refactor llm/embedding flow

app/gemini_client.py

@@ -0,0 +1,45 @@
+from google.generativeai.embedding import embed_content
+from google.generativeai.client import configure
+from google.generativeai.generative_models import GenerativeModel
+from loguru import logger
+
+class GeminiClient:
+    def __init__(self, api_key: str, model: str = "gemini-1.5-flash-latest"):
+        self.api_key = api_key
+        self.model = model
+        configure(api_key=api_key)
+        self._model = GenerativeModel(model)
+
+    def generate_text(self, prompt: str, **kwargs) -> str:
+        try:
+            response = self._model.generate_content(prompt, **kwargs)
+            logger.info(f"[GEMINI][RAW_RESPONSE] {response}")
+            # response có thể là GenerativeModelResponse, lấy text hoặc trả về str
+            if hasattr(response, 'text'):
+                return response.text
+            elif hasattr(response, 'candidates') and response.candidates:
+                return response.candidates[0].content.parts[0].text
+            return str(response)
+        except Exception as e:
+            logger.error(f"[GEMINI] Error: {e}")
+            raise
+
+    def count_tokens(self, prompt: str) -> int:
+        try:
+            return self._model.count_tokens(prompt).total_tokens
+        except Exception as e:
+            logger.error(f"[GEMINI] Token count error: {e}")
+            return 0
+
+    def create_embedding(self, text: str) -> list:
+        try:
+            response = embed_content(
+                model=self.model,
+                content=text,
+                task_type="retrieval_document"
+            )
+            logger.info(f"[GEMINI][EMBEDDING][RAW_RESPONSE] {response}")
+            return response['embedding']
+        except Exception as e:
+            logger.error(f"[GEMINI][EMBEDDING] Error: {e}")
+            raise