def test_should_set_docstring_conf_path_relatively_to_top_directory(work_dir):
""" Test load valid conf file """
conf_file = join(work_dir, '.mlvtools')
write_conf(work_dir=work_dir, conf_path=conf_file, docstring_conf='./doc_conf.yml')
conf = load_conf_or_default(conf_file, working_directory=work_dir)
assert conf.docstring_conf == join(work_dir, './doc_conf.yml')
def test_should_detect_consistency_for_notebooks(work_dir):
"""
Test check_all_scripts_consistency exit without error when there is consistency
between all notebooks and scripts from the provided directory
"""
notebook_dir = join(CURRENT_DIR, 'data', 'notebooks')
script_dir = join(CURRENT_DIR, 'data', 'script_dir')
write_conf(work_dir, conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
script_dir=script_dir, dvc_cmd_dir=work_dir)
check_call(['check_all_scripts_consistency', '-n', notebook_dir, '-w', work_dir])
def setup_with_conf(work_dir: str,
conf_path: str = None,
docstring_conf_path: str = None) -> Tuple[str, str]:
write_conf(work_dir=work_dir,
conf_path=conf_path,
ignore_keys=['# Ignore'],
dvc_cmd_dir='./dvc_cmd',
docstring_conf=docstring_conf_path)
script_path = join(work_dir, 'script_path.py')
docstring = '"""\n:param out:\n:dvc-out out: {{ conf.out_file }}\n"""' if docstring_conf_path else None
write_min_script(script_path, docstring)
return conf_path, script_path
def test_should_detect_inconsistency_if_at_least_one_inconsistent_script(work_dir):
"""
Test check_all_scripts_consistency exit with error if there is at least one inconsistency
"""
notebook_dir, script_dir = temporary_setup(work_dir)
# Replace a script content
with open(join(script_dir, listdir(script_dir)[0]), 'w') as fd:
fd.write('print("Hello world!")')
write_conf(work_dir, conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
script_dir=script_dir, dvc_cmd_dir=work_dir)
ret_code = call(['check_all_scripts_consistency', '-n', notebook_dir, '-w', work_dir])
assert ret_code != 0
def test_should_generate_python_script_with_conf_auto_detect(work_dir, mocker):
"""
Convert a Jupyter Notebook to a Python 3 script using conf
"""
mocked_check_output = mocker.patch('subprocess.check_output',
return_value=work_dir.encode())
kept_cells, dropped_cells, docstring, notebook_path = generate_test_notebook(
work_dir=work_dir, notebook_name='test_nb.ipynb')
# Create conf with knew ignore cell keywords
conf_data = write_conf(work_dir=work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
ignore_keys=['# Ignore', 'import'])
cmd_arguments = ['-n', notebook_path]
IPynbToPython().run(*cmd_arguments)
# This path is generated using the conf script_dir and the notebook name
output_script_path = join(work_dir,
conf_data['path']['python_script_root_dir'],
'mlvtools_test_nb.py')
assert exists(output_script_path)
with open(output_script_path, 'r') as fd:
file_content = fd.read()
# With those knew keywords from conf the first code cell must be ignored (due to import)
# The second no effect cell must remain because "# No effect" is no more a keyword
assert not is_in(kept_cells.code[0][1], file_content)
assert is_in(dropped_cells.no_effect[1][1], file_content)
# Check conf file has been found using git rev-parse command
assert mocked_check_output.mock_calls == [
mocker.call(['git', 'rev-parse', '--show-toplevel'], cwd=work_dir)
]
def test_should_handle_notebook_with_invalid_python_name_with_conf(
work_dir, mocker):
"""
Test invalid python filename are converted
"""
notebook_path = gen_notebook(cells=[('code', 'pass')],
tmp_dir=work_dir,
file_name='01_(test) nb.ipynb')
# Create conf in working directory
conf_data = write_conf(work_dir=work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
ignore_keys=['# Ignore', 'remove='])
cmd_arguments = ['-n', notebook_path, '-w', work_dir]
IPynbToPython().run(*cmd_arguments)
# This path is generated using the conf script_dir and the notebook name
output_script_path = join(work_dir,
conf_data['path']['python_script_root_dir'],
'mlvtools_01__test_nb.py')
assert exists(output_script_path)
with open(output_script_path, 'r') as fd:
file_content = fd.read()
# Ensure generated file syntax is right
compile(file_content, output_script_path, 'exec')
def test_should_handle_notebook_with_invalid_python_name_with_conf(work_dir, mocker):
"""
Test invalid python filename are converted
"""
mocked_check_output = mocker.patch('subprocess.check_output', return_value=work_dir.encode())
notebook_path = gen_notebook(cells=[('code', 'pass')], tmp_dir=work_dir, file_name='01_(test) nb.ipynb')
# Create conf in a freshly init git repo
conf_data = write_conf(work_dir=work_dir, conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
