fix: use AsyncClientV2 to avoid blocking the event loop

src/knowledge/ingestion.py

@@ -5,7 +5,6 @@
 """
 
 import logging
-import os
 from pathlib import Path
 
 from src.config import settings
@@ -91,7 +90,10 @@ def chunk_text(
                 chunks.append("\n\n".join(current_chunk))
                 # Keep overlap
                 overlap_text = "\n\n".join(current_chunk)
-                overlap_chars = overlap_text[-chunk_overlap:] if len(overlap_text) > chunk_overlap else overlap_text
+                if len(overlap_text) > chunk_overlap:
+                    overlap_chars = overlap_text[-chunk_overlap:]
+                else:
+                    overlap_chars = overlap_text
                 current_chunk = [overlap_chars] if overlap_chars.strip() else []
                 current_length = len(overlap_chars)
 

src/knowledge/vectorstore.py

@@ -6,7 +6,6 @@
 
 import json
 import logging
-import math
 from pathlib import Path
 
 import numpy as np

src/routes/courses.py

@@ -3,7 +3,6 @@
 import logging
 
 from fastapi import APIRouter, HTTPException
-from fastapi.responses import JSONResponse
 
 from src.models.course import Course, GenerateCourseRequest
 from src.services.generation import get_generation_service

src/services/cohere_client.py

@@ -20,7 +20,7 @@ class CohereClient:
 
     def __init__(self, api_key: str | None = None) -> None:
         self._api_key = api_key or settings.cohere_api_key
-        self._client = cohere.ClientV2(api_key=self._api_key)
+        self._client = cohere.AsyncClientV2(api_key=self._api_key)
         self._embed_model = settings.cohere_embed_model
         self._generate_model = settings.cohere_generate_model
 
@@ -54,7 +54,7 @@ async def generate(
         messages.append({"role": "user", "content": prompt})
 
         try:
-            response = self._client.chat(
+            response = await self._client.chat(
                 model=self._generate_model,
                 messages=messages,
                 max_tokens=max_tokens,
@@ -108,7 +108,11 @@ async def generate_json(
             logger.debug("Raw response: %s", raw[:500])
             raise ValueError(f"Invalid JSON response from Cohere: {exc}") from exc
 
-    async def embed(self, texts: list[str], input_type: str = "search_document") -> list[list[float]]:
+    async def embed(
+        self,
+        texts: list[str],
+        input_type: str = "search_document",
+    ) -> list[list[float]]:
         """Generate embeddings for a list of texts.
 
         Args:
@@ -125,7 +129,7 @@ async def embed(self, texts: list[str], input_type: str = "search_document") ->
             return []
 
         try:
-            response = self._client.embed(
+            response = await self._client.embed(
                 model=self._embed_model,
                 texts=texts,
                 input_type=input_type,

src/services/embedding.py

@@ -34,7 +34,8 @@ async def embed_documents(self, texts: list[str]) -> list[list[float]]:
             )
             all_embeddings.extend(embeddings)
 
-        logger.info("Embedded %d documents in %d batches", len(texts), (len(texts) - 1) // batch_size + 1)
+        num_batches = (len(texts) - 1) // batch_size + 1
+        logger.info("Embedded %d documents in %d batches", len(texts), num_batches)
         return all_embeddings
 
     async def embed_query(self, query: str) -> list[float]:

src/services/feedback_engine.py

@@ -114,7 +114,10 @@ async def quick_feedback(
             return f"Correct! {question.explanation}"
 
         correct_option = question.options[question.correct_index]
-        user_option = question.options[user_answer] if 0 <= user_answer < len(question.options) else "N/A"
+        if 0 <= user_answer < len(question.options):
+            user_option = question.options[user_answer]
+        else:
+            user_option = "N/A"
         return (
             f"Incorrect. You chose '{user_option}', but the correct answer is '{correct_option}'. "
             f"{question.explanation}"

src/services/generation.py

@@ -2,13 +2,13 @@
 
 import logging
 
+from src.knowledge.vectorstore import get_vectorstore
 from src.models.course import Course, CourseModule, GenerateCourseRequest
 from src.models.user_profile import UserProfile
 from src.prompts.course_generation import build_course_prompt, build_module_prompt
 from src.prompts.system import COURSE_SYSTEM_PROMPT
 from src.services.cohere_client import get_cohere_client
 from src.services.embedding import get_embedding_service
-from src.knowledge.vectorstore import get_vectorstore
 
 logger = logging.getLogger(__name__)
 

src/services/quiz_generator.py

@@ -3,12 +3,12 @@
 import logging
 import uuid
 
+from src.knowledge.vectorstore import get_vectorstore
 from src.models.quiz import GenerateQuizRequest, Quiz, QuizQuestion
-from src.prompts.quiz_generation import build_quiz_prompt, build_question_from_content_prompt
+from src.prompts.quiz_generation import build_question_from_content_prompt, build_quiz_prompt
 from src.prompts.system import QUIZ_SYSTEM_PROMPT
 from src.services.cohere_client import get_cohere_client
 from src.services.embedding import get_embedding_service
-from src.knowledge.vectorstore import get_vectorstore
 
 logger = logging.getLogger(__name__)