# Licensed under a 3-clause BSD style license - see LICENSE.rst
import json
import logging
import os
import re
from functools import partial
from pathlib import Path
import yaml
from astropy.table import Table
from requests.exceptions import HTTPError
from .download import download_file_cached, get_cache_path
try:
import ctapipe_resources
has_resources = True
except ImportError:
has_resources = False
from importlib.resources import files
from ..core import Provenance
logger = logging.getLogger(__name__)
__all__ = [
"get_dataset_path",
"find_in_path",
"find_all_matching_datasets",
"get_default_url",
"DEFAULT_URL",
]
#: default base URL for downloading datasets
DEFAULT_URL = "http://cccta-dataserver.in2p3.fr/data/ctapipe-test-data/v1.1.0/"
[docs]
def get_default_url():
"""Get the default download url for datasets
First tries to look-up CTAPIPE_DATASET_URL and then falls
back to ``DEFAULT_URL``
"""
return os.getenv("CTAPIPE_DATASET_URL", DEFAULT_URL)
def get_searchpath_dirs(searchpath=None, url=None):
"""returns a list of dirs in specified searchpath"""
if url is None:
url = get_default_url()
if searchpath is None:
searchpath = os.getenv("CTAPIPE_SVC_PATH")
if searchpath == "" or searchpath is None:
searchpaths = []
else:
searchpaths = [Path(p) for p in os.path.expandvars(searchpath).split(":")]
searchpaths.append(get_cache_path(url))
return searchpaths
[docs]
def find_all_matching_datasets(
pattern,
searchpath=None,
regexp_group=None,
url=None,
):
"""
Returns a list of resource names (or substrings) matching the given
pattern, searching first in searchpath (a colon-separated list of
directories) and then in the ctapipe_resources module)
Parameters
----------
pattern: str
regular expression to use for matching
searchpath: str
colon-seprated list of directories in which to search, defaulting to
CTAPIPE_SVC_PATH environment variable
regexp_group: int
if not None, return the regular expression group indicated (assuming
pattern has a group specifier in it)
Returns
-------
list(str):
resources names, use get_dataset_path() to retrieve the full filename
"""
if url is None:
url = get_default_url()
results = set()
if searchpath is None:
searchpath = os.getenv("CTAPIPE_SVC_PATH")
search_path_dirs = get_searchpath_dirs(searchpath, url=url)
# first check search path
for path in search_path_dirs:
if path.is_dir():
for entry in path.iterdir():
match = re.match(pattern, entry.name)
if match:
if regexp_group is not None:
results.add(match.group(regexp_group))
else:
results.add(entry)
# then check resources module
if has_resources:
for resource in files("ctapipe_resources").iterdir():
match = re.match(pattern, resource.name)
if match:
if regexp_group is not None:
results.add(match.group(regexp_group))
else:
results.add(resource)
return list(results)
[docs]
def find_in_path(filename, searchpath, url=None):
"""
Search in searchpath for filename, returning full path.
Parameters
----------
searchpath: str
colon-separated list of directories (like PATH)
filename:
filename to find
Returns
-------
full path to file if found, None otherwise
"""
if url is None:
url = get_default_url()
for directory in get_searchpath_dirs(searchpath, url=url):
path = directory / filename
if path.exists():
return path
return None
[docs]
def get_dataset_path(filename, url=None):
"""
Returns the full file path to an auxiliary dataset needed by
ctapipe, given the dataset's full name (filename with no directory).
This will first search for the file in directories listed in
the environment variable CTAPIPE_SVC_PATH (if set), and if not found,
will look in the ctapipe_resources module
(if installed with the ctapipe-extra package), which contains the defaults.
Parameters
----------
filename: str
name of dataset to fetch
Returns
-------
string with full path to the given dataset
"""
searchpath = os.getenv("CTAPIPE_SVC_PATH")
if url is None:
url = get_default_url()
if searchpath:
filepath = find_in_path(filename=filename, searchpath=searchpath, url=url)
if filepath:
return filepath
if has_resources:
logger.debug(
"Resource '{}' not found in CTAPIPE_SVC_PATH, looking in "
"ctapipe_resources...".format(filename)
)
return Path(ctapipe_resources.get(filename))
# last, try downloading the data
try:
return download_file_cached(filename, default_url=url, progress=True)
except HTTPError as e:
# let 404 raise the FileNotFoundError instead of HTTPError
if e.response.status_code != 404:
raise
raise FileNotFoundError(
f"Couldn't find resource: '{filename}',"
" You might want to install ctapipe_resources"
)
def try_filetypes(basename, role, file_types, url=None, **kwargs):
"""
Get the contents of dataset as an `astropy.table.Table` object from
different file types if available.
Parameters
----------
basename: str
base-name (without extension) of the dataset
role: str
Provenance role. It should be set to the CTA data hierarchy name when
possible (e.g. dl1.sub.svc.arraylayout). This will be recorded in the
provenance system.
file_types: dict
Mapping of file extensions to readers.
url : str
URL where the dataset is stored.
kwargs:
extra arguments to pass to Table.read()
Returns
-------
table : astropy.table.Table
Table containing the data in a dataser from an available file
type entry.
"""
if url is None:
url = get_default_url()
path = None
# look first in search paths (includes cache)
# so we respect user provided paths and don't have to try non-existing downloads
search_paths = get_searchpath_dirs(url=url)
for search_path in search_paths:
for ext, reader in file_types.items():
filename = basename + ext
if (search_path / filename).exists():
path = search_path / filename
break
if path is not None:
break
# no cache hit
if path is None:
for ext, reader in file_types.items():
filename = basename + ext
try:
path = get_dataset_path(filename, url=url)
break
except (FileNotFoundError, HTTPError):
pass
if path is not None:
table = reader(path, **kwargs)
Provenance().add_input_file(path, role)
return table
raise FileNotFoundError(
"Couldn't find any file: {}[{}]".format(basename, ", ".join(file_types))
)
[docs]
def get_table_dataset(table_name, role="resource", **kwargs):
"""
get a tabular dataset as an `astropy.table.Table` object
Parameters
----------
table_name: str
base name of table, without file extension
role: str
should be set to the CTA data hierarchy name when possible (e.g.
dl1.sub.svc.arraylayout). This will be recorded in the provenance
system.
kwargs:
extra arguments to pass to Table.read()
Returns
-------
Table
"""
# a mapping of types (keys) to any extra keyword args needed for
# table.read()
table_types = {
".fits.gz": Table.read,
".fits": Table.read,
".ecsv": partial(Table.read, format="ascii.ecsv"),
".ecsv.txt": partial(Table.read, format="ascii.ecsv"),
}
return try_filetypes(table_name, role, table_types, **kwargs)
def get_structured_dataset(basename, role="resource", **kwargs):
"""
find and return a YAML or JSON dataset as a dictionary
Parameters
----------
basename: str
base-name (without extension) of the dataset
role: str
provenence role
kwargs: dict
any other arguments to pass on to yaml or json loader.
Returns
-------
dict:
dictionary of data in the file
"""
def load_yaml(path, **kwargs):
with open(path, "rb") as f:
return yaml.safe_load(f, **kwargs)
def load_json(path, **kwargs):
with open(path, "rb") as f:
return json.load(f, **kwargs)
# a mapping of types (keys) to any extra keyword args needed for
# table.read()
structured_types = {".yaml": load_yaml, ".yml": load_yaml, ".json": load_json}
return try_filetypes(basename, role, structured_types, **kwargs)
[docs]
def resource_file(filename):
"""Get the absolute path of ctapipe resource files."""
return files("ctapipe").joinpath("resources", filename)
