"""Tests for embedding generation."""
from sentence_transformers import SentenceTransformer
from semantic_code_mcp.indexer.embedder import Embedder
class TestEmbedder:
"""Tests for Embedder with pre-loaded model."""
def test_creates_embedder(self, model: SentenceTransformer):
"""Can create an embedder instance with a model."""
embedder = Embedder(model)
assert embedder is not None
def test_embed_single_text(self, embedder: Embedder):
"""Can embed a single text string."""
embedding = embedder.embed_text("def hello(): pass")
assert isinstance(embedding, list)
assert len(embedding) == 384 # MiniLM embedding dimension
assert all(isinstance(x, float) for x in embedding)
def test_embed_batch(self, embedder: Embedder):
"""Can embed multiple texts in a batch."""
texts = [
"def hello(): pass",
"class Foo: pass",
"def bar(x): return x * 2",
]
embeddings = embedder.embed_batch(texts)
assert len(embeddings) == 3
assert all(len(e) == 384 for e in embeddings)
def test_embed_empty_batch_returns_empty(self, embedder: Embedder):
"""Empty batch returns empty list."""
embeddings = embedder.embed_batch([])
assert embeddings == []
def test_similar_texts_have_similar_embeddings(self, embedder: Embedder):
"""Semantically similar texts should have similar embeddings."""
# Similar code
emb1 = embedder.embed_text("def add(a, b): return a + b")
emb2 = embedder.embed_text("def sum(x, y): return x + y")
# Different code
emb3 = embedder.embed_text("class DatabaseConnection: pass")
# Compute cosine similarity
def cosine_sim(a: list[float], b: list[float]) -> float:
dot = sum(x * y for x, y in zip(a, b, strict=True))
norm_a = sum(x * x for x in a) ** 0.5
norm_b = sum(x * x for x in b) ** 0.5
return dot / (norm_a * norm_b)
sim_similar = cosine_sim(emb1, emb2)
sim_different = cosine_sim(emb1, emb3)
# Similar texts should have higher similarity
assert sim_similar > sim_different
def test_embedding_dimension(self, embedder: Embedder):
"""Embeddings have correct dimension for the model."""
embedding = embedder.embed_text("test")
assert embedder.embedding_dim == 384
assert len(embedding) == embedder.embedding_dim
