need to work on TODOs

src/pull_request_handler.py

@@ -1,7 +1,8 @@
+import json
 import logging
 import ollama
 from typing import List, Dict
-
+import os
 from config import OLLAMA_MODEL
 from src.vector_db import query_by_function_names
 from src.github_api import leave_comment
@@ -19,20 +20,21 @@ def generate_pr_feedback(
 
     prompt = f"""
     
-    You are to provide feedback on a code base pull request. Follow the Instructions to provide your feedback on the Context below.
+    You are a Pull Request Criticiser.
+    Follow the Instructions below given the Context.
     
     Start of Instructions:
     
-    1. Review the pull request's title, description, code diff, and relevant context from the codebase.
-    2. Using point-form, succinctly identify potential issues, code smells, code duplication, or downsides of 
-    the pull request. 
-    3. Consider interactions with the codebase and architectural design. 
+    1. Read the pull request's title, description, code diff, and relevant context from the codebase.
+    2. Using point-form, concisely identify potential issues, code smells, code duplication, or downsides of 
+    the pull request. Include file name in your references.
+    3. Consider interactions of the aforementioned code diff with the relevant context from codebase and architectural design. 
     4. Remember that in the code diff, '+' is code addition and '-' is code subtraction.
     5. Do not provide refactored code as part of your feedback. 
     6. End your response with the sentence "Has the PR author considered these points?"
    
-    End of Instructions.
-      
+    End of Instructions.  
+    
     Start of Context:
     
     Pull Request:
@@ -46,6 +48,8 @@ def generate_pr_feedback(
     {pr_diff}
     
     End of Context.
+    
+    
     """
     logging.info("generating feedback...")
     response = ollama.chat(
@@ -54,9 +58,23 @@ def generate_pr_feedback(
 
     return response["message"]["content"]
 
-def get_function_dependencies(changed_files: List[str]):
+
+def get_function_dependencies(changed_files: List[str]) -> List[str]:
     """Gets all internal functions that are imported into a file"""
-    pass # TODO: ensure we validate the changed_file is actially in main
+    # TODO: ensure we validate the changed_file is actially in main,
+    # TODO: modify json file so that  dots become slashes here or in gen_deps.py
+    function_set = set()
+
+    # remove file extension #todo add project path to the finction path and
+    changed_files = [os.path.splitext(x)[0] for x in changed_files]
+
+    with open("dependencies.json") as f:
+        dependency_graph = json.load(f)
+        for file in changed_files:
+            if file in dependency_graph:
+                function_set.update(dependency_graph[file])
+    return list(function_set)
+
 
 def handle_new_pull_request(
     installation_id: str,
@@ -70,12 +88,10 @@ def handle_new_pull_request(
 ):
     """Handle new pull request webhook"""
     try:
-
-        # todo: get all used function dependencies
         dependency_functions = get_function_dependencies(changed_files)
-        dependency_function_code = query_by_function_names(dependency_functions, repo_id)
+        dependency_function_context = query_by_function_names(dependency_functions, repo_id)
         # Generate feedback
-        feedback = generate_pr_feedback(dependency_function_code, pr_title, pr_body, pr_diff)
+        feedback = generate_pr_feedback(dependency_function_context, pr_title, pr_body, pr_diff)
 
         # Post comment
         leave_comment(installation_id, repo_full_name, pr_number, feedback)

src/vector_db.py

@@ -125,15 +125,20 @@ def embed_code_base(
                 functions = extract_function_info(file_path)
 
                 # Embed and store functions
-                for func in functions:
+                for func in functions: #
+                    # Get relative path of function
+                    # root_dir = os.path.dirname(os.path.abspath(__file__))
+                    # relative_path = os.path.relpath(func['function_path'], root_dir)
+                    # func["file_path"] = relative_path
+                    # TODO: done remove up to root dir ; C:\Users\amy36\PycharmProjects\doppelganger\src\webhook_handler.py:handle_pull_requests
                     # Create embedding
                     embedding = model.encode(
-                        f"{func['name']} {func['docstring']} {func['source_code']}"
+                        f"{func['source_code']}"
                     ).tolist()
 
                     # Add to ChromaDB
                     collection.add(
-                        ids=[func['name']],
+                        ids=[func["function_path"]],
                         embeddings=[embedding],
                         documents=[func["source_code"]],
                         metadatas=[{
@@ -144,7 +149,7 @@ def embed_code_base(
                     )
 
 
-def query_by_function_names(functions, repo_id):
+def query_by_function_names(function_paths, repo_id):
     collection = get_collection_for_repo_branch(repo_id)
     """
     Retrieve full code and metadata for a list of specific function name paths.
@@ -154,19 +159,19 @@ def query_by_function_names(functions, repo_id):
     :return: List of dictionaries containing function details
     """
     # Validate input
-    if not functions:
+    if not function_paths:
         return []
 
     # Retrieve functions from ChromaDB
-    results = collection.get(
-        ids=functions
-    )
+    results = collection.get(where={"function_path":{"$in": function_paths}}) # TODO: process function_paths to match something in the vectorDB
 
     # Process and format results
     function_details = []
+    # todo: make schema for metadata
+    # metadata:{'file_path': 'C:\\Users\\amy36\\PycharmProjects\\doppelganger\\src\\github_api.py', 'function_name': 'fetch_existing_issues', 'function_path': 'C:\\Users\\amy36\\PycharmProjects\\doppelganger\\src\\github_api.py:fetch_existing_issues'}
     for i in range(len(results['ids'])):
         function_info = {
-            'function_name': results['ids'][i],
+            # 'function_name': results['ids'][i],
             'source_code': results['documents'][i] if results['documents'] else None,
             'metadata': results['metadatas'][i] if results['metadatas'] else None
         }

src/webhook_handler.py

@@ -146,13 +146,14 @@ def handle_pull_requests(data, installation_id):
 
     if action == "opened" or action == "edited":
         temp_dir = None
-        ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
+        ROOT_DIR = os.path.dirname(os.path.abspath(__file__)) #todo: why is ROOT_DIR "'C:\\Users\\amy36\\PycharmProjects\\doppelganger\\src"? it should not include /src
         # try:
             # # Update main branch collection if needed
             # temp_dir = clone_repo_branch(installation_id, repo_full_name, "main")
             # code_files = index_code_files(temp_dir)
             # add_code_to_chroma(code_files, repo_id, "main")
-        embed_code_base(repo_id, ROOT_DIR) # TODO: ensure embeddings are updated after a pull request (look at actions)
+        embed_code_base(repo_id, ROOT_DIR) # TODO: 1. add embdeddings when a repository is added, ensure root dir is from the repo ofo rep_id
+        # TODO: 2. ensure embeddings are updated after a pull request (look at actions) on main
 
         # Handle the pull request
         handle_new_pull_request(