ChemaVX
2e552232f3
Los 18 waivers del ES salían como "waived: None": la línea que imprime el motivo llamaba a accepted_reason sin el sitio, así que lo buscaba en el diccionario del EN. El waiver se aplicaba bien —eso sí pasaba el sitio—, pero el informe salía mudo justo donde tiene que justificarse, que es el único propósito de esa sección. Y si algún día vuelve a no haber motivo, ahora lo dice: un waiver sin razón documentada es un fallo, no un hueco que rellenar con None.
246 lines
10 KiB
Python
Executable File
246 lines
10 KiB
Python
Executable File
#!/usr/bin/env python3
|
||
"""
|
||
Tool A — site-wide SEO auditor para los dos blogs (Ghost).
|
||
|
||
READ-ONLY. Fetches every published post via the ghst-en wrapper (Ghost Admin API,
|
||
staff token pulled on-demand from the pod — nothing written to disk), runs the
|
||
shared rule engine (seo_rules.py), and prints:
|
||
1. a summary of aggregate violations,
|
||
2. a per-post table sorted worst-first,
|
||
3. a ranked "fix this first" list.
|
||
|
||
Usage:
|
||
python3 seo_audit.py # EN (por defecto), en vivo
|
||
python3 seo_audit.py --site es # el blog ES
|
||
python3 seo_audit.py --json cache.json # audit a saved posts JSON instead of fetching
|
||
python3 seo_audit.py --save cache.json # fetch live AND save the raw JSON
|
||
|
||
Changes NOTHING in Ghost.
|
||
"""
|
||
import argparse
|
||
import json
|
||
import os
|
||
import subprocess
|
||
import sys
|
||
import tempfile
|
||
from collections import Counter
|
||
|
||
import seo_rules as R
|
||
import seo_exceptions as X
|
||
import seo_hreflang as HL
|
||
|
||
|
||
def partition(post, violations, site="en"):
|
||
"""Split one post's violations into (actionable, accepted, info).
|
||
|
||
'accepted' = an actionable-severity violation whose (rule, slug) is a
|
||
documented waiver in seo_exceptions — kept OUT of the to-fix count and shown
|
||
in its own section. The rule still ran; this is a reporting-side filter only.
|
||
"""
|
||
slug = post.get("slug")
|
||
actionable, accepted, info = [], [], []
|
||
for v in violations:
|
||
if v.severity == R.INFO:
|
||
info.append(v)
|
||
elif X.accepted_reason(v.rule, slug, site):
|
||
accepted.append(v)
|
||
else:
|
||
actionable.append(v)
|
||
return actionable, accepted, info
|
||
|
||
|
||
SITIOS = {"en": {"cli": "ghst-en", "host": "theexclusionzone.com"},
|
||
"es": {"cli": "ghst-es", "host": "zonadeexclusion.com"}}
|
||
|
||
|
||
def fetch_posts(cli="ghst-en"):
|
||
"""Pull all published posts (with html) via the site's ghst wrapper.
|
||
|
||
We redirect ghst's stdout to a temp file rather than capturing the pipe:
|
||
ghst's output is ~750 KB and a captured PIPE truncates at the 64 KB pipe
|
||
buffer, whereas a file redirect gets the whole document.
|
||
"""
|
||
fd, path = tempfile.mkstemp(suffix=".json", prefix="seo_audit_")
|
||
os.close(fd)
|
||
try:
