def cli_get_example(command, raw_args):
"""Downloads the example files to the desired directory
"""
assert command == "get-example"
# setup the arg-parsing
parser = argparse.ArgumentParser('kipoi {}'.format(command),
description='Get example files')
add_model(parser, source="kipoi")
parser.add_argument("-o", "--output", default="example", required=False,
help="Output directory where to store the examples. Default: 'example'")
args = parser.parse_args(raw_args)
# --------------------------------------------
md = kipoi.get_model_descr(args.model, args.source)
src = kipoi.get_source(args.source)
# load the default dataloader
if isinstance(md.default_dataloader, kipoi.specs.DataLoaderImport):
with cd(src.get_model_dir(args.model)):
dl_descr = md.default_dataloader.get()
else:
# load from directory
# attach the default dataloader already to the model
dl_descr = kipoi.get_dataloader_descr(os.path.join(args.model, md.default_dataloader),
source=args.source)
kwargs = dl_descr.download_example(output_dir=args.output, dry_run=False)
logger.info("Example files downloaded to: {}".format(args.output))
logger.info("use the following dataloader kwargs:")
print(json.dumps(kwargs))
def init_example(cls):
"""Instantiate the class using example_kwargs
"""
if cls.source_dir is not None:
with cd(cls.source_dir):
# always init the example in the original directory
return cls(**cls.example_kwargs)
else:
return cls(**cls.example_kwargs)
def cli_info(command, raw_args):
"""CLI interface to predict
"""
assert command == "info"
parser = argparse.ArgumentParser('kipoi {}'.format(command),
description="Prints dataloader" +
" keyword arguments.")
add_model(parser)
add_dataloader(parser, with_args=False)
args = parser.parse_args(raw_args)
# --------------------------------------------
# load model & dataloader
md = kipoi.get_model_descr(args.model, args.source)
src = kipoi.get_source(args.source)
# load the default dataloader
try:
if isinstance(md.default_dataloader, kipoi.specs.DataLoaderImport):
with cd(src.get_model_dir(args.model)):
dl_descr = md.default_dataloader.get()
else:
# load from directory
# attach the default dataloader already to the model
dl_descr = kipoi.get_dataloader_descr(os.path.join(
args.model, md.default_dataloader),
source=args.source)
# if kipoiseq is not installed you get an ImportError
except ImportError:
dl_descr = None
print("-" * 80)
print("'{0}' from source '{1}'".format(str(args.model), str(args.source)))
print("")
print("Model information")
print("-----------")
print(md.info.get_config_as_yaml())
if dl_descr:
print("Dataloader arguments")
print("--------------------")
dl_descr.print_args()
print("--------------------\n")
print("Run `kipoi get-example {} -o example` to download example files.\n".
format(args.model))
def get_dataloader(dataloader, source="kipoi"):
"""Loads the dataloader
# Arguments
dataloader (str): dataloader name
source (str): source name
# Returns
- Instance of class inheriting from `kipoi.data.BaseDataLoader` (like `kipoi.data.Dataset`)
decorated with additional attributes.
# Methods
- __batch_iter(batch_size, num_workers, **kwargs)__
- Arguments
- **batch_size**: batch size
- **num_workers**: Number of workers to use in parallel.
- ****kwargs**: Other kwargs specific to each dataloader
- Yields
- `dict` with `"inputs"`, `"targets"` and `"metadata"`
- __batch_train_iter(cycle=True, **kwargs)__
- Arguments
- **cycle**: if True, cycle indefinitely
- ****kwargs**: Kwargs passed to `batch_iter()` like `batch_size`
- Yields
- tuple of ("inputs", "targets") from the usual dict returned by `batch_iter()`
- __batch_predict_iter(**kwargs)__
- Arguments
- ****kwargs**: Kwargs passed to `batch_iter()` like `batch_size`
- Yields
- "inputs" field from the usual dict returned by `batch_iter()`
- __load_all(**kwargs)__ - load the whole dataset into memory
- Arguments
- ****kwargs**: Kwargs passed to `batch_iter()` like `batch_size`
- Returns
- `dict` with `"inputs"`, `"targets"` and `"metadata"`
- **init_example()** - instantiate the dataloader with example kwargs
- **print_args()** - print information about the required arguments
# Appended attributes
- **type** (str): dataloader type (class name)
- **defined_as** (str): path and dataloader name
- **args** (list of kipoi.specs.DataLoaderArgument): datalaoder argument description
- **info** (kipoi.specs.Info): general information about the dataloader
- **schema** (kipoi.specs.DataloaderSchema): information about the input/output
data modalities
- **dependencies** (kipoi.specs.Dependencies): class specifying the dependencies.
