def do_require_dataset(self, arg):
"""Require dataset
The argument is the name of the dataset to require.
"""
if self.get_datasets().count(arg):
with Timer(f"Downloaded dataset {arg}"):
require(self.env, arg)
self.init_benchmarks()
# FIXME CHRIS, why can't I get it to update the list of benchmarks?
# I have to restart
print("Application must be restarted to make changes visible.")
else:
print("Unknown dataset, '" + arg + "'")
print("Available datasets are listed with command, list_available_datasets")
def require_datasets(self, datasets: List[Union[str, Dataset]]) -> None:
"""Require that the given datasets are available to the environment.
Example usage:
>>> env = gym.make("llvm-v0")
>>> env.require_dataset(["npb-v0"])
>>> env.benchmarks
["npb-v0/1", "npb-v0/2", ...]
This is equivalent to calling
:meth:`require(self, dataset) <compiler_gym.datasets.require>` on
the list of datasets.
:param datasets: A list of datasets to require. Each dataset is the name
of an available dataset, the URL of a dataset to download, or a
:class:`Dataset` instance.
"""
dataset_installed = False
for dataset in datasets:
dataset_installed |= require(self, dataset)
if dataset_installed:
# Signal to the compiler service that the contents of the site data
# directory has changed.
self.service(
self.service.stub.AddBenchmark,
AddBenchmarkRequest(
benchmark=[Benchmark(uri="service://scan-site-data")]
),
)
self.make_manifest_file()
def main(argv):
"""Main entry point."""
if len(argv) != 1:
raise app.UsageError(f"Unknown command line arguments: {argv[1:]}")
env = env_from_flags()
try:
if not env.datasets_site_path:
raise app.UsageError("Environment has no benchmarks site path")
env.datasets_site_path.mkdir(parents=True, exist_ok=True)
env.inactive_datasets_site_path.mkdir(parents=True, exist_ok=True)
invalidated_manifest = False
for name_or_url in FLAGS.download:
require(env, name_or_url)
if FLAGS.download_all:
for dataset in env.available_datasets:
require(env, dataset)
for name in FLAGS.activate:
activate(env, name)
invalidated_manifest = True
if FLAGS.activate_all:
for path in env.inactive_datasets_site_path.iterdir():
activate(env, path.name)
invalidated_manifest = True
for name in FLAGS.deactivate:
deactivate(env, name)
invalidated_manifest = True
if FLAGS.deactivate_all:
for path in env.datasets_site_path.iterdir():
deactivate(env, path.name)
invalidated_manifest = True
for name in FLAGS.delete:
delete(env, name)
if invalidated_manifest:
env.make_manifest_file()
print(f"{env.spec.id} benchmarks site dir: {env.datasets_site_path}")
print()
print(enumerate_directory("Active Datasets", env.datasets_site_path), )
print(
"These benchmarks are ready for use. Deactivate them using `--deactivate=<name>`."
)
print()
print(
enumerate_directory("Inactive Datasets",
env.inactive_datasets_site_path))
print("These benchmarks may be activated using `--activate=<name>`.")
print()
print(
tabulate(
sorted([(
d.name,
d.license,
humanize.intcomma(d.file_count),
humanize.naturalsize(d.size_bytes),
) for d in env.available_datasets.values()]),
headers=(
"Downloadable Dataset",
"License",
"#. Benchmarks",
"Size on disk",
),
))
print(
"These benchmarks may be installed using `--download=<name> --activate=<name>`."
)
finally:
env.close()