def test_simple_replace(self):
"""Replace parameter references in simple template with argument values.
"""
for filename in [TEMPLATE_YAML_FILE, TEMPLATE_JSON_FILE]:
template = DefaultTemplateLoader().load(filename)
arguments = {
'code':
TemplateArgument(parameter=template.get_parameter('code'),
value=FileHandle('code/helloworld.py')),
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle('data/list-of-names.txt')),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=10)
}
spec = tmpl.replace_args(spec=template.workflow_spec,
arguments=arguments,
parameters=template.parameters)
assert spec['inputs']['files'][0] == 'helloworld.py'
assert spec['inputs']['files'][1] == 'data/names.txt'
assert spec['inputs']['parameters'][
'helloworld'] == 'code/helloworld.py'
assert spec['inputs']['parameters'][
'inputfile'] == 'data/names.txt'
assert spec['inputs']['parameters']['sleeptime'] == 10
assert spec['inputs']['parameters']['waittime'] == 5
def test_run_with_invalid_cmd(self, tmpdir):
"""Execute the helloworld example with an invalid shell command."""
# Execution fails if a file that is referenced by the workflow does not
# exist in the created workspace
repo = TemplateRepository(base_dir=os.path.join(str(tmpdir), 'repo'))
template = repo.add_template(
src_dir=WORKFLOW_DIR, template_spec_file=TEMPLATE_WITH_INVALID_CMD)
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=3)
}
# Run workflow syncronously
engine = MultiProcessWorkflowEngine(base_dir=os.path.join(
str(tmpdir), 'engine'),
run_async=False,
verbose=True)
sync_run_id, state = engine.execute(template, arguments)
assert state.is_error()
assert len(state.messages) > 0
state = engine.get_state(sync_run_id)
assert state.is_error()
assert len(state.messages) > 0
def upload_stream(self, file, file_name):
"""Create a new entry from a given file stream. Will copy the given
file to a file in the base directory.
Parameters
----------
file: werkzeug.datastructures.FileStorage
File object (e.g., uploaded via HTTP request)
file_name: string
Name of the file
Returns
-------
benchtmpl.io.files.base.FileHandle
"""
# Create a new unique identifier for the file.
identifier = util.get_unique_identifier()
file_dir = self.get_file_dir(identifier, create=True)
output_file = os.path.join(file_dir, file_name)
# Save the file object to the new file path
file.save(output_file)
f_handle = FileHandle(identifier=identifier,
filepath=output_file,
file_name=file_name)
return f_handle
def upload_file(self, filename):
"""Create a new entry from a given local file. Will make a copy of the
given file.
Raises ValueError if the given file does not exist.
Parameters
----------
filename: string
Path to file on disk
Returns
-------
benchtmpl.io.files.base.FileHandle
"""
# Ensure that the given file exists
if not os.path.isfile(filename):
raise ValueError('invalid file path \'' + str(filename) + '\'')
file_name = os.path.basename(filename)
# Create a new unique identifier for the file.
identifier = util.get_unique_identifier()
file_dir = self.get_file_dir(identifier, create=True)
output_file = os.path.join(file_dir, file_name)
# Copy the uploaded file
shutil.copyfile(filename, output_file)
# Add file to file index
f_handle = FileHandle(identifier=identifier,
filepath=output_file,
file_name=file_name)
return f_handle
def test_expand_parameters(self):
"""Test parameter expansion."""
template = DefaultTemplateLoader().load(TEMPLATE_FILE)
arguments = {
'code': TemplateArgument(
parameter=template.get_parameter('code'),
value=FileHandle(filepath='code/runme.py')
),
'names': TemplateArgument(
parameter=template.get_parameter('names'),
value=FileHandle(filepath='data/myfriends.txt')
),
'sleeptime': TemplateArgument(
parameter=template.get_parameter('sleeptime'),
value=11
),
'waittime': TemplateArgument(
parameter=template.get_parameter('waittime'),
value=22
)
}
commands = mp.get_commands(template=template, arguments=arguments)
CMDS = [
'python "runme.py" --inputfile "data/names.txt" --outputfile "results/greetings.txt" --sleeptime 11',
'wait 22',
'python "code/eval.py" --inputfile "results/greetings.txt" --outputfile results.json'
]
assert commands == CMDS
# Default values
arguments = {
'names': TemplateArgument(
parameter=template.get_parameter('names'),
value=FileHandle(filepath='data/myfriends.txt')
)
}
commands = mp.get_commands(template=template, arguments=arguments)
CMDS = [
'python "code/helloworld.py" --inputfile "data/names.txt" --outputfile "results/greetings.txt" --sleeptime 10',
'wait 5',
'python "code/eval.py" --inputfile "results/greetings.txt" --outputfile results.json'
]
assert commands == CMDS
# Error cases
del template.workflow_spec['inputs']['parameters']['inputfile']
with pytest.raises(err.InvalidTemplateError):
mp.get_commands(template=template, arguments=arguments)
def test_file_args(self):
"""Test the replace_args function for file parameters."""
