Refine DeepGit Lite description and improve error handling for GitHub API key and document embeddings

app.py

@@ -1,5 +1,4 @@
 import gradio as gr
-import os
 import time
 import threading
 import logging
@@ -60,7 +59,7 @@ title = """
 """
 
 description = """<p align="center">
-DeepGit Lite is a streamlined version of DeepGit designed to perform advanced semantic research on GitHub repositories with faster response times. It uses query enhancement, dense retrieval via FAISS, activity analysis, and a final multi-factor ranking (combining semantic similarity, activity, and popularity) to deliver the best results.
+DeepGit Lite is a streamlined version of DeepGit designed for fast semantic search on GitHub repositories. It enhances your query, retrieves repositories using dense retrieval via FAISS, filters by star count, combines scores based on semantic similarity and popularity, and then provides a concise justification for the top results.
 </p>"""
 
 consent_text = """
@@ -118,7 +117,6 @@ def parse_result_to_html(raw_result: str) -> str:
                 <th>Title</th>
                 <th>Link</th>
                 <th>Semantic Similarity</th>
-                <th>Activity Score</th>
                 <th>Final Score</th>
             </tr>
         </thead>
@@ -138,7 +136,6 @@ def parse_result_to_html(raw_result: str) -> str:
                 <td>{data.get('Title', '')}</td>
                 <td><a href="{data.get('Link', '#')}" target="_blank">GitHub</a></td>
                 <td>{format_percent(data.get('Semantic Similarity', ''))}</td>
-                <td>{float(data.get('Activity Score', 0)):.2f}</td>
                 <td>{format_percent(data.get('Final Score', ''))}</td>
             </tr>
         """

src/deepgit_lite.py

@@ -5,7 +5,6 @@ import numpy as np
 import datetime
 from sentence_transformers import SentenceTransformer
 import faiss
-import getpass
 import math
 import logging
 from dotenv import load_dotenv
@@ -20,7 +19,7 @@ dotenv_path = Path(__file__).resolve().parent.parent / ".env"
 load_dotenv(dotenv_path=str(dotenv_path))
 
 if "GITHUB_API_KEY" not in os.environ:
-    os.environ["GITHUB_API_KEY"] = getpass.getpass("Enter your GitHub API key: ")
+    raise EnvironmentError("GITHUB_API_KEY not set in environment. Please set it as an environment variable.")
 
 # ---------------------------
 # Logging Setup
@@ -49,7 +48,10 @@ Provide the refined query text."""
         ("human", prompt)
     ]
     result = llm_groq.invoke(messages)
-    return result
+    # Extract text content if available
+    if hasattr(result, "content"):
+        return result.content
+    return str(result)
 
 def justify_candidate(candidate, query):
     prompt = f"""You are a highly knowledgeable AI research assistant. In one to two lines, explain why the repository titled "{candidate['title']}" is a good match for a query on "{query}". Mention key factors such as documentation quality and community validation if relevant.
@@ -64,7 +66,9 @@ Provide a concise justification:"""
         ("human", prompt)
     ]
     result = llm_groq.invoke(messages)
-    return result
+    if hasattr(result, "content"):
+        return result.content
+    return str(result)
 
 # ---------------------------
 # GitHub API Helper Functions
@@ -172,15 +176,24 @@ def run_deepgit_lite(user_query):
     # Stage 1: Dense Retrieval with FAISS
     logger.info("Fetching repositories from GitHub...")
     repos = fetch_github_repositories(github_query)
+    if not repos:
+        logger.error("No repositories found. Please refine your query.")
+        return "\nNo repositories found for your query. Please try a different query."
+        
     docs = [repo.get("combined_doc", "") for repo in repos]
     logger.info(f"Encoding {len(docs)} documents for dense retrieval...")
     sem_model = SentenceTransformer("all-mpnet-base-v2")
     doc_embeddings = sem_model.encode(docs, convert_to_numpy=True, show_progress_bar=True, batch_size=16)
-
+    
+    # Check if embeddings array is empty or 1-dimensional
+    if doc_embeddings.ndim < 2 or doc_embeddings.shape[0] == 0:
+        logger.error("No document embeddings generated. Aborting dense retrieval.")
+        return "\nFailed to generate document embeddings. Please try again."
+    
     def normalize_embeddings(embeddings):
         norms = np.linalg.norm(embeddings, axis=1, keepdims=True)
         return embeddings / (norms + 1e-10)
-
+    
     doc_embeddings = normalize_embeddings(doc_embeddings)
     query_embedding = sem_model.encode(user_query, convert_to_numpy=True)
     query_embedding = normalize_embeddings(np.expand_dims(query_embedding, axis=0))[0]