Update app.py

app.py

@@ -4,247 +4,101 @@ import time
 import json
 import logging
 import traceback
-from contextlib import asynccontextmanager
-from typing import Dict, Any
-
 from fastapi import FastAPI, HTTPException
 from fastapi.middleware.cors import CORSMiddleware
 from pydantic import BaseModel, Field
+from contextlib import asynccontextmanager
+from typing import Dict, Any
 
-# ===================== ЛОГИРОВАНИЕ =====================
-
+# ======= ЛОГИРОВАНИЕ =======
 logging.basicConfig(
     level=logging.INFO,
     format="%(asctime)s [%(levelname)s] %(message)s",
     handlers=[logging.StreamHandler(sys.stderr)],
 )
+logger = logging.getLogger("eroha-api")
 
-logger = logging.getLogger("eroha-app")
-
-# ===================== БЕЗОПАСНЫЕ ИМПОРТЫ =====================
-
+# ======= ИМПОРТЫ =======
 try:
     from transformers import pipeline
     from langdetect import detect
-except Exception as e:  # noqa: BLE001
-    # ВАЖНО: без параметра file=...
+except Exception as e:
     logger.error("[ImportError] transformers/langdetect not available: %s", e, exc_info=True)
+    pipeline = None
+    def detect(text): return "en"
 
-    pipeline = None  # type: ignore[assignment]
-    # запасной детектор языка
-    def detect(text: str) -> str:  # type: ignore[no-redef]
-        return "en"
-
-
-# ===================== НАСТРОЙКИ HF =====================
-
+# ======= НАСТРОЙКИ =======
 HF_HOME = "/tmp/huggingface"
 os.environ["HF_HOME"] = HF_HOME
 os.makedirs(HF_HOME, exist_ok=True)
 
-# ===================== Pydantic-модели =====================
-
-
-class SummarizeRequest(BaseModel):
-    text: str = Field(..., min_length=3, max_length=1_000_000)
-
-    def clean_text(self) -> str:
-        return self.text.strip()
-
-
-class CheckRequest(BaseModel):
-    data: str = Field(..., min_length=1, max_length=500_000)
-
-    def clean_text(self) -> str:
-        return self.data.strip()
-
-
-# ===================== КЭШ МОДЕЛЕЙ =====================
-
+# ======= МОДЕЛИ =======
 _model_cache: Dict[str, Any] = {}
 
-
-def safe_detect_lang(text: str) -> str:
-    try:
-        lang = detect(text)
-        return lang or "en"
-    except Exception:  # noqa: BLE001
-        return "en"
-
-
 def get_model(lang: str):
     if pipeline is None:
         raise RuntimeError("Transformers pipeline is not available")
-
     if lang in _model_cache:
-        logger.info("[ModelCache] Using cached model for %s", lang)
         return _model_cache[lang]
-
     model_map = {
         "ru": "IlyaGusev/mbart_ru_sum_gazeta",
-        "kk": "facebook/mbart-large-50-many-to-many-mmt",
-        "de": "facebook/bart-large-cnn",
-        "es": "facebook/mbart-large-50-many-to-many-mmt",
-        "fr": "facebook/mbart-large-50-many-to-many-mmt",
         "en": "facebook/bart-large-cnn",
     }
     model_name = model_map.get(lang, "facebook/bart-large-cnn")
-
-    logger.info("[ModelLoad] Loading model for %s: %s", lang, model_name)
     model = pipeline("summarization", model=model_name, device=-1)
     _model_cache[lang] = model
-    logger.info("[ModelLoad] Cached model for %s", lang)
     return model
 
-
-# ===================== LIFESPAN (WARMUP) =====================
-
+# ======= FastAPI =======
 @asynccontextmanager
-async def lifespan(app: FastAPI):  # noqa: ARG001
+async def lifespan(app: FastAPI):
     start = time.time()
-    logger.info("[Lifespan] Application startup – warmup models...")
-
-    try:
-        if pipeline is not None:
-            for lang in ("en", "ru"):
-                try:
-                    get_model(lang)
-                except Exception as e:  # noqa: BLE001
-                    logger.error("[Lifespan] Warmup failed for %s: %s", lang, e, exc_info=True)
-        else:
-            logger.warning("[Lifespan] transformers pipeline is None – warmup skipped")
-    except Exception as e:  # noqa: BLE001
-        logger.error("[Lifespan] Warmup error: %s", e, exc_info=True)
-
-    elapsed = time.time() - start
-    logger.info("[Lifespan] Startup warmup finished in %.2f s", elapsed)
-
+    logger.info("[Startup] warming up models...")
+    for lang in ("en", "ru"):
+        try: get_model(lang)
+        except Exception as e: logger.error("Warmup failed: %s", e)
     yield
+    logger.info("[Shutdown] done")
 
-    logger.info("[Lifespan] Application shutdown")
+app = FastAPI(title="Eroha Agent API", version="v3.5", lifespan=lifespan)
+app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"], allow_headers=["*"])
 
