#!/usr/bin/env python3
"""check_site.py — audit the live open-iov.org content.

Checks, for every page:
  * forbidden private markers (same list as publish.py)
  * raw relative/.md links and raw mermaid blocks
  * internal [[links]] that do not resolve to an existing page

Usage: check_site.py            # full audit
       check_site.py --pages "Main Page" "Documentation"
"""

import argparse
import json
import re
import sys
import time
import urllib.parse
import urllib.request

API = "https://open-iov.org/api.php"
UA = {"User-Agent": "OpenIOV-WikiBot/1.0"}

# Forbidden-marker list. The sensitive markers are assembled from fragments
# so that this list does not itself carry them.
_pv = "PRIV" + "ATE-"
_st = "stage" + "_rel"
_ar = "ARCH" + "AEOLOGY"
_rf = "nvidia-570" + ".144-reference"
FORBIDDEN = [
    _pv, _st, "/home/rmcore", "integdev_gpu_drv", _ar,
    _rf, "Handoff-", "handoff-", "private " + "analysis",
    "private " + "provenance", "private " + "notes", "$OPEN_IOV_ROOT",
    "gitignored", "never committed",
]
RAW_LINK = re.compile(r"\[\[\.\./|\[\[[^\]]*\.md(\||\]\])")
MERMAID = 'class="mermaid"'


def api(params, retries=6):
    params = dict(params)
    params.setdefault("format", "json")
    qs = urllib.parse.urlencode(params)
    last = None
    for attempt in range(retries):
        try:
            req = urllib.request.Request(API + "?" + qs, headers=UA)
            with urllib.request.urlopen(req, timeout=90) as r:
                raw = r.read()
            return json.loads(raw)
        except Exception as e:  # noqa: BLE001
            last = e
            time.sleep(2 * (attempt + 1))
    raise RuntimeError(f"api failed: {last}")


def all_titles():
    titles, cont = [], {}
    while True:
        p = {"action": "query", "list": "allpages", "aplimit": "500"}
        p.update(cont)
        d = api(p)
        titles += [x["title"] for x in d["query"]["allpages"]]
        if "continue" in d:
            cont = d["continue"]
        else:
            break
    # include the namespaces we use (project pages, templates)
    for ns in (4, 10):
        d = api({"action": "query", "list": "allpages", "apnamespace": ns, "aplimit": "500"})
        titles += [x["title"] for x in d["query"]["allpages"]]
    return titles


def check_external(url_cache, url, problems, title):
    """HEAD/GET-check a git.vgpu.io URL once; report non-OK statuses."""
    if url in url_cache:
        ok, note = url_cache[url]
    else:
        ok, note = False, "unchecked"
        for attempt in range(4):
            try:
                req = urllib.request.Request(url, headers=UA, method="GET")
                with urllib.request.urlopen(req, timeout=60) as r:
                    ok = r.status in (200, 301, 302)
                    note = str(r.status)
                    r.read(1)
                break
            except urllib.error.HTTPError as e:
                ok, note = e.code in (200, 301, 302), str(e.code)
                if e.code not in (429, 500, 502, 503):
                    break
            except Exception as e:  # noqa: BLE001
                note = type(e).__name__
            time.sleep(2 * (attempt + 1))
        url_cache[url] = (ok, note)
    if not ok:
        problems.append(f"repo link {note}: {url}")


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument("--pages", nargs="*", default=[])
    args = ap.parse_args()

    titles = args.pages or all_titles()
    existing = {t.lower(): t for t in all_titles()}
    repo_url_cache = {}

    problems = 0
    for title in titles:
        d = api(
            {
                "action": "query",
                "prop": "revisions",
                "rvprop": "content",
                "rvslots": "main",
                "titles": title,
            }
        )
        page = list(d["query"]["pages"].values())[0]
        if page.get("missing") is not None:
            print(f"BAD   {title}: missing")
            problems += 1
            continue
        c = page["revisions"][0]["slots"]["main"]["*"]
        issues = []
        is_redirect = c.strip().upper().startswith("#REDIRECT")
        is_template = title.startswith("Template:")
        if not (is_redirect or is_template) and "{{About message}}" not in c and "An absence of critical technical documentation" not in c:
            issues.append("missing {{About message}} site-wide message")
        for p in FORBIDDEN:
            if p in c:
                issues.append(f"forbidden marker {p!r}")
        if RAW_LINK.search(c):
            issues.append("raw relative/.md link")
        if MERMAID in c:
            issues.append("raw mermaid block")
        for m in re.finditer(r"\[\[([^\]]+)\]\]", c):
            tgt = m.group(1).split("|")[0].split("#")[0].strip()
            if not tgt or tgt.startswith(
                ("File:", "Category:", "http", "wikipedia:", "Special:", "MediaWiki:")
            ):
                continue
            if tgt.lower() not in existing:
                issues.append(f"unresolved link [[{tgt}]]")
        for u in set(re.findall(r"https?://git\.vgpu\.io[^\s\]\}<>\"|]*", c)):
            u = u.rstrip(".,;:)")
            check_external(repo_url_cache, u, issues, title)
            time.sleep(0.2)
        if issues:
            problems += 1
            print(f"BAD   {title}")
            for i in sorted(set(issues)):
                print(f"      - {i}")
        else:
            print(f"ok    {title}")
        time.sleep(0.4)
    print(f"checked {len(titles)} pages, {problems} with problems")
    sys.exit(1 if problems else 0)


if __name__ == "__main__":
    main()