diff options
Diffstat (limited to 'tools/site/check_site.py')
| -rw-r--r-- | tools/site/check_site.py | 31 |
1 files changed, 31 insertions, 0 deletions
diff --git a/tools/site/check_site.py b/tools/site/check_site.py index e56d0c0..e79af25 100644 --- a/tools/site/check_site.py +++ b/tools/site/check_site.py @@ -72,6 +72,32 @@ def all_titles(): 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=[]) @@ -79,6 +105,7 @@ def main(): titles = args.pages or all_titles() existing = {t.lower(): t for t in all_titles()} + repo_url_cache = {} problems = 0 for title in titles: @@ -117,6 +144,10 @@ def main(): 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}") |