feat: modo diff para /watch — notifica solo si hay novedades reales

src/generator/generator.py

@@ -574,3 +574,82 @@ def generate_pdf(content: str, title: str = "ResearchOwl Output") -> bytes:
 
     doc.build(story)
     return buf.getvalue()
+
+
+async def generate_diff_summary(
+    topic: str,
+    new_urls: set,
+    old_urls: set,
+    new_chunks: list,
+    session_id: int,
+    db,
+) -> str | None:
+    from src.config import settings
+    import structlog
+    diff_logger = structlog.get_logger()
+
+    added_urls = new_urls - old_urls
+    pct_new = len(added_urls) / max(len(new_urls), 1)
+
+    diff_logger.info("Diff analysis", topic=topic,
+                     new_urls=len(new_urls), old_urls=len(old_urls),
+                     added=len(added_urls), pct_new=round(pct_new, 2))
+
+    if pct_new < 0.20 and len(added_urls) < 5:
+        diff_logger.info("Diff: no significant new sources", topic=topic)
+        return None
+
+    if not new_chunks:
+        return None
+
+    context = "\n\n---\n\n".join(
+        f"[{c.get('source_type', 'web').upper()}] {c.get('title', '')}\n{c['content'][:400]}"
+        for c in new_chunks[:20]
+    )
+
+    if not settings.anthropic_api_key:
+        return (
+            f"📊 *Novedades detectadas sobre {topic}*\n\n"
+            f"• {len(added_urls)} fuentes nuevas encontradas\n"
+            f"• {len(new_chunks)} chunks de contenido procesados\n\n"
+            f"Usa /generate report para ver el análisis completo."
+        )
+
+    try:
+        import anthropic
+        client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+        prompt = (
+            f'Analiza el siguiente material de investigación sobre "{topic}" '
+            f'y genera un resumen BREVE (máximo 300 palabras) de las novedades '
+            f'más importantes encontradas. Escribe en español.\n\n'
+            f'Si el contenido es muy similar a investigaciones anteriores o no '
+            f'contiene información genuinamente nueva, responde SOLO con: '
+            f'"SIN_NOVEDADES"\n\n'
+            f'Material nuevo:\n{context}'
+        )
+        msg = await client.messages.create(
+            model=settings.claude_model,
+            max_tokens=500,
+            messages=[{"role": "user", "content": prompt}]
+        )
+        summary = msg.content[0].text.strip()
+
+        if summary == "SIN_NOVEDADES":
+            diff_logger.info("Diff: Claude found no new information", topic=topic)
+            return None
+
+        try:
+            await db.log_api_call(session_id, "diff", settings.claude_model,
+                                  msg.usage.input_tokens, msg.usage.output_tokens)
+        except Exception:
+            pass
+
+        return f"🔔 *Novedades — {topic}*\n\n{summary}\n\nUsa /generate para report completo."
+
+    except Exception as e:
+        diff_logger.warning("Diff summary generation failed", error=str(e))
+        return (
+            f"📊 *Actualización — {topic}*\n\n"
+            f"• {len(added_urls)} fuentes nuevas\n"
+            f"Usa /generate report para ver el análisis completo."
+        )