ignore_keys=['# Ignore', 'remove='])
cmd_arguments = ['-n', notebook_path]
IPynbToPython().run(*cmd_arguments)
# This path is generated using the conf script_dir and the notebook name
output_script_path = join(work_dir, conf_data['path']['python_script_root_dir'], 'mlvtools_01__test_nb.py')
assert exists(output_script_path)
with open(output_script_path, 'r') as fd:
file_content = fd.read()
# Ensure generated file syntax is right
compile(file_content, output_script_path, 'exec')
assert mocked_check_output.mock_calls == [mocker.call(
['git', 'rev-parse', '--show-toplevel'],
cwd=work_dir)]
def test_should_generate_python_script_with_conf_auto_detect(work_dir, mocker):
"""
Convert a Jupyter Notebook to a Python 3 script using conf
"""
kept_cells, dropped_cells, docstring, notebook_path = generate_test_notebook(
work_dir=work_dir, notebook_name='test_nb.ipynb')
# Create conf with knew ignore cell keywords
conf_data = write_conf(work_dir=work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
ignore_keys=['# Ignore', 'import'])
cmd_arguments = ['-n', notebook_path, '-w', work_dir]
IPynbToPython().run(*cmd_arguments)
# This path is generated using the conf script_dir and the notebook name
output_script_path = join(work_dir,
conf_data['path']['python_script_root_dir'],
'mlvtools_test_nb.py')
assert exists(output_script_path)
with open(output_script_path, 'r') as fd:
file_content = fd.read()
# With those knew keywords from conf the first code cell must be ignored (due to import)
# The second no effect cell must remain because "# No effect" is no more a keyword
assert not is_in(kept_cells.code[0][1], file_content)
assert is_in(dropped_cells.no_effect[1][1], file_content)
def test_should_check_consistency_and_exit_with_error_if_same_but_diff_from_conf(work_dir, ref_notebook_path,
ref_script_content):
"""
Test check consistency between a notebook and its script
with diff only on blank lines and comments.
Should exit with error.
Discard a notebook cell changing conf ignore_keys
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir, conf_path, ignore_keys=['# A Tag'])
script_path = join(work_dir, 'script.py')
with open(script_path, 'w') as fd:
fd.write(ref_script_content)
arguments = ['-n', ref_notebook_path, '-s', script_path, '--conf-path', conf_path, '--working-directory', work_dir]
with pytest.raises(SystemExit) as e:
IPynbCheckScript().run(*arguments)
assert e.value.code != 0
def test_should_detect_ignore_notebooks(work_dir):
"""
Test check_all_scripts_consistency exit with error if there is at least one inconsistency
"""
notebook_dir, script_dir = temporary_setup(work_dir)
# Replace the script which correspond to notebook.ipynb
with open(join(script_dir, 'mlvtools_notebook.py'), 'w') as fd:
fd.write('print("Hello world!")')
# Duplicate a notebook with an other name so there is no associated script
shutil.copy(join(notebook_dir, 'notebook.ipynb'), join(notebook_dir, 'new_nb.ipynb'))
write_conf(work_dir, conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
script_dir=script_dir, dvc_cmd_dir=work_dir)
# Ignore notebook.ipynb and new_nb.ipynb
check_call(['check_all_scripts_consistency', '-n', notebook_dir, '-w', work_dir,
'-i', 'notebook.ipynb', '-i', 'new_nb.ipynb'])
def test_should_detect_consistency_for_notebook_and_script_with_conf(work_dir):
"""
Test check_script_consistency exit without error when there is consistency
between notebook and script with conf.
Without conf there is inconsistency but with the right ignore cell there is consistency
"""
notebook_path = join(CURRENT_DIR, 'data', 'notebooks',
'notebook_blank_and_comment_diff.ipynb')
script_path = join(CURRENT_DIR, 'data',
'script_no_blank_no_comment_disable_cell.py')
write_conf(work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
ignore_keys=['# A Tag'])
check_call([
'check_script_consistency', '-n', notebook_path, '-s', script_path,
'-w', work_dir
])
def test_should_check_consistency_for_nb_from_dir_and_exit_with_error_if_dif_from_conf(
work_dir, notebook_dir, script_dir):
"""
Test check consistency for all notebooks from a given directory.
Should exit with error.
Discard a notebook cell changing conf ignore_keys
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir,
conf_path,
script_dir=script_dir,
dvc_cmd_dir=work_dir,
ignore_keys=['# A Tag'])
arguments = [
'-n', notebook_dir, '-c', conf_path, '--working-directory', work_dir
]
with pytest.raises(SystemExit) as e:
IPynbCheckAllScripts().run(*arguments)
assert e.value.code != 0
def test_exit_with_error_if_a_script_is_missing(work_dir, notebook_dir,
script_dir):
"""
Test check consistency exits with error if a script is missing.