spec = {'input': ['$[[codeFile]]']}
parameters = pd.create_parameter_index([{
'id': 'codeFile',
'datatype': 'file',
'defaultValue': 'src/helloworld.py',
'as': 'code/helloworld.py'
}])
# Test default values (no arguments)
wf = tmpl.replace_args(spec=spec,
parameters=parameters,
arguments=dict())
assert wf['input'] == ['code/helloworld.py']
# Test default values (with arguments)
wf = tmpl.replace_args(
spec=spec,
parameters=parameters,
arguments=pr.parse_arguments(
arguments={'codeFile': FileHandle(filepath='/dev/null')},
parameters=parameters))
assert wf['input'] == ['code/helloworld.py']
# Test file parameters without constant value
parameters = pd.create_parameter_index([{
'id':
'codeFile',
'datatype':
'file',
'defaultValue':
'src/helloworld.py'
}])
# Test default values (no arguments)
wf = tmpl.replace_args(spec=spec,
parameters=parameters,
arguments=dict())
assert wf['input'] == ['src/helloworld.py']
wf = tmpl.replace_args(
spec=spec,
parameters=parameters,
arguments=pr.parse_arguments(
arguments={'codeFile': FileHandle(filepath='/dev/null')},
parameters=parameters))
assert wf['input'] == ['null']
def test_source_and_target_path(self):
"""Test source and target path methods for input file handle."""
fh = FileHandle(filepath='/home/user/files/myfile.txt')
# Input file handle without target path
f = InputFile(f_handle=fh)
assert f.source() == '/home/user/files/myfile.txt'
assert f.target() == 'myfile.txt'
# Input file handle with target path
f = InputFile(f_handle=fh, target_path='data/names.txt')
assert f.source() == '/home/user/files/myfile.txt'
assert f.target() == 'data/names.txt'
def test_run_with_missing_file(self, tmpdir):
"""Execute the helloworld example with a reference to a missing file."""
# Execution fails if a file that is referenced by the workflow does not
# exist in the created workspace
repo = TemplateRepository(base_dir=os.path.join(str(tmpdir), 'repo'))
template = repo.add_template(
src_dir=WORKFLOW_DIR,
template_spec_file=TEMPLATE_WITH_MISSING_FILE)
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=3)
}
# Run workflow syncronously
engine = MultiProcessWorkflowEngine(base_dir=os.path.join(
str(tmpdir), 'engine'),
run_async=False)
sync_run_id, state = engine.execute(template, arguments)
assert state.is_error()
assert len(state.messages) > 0
state = engine.get_state(sync_run_id)
assert state.is_error()
assert len(state.messages) > 0
# An error is raised if the input file does not exist
with pytest.raises(IOError):
engine.execute(
template=template,
arguments={
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(UNKNOWN_FILE)),
'sleeptime':
TemplateArgument(
parameter=template.get_parameter('sleeptime'), value=3)
})
def list_files(self):
"""Get list of file handles for all uploaded files.
Returns
-------
list(benchtmpl.io.files.base.FileHandle)
"""
result = list()
for f_name in os.listdir(self.directory):
dir_name = os.path.join(self.directory, f_name)
if os.path.isdir(dir_name):
file_name = os.listdir(dir_name)[0]
f_handle = FileHandle(identifier=f_name,
filepath=os.path.join(
dir_name, file_name),
file_name=file_name)
result.append(f_handle)
return result
def test_run_benchmark(self, tmpdir):
"""Test running a benchmarks."""
