def file_meta_from_file(file_):
"""Computes the file metadata for a file."""
# TODO: decoding to prevent invalid chars within archive.
file_ = file_.decode('utf8')
try:
ext = ext_to_mime(get_extension(file_))
except: # pylint: disable=W0702
ext = None
return {
'file_name': os.path.basename(file_),
'content_type': ext,
'file_size': os.path.getsize(file_),
'out_file': file_,
'md5sum': md5_for_file(file_)
}
def file_meta_from_file(file_):
"""Computes the file metadata for a file."""
# TODO: decoding to prevent invalid chars within archive.
file_ = file_.decode('utf8')
try:
ext = ext_to_mime(get_extension(file_))
except: # pylint: disable=W0702
ext = None
return {
'file_name': os.path.basename(file_),
'content_type': ext,
'file_size': os.path.getsize(file_),
'out_file': file_,
'md5sum': md5_for_file(file_)
}
def _evaluate_dispatcher(archive_item, file_meta):
"""Evaluate the dispatcher for a source and dataset couple."""
loggy = local.logger
try:
ext = ext_to_mime(get_extension(file_meta['out_file']))
except UnknownMIMETypeException:
loggy.warning("Unknown file type for %s", file_meta['out_file'])
archive_item.delete() # this is to not have useless archiveitems
return []
_file = FileMeta(
path=file_meta['out_file'],
mime_by_ext=ext,
mime_by_origin=file_meta['content_type'],
)
dispatcher_code = archive_item.dataset.source.dispatcher \
or cnsettings.CONTROLLER_DISPATCHER_DEFAULT
try:
_locals = {
'UnsupportedDatasetException': UnsupportedDatasetException,
'source': archive_item.dataset.source,
'dataset': archive_item.dataset,
'archive_item': archive_item,
'file': _file,
'recurse': __recurse
}
_globals = {}
function_name = 'dispatch'
exec to_handler_function(dispatcher_code,
function_name) in _locals, _globals
try:
result = _globals['__' + function_name]()
except UnsupportedDatasetException:
loggy.warning("Can't process file %s", file_meta['out_file'])
archive_item.delete() # this is to not have useless archiveitems
return []
else:
return result
except Exception:
loggy.exception('Error while configuring workflow')
raise
def _evaluate_dispatcher(archive_item, file_meta):
"""Evaluate the dispatcher for a source and dataset couple."""
loggy = local.logger
try:
ext = ext_to_mime(get_extension(file_meta['out_file']))
except UnknownMIMETypeException:
loggy.warning("Unknown file type for %s", file_meta['out_file'])
archive_item.delete() # this is to not have useless archiveitems
return []
_file = FileMeta(
path=file_meta['out_file'],
mime_by_ext=ext,
mime_by_origin=file_meta['content_type'],
)
dispatcher_code = archive_item.dataset.source.dispatcher \
or cnsettings.CONTROLLER_DISPATCHER_DEFAULT
try:
_locals = {
'UnsupportedDatasetException': UnsupportedDatasetException,
'source': archive_item.dataset.source,
'dataset': archive_item.dataset,
'archive_item': archive_item,
'file': _file,
'recurse': __recurse
}
_globals = {}
function_name = 'dispatch'
exec to_handler_function(
dispatcher_code, function_name) in _locals, _globals
try:
result = _globals['__' + function_name]()
except UnsupportedDatasetException:
loggy.warning("Can't process file %s", file_meta['out_file'])
archive_item.delete() # this is to not have useless archiveitems
return []
else:
return result
except Exception:
loggy.exception('Error while configuring workflow')
raise
def ext(self):
""" the extension of the file associated to this dataset (which could
be an archive file)
"""
return get_extension(self.download)
def ext(self):
""" the extension of the file associated to this dataset (which could
be an archive file)
"""
return get_extension(self.download)