fix(seo): sanitize malformed link suggestions + clean link report

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

src/generator/generator.py

@@ -328,16 +328,25 @@ def _bare_ghost_notice(ghost: "GhostPublisher", post: dict) -> str:
     )
 
 
-def _seo_link_summary(seo: dict, n_links: int) -> str:
-    pairs = [(l["phrase"], l["slug"]) for l in seo.get("internal_links", [])]
-    return f"{n_links} insertados" + (f" — {pairs}" if pairs else "")
+def _seo_link_summary(inserted_pairs: list[dict], n_suggestions: int) -> str:
+    """Report ACTUALLY-INSERTED links only ("phrase → /slug/"), with a separate
+    count of well-formed suggestions that were skipped (no verbatim body match).
+    Never surfaces raw rejected suggestions."""
+    n = len(inserted_pairs)
+    skipped = max(0, n_suggestions - n)
+    head = f"{n} insertados" + (f", {skipped} omitidos" if skipped else "")
+    if not inserted_pairs:
+        return head
+    body = "; ".join(f"{p['phrase']} → /{p['slug']}/" for p in inserted_pairs)
+    return f"{head} — {body}"
 
 
-def _seo_live_message(ghost: "GhostPublisher", post: dict, seo: dict, n_links: int) -> str:
+def _seo_live_message(ghost: "GhostPublisher", post: dict, seo: dict, inserted_pairs: list[dict]) -> str:
     """Separate Telegram message for the live autofill path (SEO written to draft)."""
     slug = post.get("slug", "")
     warns = seo.get("seo_warnings") or []
     warn_line = ("\n⚠️ revisar: " + "; ".join(warns)) if warns else ""
+    n_sug = len(seo.get("internal_links", []))
     return (
         f"📤 *Borrador en Ghost — SEO autorrelleno*\n"
         f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}\n\n"
@@ -346,7 +355,7 @@ def _seo_live_message(ghost: "GhostPublisher", post: dict, seo: dict, n_links: i
         f"• meta desc ({len(seo['meta_description'])}): {seo['meta_description']}\n"
         f"• excerpt: {len(seo['custom_excerpt'])} car.\n"
         f"• tags: {', '.join(seo['tags'])}\n"
-        f"• internal links: {_seo_link_summary(seo, n_links)}"
+        f"• internal links: {_seo_link_summary(inserted_pairs, n_sug)}"
         f"{warn_line}\n\n"
         f"🖼️ *Imagen sugerida:*\n"
         f"`imagefinder --query \"{seo['image_query']}\" --context \"{seo['image_context']}\" --lang {ghost.lang}`\n\n"
@@ -354,11 +363,12 @@ def _seo_live_message(ghost: "GhostPublisher", post: dict, seo: dict, n_links: i
     )
 
 
-def _seo_dryrun_message(ghost: "GhostPublisher", post: dict, seo: dict, n_links: int) -> str:
+def _seo_dryrun_message(ghost: "GhostPublisher", post: dict, seo: dict, inserted_pairs: list[dict]) -> str:
     """Separate Telegram message for DRY-RUN: bare draft created, SEO only proposed."""
     slug = post.get("slug", "")
     warns = seo.get("seo_warnings") or []
     warn_line = ("\n⚠️ revisar: " + "; ".join(warns)) if warns else ""
+    n_sug = len(seo.get("internal_links", []))
     return (
         f"🧪 *DRY-RUN SEO* — draft creado SIN escribir SEO (solo revisión)\n"
         f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}\n\n"
@@ -367,7 +377,7 @@ def _seo_dryrun_message(ghost: "GhostPublisher", post: dict, seo: dict, n_links:
         f"• meta desc ({len(seo['meta_description'])}): {seo['meta_description']}\n"
         f"• excerpt ({len(seo['custom_excerpt'])} car.): {seo['custom_excerpt']}\n"
         f"• tags: {', '.join(seo['tags'])}\n"
-        f"• internal links que se insertarían: {_seo_link_summary(seo, n_links)}"
+        f"• internal links que se insertarían: {_seo_link_summary(inserted_pairs, n_sug)}"
         f"{warn_line}\n\n"
         f"🖼️ *Imagen sugerida:*\n"
         f"`imagefinder --query \"{seo['image_query']}\" --context \"{seo['image_context']}\" --lang {ghost.lang}`\n\n"
@@ -523,7 +533,7 @@ class OutputGenerator:
                     raise RuntimeError("generate_seo_fields returned None")
                 # Build the SAME html that will be posted, then link it deterministically.
                 html = ghost._build_html(article_md)
-                linked_html, n_links = insert_internal_links(
+                linked_html, inserted_pairs = insert_internal_links(
                     html, seo["internal_links"], menu, lang)
 
                 if mode == "dryrun":
@@ -531,15 +541,15 @@ 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, n_links)
+                    self.last_publish_notice = _seo_dryrun_message(ghost, post, seo, inserted_pairs)
                 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, n_links)
+                    self.last_publish_notice = _seo_live_message(ghost, post, seo, inserted_pairs)
                 logger.info("Auto-published blog to Ghost",
-                            mode=mode, post_id=post["id"], links=n_links)
+                            mode=mode, post_id=post["id"], links=len(inserted_pairs))
                 return ""
             except Exception as e:
                 logger.warning("SEO autofill failed; falling back to bare draft",