# Initialize the BASEDIR environment variable
os.environ[config.ENV_BASEDIR] = os.path.abspath(str(tmpdir))
# Create a new database and open a connection
connect_string = 'sqlite:{}/auth.db'.format(str(tmpdir))
DatabaseDriver.init_db(connect_string=connect_string)
con = DatabaseDriver.connect(connect_string=connect_string)
# Create repository and engine instances
repository = BenchmarkRepository(
con=con,
template_store=TemplateRepository(
base_dir=config.get_template_dir(),
loader=BenchmarkTemplateLoader(),
filenames=['benchmark', 'template', 'workflow']))
engine = BenchmarkEngine(con)
# Add with minimal information
benchmark = repository.add_benchmark(name='My benchmark',
src_dir=TEMPLATE_DIR)
template = benchmark.template
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=1),
'greeting':
TemplateArgument(parameter=template.get_parameter('greeting'),
value='Welcome')
}
run_id, state = engine.run(benchmark, arguments, 'USERID')
assert state.is_success()
sql = 'SELECT * FROM benchmark_run WHERE run_id = ?'
rs = con.execute(sql, (run_id, )).fetchone()
assert rs['benchmark_id'] == benchmark.identifier
assert rs['user_id'] == 'USERID'
assert rs['state'] == state.type_id
table_name = bm.PREFIX_RESULT_TABLE + benchmark.identifier
sql = 'SELECT * FROM {} WHERE run_id = ?'.format(table_name)
rs = con.execute(sql, (run_id, )).fetchone()
assert rs['max_line'] == 'Welcome Alice!'
def get_file(self, identifier):
"""Get handle for file with given identifier. Returns None if no file
with given identifier exists.
Parameters
----------
identifier: string
Unique file identifier
Returns
-------
benchtmpl.io.files.base.FileHandle
"""
file_dir = self.get_file_dir(identifier)
if os.path.isdir(file_dir):
# The uploaded file is the only file in the directory
file_name = os.listdir(file_dir)[0]
return FileHandle(identifier=identifier,
filepath=os.path.join(file_dir, file_name),
file_name=file_name)
return None
def read_parameter(para, scanner, prompt_prefix=''):
"""Read value for a given template parameter declaration. Prompts the
user to enter a value for the given parameter and returns the converted
value that was entered by the user.
Parameters
----------
para: benchtmpl.workflow.parameter.TemplateParameter
Workflow template parameter declaration
Returns
-------
bool or float or int or string or benchtmpl.io.files.base.InputFile
"""
done = False
while not done:
done = True
print(prompt_prefix + para.prompt(), end='')
try:
if para.is_bool():
return scanner.next_bool(default_value=para.default_value)
elif para.is_file():
filename = scanner.next_file(default_value=para.default_value)
target_path = None
if para.has_constant() and para.as_input():
print('Target Path:', end='')
target_path = scanner.next_string()
return InputFile(f_handle=FileHandle(filepath=filename),
target_path=target_path)
elif para.is_float():
return scanner.next_float(default_value=para.default_value)
elif para.is_int():
return scanner.next_int(default_value=para.default_value)
else:
return scanner.next_string(default_value=para.default_value)
except ValueError as ex:
print(ex)
done = False
def test_prepare_inputs_for_local_run(self, tmpdir):
"""Test copying input files for a local workflow run."""
# Load template
store = TemplateRepository(base_dir=str(tmpdir))
template = store.add_template(src_dir=WORKFLOW_DIR,
template_spec_file=SPEC_FILE)
# Create run directory
run_dir = os.path.join(str(tmpdir), 'run')
os.makedirs(run_dir)
# Copy input files to run directory
backend.upload_files(
template=template,
files=template.workflow_spec.get('inputs', {}).get('files', []),
arguments={
'names':
TemplateArgument(template.get_parameter('names'),
value=FileHandle(filepath=DATA_FILE))
},
loader=FileCopy(run_dir))
# We should have the following files in the run directory:
# code/helloworld.py
# data/persons.txt
# data/friends.txt
assert os.path.isfile(os.path.join(run_dir, 'code', 'helloworld.py'))
assert os.path.isfile(os.path.join(run_dir, 'code', 'script.sh'))
assert os.path.isfile(os.path.join(run_dir, 'data', 'persons.txt'))
assert os.path.isfile(os.path.join(run_dir, 'data', 'friends.txt'))
assert not os.path.isfile(os.path.join(run_dir, 'code', 'dontcopy.me'))
# data/persons.txt should contain Alice and Bob
names = set()
with open(os.path.join(run_dir, 'data', 'persons.txt'), 'r') as f:
for line in f:
names.add(line.strip())