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir,
conf_path,
script_dir=script_dir,
dvc_cmd_dir=work_dir)
# Remove a valid script
remove(join(script_dir, 'mlvtools_bye.py'))
arguments = [
'-n', notebook_dir, '-c', conf_path, '--working-directory', work_dir
]
with pytest.raises(SystemExit) as e:
IPynbCheckAllScripts().run(*arguments)
assert e.value.code != 0
def test_exit_without_error_if_a_script_is_inconsistent_but_ignored(
work_dir, notebook_dir, script_dir):
"""
Test check consistency exits without error if inconsistent script is ignored
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir,
conf_path,
script_dir=script_dir,
dvc_cmd_dir=work_dir)
# Overwrite one valid script
write_script(join(script_dir, 'mlvtools_bye.py'),
'print("A different thing")')
arguments = [
'-n', notebook_dir, '-c', conf_path, '--working-directory', work_dir,
'-i', 'bye.ipynb'
]
with pytest.raises(SystemExit) as e:
IPynbCheckAllScripts().run(*arguments)
assert e.value.code == 0
def test_should_check_consistency_for_all_notebooks_from_directory_and_exit_with_error(
work_dir, notebook_dir, script_dir):
"""
Test check consistency for all notebooks from a given directory.
Should exit with error due to one inconsistency.
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir,
conf_path,
script_dir=script_dir,
dvc_cmd_dir=work_dir)
# Overwrite one valid script
write_script(join(script_dir, 'mlvtools_bye.py'),
'print("A different thing")')
arguments = [
'-n', notebook_dir, '-c', conf_path, '--working-directory', work_dir
]
with pytest.raises(SystemExit) as e:
IPynbCheckAllScripts().run(*arguments)
assert e.value.code != 0
def test_should_convert_to_python_script_using_command_line(work_dir):
"""
Test ipynb_to_python using command line
"""
dvc_dir = join(work_dir, 'dvc')
script_dir = join(work_dir, 'scripts')
write_conf(work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
script_dir=script_dir,
dvc_cmd_dir=dvc_dir)
notebook_path = join(CURRENT_DIR, 'data', 'notebook.ipynb')
check_call(['ipynb_to_dvc', '-n', notebook_path, '-w', work_dir])
assert exists(script_dir)
script_dir_content = glob.glob(join(script_dir, 'mlvtools*.py'))
assert len(script_dir_content) == 1
assert stat.S_IMODE(os_stat(script_dir_content[0]).st_mode) == 0o755
assert exists(script_dir)
dvc_dir_content = glob.glob(join(dvc_dir, 'mlvtools*_dvc'))
assert len(dvc_dir_content) == 1
assert stat.S_IMODE(os_stat(dvc_dir_content[0]).st_mode) == 0o755
def test_should_check_consistency_for_all_notebooks_from_directory_and_exit_without_error(
work_dir, notebook_dir, script_dir):
"""
Test check consistency for all notebooks from a given directory.
Should exit without error.
Ignore other file from the given directory.
"""
conf_path = join(work_dir, 'conf.json')
write_conf(work_dir,
conf_path,
script_dir=script_dir,
dvc_cmd_dir=work_dir)
# Add noise: a text file in notebook directory
with open(join(notebook_dir, 'text_file.txt'), 'w') as fd:
fd.write('some text')
arguments = [
'-n', notebook_dir, '-c', conf_path, '--working-directory', work_dir
]
with pytest.raises(SystemExit) as e:
IPynbCheckAllScripts().run(*arguments)
assert e.value.code == 0
def test_should_load_conf_file(work_dir):
""" Test load valid conf file """
conf_file = join(work_dir, '.mlvtools')
write_conf(work_dir=work_dir, conf_path=conf_file, ignore_keys=['# No effect', "# Ignore"],
script_dir='./scripts', dvc_cmd_dir='./dvc_cmd',
dvc_py_cmd_name='VAR_Name', dvc_py_cmd_path='var_PATh3',
dvc_meta_file_name='mETA_VAR_NaME')
conf = load_conf_or_default(conf_file, working_directory=work_dir)
assert conf.path.python_script_root_dir == './scripts'
assert conf.path.dvc_cmd_root_dir == './dvc_cmd'
assert '# No effect' in conf.ignore_keys
assert '# Ignore' in conf.ignore_keys
assert conf.dvc_var_python_cmd_path == 'var_PATh3'
assert conf.dvc_var_python_cmd_name == 'VAR_Name'
assert conf.dvc_var_meta_filename == 'mETA_VAR_NaME'
script_path = join(conf.path.python_script_root_dir, 'mlvtools_pipeline_part1.py')
assert get_script_output_path('./data/Pipeline Part1.ipynb', conf) == join(work_dir, script_path)
dvc_cmd_path = join(conf.path.dvc_cmd_root_dir, 'pipeline_part1_dvc')
assert get_dvc_cmd_output_path('./data/pipeline_part1.py', conf) == join(work_dir, dvc_cmd_path)
def setup_with_conf(work_dir: str, conf_path: str) -> Tuple[str, str]:
write_conf(work_dir=work_dir,
conf_path=conf_path,
script_dir='./test_scripts')
nb_path = gen_notebook([('markdown', '# test')], work_dir, 'nb_test.ipynb')
return conf_path, nb_path
