src/ai_generators/blog_flow.py

@@ -17,19 +17,14 @@ journalist draft is chunked, embedded, and stored in a collection; the editor
 receives the top-N most relevant chunks as context.
 """
 
-from __future__ import annotations
-
 import json
 import os
 import random
 import re
 import string
 from datetime import datetime
-from typing import TYPE_CHECKING
-
-if TYPE_CHECKING:
-    import chromadb  # noqa: F811
 
+import chromadb
 from crewai.flow.flow import Flow, listen, start
 from ollama import Client
 from pydantic import BaseModel, ConfigDict
@@ -90,21 +85,7 @@ class BlogFlow(Flow[BlogFlowState]):
         )
 
     @staticmethod
-    def _get_chroma_client() -> "chromadb.HttpClient":
-        # Lazily import chromadb here rather than at module level.
-        # chromadb unconditionally loads hnswlib (a native C++ library
-        # compiled with AVX instructions) even when using HttpClient
-        # to talk to an external ChromaDB server.  Deferring the import
-        # avoids "Illegal instruction" (SIGILL) crashes in environments
-        # that lack AVX support (e.g. act, older CI runners).
-        try:
-            import chromadb
-        except ImportError as exc:
-            raise RuntimeError(
-                "chromadb is required for the editor phase but could not be "
-                f"imported: {exc}"
-            ) from exc
-
+    def _get_chroma_client() -> chromadb.HttpClient:
         chroma_port = int(os.environ["CHROMA_PORT"])
         return chromadb.HttpClient(host=os.environ["CHROMA_HOST"], port=chroma_port)
 
@@ -146,7 +127,7 @@ class BlogFlow(Flow[BlogFlowState]):
             print(f"Error generating embeddings: {exc}")
             return []
 
-    def _load_drafts_to_vector_db(self, drafts: list[str]) -> "chromadb.Collection":
+    def _load_drafts_to_vector_db(self, drafts: list[str]) -> chromadb.Collection:
         """Load journalist drafts into a new ChromaDB collection and return it."""
         chroma = self._get_chroma_client()
         collection_name = (
@@ -184,7 +165,7 @@ class BlogFlow(Flow[BlogFlowState]):
         return collection
 
     @staticmethod
-    def _query_vector_db(collection: "chromadb.Collection", query_text: str) -> str:
+    def _query_vector_db(collection: chromadb.Collection, query_text: str) -> str:
         """Query the ChromaDB collection and return the most relevant
         document chunks joined as a single string."""
         ollama_client = BlogFlow._get_ollama_client()