def test_generate_commands_should_raise_if_missing_variable(work_dir):
"""
Test dvc bash command generation fails if a docstring template variable is missing
"""
python_script = 'def my_funct(subset: str, rate: int):\n' \
'\t"""\n' \
':param str locale: The data locale\n' \
':param output_file: the output_file\n' \
':dvc-out output_file: {{ conf.output_file }}\n' \
':dvc-extra: --locale {{ conf.locale }}\n' \
'\t"""\n' \
'\tprint(\'toto\')\n'
# Write python script
script_path = join(work_dir, 'script_python.py')
with open(script_path, 'w') as fd:
fd.write(python_script)
# Write docstring conf
dc_conf_path = join(work_dir, 'dc_conf.yml')
with open(dc_conf_path, 'w') as fd:
yaml.dump({'output_file': './data/other.txt'}, fd)
dvc_cmd_path = join(work_dir, 'dvc_cmd')
arguments = ['-i', script_path, '--out-dvc-cmd', dvc_cmd_path, '--working-directory', work_dir,
'--docstring-conf', dc_conf_path]
with pytest.raises(MlVToolException):
MlScriptToCmd().run(*arguments)
def test_should_overwrite_conf_for_docstring_conf(work_dir):
"""
Test main configuration is overwritten for docstring conf selection
"""
conf_dc_conf_path = join(work_dir, 'conf_dc_path.yml')
write_docstring_conf(conf_dc_conf_path, './output_file_base.txt')
conf_path, script_path = setup_with_conf(
work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
docstring_conf_path=conf_dc_conf_path)
new_dc_conf_path = join(work_dir, 'new_dc_path.yml')
write_docstring_conf(new_dc_conf_path, './output_file_overwritten.txt')
dvc_cmd_path = join(work_dir, 'new_place_dvc')
arguments = [
'-i', script_path, '--working-directory', work_dir, '--docstring-conf',
new_dc_conf_path, '--out-dvc-cmd', dvc_cmd_path
]
MlScriptToCmd().run(*arguments)
# Assert docstring template value are replaced with the right conf content
with open(dvc_cmd_path, 'r') as fd:
content = fd.read()
assert './output_file_base.txt' not in content
assert './output_file_overwritten.txt' in content
def test_should_raise_if_missing_dvc_command_output_path_argument_and_no_conf(
):
"""
Test command raise if dvc command output path is not provided when no conf
"""
arguments = ['-i', './test.py', '--working-directory', './']
with pytest.raises(MlVToolException):
MlScriptToCmd().run(*arguments)
def test_should_raise_if_dvc_command_output_path_exist_and_no_force(work_dir):
"""
Test command raise if dvc output path already exists and no force argument
"""
dvc_cmd_output = join(work_dir, 'dvc_out')
with open(dvc_cmd_output, 'w') as fd:
fd.write('')
arguments = [
'-i', './test.py', '--working-directory', work_dir, '--out-dvc-cmd',
dvc_cmd_output
]
with pytest.raises(MlVToolException):
MlScriptToCmd().run(*arguments)
def test_should_get_output_path_from_auto_detected_conf(work_dir):
"""
Test commands are generated from python script using auto detected configuration
"""
conf_path, script_path = setup_with_conf(work_dir,
conf_path=join(
work_dir,
DEFAULT_CONF_FILENAME))
arguments = ['-i', script_path, '--working-directory', work_dir]
MlScriptToCmd().run(*arguments)
# This path is generated using conf path and the script name
dvc_cmd_path = join(work_dir, 'dvc_cmd', 'script_path_dvc')
assert exists(dvc_cmd_path)
def test_should_generate_commands(work_dir):
"""
Test dvc bash command is generated from python script with param specified
in docstring.
