def convert_nexson_format(blob,
out_nexson_format,
current_format=None,
remove_old_structs=True,
pristine_if_invalid=False,
sort_arbitrary=False):
'''Take a dict form of NexSON and converts its datastructures to
those needed to serialize as out_nexson_format.
If current_format is not specified, it will be inferred.
If `remove_old_structs` is False and different honeybadgerfish varieties
are selected, the `blob` will be 'fat" containing both types
of lookup structures.
If pristine_if_invalid is False, then the object may be corrupted if it
is an invalid nexson struct. Setting this to False can result in
faster translation, but if an exception is raised the object may
be polluted with partially constructed fields for the out_nexson_format.
'''
if not current_format:
current_format = detect_nexson_version(blob)
out_nexson_format = resolve_nexson_format(out_nexson_format)
if current_format == out_nexson_format:
if sort_arbitrary:
sort_arbitrarily_ordered_nexson(blob)
return blob
two2zero = _is_by_id_hbf(out_nexson_format) and _is_badgerfish_version(current_format)
zero2two = _is_by_id_hbf(current_format) and _is_badgerfish_version(out_nexson_format)
if two2zero or zero2two:
# go from 0.0 -> 1.0 then the 1.0->1.2 should succeed without nexml...
blob = convert_nexson_format(blob,
DIRECT_HONEY_BADGERFISH,
current_format=current_format,
remove_old_structs=remove_old_structs,
pristine_if_invalid=pristine_if_invalid)
current_format = DIRECT_HONEY_BADGERFISH
ccdict = {'output_format':out_nexson_format,
'input_format':current_format,
'remove_old_structs': remove_old_structs,
'pristine_if_invalid': pristine_if_invalid}
ccfg = ConversionConfig(ccdict)
if _is_badgerfish_version(current_format):
converter = Badgerfish2DirectNexson(ccfg)
elif _is_badgerfish_version(out_nexson_format):
assert _is_direct_hbf(current_format)
converter = Direct2BadgerfishNexson(ccfg)
elif _is_direct_hbf(current_format) and (out_nexson_format == BY_ID_HONEY_BADGERFISH):
converter = Direct2OptimalNexson(ccfg)
elif _is_direct_hbf(out_nexson_format) and (current_format == BY_ID_HONEY_BADGERFISH):
converter = Optimal2DirectNexson(ccfg)
else:
raise NotImplementedError('Conversion from {i} to {o}'.format(i=current_format, o=out_nexson_format))
blob = converter.convert(blob)
if sort_arbitrary:
sort_arbitrarily_ordered_nexson(blob)
return blob
def convert_nexson_format(blob,
out_nexson_format,
current_format=None,
remove_old_structs=True,
pristine_if_invalid=False,
sort_arbitrary=False):
'''Take a dict form of NexSON and converts its datastructures to
those needed to serialize as out_nexson_format.
If current_format is not specified, it will be inferred.
If `remove_old_structs` is False and different honeybadgerfish varieties
are selected, the `blob` will be 'fat" containing both types
of lookup structures.
If pristine_if_invalid is False, then the object may be corrupted if it
is an invalid nexson struct. Setting this to False can result in
faster translation, but if an exception is raised the object may
be polluted with partially constructed fields for the out_nexson_format.
'''
if not current_format:
current_format = detect_nexson_version(blob)
out_nexson_format = resolve_nexson_format(out_nexson_format)
if current_format == out_nexson_format:
if sort_arbitrary:
sort_arbitrarily_ordered_nexson(blob)
return blob
two2zero = _is_by_id_hbf(out_nexson_format) and _is_badgerfish_version(current_format)
zero2two = _is_by_id_hbf(current_format) and _is_badgerfish_version(out_nexson_format)
if two2zero or zero2two:
# go from 0.0 -> 1.0 then the 1.0->1.2 should succeed without nexml...
blob = convert_nexson_format(blob,
DIRECT_HONEY_BADGERFISH,
current_format=current_format,
remove_old_structs=remove_old_structs,
pristine_if_invalid=pristine_if_invalid)
current_format = DIRECT_HONEY_BADGERFISH
ccdict = {'output_format':out_nexson_format,
'input_format':current_format,
'remove_old_structs': remove_old_structs,
'pristine_if_invalid': pristine_if_invalid}
ccfg = ConversionConfig(ccdict)
if _is_badgerfish_version(current_format):
converter = Badgerfish2DirectNexson(ccfg)
elif _is_badgerfish_version(out_nexson_format):
assert _is_direct_hbf(current_format)
converter = Direct2BadgerfishNexson(ccfg)
elif _is_direct_hbf(current_format) and (out_nexson_format == BY_ID_HONEY_BADGERFISH):
converter = Direct2OptimalNexson(ccfg)
elif _is_direct_hbf(out_nexson_format) and (current_format == BY_ID_HONEY_BADGERFISH):
converter = Optimal2DirectNexson(ccfg)
else:
raise NotImplementedError('Conversion from {i} to {o}'.format(i=current_format, o=out_nexson_format))
blob = converter.convert(blob)
if sort_arbitrary:
sort_arbitrarily_ordered_nexson(blob)
return blob
def sort_meta_elements(blob):
'''For v0.0 (which has meta values in a list), this
function recursively walks through the object
and sorts each meta by @property or @rel values.