(implements `install` method for running the installation)
- **name** (str): model name
- **source** (str): model source
- **source_dir** (str): local path to model source storage
- **postprocessing** (dict): dictionary of loaded plugin specifications
- **example_kwargs** (dict): kwargs for running the provided example
"""
# if source == 'py':
# # load it from the python object
# sys.path.append(os.path.getcwd())
# return DataLoaderImport(defined_as=dataloader).get()
# TODO - allow source=py
# pull the dataloader & get the dataloader directory
if isinstance(source, str):
source = kipoi.config.get_source(source)
source.pull_dataloader(dataloader)
dataloader_dir = source.get_dataloader_dir(dataloader)
# --------------------------------------------
# Setup dataloader description
descr = source.get_dataloader_descr(dataloader)
with cd(dataloader_dir): # move to the dataloader directory temporarily
if "::" in descr.defined_as:
# old API
file_path, obj_name = tuple(descr.defined_as.split("::"))
CustomDataLoader = getattr(load_module(file_path), obj_name)
else:
# new API - directly specify the object
CustomDataLoader = load_obj(descr.defined_as)
# download util links if specified under default & override the default parameters
override = download_default_args(
descr.args, source.get_dataloader_download_dir(dataloader))
if override:
# override default arguments specified under default
CustomDataLoader = override_default_kwargs(CustomDataLoader, override)
# infer the type
if descr.type is None:
if inspect.isfunction(CustomDataLoader):
raise ValueError(
"Dataloaders implemented as functions/generator need to specify the type flag in dataloader.yaml"
)
else:
# figure out the right dataloader type
descr.type = infer_parent_class(CustomDataLoader,
AVAILABLE_DATALOADERS)
if descr.type is None:
raise ValueError(
"Dataloader needs to inherit from one of the available dataloaders {}"
.format(list(AVAILABLE_DATALOADERS)))
# check that descr.type is correct
if descr.type not in AVAILABLE_DATALOADERS:
raise ValueError(
"dataloader type: {0} is not in supported dataloaders:{1}".format(
descr.type, list(AVAILABLE_DATALOADERS.keys())))
# check that the extractor arguments match yaml arguments
if not getargs(CustomDataLoader) == set(descr.args.keys()):
# if kwargs is defined, we can assume that the dataloader accepts more than its explicitly defined arguments
if "kwargs" in getargs(CustomDataLoader):
missing_required_args = getargs(CustomDataLoader) - set(
descr.args.keys()) - {"kwargs"}
if len(missing_required_args) > 0:
raise ValueError(
"DataLoader arguments:\n\t{args}\n are missing required arguments:\n\t{missing}"
.format(
args=set(descr.args.keys()),
missing=missing_required_args,
))
else:
raise ValueError(
"DataLoader arguments: \n{0}\n don't match ".format(
set(getargs(CustomDataLoader))) +
"the specification in the dataloader.yaml file:\n{0}".format(
set(descr.args.keys())))
# check that CustomDataLoader indeed interits from the right DataLoader
if descr.type in DATALOADERS_AS_FUNCTIONS:
# transform the functions into objects
assert isinstance(CustomDataLoader, types.FunctionType)
CustomDataLoader = AVAILABLE_DATALOADERS[descr.type].from_fn(
CustomDataLoader)
else:
if not issubclass(CustomDataLoader, AVAILABLE_DATALOADERS[descr.type]):
raise ValueError(
"DataLoader does't inherit from the specified dataloader: {0}".
format(AVAILABLE_DATALOADERS[descr.type].__name__))
logger.info('successfully loaded the dataloader {} from {}'.format(
dataloader,
os.path.normpath(os.path.join(dataloader_dir, descr.defined_as))))
# enrich the original dataloader class with description
Dl = CustomDataLoader._add_description_factory(descr)
# add other fields
Dl.source = source
Dl.source_dir = dataloader_dir
return Dl