|
||
# ⚠️ `--status published` NO filtra: ghst devuelve published Y scheduled
|
||
# (medido el 2026-07-28: 33 + 9 = los 42 que el informe llamaba
|
||
# "published"). Se pide sin filtro y se separa aquí, que es lo honesto.
|
||
cmd = (f"{cli} post list --limit all "
|
||
"--formats html,lexical --json > " + path)
|
||
r = subprocess.run(cmd, shell=True, stderr=subprocess.PIPE, text=True)
|
||
if r.returncode != 0:
|
||
sys.exit(f"{cli} fetch failed:\n{r.stderr}")
|
||
data = json.load(open(path))
|
||
posts = data["posts"] if isinstance(data, dict) and "posts" in data else data
|
||
return [x for x in posts if x.get("status") in ("published", "scheduled")]
|
||
finally:
|
||
os.unlink(path)
|
||
|
||
|
||
def load_posts(path):
|
||
data = json.load(open(path))
|
||
return data["posts"] if isinstance(data, dict) and "posts" in data else data
|
||
|
||
|
||
def main():
|
||
ap = argparse.ArgumentParser()
|
||
ap.add_argument("--site", choices=("en", "es"), default="en")
|
||
ap.add_argument("--json", help="audit a saved posts JSON instead of fetching live")
|
||
ap.add_argument("--save", help="save fetched raw posts JSON to this path")
|
||
args = ap.parse_args()
|
||
|
||
if args.json:
|
||
posts = load_posts(args.json)
|
||
src = f"file {args.json}"
|
||
else:
|
||
posts = fetch_posts(SITIOS[args.site]["cli"])
|
||
src = f"live ({SITIOS[args.site]['cli']}, Ghost Admin API)"
|
||
if args.save:
|
||
json.dump({"posts": posts}, open(args.save, "w"), indent=2)
|
||
|
||
# El motor decide qué href es interno con R.SITE_HOST: sin esto, auditar
|
||
# el ES contaría cero enlaces internos en los 31 posts.
|
||
R.usar_sitio(args.site)
|
||
|
||
# Las parejas espejo son una decisión editorial curada, no una regla del
|
||
# motor: el auditor las inyecta y seo_rules sigue sin saber que existen.
|
||
try:
|
||
parejas = HL.lee_parejas()
|
||
except OSError:
|
||
parejas = []
|
||
|
||
results = [] # (post, actionable, accepted, info, score)
|
||
for p in posts:
|
||
v = R.check_post(p) + HL.violaciones(p, args.site, parejas)
|
||
actionable, accepted, info = partition(p, v, args.site)
|
||
# score ranks worst-first on ACTIONABLE severity only — accepted waivers
|
||
# and INFO don't push a post up the list.
|
||
results.append((p, actionable, accepted, info, R.score(actionable)))
|
||
|
||
# worst-first: by actionable severity score, then actionable count, then slug
|
||
results.sort(key=lambda t: (-t[4], -len(t[1]), t[0]["slug"]))
|
||
|
||
total_posts = len(posts)
|
||
# SUMMARY counts reflect ACTIONABLE violations only (waived items excluded).
|
||
posts_with_rule = Counter()
|
||
for _, actionable, _, _, _ in results:
|
||
for rule in {v.rule for v in actionable}:
|
||
posts_with_rule[rule] += 1
|
||
|
||
clean = sum(1 for _, actionable, _, _, _ in results if not actionable)
|
||
accepted_total = sum(len(acc) for _, _, acc, _, _ in results)
|
||
|
||
# ---- header ----
|
||
print("=" * 78)
|
||
print(f" SEO AUDIT — {SITIOS[args.site]['host']} source: {src}")
|
||
n_pub = sum(1 for p in posts if p.get("status") == "published")
|
||
n_prog = total_posts - n_pub
|
||
print(f" {n_pub} publicados"
|
||
+ (f" + {n_prog} programados" if n_prog else "")
|
||
+ f" | {clean} clean (no actionable issues)"
|
||
+ (f" | {accepted_total} accepted exception(s) waived" if accepted_total else ""))
|
||
print(f" JSON-LD: handled globally by Edition-main theme — not flagged per post")
|
||
print("=" * 78)
|
||
|
||
# ---- summary ----
|
||
print("\nSUMMARY — violations by rule (most posts affected first):\n")
|
||
LABEL = {
|
||
"meta_description.missing": "meta_description MISSING",
|
||
"meta_description.too_long": f"meta_description > {R.META_DESC_MAX}",
|
||
"meta_title.missing": "meta_title missing (falls back to title)",
|
||
"meta_title.too_long": f"meta_title > {R.META_TITLE_MAX}",
|
||
"custom_excerpt.too_long": f"custom_excerpt > {R.CUSTOM_EXCERPT_MAX}",
|
||
"og_title.empty": "og_title empty",
|
||
"og_description.empty": "og_description empty",
|
||
