test: add coverage for text_embedding_base.py (#476)

- Create tests/test_text_embedding_base.py
- Update pyproject.toml to require Python >= 3.11 to resolve numpy conflicts
- Verify delegation and error handling in abstract base class

Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>

Author: n24q02m

tests/test_text_embedding_base.py

@@ -0,0 +1,124 @@
+from collections.abc import Iterable
+from typing import Any
+
+import numpy as np
+import pytest
+
+from qwen3_embed.common.model_description import DenseModelDescription
+from qwen3_embed.common.types import NumpyArray
+from qwen3_embed.text.text_embedding_base import TextEmbeddingBase
+
+
+class MockEmbedding(TextEmbeddingBase):
+    """A concrete implementation of TextEmbeddingBase for testing purposes."""
+
+    @classmethod
+    def _list_supported_models(cls) -> list[DenseModelDescription]:
+        return []
+
+    def embed(
+        self,
+        documents: str | Iterable[str],
+        batch_size: int = 256,
+        parallel: int | None = None,
+        **kwargs: Any,
+    ) -> Iterable[NumpyArray]:
+        # Simple mock implementation that yields dummy arrays
+        if isinstance(documents, str):
+            documents = [documents]
+        for _ in documents:
+            yield np.zeros(10)
+
+    @classmethod
+    def get_embedding_size(cls, model_name: str) -> int:
+        return 10
+
+    @property
+    def embedding_size(self) -> int:
+        return 10
+
+    def token_count(self, texts: str | Iterable[str], **kwargs: Any) -> int:
+        return 5
+
+
+def test_text_embedding_base_init():
+    """Test that TextEmbeddingBase initializes its attributes correctly."""
+    model_name = "test-model"
+    cache_dir = "/tmp/cache"
+    threads = 4
+    local_files_only = True
+
+    # We instantiate the mock because the base class is abstract (uses NotImplementedError)
+    model = MockEmbedding(
+        model_name=model_name,
+        cache_dir=cache_dir,
+        threads=threads,
+        local_files_only=local_files_only,
+    )
+
+    assert model.model_name == model_name
+    assert model.cache_dir == cache_dir
+    assert model.threads == threads
+    assert model._local_files_only is True
+    assert model._embedding_size is None
+
+
+def test_passage_embed_delegation():
+    """Test that passage_embed correctly delegates to the embed method."""
+    model = MockEmbedding(model_name="test-model")
+    texts = ["text1", "text2"]
+
+    embeddings = list(model.passage_embed(texts))
+
+    assert len(embeddings) == 2
+    for emb in embeddings:
+        assert isinstance(emb, np.ndarray)
+        assert emb.shape == (10,)
+
+
+def test_query_embed_single_string():
+    """Test that query_embed handles a single string correctly."""
+    model = MockEmbedding(model_name="test-model")
+    query = "test query"
+
+    embeddings = list(model.query_embed(query))
+
+    assert len(embeddings) == 1
+    assert isinstance(embeddings[0], np.ndarray)
+    assert embeddings[0].shape == (10,)
+
+
+def test_query_embed_iterable():
+    """Test that query_embed handles an iterable of strings correctly."""
+    model = MockEmbedding(model_name="test-model")
+    queries = ["query1", "query2"]
+
+    embeddings = list(model.query_embed(queries))
+
+    assert len(embeddings) == 2
+    for emb in embeddings:
+        assert isinstance(emb, np.ndarray)
+        assert emb.shape == (10,)
+
+
+def test_base_class_raises_not_implemented():
+    """Test that the base class methods raise NotImplementedError when called directly or via incomplete subclass."""
+
+    class IncompleteEmbedding(TextEmbeddingBase):
+        @classmethod
+        def _list_supported_models(cls) -> list[DenseModelDescription]:
+            return []
+
+    model = IncompleteEmbedding(model_name="test-model")
+
+    with pytest.raises(NotImplementedError):
+        list(model.embed("test"))
+
+    with pytest.raises(NotImplementedError, match="Subclasses must implement this method"):
+        IncompleteEmbedding.get_embedding_size("test-model")
+
+    with pytest.raises(NotImplementedError, match="Subclasses must implement this method"):
+        _ = model.embedding_size
+
+    with pytest.raises(NotImplementedError, match="Subclasses must implement this method"):
+        model.token_count("test")