assert len(names) == 2
assert 'Alice' in names
assert 'Bob' in names
# data/friends contains Jane Doe and Joe Bloggs
friends = set()
with open(os.path.join(run_dir, 'data', 'friends.txt'), 'r') as f:
for line in f:
friends.add(line.strip())
assert len(friends) == 2
assert 'Jane Doe' in friends
assert 'Joe Bloggs' in friends
# Error cases
with pytest.raises(err.MissingArgumentError):
backend.upload_files(template=template,
files=template.workflow_spec.get(
'inputs', {}).get('files', []),
arguments={},
loader=FileCopy(run_dir))
# Error when copying non-existing file
template = store.add_template(src_dir=WORKFLOW_DIR,
template_spec_file=SPEC_FILE)
shutil.rmtree(run_dir)
os.makedirs(run_dir)
with pytest.raises(IOError):
backend.upload_files(
template=template,
files=template.workflow_spec.get('inputs',
{}).get('files', []),
arguments={
'names':
TemplateArgument(
template.get_parameter('names'),
value=FileHandle(
filepath=os.path.join(str(tmpdir), 'no.file')))
},
loader=FileCopy(run_dir))
assert not os.path.isdir(os.path.join(run_dir, 'data'))
# If the constant value for the names parameter is removed the names
# file is copied to the run directory and not to the data folder
para = template.get_parameter('names')
para.as_constant = None
shutil.rmtree(run_dir)
os.makedirs(run_dir)
backend.upload_files(
template=template,
files=template.workflow_spec.get('inputs', {}).get('files', []),
arguments={
'names':
TemplateArgument(parameter=para,
value=FileHandle(filepath=DATA_FILE))
},
loader=FileCopy(run_dir))
# We should have the following files in the run directory:
# code/helloworld.py
# names.txt
# data/friends.txt
assert os.path.isfile(os.path.join(run_dir, 'code', 'helloworld.py'))
assert os.path.isfile(os.path.join(run_dir, 'names.txt'))
assert not os.path.isfile(os.path.join(run_dir, 'data', 'persons.txt'))
assert os.path.isfile(os.path.join(run_dir, 'data', 'friends.txt'))
# Template with input file parameter that is not of type file
template = store.add_template(src_dir=WORKFLOW_DIR,
template_spec_file=SPEC_FILE_ERR)
shutil.rmtree(run_dir)
os.makedirs(run_dir)
# Copy input files to run directory
with pytest.raises(err.InvalidTemplateError):
backend.upload_files(
template=template,
files=template.workflow_spec.get('inputs',
{}).get('files', []),
arguments={
'sleeptime':
TemplateArgument(template.get_parameter('names'),
value=FileHandle(filepath=DATA_FILE))
},
loader=FileCopy(run_dir))
def test_flat_parse(self):
"""Test parsing arguments for a flat (un-nested) parameter declaration.
"""
template = TemplateHandle(
workflow_spec=dict(),
parameters=[
TemplateParameter(
pd.parameter_declaration('A', data_type=pd.DT_INTEGER)),
TemplateParameter(
pd.parameter_declaration('B', data_type=pd.DT_BOOL)),
TemplateParameter(
pd.parameter_declaration('C', data_type=pd.DT_DECIMAL)),
TemplateParameter(
pd.parameter_declaration('D',
data_type=pd.DT_FILE,
required=False)),
TemplateParameter(
pd.parameter_declaration('E',
data_type=pd.DT_STRING,
required=False))
])
params = template.parameters
fh = InputFile(f_handle=FileHandle(filepath=LOCAL_FILE))
# Valid argument set
args = values.parse_arguments(arguments={
'A': 10,
'B': True,
'C': 12.5,
'D': fh,
'E': 'ABC'
},
parameters=params,
validate=True)
assert len(args) == 5
for key in params.keys():
assert key in args
values.parse_arguments(arguments=args,
parameters=params,
validate=False)
# Error cases
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'Z': 0
},
parameters=params)
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'B': True
},
parameters=params)
# Validate data type
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': '10',
'B': True,
'C': 12.3,
'D': fh,
'E': 'ABC'
},
parameters=params,
validate=True)
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'B': 23,
'C': 12.3,
'D': fh,
'E': 'ABC'
},
parameters=params,
validate=True)
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'B': True,
'C': '12.3',
'D': fh,
'E': 'ABC'
},
parameters=params,
validate=True)
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'B': True,
'C': 12.3,
'D': 'fh',
'E': 'ABC'
},
parameters=params,
validate=True)
with pytest.raises(ValueError):
values.parse_arguments(arguments={
'A': 10,
'B': True,
'C': 12.3,
'D': fh,
'E': 12
},
parameters=params,
validate=True)
def test_run_helloworld(self, tmpdir):
"""Execute the helloworld example."""
