#!/usr/bin/env python3
"""verify-blog-titles - assert generated blog posts carry the FULL database title.
Catches the deblog truncation bug: the 55-char Page Title bleeding into the
on-page heading, og:title and the JSON-LD headline. The database title has four
destinations and exactly one of them may differ - the Page Title - and it
differs by REWRITE, never by slicing at a word boundary.
Checks, per published post in the DB dump:
heading (h1/h2 in .post) == database title, verbatim
og:title == database title, verbatim
JSON-LD "headline" == database title, verbatim, no HTML entities
Page Title "> 1 else "posts.json"
BLOG = sys.argv[2] if len(sys.argv) > 2 else "blog"
CATS = sys.argv[3] if len(sys.argv) > 3 else None
PAGE_TITLE_MAX = 55
def canonical(title):
"""Database title -> plain text.
DB titles arrive HTML-escaped ('&'), so unescape first, then apply the
house convention: straight quotes, no non-breaking spaces.
"""
title = html.unescape(title)
return (title.replace("’", "'").replace("‘", "'")
.replace("“", '"').replace("”", '"')
.replace("\xa0", " ").strip())
def load_posts(path):
data = json.load(open(path, encoding="utf-8"))
posts = data["posts"] if isinstance(data, dict) else data
return [p for p in posts if p.get("status", "publish") == "publish"]
def find_file(files, post, claimed=None):
"""Match a DB post to its generated file. Returns (name, warn).
Convention: dates are stripped from filenames; the date lives in the
JSON-LD datePublished. A -YYMMDD suffix on the file is the old convention
and only triggers a warning. Duplicate stripped slugs disambiguate with a
-YYYY suffix on the older copy; the title decides which candidate is the
post's file.
"""
plain = post["slug"] + ".aspx"
if plain in files:
if re.search(r"-\d{6}$", post["slug"]):
return plain, "dated filename (convention is date-stripped; the date lives in JSON-LD datePublished)"
return plain, None
dated = re.compile("^" + re.escape(post["slug"]) + r"-\d{6}\.aspx$")
for name in files:
if dated.match(name):
return name, "dated filename (convention is date-stripped; the date lives in JSON-LD datePublished)"
stripped = re.sub(r"-aNaNaN$", "", post["slug"])
stripped = re.sub(r"-\d{6}$", "", stripped)
stripped = stripped.rstrip("-")
candidates = [stripped + ".aspx"]
if "date" in post and post["date"]:
candidates.append(stripped + "-" + post["date"][:4] + ".aspx")
full = canonical(post["title"])
claimed = claimed or set()
for cand in candidates:
if cand not in files or cand in claimed:
continue
og = re.search(r'og:title"\s+content="(.*?)"\s*/?>', files[cand])
if not og or html.unescape(og.group(1)).strip() == full:
return cand, None
return None, None
def load_categories(path):
data = json.load(open(path, encoding="utf-8"))
valid = set()
for category in data["categories"]:
valid.add(category["slug"])
valid.update(category.get("aliases", []))
return valid
def check_categories(src, valid):
m = re.search(r'', src)
if not og:
out.append(("og:title", "missing"))
elif html.unescape(og.group(1)).strip() != full:
out.append(("og:title", html.unescape(og.group(1)).strip()))
ld = re.search(r'"headline":\s*"((?:[^"\\]|\\.)*)"', src)
if not ld:
out.append(("json-ld", "missing"))
else:
value = json.loads('"%s"' % ld.group(1)).strip()
if value != full:
out.append(("json-ld", value))
elif "&" in value or "&#" in value or """ in value:
#