def test_generate_task_creates_a_file(monkeypatch, tmpdir):
source_image = tmpdir / "dummy.jp2"
out_text = tmpdir / "dummy.txt"
tessdata_dir = tmpdir / "tessdata"
tessdata_dir.ensure_dir()
(tessdata_dir / "eng.traineddata").ensure()
(tessdata_dir / "osd.traineddata").ensure()
def mock_read(*args, **kwargs):
return "Spam bacon eggs"
mock_reader = Mock()
with monkeypatch.context() as patcher:
patcher.setattr(reader.Reader, "read", mock_read)
patcher.setattr(tesseractwrap, "Reader", mock_reader)
task = workflow_ocr.GenerateOCRFileTask(
source_image=source_image.strpath,
out_text_file=out_text.strpath,
tesseract_path=tessdata_dir.strpath)
task.log = MagicMock()
task.work()
assert os.path.exists(out_text.strpath)
with open(out_text.strpath, "r") as f:
assert f.read() == "Spam bacon eggs"
def test_work(self, monkeypatch):
source_image = os.path.join("12345", "sample.jp2")
out_text_file = os.path.join("12345", "sample.txt")
lang = "eng"
tesseract_path = "tesspath"
workflow_ocr.GenerateOCRFileTask.set_tess_path = Mock()
workflow_ocr.GenerateOCRFileTask.engine = Mock()
task = workflow_ocr.GenerateOCRFileTask(source_image=source_image,
out_text_file=out_text_file,
lang=lang,
tesseract_path=tesseract_path)
m = mock_open()
with patch('speedwagon.workflows.workflow_ocr.open', m):
assert task.work() is True
assert m.called is True
def test_read_image(self, monkeypatch):
source_image = os.path.join("12345", "sample.jp2")
out_text_file = os.path.join("12345", "sample.txt")
lang = "eng"
tesseract_path = "tesspath"
workflow_ocr.GenerateOCRFileTask.set_tess_path = Mock()
workflow_ocr.GenerateOCRFileTask.engine = Mock()
reader = Mock()
workflow_ocr.GenerateOCRFileTask.engine.get_reader = \
lambda args: reader
task = workflow_ocr.GenerateOCRFileTask(source_image=source_image,
out_text_file=out_text_file,
lang=lang,
tesseract_path=tesseract_path)
task.read_image(source_image, "eng")
assert reader.read.called is True
def test_read_image(self, monkeypatch):
source_image = os.path.join("12345", "sample.jp2")
out_text_file = os.path.join("12345", "sample.txt")
lang = "eng"
tesseract_path = "tesspath"
workflow_ocr.GenerateOCRFileTask.set_tess_path = Mock()
workflow_ocr.GenerateOCRFileTask.engine = Mock()
reader = Mock()
workflow_ocr.GenerateOCRFileTask.engine.get_reader = \
lambda args: reader
task = workflow_ocr.GenerateOCRFileTask(source_image=source_image,
out_text_file=out_text_file,
lang=lang,
tesseract_path=tesseract_path)
task.read_image(source_image, "eng")
assert reader.read.called is True
@pytest.mark.parametrize("task", [
workflow_ocr.GenerateOCRFileTask(source_image="source_image",
out_text_file="out_text_file",
lang="lang",
tesseract_path="tesseract_path"),
workflow_ocr.FindImagesTask(root="root", file_extension=".tif")
])
def test_tasks_have_description(task):
assert task.task_description() is not None