feat: trackeo de coste por llamada Claude — tabla api_usage + /costs

src/generator/generator.py

@@ -179,7 +179,7 @@ class OutputGenerator:
         system = self._get_system(output_type)
         prompt = PROMPTS[output_type].format(topic=topic, context=context)
 
-        output = await self._generate(prompt, system, output_type)
+        output = await self._generate(prompt, system, output_type, session_id)
 
         # Add metadata header
         stats = await self.db.get_session_stats(session_id)
@@ -192,12 +192,14 @@ class OutputGenerator:
         logger.info("Output generated", type=output_type, length=len(full_output))
         return full_output
 
-    async def _generate(self, prompt: str, system: str, output_type: OutputType) -> str:
+    async def _generate(self, prompt: str, system: str, output_type: OutputType,
+                         session_id: int | None = None) -> str:
         if settings.anthropic_api_key:
-            return await self._generate_with_claude(prompt, system, output_type)
+            return await self._generate_with_claude(prompt, system, output_type, session_id)
         return await self._generate_with_ollama(prompt, system)
 
-    async def _generate_with_claude(self, prompt: str, system: str, output_type: OutputType) -> str:
+    async def _generate_with_claude(self, prompt: str, system: str, output_type: OutputType,
+                                     session_id: int | None = None) -> str:
         import anthropic
         max_tokens = 4096 if output_type == OutputType.THREAD else 8192
         try:
@@ -208,6 +210,14 @@ class OutputGenerator:
                 system=system,
                 messages=[{"role": "user", "content": prompt}],
             )
+            if session_id is not None:
+                try:
+                    await self.db.log_api_call(
+                        session_id, "generation", settings.claude_model,
+                        msg.usage.input_tokens, msg.usage.output_tokens
+                    )
+                except Exception as log_err:
+                    logger.warning("Failed to log API usage", error=str(log_err))
             return msg.content[0].text.strip()
         except Exception as e:
             logger.warning("Claude generation failed, falling back to Ollama", error=str(e))