def get_root_data_dir() -> str:
"""
Return the directory where the mephisto data is expected to go
"""
global loaded_data_dir
if loaded_data_dir is None:
default_data_dir = os.path.join(get_root_dir(), "data")
actual_data_dir = get_config_arg(CORE_SECTION, DATA_STORAGE_KEY)
if actual_data_dir is None:
data_dir_location = input(
"Please enter the full path to a location to store Mephisto run data. By default this "
f"would be at '{default_data_dir}'. This dir should NOT be on a distributed file "
"store. Press enter to use the default: "
).strip()
if len(data_dir_location) == 0:
data_dir_location = default_data_dir
data_dir_location = os.path.expanduser(data_dir_location)
os.makedirs(data_dir_location, exist_ok=True)
# Check to see if there is existing data to possibly move to the data dir:
database_loc = os.path.join(default_data_dir, "database.db")
if os.path.exists(database_loc) and data_dir_location != default_data_dir:
should_migrate = (
input(
"We have found an existing database in the default data directory, do you want to "
f"copy any existing data from the default location to {data_dir_location}? (y)es/no: "
)
.lower()
.strip()
)
if len(should_migrate) == 0 or should_migrate[0] == "y":
copy_tree(default_data_dir, data_dir_location)
print(
"Mephisto data successfully copied, once you've confirmed the migration worked, "
"feel free to remove all of the contents in "
f"{default_data_dir} EXCEPT for `README.md`."
)
add_config_arg(CORE_SECTION, DATA_STORAGE_KEY, data_dir_location)
loaded_data_dir = get_config_arg(CORE_SECTION, DATA_STORAGE_KEY)
if not os.path.isdir(loaded_data_dir):
raise NotADirectoryError(
f"The provided Mephisto data directory {loaded_data_dir} as set in "
f"{DEFAULT_CONFIG_FILE} is not a directory! Please locate your Mephisto "
f"data directory and update {DEFAULT_CONFIG_FILE} to point to it."
)
return loaded_data_dir
def create_hit_type(
client: MTurkClient,
task_config: "TaskConfig",
qualifications: List[Dict[str, Any]],
auto_approve_delay: Optional[int] = 7 * 24 * 3600, # default 1 week
) -> str:
"""Create a HIT type to be used to generate HITs of the requested params"""
hit_title = task_config.task_title
hit_description = task_config.task_description
hit_keywords = ",".join(task_config.task_tags)
hit_reward = task_config.task_reward
assignment_duration_in_seconds = task_config.assignment_duration_in_seconds
existing_qualifications = convert_mephisto_qualifications(client, qualifications)
# If the user hasn't specified a location qualification, we assume to
# restrict the HIT to some english-speaking countries.
locale_requirements: List[Any] = []
has_locale_qual = False
if existing_qualifications is not None:
for q in existing_qualifications:
if q["QualificationTypeId"] == MTURK_LOCALE_REQUIREMENT:
has_locale_qual = True
locale_requirements += existing_qualifications
if not has_locale_qual and not client_is_sandbox(client):
allowed_locales = get_config_arg("mturk", "allowed_locales")
if allowed_locales is None:
allowed_locales = [
{"Country": "US"},
{"Country": "CA"},
{"Country": "GB"},
{"Country": "AU"},
{"Country": "NZ"},
]
locale_requirements.append(
{
"QualificationTypeId": MTURK_LOCALE_REQUIREMENT,
"Comparator": "In",
"LocaleValues": allowed_locales,
"ActionsGuarded": "DiscoverPreviewAndAccept",
}
)
# Create the HIT type
response = client.create_hit_type(
AutoApprovalDelayInSeconds=auto_approve_delay,
AssignmentDurationInSeconds=assignment_duration_in_seconds,
Reward=str(hit_reward),
Title=hit_title,
Keywords=hit_keywords,
Description=hit_description,
QualificationRequirements=locale_requirements,
)
hit_type_id = response["HITTypeId"]
return hit_type_id