def parse_filename(project, filename):
"""Parses a filename.
:param str project: Project code.
:param str filename: Filename.
:returns: Set of terms extracted from the filename.
:rtype: set
"""
assert isinstance(project, basestring), 'Invalid project'
assert isinstance(filename, basestring), 'Invalid filename'
global _PROJECT, _PARSER, _TEMPLATE, _COLLECTIONS
if _PROJECT != project:
# Get scope corresponding to the project code.
scopes = all_scopes()
assert project in [scope.name
for scope in scopes], 'Unsupported project'
scope = [scope for scope in scopes if scope.name == project][0]
assert 'filename' in scope.data.keys(), 'Filename parser not found'
assert 'template' in scope.data['filename'].keys(
), 'Filename parser template not found'
assert 'collections' in scope.data['filename'].keys(
), 'Filename parser template collections not found'
# Get template from data scope.
_TEMPLATE = scope.data['filename']['template']
assert isinstance(_TEMPLATE, basestring), 'Invalid template'
# Get template collections from data scope.
_COLLECTIONS = list()
for name in scope.data['filename']['collections']:
_COLLECTIONS.append([
collection.namespace for collection in scope.collections
if collection.name == name.replace('_', '-')
][0])
assert _COLLECTIONS, 'Invalid collections'
# Instantiate parser JIT.
_PARSER = create_template_parser(_TEMPLATE,
tuple(_COLLECTIONS),
PARSING_STRICTNESS_1,
separator='_')
# Cached project.
_PROJECT = project
# Strip file extension.
filename = splitext(filename)[0]
try:
return _PARSER.parse(filename)
except TemplateParsingError:
# Add suffix to filename without file period.
return _PARSER.parse(filename + '_fixed')
def parse_directory(project, directory):
"""Parses a directory.
:param str project: Project code.
:param str directory: Data directory.
:returns: Set of terms extracted from the directory.
:rtype: set
"""
assert isinstance(project, basestring), 'Invalid project'
assert isinstance(directory, basestring), 'Invalid directory'
global _PROJECT, _PARSER, _TEMPLATE, _COLLECTIONS
if _PROJECT != project:
# Instantiated template
_TEMPLATE = None
# Instantiated template collections
_COLLECTIONS = None
# Get scope corresponding to the project code.
scopes = all_scopes()
assert project in [scope.name
for scope in scopes], 'Unsupported project'
scope = [scope for scope in scopes if scope.name == project][0]
assert 'directory_structure' in scope.data.keys(
), 'Directory parser not found'
assert 'template' in scope.data['directory_structure'].keys(
), 'Directory parser template not found'
assert 'collections' in scope.data['directory_structure'].keys(
), 'Directory parser template collections not found'
# Get template from data scope.
_TEMPLATE = scope.data['directory_structure']['template']
assert isinstance(_TEMPLATE, basestring), 'Invalid template'
# Get template collections from data scope.
_COLLECTIONS = list()
for name in scope.data['directory_structure']['collections']:
_COLLECTIONS.append([
collection.namespace for collection in scope.collections
if collection.name == name.replace('_', '-')
][0])
assert _COLLECTIONS, 'Invalid collections'
# Instantiate parser JIT.
_PARSER = create_template_parser(_TEMPLATE,
tuple(_COLLECTIONS),
PARSING_STRICTNESS_1,
separator='/')
# Cached project.
_PROJECT = project
return _PARSER.parse(directory)
def parse(identifier):
"""Parses a CMIP6 dataset identifier.
"""
parser = create_template_parser(_TEMPLATE, _COLLECTIONS, PARSING_STRICTNESS_1)
# Strip version suffix.
if '#' in identifier:
identifier = identifier.split('#')[0]
return parser.parse(identifier.split('#')[0])
def parse_dataset_identifier(project, identifier):
"""Parses a dataset identifier.
:param str project: Project code.
:param str identifier: Dataset identifier.
:returns: Set of terms extracted from the identifier.
:rtype: set
"""
assert isinstance(project, basestring), 'Invalid project'
assert isinstance(identifier, basestring), 'Invalid identifier'
global _PROJECT, _PARSER, _TEMPLATE, _COLLECTIONS
if _PROJECT != project:
# Get scope corresponding to the project code.
scopes = all_scopes()
assert project in [scope.name
for scope in scopes], 'Unsupported project'
scope = [scope for scope in scopes if scope.name == project][0]
assert 'dataset_id' in scope.data.keys(), 'Dataset ID parser not found'
assert 'template' in scope.data['dataset_id'].keys(
), 'Dataset ID parser template not found'
assert 'collections' in scope.data['dataset_id'].keys(
), 'Dataset ID parser template collections not found'
# Get template from data scope.
_TEMPLATE = scope.data['dataset_id']['template']
assert isinstance(_TEMPLATE, basestring), 'Invalid template'
# Get template collections from data scope.
_COLLECTIONS = list()
for name in scope.data['dataset_id']['collections']:
_COLLECTIONS.append([
collection.namespace for collection in scope.collections
if collection.name == name.replace('_', '-')
][0])
assert _COLLECTIONS, 'Invalid collections'
# Instantiate parser JIT.
_PARSER = create_template_parser(_TEMPLATE, tuple(_COLLECTIONS),
PARSING_STRICTNESS_1)
# Cached project.
_PROJECT = project
# Convert version suffix to an identifier element.
identifier = identifier.replace('#', '.v')
return _PARSER.parse(identifier)