feat(seo): aviso de colisión de tema en /generate blog en
Build & Deploy ResearchOwl / build-and-push (push) Successful in 7s

Al publicar el draft EN en Ghost, el título propuesto se compara contra
los posts published+scheduled del sitio (corpus vía Admin API — incluye
la cola programada, justo el caso del doble Kecksburg del 2026-07-10) con
el topic_collision vendorizado. Si colisiona, el notice de Telegram lleva
un bloque "🚨 Posible colisión de tema" en las tres rutas (live, dryrun,
bare). Nunca bloquea: el draft se crea igual, el humano decide (fusionar,
retitular o enlazar a propósito). Solo lang=en (stopwords inglesas).
Títulos ajenos saneados de entidades Markdown (regla de _safe_send).
Aislamiento: cualquier fallo del check → notice sin bloque y warning en
logs, jamás rompe la publicación.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
ChemaVX
2026-07-10 11:17:33 +00:00
co-authored by Claude Fable 5
parent 0e9a2e5b36
commit dadc030d16
3 changed files with 127 additions and 3 deletions
+24 -3
View File
@@ -612,6 +612,7 @@ class OutputGenerator:
return ""
title = _extract_title(full_output) or topic
mode = _resolve_seo_mode(seo_override)
collision_note = await self._collision_note(lang, title)
if mode in ("on", "dryrun"):
try:
@@ -636,13 +637,17 @@ class OutputGenerator:
# surface the proposal so Jose can inspect before trusting writes.
result = await ghost.publish_draft(title, full_output)
post = result["posts"][0]
self.last_publish_notice = _seo_dryrun_message(ghost, post, seo, inserted_pairs)
self.last_publish_notice = (
_seo_dryrun_message(ghost, post, seo, inserted_pairs)
+ collision_note)
else:
result = await ghost.publish_draft(
title, full_output, tags=seo["tags"], seo=seo,
body_html=linked_html)
post = result["posts"][0]
self.last_publish_notice = _seo_live_message(ghost, post, seo, inserted_pairs)
self.last_publish_notice = (
_seo_live_message(ghost, post, seo, inserted_pairs)
+ collision_note)
logger.info("Auto-published blog to Ghost",
mode=mode, post_id=post["id"], links=len(inserted_pairs))
return ""
@@ -661,11 +666,27 @@ class OutputGenerator:
result = await ghost.publish_draft(title, full_output)
post = result["posts"][0]
logger.info("Auto-published blog to Ghost (bare)", post_id=post["id"])
return _bare_ghost_notice(ghost, post)
return _bare_ghost_notice(ghost, post) + collision_note
except Exception as e:
logger.warning("Auto-publish to Ghost failed", error=str(e))
return ""
async def _collision_note(self, lang: str, title: str) -> str:
"""Aviso de canibalización del título propuesto contra los posts
published+scheduled del sitio. Solo EN (las stopwords del motor son
inglesas). Nunca lanza y nunca bloquea: el draft se publica igual y
el aviso viaja en el notice de Telegram.
"""
if lang != "en":
return ""
try:
from src.seo.autofill import collision_notice, fetch_collision_corpus
corpus = await fetch_collision_corpus(lang)
return collision_notice(title, corpus) or ""
except Exception as e:
logger.warning("Topic collision check failed — skipped", error=str(e))
return ""
async def generate(self, session_id: int, output_type: OutputType,
progress_callback=None, lang: str = "es",
seo_override: str | None = None) -> str: