don't run extras on 3.13

datastax · Feb 5, 2025 · 4e2422e · 4e2422e
1 parent f05141d
commit 4e2422e
Show file tree

Hide file tree

Showing 7 changed files with 83 additions and 49 deletions.
diff --git a/.github/workflows/main.yml b/.github/workflows/main.yml
@@ -69,7 +69,13 @@ jobs:
         with:
           python-version: ${{ matrix.python-version }}
 
-      - name: Install Python dependencies
+      - name: (non-3.12) Install Python dependencies (without extras)
+        if: ${{ success() && matrix.python-version != '3.12' }}
+        run: uv sync --frozen --all-packages
+        shell: bash
+
+      - name: (3.12 only) Install Python dependencies (including all extras)
+        if: ${{ success() && matrix.python-version == '3.12' }}
         run: uv sync --frozen --all-packages --all-extras
         shell: bash
 
@@ -78,21 +84,26 @@ jobs:
           uv run coverage run -m pytest -vs packages/graph-retriever
               --junitxml=junit/test-results-gr-${{ matrix.python-version }}.xml
 
-      - name: (non-3.12) Test langchain-graph-retriever (In-Memory Stores)
-        if: ${{ success() && matrix.python-version != '3.12' }}
-        run: uv run pytest -vs packages/langchain-graph-retriever
+      - name: (3.10, 3.11) Test langchain-graph-retriever (In-Memory Stores) with extras
+        if: ${{ success() && matrix.python-version == '3.10' || matrix.python-version == '3.11' }}
+        run: uv run pytest -vs --runextras packages/langchain-graph-retriever
                 --junitxml=junit/test-results-lgr-${{ matrix.python-version }}.xml
 
-      - name: (3.12 only) Test langchain-graph-retriever (All Stores)
+      - name: (3.12) Test langchain-graph-retriever (All Stores) with extras
         if: ${{ success() && matrix.python-version == '3.12' }}
         id: test
-        run: uv run coverage run -a -m pytest -vs packages/langchain-graph-retriever --stores=all
+        run: uv run coverage run -a -m pytest -vs --runextras packages/langchain-graph-retriever --stores=all
                 --junitxml=junit/test-results-lgr-${{ matrix.python-version }}.xml
         env:
           ASTRA_DB_APPLICATION_TOKEN: ${{ secrets.ASTRA_DB_APPLICATION_TOKEN }}
           ASTRA_DB_API_ENDPOINT: ${{ secrets.ASTRA_DB_API_ENDPOINT }}
           ASTRA_DB_KEYSPACE: ci_${{ github.run_id }}_${{ strategy.job-index }}_${{ github.run_attempt }}
 
+      - name: (3.13) Test langchain-graph-retriever (In-Memory Stores) without extras
+        if: ${{ success() && matrix.python-version == '3.13' }}
+        run: uv run pytest -vs packages/langchain-graph-retriever
+                --junitxml=junit/test-results-lgr-${{ matrix.python-version }}.xml
+
       - name: Drop Astra Keyspace
         # Even though it seems redundant, the `always() &&` is necessary to signal to
         # GitHub actions that we want this to run even if the job is cancelled.

diff --git a/packages/langchain-graph-retriever/tests/conftest.py b/packages/langchain-graph-retriever/tests/conftest.py
@@ -42,6 +42,25 @@ def pytest_addoption(parser: Parser):
         choices=TESTCONTAINER_STORES + ["none"],
         help="which stores to run testcontainers for (default: 'all')",
     )
+    parser.addoption(
+        "--runextras", action="store_true", default=False, help="run tests for extras"
+    )
+
+
+def pytest_configure(config):
+    config.addinivalue_line(
+        "markers", "extra: mark test as requiring an `extra` package"
+    )
+
+
+def pytest_collection_modifyitems(config, items):
+    if config.getoption("--runextras"):
+        # --runextras given in cli: do not skip extras
+        return
+    skip_extras = pytest.mark.skip(reason="need --runextras option to run")
+    for item in items:
+        if "extra" in item.keywords:
+            item.add_marker(skip_extras)
 
 
 @pytest.fixture(scope="session")

diff --git a/packages/langchain-graph-retriever/tests/transformers/test_gliner.py b/packages/langchain-graph-retriever/tests/transformers/test_gliner.py
@@ -1,22 +1,23 @@
 from typing import Any
 
 import pytest
-from gliner import GLiNER  # type: ignore
 from langchain_core.documents import Document
 from langchain_graph_retriever.transformers.gliner import GLiNERTransformer
 
 
-class FakeGLiNER(GLiNER):
-    def __init__(self):
-        pass
+@pytest.mark.extra
+def test_transform_documents(animal_docs: list[Document]):
+    from gliner import GLiNER  # type: ignore
 
-    def batch_predict_entities(
-        self, texts: list[str], **kwargs: Any
-    ) -> list[list[dict[str, str]]]:
-        return [[{"text": text.split()[0], "label": "first"}] for text in texts]
+    class FakeGLiNER(GLiNER):
+        def __init__(self):
+            pass
 
