som github-slugger, som den gamle build brugte."""
    s = html.unescape(re.sub(r"<[^>]+>", "", text)).strip().lower()
    s = re.sub(r"[^\w\- ]", "", s)
    return s.replace(" ", "-")


def heading_ids():
    for f in pages():
        src = f.read_text()
        if 'class="prose' not in src:
            continue
        seen = {}

        def fix(m):
            s = slug(m[3])
            n = seen.get(s, 0)
            seen[s] = n + 1
            return f'<h{m[1]} id="{s}-{n}"{m[2]}>' if n else f'<h{m[1]} id="{s}"{m[2]}>'

        f.write_text(re.sub(r'<h([1-6]) id="[^"]*"([^>]*)>(.*?)(?=</h\1>)',
                            lambda m: fix(m) + m[3], src, flags=re.S))


def git_date(*files):
    if int(subprocess.run(["git", "rev-list", "--count", "HEAD"], capture_output=True, text=True).stdout or 0) < 2:
        return None
    out = subprocess.run(["git", "log", "-1", "--format=%cI", "--", *files], capture_output=True, text=True).stdout
    return out.strip() or None


def frontmatter(path):
    text = path.read_text()
    head = text.split("---\n", 2)[1]
    return {k: v.strip().strip("'\"") for k, v in re.findall(r"^\s*(\w+):(.*)$", head, re.M)}


def iso(date):
    return f"{date}T00:00:00.000Z"


def lastmods():
    """Rute -> lastmod: sidste commit for statiske sider, frontmatter for blogindlæg."""
    routes = {}
    for md in pathlib.Path("content").rglob("*.md"):
        fm = frontmatter(md)
        if "blog" in md.parts or md.parts[1] == "projekter" or "projects" in md.parts:
            continue
        route = "/" + str(md.relative_to("content").with_suffix("")).removesuffix("_index")
        route = route.rstrip("/") or "/"
        tpl = fm.get("template")
        d = git_date(str(md), f"templates/{tpl}") if tpl else git_date(str(md))
        if d:
            routes[route] = d
    posts = []
    for md in pathlib.Path("content/blog").glob("*/index.md"):
        fm = frontmatter(md)
        d = iso(fm.get("updated") or fm["date"])
        routes[f"/blog/{md.parent.name}"] = d
        posts.append(d)
    if posts:
        routes["/blog"] = max(posts)
    return routes


def sitemap():
    skip = ("/tags", "/peppol")
    skip_exact = ("/cookies", "/status", "/chat")
    routes = []
    for f in pages():
        rel = f.relative_to(OUT)
        if rel.name != "index.html":
            continue
        r = "/" + str(rel.parent).replace(".", "", 1) if str(rel.parent) == "." else f"/{rel.parent}"
        if r.startswith(skip) or r in skip_exact:
            continue
        routes.append(r)
    routes.append("/tarballz/")
    mods = lastmods()
    urls = []
    for r in sorted(routes):
        mod = mods.get(r)
        urls.append(f"<url><loc>{SITE}{r}</loc>" + (f"<lastmod>{mod}</lastmod>" if mod else "")
                    + f"<changefreq>weekly</changefreq><priority>{'1.0' if r == '/' else '0.7'}</priority></url>")
    ns = 'xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"'
    (OUT / "sitemap-0.xml").write_text(f'<?xml version="1.0" encoding="UTF-8"?><urlset {ns}>{"".join(urls)}</urlset>')
    newest = max(mods.values(), default=None)
    (OUT / "sitemap-index.xml").write_text(
        f'<?xml version="1.0" encoding="UTF-8"?><sitemapindex {ns}><sitemap><loc>{SITE}/sitemap-0.xml</loc>'
        + (f"<lastmod>{newest}</lastmod>" if newest else "") + "</sitemap></sitemapindex>")


def service_worker():
    """Precache-listen afhænger af buildet, resten af sw.js ligger i scripts/sw.js."""
    exts = {".html", ".css", ".js", ".svg", ".png", ".woff2", ".xml", ".txt", ".webmanifest"}
    files = sorted(p for p in OUT.rglob("*") if p.is_file() and p.suffix in exts
                   and p.name not in ("sw.js", "mermaid.min.js"))
    h = hashlib.sha256()
    for p in files:
        h.update(p.read_bytes())
    urls = ["/" + str(p.relative_to(OUT)) for p in files]
    src = pathlib.Path("scripts/sw.js").read_text()
    src = src.replace("__VERSION__", h.hexdigest()[:12]).replace("__URLS__", json.dumps(urls, ensure_ascii=False))
    (OUT / "sw.js").write_text(src)


def rebase():
    """Rod-relative links og billeder i HTML peger ind under /udkast."""
    attr = re.compile(r'\b(href|src|action)="/(?!/|udkast)')
    srcset = re.compile(r'\bsrcset="([^"]*)"')

    def fix_set(m):
        return 'srcset="' + re.sub(r"(^|, )/(?!/)", rf"\1{BASE}/", m[1]) + '"'

    for f in pages():
        src = f.read_text()
        f.write_text(srcset.sub(fix_set, attr.sub(rf'\1="{BASE}/', src)))


def main():
    (OUT / "sitemap.xml").unlink(missing_ok=True)
    heading_ids()
    if PREVIEW:
        rebase()
    else:
        sitemap()
        service_worker()


if __name__ == "__main__":
    main()