Feat/extras matrix for tests (#269)

* add dependencies matrix

* add pipeline tests for each scoring module

* bug fix

* upd extras installation in gh actions

* upd extras in mypy ci

* fix issues with dependencies for bert scorer test

* fix gcn scorer tests

* fix typing errors

* upd assertions about catboost predictions

* upd ci with presets tests

* try to fix unit tests

Author: voorhs

.github/workflows/reusable-test.yaml

@@ -7,6 +7,11 @@ on:
         required: true
         type: string
         description: 'Command to run tests'
+      extras:
+        required: false
+        type: string
+        default: ''
+        description: 'Space-separated --extra flags (e.g., "--extra transformers --extra peft")'
 
 jobs:
   test:
@@ -39,7 +44,7 @@ jobs:
     - name: Install dependencies for Python ${{ matrix.python-version }}
       run: |
         uv python pin ${{ matrix.python-version }}
-        uv sync --group test
+        uv sync --group test ${{ inputs.extras }}
 
     - name: Run tests
       run: |

.github/workflows/test-embedder.yaml

@@ -0,0 +1,15 @@
+name: test embedder
+
+on:
+  push:
+    branches:
+      - dev
+  pull_request:
+
+jobs:
+  test:
+    uses: ./.github/workflows/reusable-test.yaml
+    with:
+      test_command: pytest -n auto tests/embedder/
+      extras: --extra sentence-transformers
+

.github/workflows/test-inference.yaml

@@ -11,3 +11,4 @@ jobs:
     uses: ./.github/workflows/reusable-test.yaml
     with:
       test_command: pytest -n auto tests/pipeline/test_inference.py
+      extras: --extra catboost --extra peft --extra transformers --extra sentence-transformers

.github/workflows/test-optimization.yaml

@@ -11,3 +11,4 @@ jobs:
     uses: ./.github/workflows/reusable-test.yaml
     with:
       test_command: pytest -n auto tests/pipeline/test_optimization.py
+      extras: --extra catboost --extra peft --extra transformers --extra sentence-transformers

.github/workflows/test-presets.yaml

@@ -11,3 +11,4 @@ jobs:
     uses: ./.github/workflows/reusable-test.yaml
     with:
       test_command: pytest -n auto tests/pipeline/test_presets.py
+      extras: --extra catboost --extra peft --extra transformers --extra sentence-transformers

.github/workflows/test-scorers.yaml

@@ -0,0 +1,47 @@
+name: test scorers
+
+on:
+  push:
+    branches:
+      - dev
+  pull_request:
+
+jobs:
+  test:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ ubuntu-latest ]
+        python-version: [ "3.10", "3.11", "3.12" ]
+        dependency-group: [ "base", "transformers", "peft", "catboost" ]
+        include:
+          - os: windows-latest
+            python-version: "3.10"
+            dependency-group: "base"
+
+    steps:
+    - name: Checkout code
+      uses: actions/checkout@v4
+
+    - name: Cache Hugging Face
+      id: cache-hf
+      uses: actions/cache@v4
+      with:
+        path: ~/.cache/huggingface
+        key: ${{ runner.os }}-hf
+
+    - name: Install uv
+      uses: astral-sh/setup-uv@v6
+      with:
+        version: "0.8.8"
+
+    - name: Install dependencies for Python ${{ matrix.python-version }}
+      run: |
+        uv python pin ${{ matrix.python-version }}
+        uv sync --group test ${{ matrix.dependency-group != 'base' && format('--extra {0}', matrix.dependency-group) || '' }}
+
+    - name: Run scorer tests
+      run: |
+        uv run pytest -n auto tests/modules/scoring/
+

.github/workflows/typing.yml

@@ -18,7 +18,7 @@ jobs:
       - name: Install dependencies
         run: |
           uv lock
-          uv sync --group typing
+          uv sync --group typing --extra peft --extra sentence-transformers
 
       - name: Run mypy
         run: uv run mypy src/autointent

.github/workflows/unit-tests.yaml

@@ -10,4 +10,4 @@ jobs:
   test:
     uses: ./.github/workflows/reusable-test.yaml
     with:
-      test_command: pytest -n auto --ignore=tests/nodes --ignore=tests/pipeline
+      test_command: pytest -n auto --ignore=tests/modules/scoring/ --ignore=tests/pipeline --ignore=tests/embedder

pyproject.toml

@@ -51,7 +51,9 @@ dependencies = [
 [project.optional-dependencies]
 catboost = ["catboost (>=1.2.8,<2.0.0)"]
 peft = ["peft (>= 0.10.0, !=0.15.0, !=0.15.1, <1.0.0)"]
-transformers = ["transformers (>=4.49.0,<5.0.0)"]
+transformers = [
+    "transformers[torch] (>=4.49.0,<5.0.0)",
+]
 sentence-transformers = ["sentence-transformers (>=3,<4)"]
 dspy = [
     "dspy (>=2.6.5,<3.0.0)",

src/autointent/_wrappers/embedder/hashing_vectorizer.py

@@ -67,7 +67,7 @@ def get_hash(self) -> int:
         hasher.update(self.config.norm if self.config.norm is not None else "None")
         hasher.update(self.config.binary)
         hasher.update(self.config.dtype)
-        return hasher.hexdigest()
+        return int(hasher.hexdigest(), 16)
 
     @overload
     def embed(
@@ -97,7 +97,7 @@ def embed(
         """
         # Transform texts to sparse matrix, then convert to dense
         embeddings_sparse = self._vectorizer.transform(utterances)
-        embeddings = embeddings_sparse.toarray().astype(np.float32)
+        embeddings: npt.NDArray[np.float32] = embeddings_sparse.toarray().astype(np.float32)
 
         if return_tensors:
             return torch.from_numpy(embeddings)
@@ -115,7 +115,8 @@ def similarity(
         Returns:
             Similarity matrix with shape (n_samples, m_samples).
         """
-        return cosine_similarity(embeddings1, embeddings2).astype(np.float32)
+        similarity_matrix: npt.NDArray[np.float32] = cosine_similarity(embeddings1, embeddings2).astype(np.float32)
+        return similarity_matrix
 
     def dump(self, path: Path) -> None:
         """Save the backend state to disk.
@@ -157,7 +158,7 @@ def load(cls, path: Path) -> "HashingVectorizerEmbeddingBackend":
         logger.debug("Loaded HashingVectorizer backend from %s", path)
         return instance
 
-    def train(self, utterances: list[str], labels: list[int], config) -> None:  # noqa: ANN001
+    def train(self, utterances: list[str], labels: list[int], config) -> None:  # noqa: ANN001  # type: ignore[no-untyped-def]
         """Train the backend.
 
         HashingVectorizer is stateless and doesn't support training.