fix: resolve ruff 0.15.0 lint violations (#178)

* fix: resolve ruff 0.15.0 lint violations

* chore: add onnxruntime dependency for Python 3.10 compatibility

Author: r-dh

pyproject.toml

@@ -20,6 +20,7 @@ dependencies = [
   # Sentence and chunk splitting:
   "markdown-it-py (>=3.0.0)",
   "numpy (>=1.26.4,<2.0.0)",
+  "onnxruntime (<1.24.0); python_version < '3.11'",  # 1.24+ drops Python 3.10 support
   "scipy (>=1.11.2,!=1.15.0.*,!=1.15.1,!=1.15.2)",
   "wtpsplit-lite (>=0.1.0)",
   # Large Language Models:

src/raglite/_chainlit.py

@@ -2,6 +2,7 @@
 
 import os
 from pathlib import Path
+from typing import TYPE_CHECKING
 
 import chainlit as cl
 from chainlit.input_widget import Switch, TextInput
@@ -16,6 +17,9 @@
 )
 from raglite._markdown import document_to_markdown
 
+if TYPE_CHECKING:
+    from raglite._database import ChunkSpan
+
 async_insert_documents = cl.make_async(insert_documents)
 async_hybrid_search = cl.make_async(hybrid_search)
 async_rerank_chunks = cl.make_async(rerank_chunks)
@@ -91,12 +95,10 @@ async def handle_message(user_message: cl.Message) -> None:
     ).strip()
     # Stream the LLM response.
     assistant_message = cl.Message(content="")
-    chunk_spans = []
+    chunk_spans: list[ChunkSpan] = []
     messages: list[dict[str, str]] = cl.chat_context.to_openai()[:-1]  # type: ignore[no-untyped-call]
     messages.append({"role": "user", "content": user_prompt})
-    async for token in async_rag(
-        messages, on_retrieval=lambda x: chunk_spans.extend(x), config=config
-    ):
+    async for token in async_rag(messages, on_retrieval=chunk_spans.extend, config=config):
         await assistant_message.stream_token(token)
     # Append RAG sources, if any.
     if chunk_spans:

src/raglite/_database.py

@@ -349,7 +349,7 @@ def to_xml(self, index: int | None = None) -> str:
         xml_document = "\n".join(
             [
                 f'<document{index_attribute} id="{self.document.id}">',
-                f"<source>{self.document.url if self.document.url else self.document.filename}</source>",
+                f"<source>{self.document.url or self.document.filename}</source>",
                 f'<span from_chunk_id="{self.chunks[0].id}" to_chunk_id="{self.chunks[-1].id}">',
                 f"<headings>\n{escape(self.chunks[0].headings.strip())}\n</headings>",
                 f"<content>\n{escape(''.join(chunk.body for chunk in self.chunks).strip())}\n</content>",
@@ -372,7 +372,7 @@ def to_json(self, index: int | None = None) -> str:
         json_document = {
             **index_attribute,
             "id": self.document.id,
-            "source": self.document.url if self.document.url else self.document.filename,
+            "source": self.document.url or self.document.filename,
             "span": {
                 "from_chunk_id": self.chunks[0].id,
                 "to_chunk_id": self.chunks[-1].id,

src/raglite/_split_chunklets.py

@@ -129,7 +129,7 @@ def split_chunklets(
     if not use_optimized:
         # Fall back to default lambdas if only one is provided.
         boundary_cost = boundary_cost or (lambda p: (1.0 - p[0]) + np.sum(p[1:]))
-        statement_cost = statement_cost or (lambda s: ((s - 3) ** 2 / np.sqrt(max(s, 1e-6)) / 2))
+        statement_cost = statement_cost or (lambda s: (s - 3) ** 2 / np.sqrt(max(s, 1e-6)) / 2)
     else:
         prefix_boundary = np.concatenate(([0.0], np.cumsum(boundary_probas)), dtype=np.float64)
         prefix_statements = np.concatenate(([0.0], np.cumsum(num_statements_arr)), dtype=np.float64)

tests/test_rag.py

@@ -32,8 +32,8 @@ def test_rag_auto_with_retrieval(raglite_test_config: RAGLiteConfig) -> None:
     # Answer a question that requires RAG.
     user_prompt = "How does Einstein define 'simultaneous events' in his special relativity paper?"
     messages = [{"role": "user", "content": user_prompt}]
-    chunk_spans = []
-    stream = rag(messages, on_retrieval=lambda x: chunk_spans.extend(x), config=raglite_test_config)
+    chunk_spans: list[ChunkSpan] = []
+    stream = rag(messages, on_retrieval=chunk_spans.extend, config=raglite_test_config)
     answer = ""
     for update in stream:
         assert isinstance(update, str)
@@ -52,8 +52,8 @@ def test_rag_auto_without_retrieval(raglite_test_config: RAGLiteConfig) -> None:
     # Answer a question that does not require RAG.
     user_prompt = "Is 7 a prime number?"
     messages = [{"role": "user", "content": user_prompt}]
-    chunk_spans = []
-    stream = rag(messages, on_retrieval=lambda x: chunk_spans.extend(x), config=raglite_test_config)
+    chunk_spans: list[ChunkSpan] = []
+    stream = rag(messages, on_retrieval=chunk_spans.extend, config=raglite_test_config)
     answer = ""
     for update in stream:
         assert isinstance(update, str)