repo = TemplateRepository(base_dir=os.path.join(str(tmpdir), 'repo'))
template = repo.add_template(src_dir=WORKFLOW_DIR,
template_spec_file=TEMPLATE_FILE)
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=3)
}
engine_dir = os.path.join(str(tmpdir), 'engine')
engine = MultiProcessWorkflowEngine(base_dir=engine_dir,
run_async=True)
# Run workflow asyncronously
run_id, _ = engine.execute(template, arguments)
while engine.get_state(run_id).is_active():
time.sleep(1)
state = engine.get_state(run_id)
self.validate_run_result(state)
# Remove run resources (can be called multiple times as long as the
# task is not active)
engine.remove_run(run_id)
engine.remove_run(run_id)
# Getting the state of a removed run raises an error
with pytest.raises(err.UnknownRunError):
engine.get_state(run_id)
# Cancel run
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=30)
}
run_id, _ = engine.execute(template, arguments)
while engine.get_state(run_id).is_active():
# Removing resources will raise an error
with pytest.raises(RuntimeError):
engine.remove_run(run_id)
# Cancel the run
engine.cancel_run(run_id)
break
with pytest.raises(err.UnknownRunError):
engine.get_state(run_id)
# Remove run resources does not raise an error now
engine.remove_run(run_id)
# Run workflow syncronously
arguments = {
'names':
TemplateArgument(parameter=template.get_parameter('names'),
value=FileHandle(DATA_FILE)),
'sleeptime':
TemplateArgument(parameter=template.get_parameter('sleeptime'),
value=1)
}
engine = MultiProcessWorkflowEngine(base_dir=engine_dir,
run_async=False)
sync_run_id, state = engine.execute(template, arguments)
assert run_id != sync_run_id
self.validate_run_result(state)
self.validate_run_result(state)
state = engine.get_state(sync_run_id)
# Remove run resources (can be called multiple times as long as the
# task is not active)
engine.remove_run(run_id)
def upload_files(template, files, arguments, loader):
"""Upload all references to local files in a given list of file names of
parameter references. THe list of files, for example corresponds to the
entries in the 'inputs.files' section of a REANA workflow specification.
Uses a loader function to allow use of this method in cases where the
workflow is executed locally or remote using a REANA cluster instance.
Raises errors if (i) an unknown parameter is referenced or (ii) if the type
of a referenced parameter in the input files section is not of type file.
Parameters
----------
template: benchtmpl.workflow.template.base.TemplateHandle
Workflow template containing the parameterized specification and the
parameter declarations
files: list(string)
List of file references
arguments: dict(benchtmpl.workflow.parameter.value.TemplateArgument)
Dictionary of argument values for parameters in the template
loader: func
File (up)load function that takes a filepath as the first argument and
a (remote) target path as the second argument
Raises
------
benchtmpl.error.InvalidTemplateError
benchtmpl.error.MissingArgumentError
benchtmpl.error.UnknownParameterError
"""
for val in files:
# Set source and target values depending on whether the list
# entry references a template parameter or not
if tmpl.is_parameter(val):
var = tmpl.get_parameter_name(val)
# Raise error if the type of the referenced parameter is
# not file
para = template.get_parameter(var)
if not para.is_file():
raise err.InvalidTemplateError('expected file parameter for \'{}\''.format(var))
arg = arguments.get(var)
if arg is None:
if para.default_value is None:
raise err.MissingArgumentError(var)
else:
# Set argument to file handle using the default value
# (assuming that the default points to a file in the
# template base directory).
if para.has_constant() and not para.as_input():
target_path = para.get_constant()
else:
target_path = para.default_value
arg = TemplateArgument(
parameter=para,
value=InputFile(
f_handle=FileHandle(
filepath=os.path.join(
template.base_dir,
para.default_value
)
),
target_path=target_path
)
)
# Get path to source file and the target path from the input
# file handle
source = arg.value.source()
target = arg.value.target()
else:
source = os.path.join(template.base_dir, val)
target = val
# Upload source file
loader(source, target)
