'changes'

app/config.py

@@ -3,18 +3,15 @@ from pydantic_settings import BaseSettings
 
 class Settings(BaseSettings):
     OPENAI_API_KEY: str
-    GEMINI_API_KEY: str
-    # Set to "openai" or "gemini" to choose the AI provider
-    AI_PROVIDER: str = "gemini"
+    GEMINI_API_KEY: str = ""
+    GEMINI_OPENAI_COMPATIBLE_BASE_URL: str = "https://generativelanguage.googleapis.com/v1beta"
+    AI_PROVIDER: str = "openai"  # "openai" or "gemini"
     DATABASE_URL: str = os.getenv("DATABASE_URL", "")
     NEON_DATABASE_URL: str = os.getenv("NEON_DATABASE_URL", "")
     QDRANT_URL: str = os.getenv("QDRANT_URL", "http://localhost:6333")
     QDRANT_API_KEY: str = os.getenv("QDRANT_API_KEY", "")
     OPENAI_MODEL_CHAT: str = "gpt-4o-mini"
     OPENAI_MODEL_EMBEDDING: str = "text-embedding-3-small"
-    GEMINI_MODEL_CHAT: str = "gemini-2.5-flash"
-    GEMINI_MODEL_EMBEDDING: str = "embedding-001"
-    GEMINI_OPENAI_COMPATIBLE_BASE_URL: str = "https://generativelanguage.googleapis.com/v1beta/openai/"
     
     class Config:
         env_file = ".env"

app/qdrant_client.py

@@ -11,13 +11,6 @@ qdrant_client = QdrantClient(
 
 COLLECTION_NAME = "book_embeddings"
 
-def get_vector_size():
-    """Get the appropriate vector size based on the AI provider"""
-    if settings.AI_PROVIDER.lower() == "gemini":
-        return 768  # Gemini embedding dimension
-    else:
-        return 1536  # OpenAI text-embedding-3-small dimension
-
 def init_qdrant_collection():
     """Initialize Qdrant collection if it doesn't exist"""
     try:
@@ -30,11 +23,11 @@ def init_qdrant_collection():
             qdrant_client.create_collection(
                 collection_name=COLLECTION_NAME,
                 vectors_config=VectorParams(
-                    size=get_vector_size(),  # Dynamic vector size based on provider
+                    size=1536,  # OpenAI text-embedding-3-small dimension
                     distance=Distance.COSINE
                 )
             )
-            print(f"✅ Created Qdrant collection: {COLLECTION_NAME} with {get_vector_size()} dimensions")
+            print(f"✅ Created Qdrant collection: {COLLECTION_NAME}")
         else:
             print(f"✅ Qdrant collection already exists: {COLLECTION_NAME}")
     except Exception as e:

app/routes/chat.py

@@ -3,9 +3,8 @@ from qdrant_client import QdrantClient
 from app.qdrant_client import get_qdrant_client
 from app.schemas.chat import ChatRequest, ChatResponse, ChatSelectionRequest
 from app.services.rag_service import RAGService
-from app.services.embeddings_service import EmbeddingsService, GeminiEmbeddingsService
-from app.services.openai_service import OpenAIService, GeminiOpenAIService
-from app.config import settings
+from app.services.embeddings_service import get_embeddings_service
+from app.services.openai_service import get_openai_service
 import logging
 
 logger = logging.getLogger(__name__)
@@ -15,15 +14,9 @@ router = APIRouter(prefix="/api", tags=["chat"])
 def get_rag_service(
     qdrant_client: QdrantClient = Depends(get_qdrant_client)
 ):
-    # Choose the appropriate services based on AI_PROVIDER setting
-    if settings.AI_PROVIDER.lower() == "gemini":
-        embeddings_service = GeminiEmbeddingsService()
-        ai_service = GeminiOpenAIService()
-    else:
-        embeddings_service = EmbeddingsService()
-        ai_service = OpenAIService()
-    
-    return RAGService(qdrant_client, embeddings_service, ai_service)
+    embeddings_service = get_embeddings_service()
+    openai_service = get_openai_service()
+    return RAGService(qdrant_client, embeddings_service, openai_service)
 
 @router.post("/chat", response_model=ChatResponse)
 async def chat(
@@ -34,7 +27,7 @@ async def chat(
         # Retrieve context from vector database
         context = await rag_service.retrieve_context(request.question, top_k=3)
         