"""
python_script = 'def my_funct(subset, rate):\n' \
'\t"""\n' \
':param str input_file: the input file\n' \
':param output_file: the output_file\n' \
':param rate: the rate\n' \
':param int retry:\n' \
':param List[int] threshold:\n' \
':dvc-in input_file: ./data/train_set.csv\n' \
':dvc-out output_file: {{ conf.output_file }}\n' \
':dvc-out: ./data/other.txt\n' \
'\t"""\n' \
'\tprint(\'toto\')\n'
# Write python script
script_path = join(work_dir, 'script_python.py')
with open(script_path, 'w') as fd:
fd.write(python_script)
# Write docstring conf
dc_conf_path = join(work_dir, 'dc_conf.yml')
with open(dc_conf_path, 'w') as fd:
yaml.dump({'output_file': './data/other.txt'}, fd)
dvc_cmd_path = join(work_dir, 'dvc_cmd')
arguments = [
'-i', script_path, '--out-dvc-cmd', dvc_cmd_path,
'--working-directory', work_dir, '--docstring-conf', dc_conf_path
]
MlScriptToCmd().run(*arguments)
assert exists(dvc_cmd_path)
assert stat.S_IMODE(os_stat(dvc_cmd_path).st_mode) == 0o755
# Ensure dvc command is in dvc bash command
with open(dvc_cmd_path, 'r') as fd:
dvc_bash_content = fd.read()
assert 'OUTPUT_FILE="./data/other.txt"' in dvc_bash_content
assert 'MLV_DVC_META_FILENAME="script_python.dvc"' in dvc_bash_content
assert 'dvc run${NO_CACHE_OPT} --overwrite-dvcfile -f $MLV_DVC_META_FILENAME' in dvc_bash_content
assert '-o $OUTPUT_FILE' in dvc_bash_content
assert '-o ./data/other.txt' in dvc_bash_content
assert '-d $INPUT_FILE' in dvc_bash_content
def test_should_get_output_path_from_conf(work_dir):
"""
Test commands are generated from python script using provided configuration
"""
conf_path, script_path = setup_with_conf(work_dir,
conf_path=join(
work_dir, 'my_conf'))
arguments = [
'-i', script_path, '--working-directory', work_dir, '--conf-path',
conf_path
]
MlScriptToCmd().run(*arguments)
# This path is generated using conf path and the script name
dvc_cmd_path = join(work_dir, 'dvc_cmd', 'script_path_dvc')
assert exists(dvc_cmd_path)
def test_should_overwrite_with_force_argument(work_dir):
"""
Test output paths are overwritten with force argument
"""
script_path = join(work_dir, 'script_path.py')
write_min_script(script_path)
dvc_cmd_path = join(work_dir, 'dvc_cmd')
with open(dvc_cmd_path, 'w') as fd:
fd.write('')
arguments = [
'-i', script_path, '--out-dvc-cmd', dvc_cmd_path,
'--working-directory', work_dir, '--force'
]
MlScriptToCmd().run(*arguments)
with open(dvc_cmd_path, 'r') as fd:
assert fd.read()
def test_should_overwrite_conf_for_path(work_dir):
"""
Test output paths argument overwrite conf
"""
conf_path, script_path = setup_with_conf(work_dir,
conf_path=join(
work_dir,
DEFAULT_CONF_FILENAME))
arg_dvc_cmd_path = join(work_dir, 'new_place_dvc')
arguments = [
'-i', script_path, '--working-directory', work_dir, '--out-dvc-cmd',
arg_dvc_cmd_path
]
MlScriptToCmd().run(*arguments)
# Assert output path is the one from command argument not from conf
assert exists(arg_dvc_cmd_path)
conf_dvc_cmd_path = join(work_dir, 'dvc_cmd', 'script_path_dvc')
assert not exists(conf_dvc_cmd_path)
def test_should_generate_dvc_with_whole_cmd(work_dir):
"""
Test dvc bash command is generated from python script with whole dvc command
ad specified in docstring
"""
cmd = 'dvc run -o ./out_train.csv \n' \
'-o ./out_test.csv\n' \
'$MLV_PY_CMD_PATH -m train --out ./out_train.csv &&\n' \
'./python/${MLV_PY_CMD_NAME} -m test --out ./out_test.csv'
python_script = 'def my_funct(subset: str, rate: int):\n' \
'\t"""\n' \
':param str input_file: the input file\n' \
':param output_file: the output_file\n' \
f':dvc-cmd:{cmd}\n' \
'\t"""\n' \
'\tprint(\'toto\')\n'
script_path = join(work_dir, 'script_python.py')
with open(script_path, 'w') as fd:
fd.write(python_script)
makedirs(join(work_dir, 'python'))
dvc_cmd_path = join(work_dir, 'dvc_cmd')
arguments = [
'-i', script_path, '--out-dvc-cmd', dvc_cmd_path,
'--working-directory', work_dir
]
MlScriptToCmd().run(*arguments)
assert exists(dvc_cmd_path)
# Ensure whole command is in dvc bash command
with open(dvc_cmd_path, 'r') as fd:
dvc_bash_content = fd.read()
relative_py_cmd_path = relpath(script_path, work_dir)
assert f'MLV_PY_CMD_PATH="{relative_py_cmd_path}"' in dvc_bash_content
assert f'MLV_PY_CMD_NAME="{basename(relative_py_cmd_path)}"' in dvc_bash_content
assert cmd.replace('\n', ' \\\n') in dvc_bash_content
def test_should_get_docstring_conf_from_main_conf(work_dir):
"""
Test docstring template values are replaced with docstring conf provided in main conf file
"""
conf_dc_conf_path = join(work_dir, 'conf_dc_path.yml')
write_docstring_conf(conf_dc_conf_path, './output_file_base.txt')
conf_path, script_path = setup_with_conf(
work_dir,
conf_path=join(work_dir, DEFAULT_CONF_FILENAME),
docstring_conf_path=conf_dc_conf_path)
dvc_cmd_path = join(work_dir, 'new_place_dvc')
arguments = [
'-i', script_path, '--working-directory', work_dir, '--out-dvc-cmd',
dvc_cmd_path
]
MlScriptToCmd().run(*arguments)
# Assert docstring template value are replaced with the docstring conf content
with open(dvc_cmd_path, 'r') as fd:
content = fd.read()
assert './output_file_base.txt' in content