Add application file

app/routes/embedding_routes.py

@@ -11,7 +11,7 @@ router = APIRouter()
 @router.post("/create-embeddings")
 async def create_embeddings(file: UploadFile = File(...)):
     """
-    Create embeddings from an uploaded Excel or CSV file.
+    Create embeddings from an uploaded Excel or CSV file containing question-answer pairs.
 
     - **file**: The Excel or CSV file containing questions and answers
 
@@ -31,8 +31,15 @@ async def create_embeddings(file: UploadFile = File(...)):
             detail="Unsupported file type. Please upload an Excel (.xlsx, .xls) or CSV (.csv) file.",
         )
 
+    # Ensure temp directory exists
+    temp_dir = "/app/temp"
+    if not os.path.exists(temp_dir):
+        os.makedirs(temp_dir, exist_ok=True)
+
     # Create a temporary file to store the uploaded file
-    temp_file_path = f"temp_{file.filename}"
+    safe_filename = os.path.basename(file.filename).replace(" ", "_")
+    temp_file_path = os.path.join(temp_dir, f"temp_{safe_filename}")
+
     try:
         # Save the uploaded file
         with open(temp_file_path, "wb") as buffer:
@@ -50,4 +57,9 @@ async def create_embeddings(file: UploadFile = File(...)):
     finally:
         # Clean up the temporary file
         if os.path.exists(temp_file_path):
-            os.remove(temp_file_path)
+            try:
+                os.remove(temp_file_path)
+            except Exception as e:
+                print(
+                    f"Warning: Could not remove temporary file {temp_file_path}: {str(e)}"
+                )

app/services/embedding_service.py

@@ -7,8 +7,11 @@ from typing import List, Dict, Tuple, Any
 
 from app.services.model_service import get_model, reload_embeddings
 
+# Define data directory path
+DATA_DIR = "/app/data"
+
 # Ensure data directory exists
-os.makedirs("data", exist_ok=True)
+os.makedirs(DATA_DIR, exist_ok=True)
 
 
 def remove_prefix(text: str, prefix_pattern: str) -> str:
@@ -55,7 +58,8 @@ def save_raw_data(qa_list: List[Dict[str, str]]) -> None:
     """
     Save the raw question-answer pairs to a JSON file.
     """
-    with open("data/raw.json", "w", encoding="utf-8") as json_file:
+    raw_path = os.path.join(DATA_DIR, "raw.json")
+    with open(raw_path, "w", encoding="utf-8") as json_file:
         json.dump(qa_list, json_file, ensure_ascii=False, indent=2)
 
 
@@ -74,11 +78,15 @@ def create_and_save_embeddings(qa_list: List[Dict[str, str]]) -> None:
     answer_embeddings = model.encode(answers, convert_to_numpy=True)
 
     # Save embeddings as numpy arrays
-    np.save("data/question_embeddings.npy", question_embeddings)
-    np.save("data/answer_embeddings.npy", answer_embeddings)
+    q_emb_path = os.path.join(DATA_DIR, "question_embeddings.npy")
+    a_emb_path = os.path.join(DATA_DIR, "answer_embeddings.npy")
+    qa_data_path = os.path.join(DATA_DIR, "qa_data.json")
+
+    np.save(q_emb_path, question_embeddings)
+    np.save(a_emb_path, answer_embeddings)
 
     # Save the original data
-    with open("data/qa_data.json", "w", encoding="utf-8") as f:
+    with open(qa_data_path, "w", encoding="utf-8") as f:
         json.dump(qa_list, f, ensure_ascii=False, indent=2)
 
 

app/services/model_service.py

@@ -1,8 +1,12 @@
 import json
 import numpy as np
+import os
 from sentence_transformers import SentenceTransformer
 from typing import List, Dict, Tuple, Any, Optional
 
+# Define data directory path
+DATA_DIR = "/app/data"
+
 # Global variables to store model and data
 _model = None
 _question_embeddings = None
@@ -37,10 +41,14 @@ def load_embeddings() -> Tuple[np.ndarray, np.ndarray, List[Dict[str, str]]]:
     global _question_embeddings, _answer_embeddings, _qa_data
 
     try:
-        _question_embeddings = np.load("data/question_embeddings.npy")
-        _answer_embeddings = np.load("data/answer_embeddings.npy")
+        q_emb_path = os.path.join(DATA_DIR, "question_embeddings.npy")
+        a_emb_path = os.path.join(DATA_DIR, "answer_embeddings.npy")
+        qa_data_path = os.path.join(DATA_DIR, "qa_data.json")
+
+        _question_embeddings = np.load(q_emb_path)
+        _answer_embeddings = np.load(a_emb_path)
 
-        with open("data/qa_data.json", "r", encoding="utf-8") as f:
+        with open(qa_data_path, "r", encoding="utf-8") as f:
             _qa_data = json.load(f)
 
         return _question_embeddings, _answer_embeddings, _qa_data