"twitter_title.empty": "twitter_title empty",
|
||
"twitter_description.empty": "twitter_description empty",
|
||
"feature_image.missing": "feature_image missing",
|
||
"feature_image_alt.missing": "feature_image_alt missing",
|
||
"internal_links.too_few": f"internal links < {R.MIN_INTERNAL_LINKS}",
|
||
"title_eq_meta_title": "meta_title == title (info)",
|
||
}
|
||
# severity per rule (look one up from any occurrence, incl. waived ones)
|
||
sev_of = {}
|
||
for _, actionable, accepted, info, _ in results:
|
||
for v in actionable + accepted + info:
|
||
sev_of[v.rule] = v.severity
|
||
for rule, _ in sorted(posts_with_rule.items(),
|
||
key=lambda kv: (-sev_of.get(kv[0], 0), -kv[1])):
|
||
sev = R.SEV_NAME[sev_of.get(rule, 0)]
|
||
print(f" [{sev:4}] {posts_with_rule[rule]:2}/{total_posts} posts "
|
||
f"{LABEL.get(rule, rule)}")
|
||
|
||
# ---- per-post table ----
|
||
print("\n" + "-" * 78)
|
||
print("PER-POST (worst first) — score = summed severity (HIGH 3 / MED 2 / LOW 1)\n")
|
||
print(f"{'score':>5} {'slug':45} issues")
|
||
print(f"{'-'*5} {'-'*45} {'-'*20}")
|
||
for p, actionable, accepted, info, sc in results:
|
||
if not actionable:
|
||
note = "✓ clean" + (f" ({len(accepted)} accepted exception)" if accepted else "")
|
||
print(f"{sc:5} {p['slug'][:45]:45} {note}")
|
||
continue
|
||
tags = ", ".join(sorted({v.rule for v in actionable}))
|
||
print(f"{sc:5} {p['slug'][:45]:45} {tags}")
|
||
|
||
# ---- detail per non-clean post ----
|
||
print("\n" + "-" * 78)
|
||
print("DETAIL\n")
|
||
for p, actionable, accepted, info, sc in results:
|
||
if not actionable and not info:
|
||
continue
|
||
print(f"• {p['slug']} (score {sc})")
|
||
for v in sorted(actionable, key=lambda x: -x.severity):
|
||
print(f" [{R.SEV_NAME[v.severity]:4}] {v.message}"
|
||
+ (f" → {v.fix}" if v.fix else ""))
|
||
for v in info:
|
||
print(f" [INFO] {v.message}")
|
||
print()
|
||
|
||
# ---- accepted exceptions (won't fix — documented) ----
|
||
if accepted_total:
|
||
print("-" * 78)
|
||
print("ACCEPTED EXCEPTIONS (won't fix — documented decisions, NOT in the count)\n")
|
||
for p, actionable, accepted, info, sc in sorted(results, key=lambda t: t[0]["slug"]):
|
||
for v in accepted:
|
||
# ⚠️ CON el sitio: sin él busca en el diccionario del EN y
|
||
# devuelve None para los 18 waivers del ES, dejando el informe
|
||
# mudo justo donde tiene que justificarse.
|
||
reason = X.accepted_reason(v.rule, p["slug"], args.site)
|
||
print(f"• {p['slug']}")
|
||
print(f" [{R.SEV_NAME[v.severity]:4}] {v.message}")
|
||
print(f" waived: {reason or '⚠ SIN MOTIVO DOCUMENTADO — esto es un fallo'}")
|
||
print()
|
||
|
||
# ---- ranked guidance ----
|
||
print("-" * 78)
|
||
print("FIX FIRST (by impact / effort):\n")
|
||
md_missing = posts_with_rule.get("meta_description.missing", 0)
|
||
md_long = posts_with_rule.get("meta_description.too_long", 0)
|
||
links = posts_with_rule.get("internal_links.too_few", 0)
|
||
og = posts_with_rule.get("og_title.empty", 0)
|
||
alt = posts_with_rule.get("feature_image_alt.missing", 0)
|
||
rank = []
|
||
if md_missing:
|
||
rank.append(f"meta_description MISSING on {md_missing} post(s) — HIGH impact, "
|
||
f"controls the Google snippet. Write unique 140–145-char descriptions.")
|
||
if md_long:
|
||
rank.append(f"meta_description > {R.META_DESC_MAX} on {md_long} post(s) — trim so the "
|
||
f"snippet isn't truncated (real ranking/CTR impact).")
|
||
if links:
|
||
rank.append(f"<{R.MIN_INTERNAL_LINKS} internal links on {links} post(s) — add 2–3 contextual "
|
||
f"links to related articles (helps crawl + topical authority).")
|
||
if og:
|
||
rank.append(f"Empty OG/Twitter fields on {og} post(s) — SAFEST AUTO-FIX: mirror "
|
||
f"meta_title/description into og_* and twitter_*. Highest count, lowest risk.")
|
||
if alt:
|
||
rank.append(f"Missing feature_image_alt on {alt} post(s) — quick accessibility + image-SEO win.")
|
||
for i, line in enumerate(rank, 1):
|
||
print(f" {i}. {line}")
|
||
print()
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main()
|