def test_discover_task_metadata(monkeypatch, tmpdir):
user_options = {"tessdata": "/some/path"}
monkeypatch.setattr(os.path, "exists", lambda args: True)
workflow = workflow_ocr.OCRWorkflow(global_settings=user_options)
tessdata_dir = tmpdir / "tessdata"
image_dir = tmpdir / "images"
tessdata_dir.ensure_dir()
user_options = {
"tessdata": tessdata_dir.strpath,
'Image File Type': 'JPEG 2000',
'Language': 'English',
'Path': image_dir.strpath
}
initial_results = [
Result(source=workflow_ocr.FindImagesTask,
data=[(image_dir / "dummy.jp2").strpath])
]
new_tasks = workflow.discover_task_metadata(initial_results, None,
**user_options)
assert len(new_tasks) == 1
new_task = new_tasks[0]
assert new_task == {
'source_file_path': (image_dir / "dummy.jp2").strpath,
'destination_path': image_dir.strpath,
'output_file_name': 'dummy.txt',
'lang_code': 'eng',
}
def test_discover_task_metadata_raises_with_no_tessdata(monkeypatch):
user_options = {"tessdata": "/some/path"}
monkeypatch.setattr(os.path, "exists", lambda args: True)
workflow = workflow_ocr.OCRWorkflow(global_settings=user_options)
monkeypatch.setattr(os.path, "exists", lambda args: False)
with pytest.raises(SpeedwagonException):
user_options = {"tessdata": None}
workflow.discover_task_metadata([], None, **user_options)
def test_no_config():
with pytest.raises(MissingConfiguration):
workflow_ocr.OCRWorkflow()
def workflow(self, monkeypatch):
global_settings = {"tessdata": os.path.join("some", "path")}
monkeypatch.setattr(workflow_ocr.os.path, "exists",
lambda path: path == global_settings["tessdata"])
return \
workflow_ocr.OCRWorkflow(global_settings)