+        def batch_predict_entities(
+            self, texts: list[str], **kwargs: Any
+        ) -> list[list[dict[str, str]]]:
+            return [[{"text": text.split()[0], "label": "first"}] for text in texts]
 
-def test_transform_documents(animal_docs: list[Document]):
     fake_model = FakeGLiNER()
     transformer = GLiNERTransformer(
         ["first"], model=fake_model, metadata_key_prefix="prefix_"

diff --git a/packages/langchain-graph-retriever/tests/transformers/test_html_hyperlink.py b/packages/langchain-graph-retriever/tests/transformers/test_html_hyperlink.py
@@ -26,6 +26,7 @@
 """
 
 
+@pytest.mark.extra
 def test_transform_documents():
     doc = Document(
         id="animal_html",

diff --git a/packages/langchain-graph-retriever/tests/transformers/test_keybert.py b/packages/langchain-graph-retriever/tests/transformers/test_keybert.py
@@ -1,25 +1,26 @@
 from typing import Any
 
 import pytest
-from keybert import KeyBERT  # type: ignore
 from langchain_core.documents import Document
 from langchain_graph_retriever.transformers.keybert import KeyBERTTransformer
 
 
-class FakeKeyBERT(KeyBERT):
-    def __init__(self):
-        pass
+@pytest.mark.extra
+def test_transform_documents(animal_docs: list[Document]):
+    from keybert import KeyBERT  # type: ignore
 
-    def extract_keywords(
-        self, docs: list[str], **kwargs: Any
-    ) -> list[list[tuple[str, float]]]:
-        return [
-            [(word, len(word)) for word in set(doc.split()) if len(word) > 5]
-            for doc in docs
-        ]
+    class FakeKeyBERT(KeyBERT):
+        def __init__(self):
+            pass
 
+        def extract_keywords(
+            self, docs: list[str], **kwargs: Any
+        ) -> list[list[tuple[str, float]]]:
+            return [
+                [(word, len(word)) for word in set(doc.split()) if len(word) > 5]
+                for doc in docs
+            ]
 
-def test_transform_documents(animal_docs: list[Document]):
     fake_model = FakeKeyBERT()
     transformer = KeyBERTTransformer(model=fake_model, metadata_key="keybert")
 

diff --git a/packages/langchain-graph-retriever/tests/transformers/test_spacy.py b/packages/langchain-graph-retriever/tests/transformers/test_spacy.py
@@ -1,28 +1,29 @@
 import pytest
 from langchain_core.documents import Document
 from langchain_graph_retriever.transformers.spacy import SpacyNERTransformer
-from spacy.language import Language
-from spacy.tokens import Doc, Span
-from spacy.vocab import Vocab
-
-
-class FakeLanguage(Language):
-    def __init__(self):
-        pass
-
-    def __call__(self, text: str | Doc, **kwargs) -> Doc:
-        vocab = Vocab()
-        assert isinstance(text, str)
-        doc = Doc(vocab=vocab, words=text.split())
-        doc.ents = [
-            Span(doc, start=0, end=1, label="first"),
-            Span(doc, start=1, end=2, label="second"),
-            Span(doc, start=2, end=3, label="third"),
-        ]
-        return doc
 
 
+@pytest.mark.extra
 def test_transform_documents(animal_docs: list[Document]):
+    from spacy.language import Language
+    from spacy.tokens import Doc, Span
+    from spacy.vocab import Vocab
+
+    class FakeLanguage(Language):
+        def __init__(self):
+            pass
+
+        def __call__(self, text: str | Doc, **kwargs) -> Doc:
+            vocab = Vocab()
+            assert isinstance(text, str)
+            doc = Doc(vocab=vocab, words=text.split())
+            doc.ents = [
+                Span(doc, start=0, end=1, label="first"),
+                Span(doc, start=1, end=2, label="second"),
+                Span(doc, start=2, end=3, label="third"),
+            ]
+            return doc
+
     fake_model = FakeLanguage()
 
     transformer = SpacyNERTransformer(model=fake_model, metadata_key="spacey")

diff --git a/pyproject.toml b/pyproject.toml
@@ -139,14 +139,14 @@ help = "Runs all formatting, lints, and checks (fixing where possible)"
 sequence = [ "lock-fix", "fmt-fix", "lint-fix", "type-check", "dep-check"]
 
 [tool.poe.tasks.test-gr]
-help = "Runs all tests (against in-memory stores)"
+help = "Runs graph-retriever tests (against in-memory stores)"
 cwd = "packages/graph-retriever"
 cmd = "uv run pytest -vs ."
 
 [tool.poe.tasks.test-lgr]
-help = "Runs all tests (against in-memory stores)"
+help = "Runs langchain-graph-retriever tests (against in-memory stores)"
 cwd = "packages/langchain-graph-retriever"
-cmd = "uv run pytest -vs ."
+cmd = "uv run pytest -vs --runextras ."
 
 [tool.poe.tasks.test]
 help = "Runs all tests (against in-memory stores)"