'''
v = detect_nexson_version(blob)
if _is_badgerfish_version(v):
_recursive_sort_meta(blob, '')
return blob
def sort_meta_elements(blob):
'''For v0.0 (which has meta values in a list), this
function recursively walks through the object
and sorts each meta by @property or @rel values.
'''
v = detect_nexson_version(blob)
if _is_badgerfish_version(v):
_recursive_sort_meta(blob, '')
return blob
def add_resource_meta(obj, rel, href, version):
if _is_badgerfish_version(version):
m = obj.setdefault('meta', [])
if not isinstance(m, list):
m = [m]
obj['meta'] = m
m.append({'@href': href, '@rel': rel, '@xsi:type': 'nex:ResourceMeta'})
else:
k = '^' + rel
_add_value_to_dict_bf(obj, k, {'@href': href})
def add_resource_meta(obj, rel, href, version):
if _is_badgerfish_version(version):
m = obj.setdefault('meta', [])
if not isinstance(m, list):
m = [m]
obj['meta'] = m
m.append({'@href':href,
'@rel': rel,
'@xsi:type': 'nex:ResourceMeta'})
else:
k = '^' + rel
_add_value_to_dict_bf(obj, k, {'@href': href})
def __init__(self, conv_cfg):
NexsonConverter.__init__(self, conv_cfg)
self.input_format = conv_cfg.input_format
self.use_default_root_atts = conv_cfg.get('use_default_root_atts', True)
self.otu_label = conv_cfg.get('otu_label', 'ot:originalLabel')
if self.otu_label.startswith('^'):
self.otu_label = self.otu_label[1:]
self._migrating_from_bf = _is_badgerfish_version(self.input_format)
# TreeBase and phylografter trees often lack the tree xsi:type
self._adding_tree_xsi_type = True
# we have started using ot:ottTaxonName, ot:originalLabel or ot:ottId
self._creating_otu_label = True
def __init__(self, conv_cfg):
NexsonConverter.__init__(self, conv_cfg)
self.input_format = conv_cfg.input_format
self.use_default_root_atts = conv_cfg.get('use_default_root_atts',
True)
self.otu_label = conv_cfg.get('otu_label', 'ot:originalLabel')
if self.otu_label.startswith('^'):
self.otu_label = self.otu_label[1:]
self._migrating_from_bf = _is_badgerfish_version(self.input_format)
# TreeBase and phylografter trees often lack the tree xsi:type
self._adding_tree_xsi_type = True
# we have started using ot:ottTaxonName, ot:originalLabel or ot:ottId
self._creating_otu_label = True
def create_validation_adaptor(obj, logger, **kwargs):
try:
nexson_version = detect_nexson_version(obj)
except:
return BadgerFishValidationAdaptor(obj, logger, **kwargs)
if _is_by_id_hbf(nexson_version):
# _LOG.debug('validating as ById...')
return ByIdHBFValidationAdaptor(obj, logger, **kwargs)
elif _is_badgerfish_version(nexson_version):
# _LOG.debug('validating as BadgerFish...')
return BadgerFishValidationAdaptor(obj, logger, **kwargs)
elif _is_direct_hbf(nexson_version):
# _LOG.debug('validating as DirectHBF...')
return DirectHBFValidationAdaptor(obj, logger, **kwargs)
raise NotImplementedError('nexml2json version {v}'.format(v=nexson_version))
def add_schema_attributes(container, nexson_version):
"""Adds several attributes to `container`:
_using_hbf_meta - boolean. True for HoneyBadgerFish v1-style meta elements ('^prop': value
rather than 'meta': {'$':value})
and the following _SchemaFragment instances:
_NexmlEl_Schema
"""
if _is_by_id_hbf(nexson_version):
_add_by_id_nexson_schema_attributes(container)
elif _is_badgerfish_version(nexson_version):
_add_badgerfish_nexson_schema_attributes(container)
elif _is_direct_hbf(nexson_version):
_add_direct_nexson_schema_attributes(container)
else:
raise NotImplementedError('unrecognized nexson variant {}'.format(nexson_version))
def _nexson_directly_translatable_to_nexml(vers):
'TEMP: until we refactor nexml writing code to be more general...'
return (_is_badgerfish_version(vers)
or _is_direct_hbf(vers)
or vers == 'nexml')
def _nexson_directly_translatable_to_nexml(vers):
'TEMP: until we refactor nexml writing code to be more general...'
return (_is_badgerfish_version(vers)
or _is_direct_hbf(vers)
or vers == 'nexml')
def __init__(self, conv_cfg):
NexsonConverter.__init__(self, conv_cfg)
self.output_format = conv_cfg.output_format
self._badgerfish_style_conversion = _is_badgerfish_version(
conv_cfg.output_format)
def __init__(self, conv_cfg):
NexsonConverter.__init__(self, conv_cfg)
self.output_format = conv_cfg.output_format
self._badgerfish_style_conversion = _is_badgerfish_version(conv_cfg.output_format)