Solution published: update src/utils.py

src/utils.py

@@ -1,47 +1,35 @@
 import os
-from pathlib import Path
+import pathlib
-import json
+from typing import List
-import httpx
+
-from langchain_ollama import OllamaEmbeddings
+from langchain_community.document_loaders import TextLoader
-from langchain_chroma import Chroma
 from langchain_text_splitters import RecursiveCharacterTextSplitter
+from langchain_chroma import Chroma
+from langchain_ollama import OllamaEmbeddings
 
-CHROMA_PATH = Path("./chroma_faq")
+CHROMA_PATH = pathlib.Path("./chroma_faq")
+CHROMA_PATH.mkdir(parents=True, exist_ok=True)
 
-# Load FAQ markdown files into ChromaDB
+
-def load_faq_to_chroma(md_dir: str = "data"):
+def load_faq_to_chroma(md_dir: str = "data") -> None:
-    Path(md_dir).mkdir(parents=True, exist_ok=True)
+    """Load all .md files from md_dir into a Chroma vector store.
-    md_files = list(Path(md_dir).glob("*.md"))
+    The store is persisted at CHROMA_PATH.
-    if not md_files:
+    """
-        raise FileNotFoundError(f"No .md files found in {md_dir}")
+    loader = TextLoader
-    texts = []
+    all_docs = []
-    for md_file in md_files:
+    for md_file in pathlib.Path(md_dir).glob("*.md"):
-        text = md_file.read_text(encoding="utf-8")
+        loader_obj = loader(str(md_file))
-        splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+        docs = loader_obj.load()
-        texts.extend(splitter.split_text(text))
+        all_docs.extend(docs)
+    splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
+    docs = splitter.split_documents(all_docs)
+    embeddings = OllamaEmbeddings(model="nomic-embed-text")
+    Chroma.from_documents(docs, embeddings, persist_directory=str(CHROMA_PATH))
+
+
+def search_course_docs(query: str, k: int = 3) -> List[str]:
+    """Return top k document snippets from the persisted Chroma store."""
     embeddings = OllamaEmbeddings(model="nomic-embed-text")
     chroma = Chroma(persist_directory=str(CHROMA_PATH), embedding_function=embeddings)
-    chroma.add_texts(texts)
+    results = chroma.similarity_search(query, k=k)
-    chroma.persist()
+    return [doc.page_content for doc in results]
-    return chroma
-
-# Search the ChromaDB for relevant documents
-def search_course_docs(query: str, k: int = 3):
-    chroma = Chroma(persist_directory=str(CHROMA_PATH), embedding_function=OllamaEmbeddings(model="nomic-embed-text"))
-    retriever = chroma.as_retriever(search_kwargs={"k": k})
-    return retriever.get_relevant_documents(query)
-
-# MCP‑style tool: fetch metadata from a mock HTTP endpoint
-# In production this would be a real MCP server. Here we use a local JSON file served by http.server.
-def fetch_course_meta(query: str):
-    url = f"http://localhost:8000/course_meta.json"
-    try:
-        response = httpx.get(url, timeout=5.0)
-        response.raise_for_status()
-        data = response.json()
-    except Exception:
-        # Fallback to local static file if server not running
-        data = json.loads(Path("data/course_meta.json").read_text(encoding="utf-8"))
-    # Simple filtering: return items where query is in title or description
-    results = [item for item in data if query.lower() in item.get("title", "").lower() or query.lower() in item.get("description", "").lower()]
-    return results