"""Unit tests for batch file validator service.""" import io from unittest.mock import AsyncMock import pytest from fastapi import UploadFile from src.services.batch_validator import ( MAX_BATCH_SIZE_BYTES, MAX_FILES_PER_BATCH, BatchValidationError, validate_batch, validate_batch_file_count, validate_batch_total_size, ) def create_mock_upload_file(filename: str, content: bytes) -> UploadFile: """Helper to create a mock UploadFile with specified content. Args: filename: Name of the file. content: File content as bytes. Returns: UploadFile instance with mock read/seek methods. """ file_obj = io.BytesIO(content) upload_file = UploadFile(filename=filename, file=file_obj) # Override read and seek to work with BytesIO async def async_read(): return file_obj.read() async def async_seek(position: int): return file_obj.seek(position) upload_file.read = async_read upload_file.seek = async_seek return upload_file class TestValidateBatchFileCount: """Test file count validation.""" @pytest.mark.asyncio async def test_valid_single_file(self): """Should pass for single file.""" files = [create_mock_upload_file("file1.xlsx", b"test")] await validate_batch_file_count(files) # Should not raise @pytest.mark.asyncio async def test_valid_multiple_files(self): """Should pass for multiple files within limit.""" files = [ create_mock_upload_file(f"file{i}.xlsx", b"test") for i in range(10) ] await validate_batch_file_count(files) # Should not raise @pytest.mark.asyncio async def test_valid_max_files(self): """Should pass for exactly 20 files.""" files = [ create_mock_upload_file(f"file{i}.xlsx", b"test") for i in range(MAX_FILES_PER_BATCH) ] await validate_batch_file_count(files) # Should not raise @pytest.mark.asyncio async def test_too_many_files(self): """Should raise TOO_MANY_FILES for 21 files.""" files = [ create_mock_upload_file(f"file{i}.xlsx", b"test") for i in range(MAX_FILES_PER_BATCH + 1) ] with pytest.raises(BatchValidationError) as exc_info: await validate_batch_file_count(files) assert exc_info.value.code == "TOO_MANY_FILES" assert "21 files" in exc_info.value.message assert "maximum allowed is 20" in exc_info.value.message @pytest.mark.asyncio async def test_empty_batch(self): """Should raise EMPTY_BATCH for zero files.""" files = [] with pytest.raises(BatchValidationError) as exc_info: await validate_batch_file_count(files) assert exc_info.value.code == "EMPTY_BATCH" assert "No files provided" in exc_info.value.message class TestValidateBatchTotalSize: """Test total batch size validation.""" @pytest.mark.asyncio async def test_valid_small_batch(self): """Should pass for small batch.""" files = [ create_mock_upload_file("file1.xlsx", b"x" * 1024), # 1KB create_mock_upload_file("file2.xlsx", b"y" * 2048), # 2KB ] total_size = await validate_batch_total_size(files) assert total_size == 1024 + 2048 @pytest.mark.asyncio async def test_valid_large_batch_under_limit(self): """Should pass for batch just under 500MB limit.""" # Create 10 files of ~49MB each = ~490MB total file_size = 49 * 1024 * 1024 files = [ create_mock_upload_file(f"file{i}.xlsx", b"x" * file_size) for i in range(10) ] total_size = await validate_batch_total_size(files) assert total_size == file_size * 10 assert total_size < MAX_BATCH_SIZE_BYTES @pytest.mark.asyncio async def test_valid_batch_at_limit(self): """Should pass for batch exactly at 500MB limit.""" files = [create_mock_upload_file("file1.xlsx", b"x" * MAX_BATCH_SIZE_BYTES)] total_size = await validate_batch_total_size(files) assert total_size == MAX_BATCH_SIZE_BYTES @pytest.mark.asyncio async def test_batch_exceeds_limit(self): """Should raise BATCH_TOO_LARGE for batch over 500MB.""" oversized = MAX_BATCH_SIZE_BYTES + 1024 # 500MB + 1KB files = [create_mock_upload_file("huge.xlsx", b"x" * oversized)] with pytest.raises(BatchValidationError) as exc_info: await validate_batch_total_size(files) assert exc_info.value.code == "BATCH_TOO_LARGE" assert "500.00MB" in exc_info.value.message assert "exceeds maximum allowed size of 500MB" in exc_info.value.message @pytest.mark.asyncio async def test_multiple_files_exceed_limit(self): """Should raise BATCH_TOO_LARGE when multiple files exceed limit.""" # 5 files of 110MB each = 550MB total (over 500MB limit) file_size = 110 * 1024 * 1024 files = [ create_mock_upload_file(f"file{i}.xlsx", b"x" * file_size) for i in range(5) ] with pytest.raises(BatchValidationError) as exc_info: await validate_batch_total_size(files) assert exc_info.value.code == "BATCH_TOO_LARGE" @pytest.mark.asyncio async def test_file_pointer_reset_after_read(self): """Should reset file pointers after reading for size check.""" content = b"test content" files = [create_mock_upload_file("file1.xlsx", content)] await validate_batch_total_size(files) # File should be readable again after validation read_content = await files[0].read() assert read_content == content class TestValidateBatch: """Test orchestrated batch validation.""" @pytest.mark.asyncio async def test_valid_batch_all_checks_pass(self): """Should pass when all validations succeed.""" files = [ create_mock_upload_file("file1.xlsx", b"x" * (10 * 1024 * 1024)), # 10MB create_mock_upload_file("file2.xlsx", b"y" * (20 * 1024 * 1024)), # 20MB ] total_size = await validate_batch(files) assert total_size == 30 * 1024 * 1024 @pytest.mark.asyncio async def test_fails_on_too_many_files(self): """Should fail on file count before checking size.""" # 21 small files (file count violation) files = [ create_mock_upload_file(f"file{i}.xlsx", b"x" * 1024) for i in range(MAX_FILES_PER_BATCH + 1) ] with pytest.raises(BatchValidationError) as exc_info: await validate_batch(files) assert exc_info.value.code == "TOO_MANY_FILES" @pytest.mark.asyncio async def test_fails_on_batch_size(self): """Should fail on total size after file count passes.""" # 10 files of 60MB each = 600MB (size violation) file_size = 60 * 1024 * 1024 files = [ create_mock_upload_file(f"file{i}.xlsx", b"x" * file_size) for i in range(10) ] with pytest.raises(BatchValidationError) as exc_info: await validate_batch(files) assert exc_info.value.code == "BATCH_TOO_LARGE" @pytest.mark.asyncio async def test_empty_batch_fails(self): """Should fail for empty batch.""" files = [] with pytest.raises(BatchValidationError) as exc_info: await validate_batch(files) assert exc_info.value.code == "EMPTY_BATCH"