ci: add lint and test workflows with uv setup

.github/actions/setup-uv/action.yml

@@ -0,0 +1,33 @@
+name: 'Setup uv'
+description: 'Setup uv with Python and install dependencies'
+inputs:
+  python-version:
+    description: 'Python version to use'
+    required: false
+    default: '3.12'
+  install-dev:
+    description: 'Install dev dependencies'
+    required: false
+    default: 'true'
+
+runs:
+  using: 'composite'
+  steps:
+    - name: Install uv
+      uses: astral-sh/setup-uv@v4
+      with:
+        enable-cache: true
+        cache-dependency-glob: "uv.lock"
+
+    - name: Set up Python
+      run: uv python install ${{ inputs.python-version }}
+      shell: bash
+
+    - name: Install dependencies
+      run: |
+        if [ "${{ inputs.install-dev }}" = "true" ]; then
+          uv sync --all-extras --dev
+        else
+          uv sync --all-extras
+        fi
+      shell: bash

.github/workflows/lint.yml

@@ -0,0 +1,26 @@
+name: Lint
+
+on:
+  push:
+    branches: [ "*" ]
+  workflow_dispatch:
+
+jobs:
+  lint:
+    runs-on: ubuntu-latest
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Setup uv
+      uses: ./.github/actions/setup-uv
+      with:
+        python-version: '3.12'
+        install-dev: 'true'
+
+    - name: Run ruff check
+      run: uv run ruff check .
+
+    - name: Run ruff format check
+      run: uv run ruff format --check .

.github/workflows/test.yml

@@ -0,0 +1,35 @@
+name: Test
+
+on:
+  pull_request:
+    branches: [ "main", "dev" ]
+  workflow_dispatch:
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ['3.12']
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Setup uv
+      uses: ./.github/actions/setup-uv
+      with:
+        python-version: ${{ matrix.python-version }}
+        install-dev: 'true'
+
+    - name: Run tests
+      run: uv run pytest -v --cov=rocketrag --cov-report=term-missing
+
+    - name: Run type checking (if available)
+      run: |
+        if uv run python -c "import mypy" 2>/dev/null; then
+          uv run mypy rocketrag
+        else
+          echo "Type checker not available, skipping"
+        fi
+      continue-on-error: true

rocketrag/rocketrag.py

@@ -79,6 +79,8 @@ def __init__(
             self.metadata,
         )
 
+        self.rag = RAG(self.db, self.llm)
+
     def prepare(self, recreate: bool = False):
         if self.loader is None:
             raise ValueError("Loader is not defined.")
@@ -100,13 +102,11 @@ def stream_llm(self, messages: list[dict]) -> str:
     @ensure_llm_loaded
     def ask(self, question: str) -> tuple[str, list[SearchResult]]:
         # TODO: Should print the error on vectorizer mismach
-        rag = RAG(self.db, self.llm)
-        stream, sources = rag.run(question)
+        stream, sources = self.rag.run(question)
         return stream, sources
 
     @ensure_llm_loaded
     def stream_ask(self, question: str) -> tuple[str, list[SearchResult]]:
         # TODO: Should print the error on vectorizer mismach
-        rag = RAG(self.db, self.llm)
-        stream, sources = rag.stream(question)
+        stream, sources = self.rag.stream(question)
         return stream, sources

rocketrag/vectors.py

@@ -1,3 +1,6 @@
+import os
+
+os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"
 from sentence_transformers import SentenceTransformer
 from .base import BaseVectorizer
 

tests/test_integration.py

@@ -115,11 +115,18 @@ def test_stream_ask_question(self, integration_rocketrag):
 
         # Ask a question using streaming
         question = "What are the main topics covered?"
-        response, sources = integration_rocketrag.stream_ask(question)
+        stream, sources = integration_rocketrag.stream_ask(question)
+
+        # Consume the stream to get the complete response
+        full_response = ""
+        for output in stream:
+            delta = output["choices"][0]["delta"]
+            if "content" in delta:
+                full_response += delta["content"]
 
         # Verify response (streaming should still return complete response)
-        assert isinstance(response, str)
-        assert len(response) > 0
+        assert isinstance(full_response, str)
+        assert len(full_response) > 0
 
         # Verify sources
         assert isinstance(sources, list)
@@ -204,7 +211,7 @@ def test_recreate_database(self, integration_rocketrag):
         """Test recreating the database."""
         # Prepare the database
         integration_rocketrag.prepare()
-        initial_count = integration_rocketrag.db.count_records()
+        initial_count = integration_rocketrag.db.get_total_count()
         assert initial_count > 0
 
         # Recreate the database