-        # Generate response using the configured AI service
+        # Generate response using OpenAI
         answer = await rag_service.generate_response(request.question, context)
         
         # Extract sources from context

app/services/embeddings_service.py

@@ -34,4 +34,11 @@ class GeminiEmbeddingsService:
             input=[text],
             model=self.model
         )
-        return response.data[0].embedding
+        return response.data[0].embedding
+
+# Factory function to get the appropriate embeddings service
+def get_embeddings_service():
+    if settings.AI_PROVIDER.lower() == "gemini":
+        return GeminiEmbeddingsService()
+    else:
+        return EmbeddingsService()

app/services/openai_service.py

@@ -43,4 +43,11 @@ class GeminiOpenAIService:
             model=self.model,
             messages=messages
         )
-        return response.choices[0].message.content
+        return response.choices[0].message.content
+
+# Factory function to get the appropriate OpenAI service
+def get_openai_service():
+    if settings.AI_PROVIDER.lower() == "gemini":
+        return GeminiOpenAIService()
+    else:
+        return OpenAIService()

app/services/rag_service.py

@@ -3,18 +3,22 @@ from qdrant_client import QdrantClient
 from qdrant_client.models import NamedVector
 from typing import List
 
+# We'll use the base classes for type hints
 from app.services.openai_service import OpenAIService, GeminiOpenAIService
 from app.services.embeddings_service import EmbeddingsService, GeminiEmbeddingsService
 
+# Type alias for either service
+AIEmbeddingService = EmbeddingsService | GeminiEmbeddingsService
+AIOpenAIService = OpenAIService | GeminiOpenAIService
+
 class RAGService:
-    def __init__(self, qdrant_client: QdrantClient, embeddings_service: EmbeddingsService, ai_service: OpenAIService):
+    def __init__(self, qdrant_client: QdrantClient, embeddings_service: AIEmbeddingService, openai_service: AIOpenAIService):
         self.qdrant_client = qdrant_client
         self.embeddings_service = embeddings_service
-        self.ai_service = ai_service
+        self.openai_service = openai_service
         self.collection_name = os.getenv("QDRANT_COLLECTION_NAME", "book_embeddings")
 
     async def retrieve_context(self, query: str, top_k: int = 3) -> List[str]:
-        # Handle both OpenAI and Gemini embeddings
         query_vector = await self.embeddings_service.create_embedding(query)
         
         search_result = self.qdrant_client.search(
@@ -33,6 +37,5 @@ class RAGService:
 Question: {query}
 
 Answer:"""
-        # Handle both OpenAI and Gemini services
-        response = await self.ai_service.get_chat_response(full_prompt)
+        response = await self.openai_service.get_chat_response(full_prompt)
         return response

scripts/ingest_content.py

@@ -11,9 +11,8 @@ from dotenv import load_dotenv
 import sys
 sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
 
-from app.services.embeddings_service import EmbeddingsService, GeminiEmbeddingsService
-from app.qdrant_client import get_qdrant_client, get_vector_size
-from app.config import settings
+from app.services.embeddings_service import get_embeddings_service
+from app.qdrant_client import get_qdrant_client
 
 load_dotenv(dotenv_path=Path(__file__).resolve().parent.parent / ".env")
 
@@ -36,15 +35,12 @@ def chunk_text(text: str, chunk_size: int = 1000, overlap: int = 200) -> list[st
 async def ingest_content(
     docs_path: Path,
     qdrant_client: QdrantClient,
-    embeddings_service: EmbeddingsService,
+    embeddings_service,
     collection_name: str,
 ):
-    # Determine vector size based on the embedding service
-    vector_size = get_vector_size()
-    
     qdrant_client.recreate_collection(
         collection_name=collection_name,
-        vectors_config=VectorParams(size=vector_size, distance=Distance.COSINE),
+        vectors_config=VectorParams(size=1536, distance=Distance.COSINE),  # OpenAI embeddings size
     )
 
     points = []
@@ -96,12 +92,7 @@ if __name__ == "__main__":
     args = parser.parse_args()
 
     qdrant_client = get_qdrant_client()
-    
-    # Choose the appropriate embedding service based on AI_PROVIDER setting
-    if settings.AI_PROVIDER.lower() == "gemini":
-        embeddings_service = GeminiEmbeddingsService()
-    else:
-        embeddings_service = EmbeddingsService()
+    embeddings_service = get_embeddings_service()
 
     # Run the async ingestion
     import asyncio