+# ======= МОДЕЛИ ЗАПРОСОВ =======
+class SummarizeRequest(BaseModel):
+    text: str = Field(..., min_length=3, max_length=1_000_000)
 
-app = FastAPI(
-    title="Eroha AI Summarizer PRO",
-    version="3.4",
-    lifespan=lifespan,
-)
-
-app.add_middleware(
-    CORSMiddleware,
-    allow_origins=["*"],
-    allow_methods=["*"],
-    allow_headers=["*"],
-)
-
-# ===================== ЭНДПОИНТЫ =====================
-
+class MemoryRequest(BaseModel):
+    key: str
+    content: str
 
+# ======= ЭНДПОИНТЫ =======
 @app.get("/")
 async def root():
-    return {
-        "status": "ok",
-        "version": "v3.4",
-        "cached_models": list(_model_cache.keys()),
-        "endpoints": ["/ping", "/check", "/summarize", "/warmup"],
-    }
-
+    return {"status": "ok", "version": "v3.5"}
 
 @app.get("/ping")
 async def ping():
-    return {
-        "status": "healthy",
-        "cached_models": list(_model_cache.keys()),
-    }
-
-
-@app.get("/warmup")
-async def warmup():
-    try:
-        if pipeline is None:
-            return {
-                "status": "skipped",
-                "reason": "transformers pipeline is not available",
-            }
-
-        loaded = []
-        for lang in ("en", "ru"):
-            try:
-                get_model(lang)
-                loaded.append(lang)
-            except Exception as e:  # noqa: BLE001
-                logger.error("[Warmup] Failed for %s: %s", lang, e, exc_info=True)
-
-        return {
-            "status": "ok",
-            "preloaded": loaded,
-            "cache_size": len(_model_cache),
-        }
-    except Exception as e:  # noqa: BLE001
-        logger.error("[Warmup] Error: %s", e, exc_info=True)
-        raise HTTPException(status_code=500, detail="Warmup failed") from e
-
-
-@app.post("/check")
-async def check_text(req: CheckRequest):
-    try:
-        text = req.clean_text()
-        lang = safe_detect_lang(text)
-        return {
-            "status": "success",
-            "preview": text[:150],
-            "length": len(text),
-            "detected_language": lang,
-        }
-    except Exception as e:  # noqa: BLE001
-        logger.error("/check error: %s", traceback.format_exc())
-        raise HTTPException(status_code=500, detail=str(e)) from e
-
+    return {"status": "healthy", "cache": list(_model_cache.keys())}
 
 @app.post("/summarize")
 async def summarize(req: SummarizeRequest):
-    try:
-        text = req.clean_text()
-        if not text:
-            raise HTTPException(status_code=400, detail="Text cannot be empty")
-
-        lang = safe_detect_lang(text)
-        model = get_model(lang)
-
-        max_input_length = 3000
-        input_text = text[:max_input_length]
-
-        result = model(input_text, max_length=180, min_length=40, do_sample=False)
-        summary = result[0]["summary_text"].replace("▁", " ").strip()
-
-        seo_json_ld = {
-            "@context": "https://schema.org",
-            "@type": "NewsArticle",
-            "headline": summary[:80],
-            "inLanguage": lang,
-            "publisher": {
-                "@type": "Organization",
-                "name": "Eroha AI Publisher",
-            },
-        }
-
-        return {
-            "status": "success",
-            "language": lang,
-            "summary": summary,
-            "summary_length": len(summary),
-            "original_length": len(text),
-            "truncated": len(text) > max_input_length,
-            "seo_json_ld": seo_json_ld,
-        }
-    except HTTPException:
-        raise
-    except Exception as e:  # noqa: BLE001
-        logger.error("/summarize error: %s", traceback.format_exc())
-        raise HTTPException(status_code=500, detail=str(e)) from e
-
-
-if __name__ == "__main__":
-    import uvicorn
-
-    uvicorn.run(app, host="0.0.0.0", port=7860)
+    lang = "ru" if "а" in req.text.lower() else "en"
+    model = get_model(lang)
+    result = model(req.text[:2000], max_length=180, min_length=50, do_sample=False)
+    return {"summary": result[0]["summary_text"].strip(), "lang": lang}
+
+# ======= MEMORY API =======
+@app.post("/memorize")
+async def memorize(req: MemoryRequest):
+    with open("memory.json", "a") as f:
+        f.write(json.dumps(req.dict(), ensure_ascii=False) + "\\n")
+    return {"status": "saved"}
+
+@app.post("/retrieve")
+async def retrieve(req: MemoryRequest):
+    if not os.path.exists("memory.json"):
+        return {"found": []}
+    with open("memory.json", "r") as f:
+        lines = [json.loads(l) for l in f]
+    found = [l for l in lines if req.key.lower() in l["key"].lower()]
+    return {"found": found}