from unittest.mock import MagicMock
import pytest
from src.ingestion.embedding.batch_processor import BatchProcessor
from src.ingestion.models import Chunk
def test_batch_processor_splits_into_stable_batches_and_preserves_order():
chunks = [Chunk(text=t) for t in ["a", "b", "c", "d", "e"]]
dense = MagicMock()
sparse = MagicMock()
def dense_encode(batch, trace=None):
return [[ord(c.text)] for c in batch]
def sparse_encode(batch, trace=None):
return [{"term": c.text} for c in batch]
dense.encode.side_effect = dense_encode
sparse.encode.side_effect = sparse_encode
processor = BatchProcessor(batch_size=2)
result = processor.process(chunks, dense_encoder=dense, sparse_encoder=sparse)
assert [m.size for m in result.batches] == [2, 2, 1]
assert result.dense_vectors == [[97], [98], [99], [100], [101]]
assert result.sparse_vectors == [
{"term": "a"},
{"term": "b"},
{"term": "c"},
{"term": "d"},
{"term": "e"},
]
assert dense.encode.call_count == 3
assert sparse.encode.call_count == 3
assert [call.args[0] for call in dense.encode.call_args_list] == [
chunks[0:2],
chunks[2:4],
chunks[4:5],
]
def test_batch_processor_empty_input_returns_empty():
dense = MagicMock()
sparse = MagicMock()
processor = BatchProcessor(batch_size=2)
result = processor.process([], dense_encoder=dense, sparse_encoder=sparse)
assert result.dense_vectors == []
assert result.sparse_vectors == []
assert result.batches == []
dense.encode.assert_not_called()
sparse.encode.assert_not_called()
def test_batch_processor_rejects_non_positive_batch_size():
with pytest.raises(ValueError, match="batch_size must be a positive integer"):
BatchProcessor(batch_size=0)
