fix false/404 links in reference links

Author: smutyala1at

apps/meshjs-rag/app/api/v1/ask_mesh_ai.py

@@ -52,7 +52,6 @@ async def ask_mesh_ai(body: ChatCompletionRequest, credentials: HTTPAuthorizatio
 
   try:
     question = body.messages[-1].content
-
     embedded_query = await openai_service.embed_query(question)
     context = await get_context(embedded_query, supabase)
     generator = openai_service.get_answer(question=question, context=context)

apps/meshjs-rag/app/services/openai.py

@@ -19,6 +19,22 @@
 Answer only with the succinct context and nothing else.
 """
 
+AI_PROMPT = """
+You are a MeshJS expert assistant. Help developers with MeshJS questions using the provided context. Use the documentation context to answer questions about MeshJS and Cardano development. Provide accurate code examples and explanations based on the context provided.
+
+When answering:
+- Give direct, helpful answers based on the context
+- Include relevant code examples when available
+- Explain concepts clearly for developers
+- Include any documentation or resource links found in the context. Handle them as follows:
+  - When you find "location: ...", include the link in your answer as a reference.
+  - If multiple links are present, include all relevant ones.
+- If the context doesn't cover the question, say so clearly.
+- Do not invent or assume APIs, methods, or functionality not in the documentation.
+
+Be concise but thorough. Focus on practical, actionable guidance for MeshJS development.
+"""
+
 class OpenAIService:
   def __init__(self, openai_api_key):
     self.client = AsyncOpenAI(api_key=openai_api_key)
@@ -79,7 +95,7 @@ async def get_answer(self, question: str, context: str, model="gpt-4o-mini"):
     messages = [
       {
         "role": "system",
-        "content": "You are a MeshJS expert assistant. Help developers with MeshJS questions using the provided context.\n\nUse the documentation context to answer questions about MeshJS and Cardano development. Provide accurate code examples and explanations based on the context provided.\n\nWhen answering:\n- Give direct, helpful answers based on the context\n- Include relevant code examples when available\n- Explain concepts clearly for developers\n- Include any links present in the context for additional resources\n- If the context doesn't cover something, say so\n- Don't make up APIs or methods not in the documentation\n\nBe concise but thorough. Focus on practical, actionable guidance for MeshJS development."
+        "content": AI_PROMPT
       },
       {
         "role": "user",
@@ -99,7 +115,7 @@ async def get_mcp_answer(self, question: str, context: str, model="gpt-4o-mini")
     messages = [
       {
         "role": "system",
-        "content": "You are a MeshJS expert assistant. Help developers with MeshJS questions using the provided context.\n\nUse the documentation context to answer questions about MeshJS and Cardano development. Provide accurate code examples and explanations based on the context provided.\n\nWhen answering:\n- Give direct, helpful answers based on the context\n- Include relevant code examples when available\n- Explain concepts clearly for developers\n- Include any links present in the context for additional resources\n- If the context doesn't cover something, say so\n- Don't make up APIs or methods not in the documentation\n\nBe concise but thorough. Focus on practical, actionable guidance for MeshJS development."
+        "content": AI_PROMPT
       },
       {
         "role": "user",

apps/meshjs-rag/app/utils/extract_github_links.py

@@ -0,0 +1,11 @@
+from typing import List
+
+def extract_github_links(text: str) -> List[str]:
+    lines = text.splitlines()
+    github_links = []
+
+    for line in lines:
+        if "https://github.com/MeshJS/" in line:
+            github_links.append(line.strip())
+
+    return github_links

apps/meshjs-rag/app/utils/get_context.py

@@ -1,7 +1,9 @@
 from typing import List
 from supabase import AsyncClient
 
+from app.utils.extract_github_links import extract_github_links
 
+PACKAGES = ["mesh-common", "mesh-core-csl", "mesh-contract", "mesh-provider", "mesh-transaction", "mesh-wallet"]
 
 async def get_context(embedded_query: List[float], supabase: AsyncClient) -> str:
   response = await supabase.rpc("match_docs", {
@@ -10,9 +12,19 @@ async def get_context(embedded_query: List[float], supabase: AsyncClient) -> str
     "match_count": 5
   }).execute()
 
-  contextual_text = "\n\n".join([data["contextual_text"] for data in response.data]) if response.data else None
+  final_contextual_data = ""
+  if response.data:
+    for data in response.data:
+      if not str(data["filepath"]).startswith(tuple(PACKAGES)):
+        file_location = f"location: https://meshjs.dev/{data["filepath"].replace(".mdx", "")}"
+      else:
+        links = "\n".join(extract_github_links(data["contextual_text"]))
+        file_location = f"location: {links}" if links else ""
 
-  if contextual_text:
-    return contextual_text
+      contextual_data = data["contextual_text"] + "\n\n" + file_location + "\n\n" if file_location else data["contextual_text"] + "\n\n"
+      final_contextual_data += contextual_data if contextual_data else ""
+
+  if final_contextual_data:
+    return final_contextual_data
   else:
-    return None
+    return "No relevant context found."