def _move_label_properties_for_otu(otu, syntax_version):
ol = find_val_literal_meta_first(otu, 'ot:originalLabel', syntax_version)
assert ol is not None
label_att = otu.get('@label')
if label_att is not None:
del otu['@label']
if label_att != ol:
ml = find_val_literal_meta_first(otu, 'ot:ottTaxonName', syntax_version)
if (not ml) or (ml != label_att):
add_literal_meta(otu, 'ot:altLabel', label_att, syntax_version)
def add_or_replace_annotation(
self, #pylint: disable=R0201
obj,
annotation,
agent,
add_agent_only=False):
'''Takes an `annotation` dictionary which is
expected to have a string as the value of annotation['author']['name']
This function will remove all annotations from obj that:
1. have the same author/name, and
2. have no messages that are flagged as messages to be preserved (values for 'preserve'
that evaluate to true)
'''
nex = get_nexml_el(obj)
nvers = detect_nexson_version(obj)
_LOG.debug('detected version as ' + nvers)
agents_obj = find_val_literal_meta_first(nex, 'ot:agents', nvers)
if not agents_obj:
agents_obj = add_literal_meta(nex, 'ot:agents', {'agent': []},
nvers)
agents_list = agents_obj.setdefault('agent', [])
found_agent = False
aid = agent['@id']
for a in agents_list:
if a.get('@id') == aid:
found_agent = True
break
if not found_agent:
agents_list.append(agent)
if add_agent_only:
delete_same_agent_annotation(obj, annotation)
else:
replace_same_agent_annotation(obj, annotation)
def add_or_replace_annotation(self, # pylint: disable=R0201
obj,
annotation,
agent,
add_agent_only=False):
"""Takes an `annotation` dictionary which is
expected to have a string as the value of annotation['author']['name']
This function will remove all annotations from obj that:
1. have the same author/name, and
2. have no messages that are flagged as messages to be preserved (values for 'preserve'
that evaluate to true)
"""
nex = get_nexml_el(obj)
nvers = detect_nexson_version(obj)
_LOG.debug('detected version as ' + nvers)
agents_obj = find_val_literal_meta_first(nex, 'ot:agents', nvers)
if not agents_obj:
agents_obj = add_literal_meta(nex, 'ot:agents', {'agent': []}, nvers)
agents_list = agents_obj.setdefault('agent', [])
found_agent = False
aid = agent['@id']
for a in agents_list:
if a.get('@id') == aid:
found_agent = True
break
if not found_agent:
agents_list.append(agent)
if add_agent_only:
delete_same_agent_annotation(obj, annotation)
else:
replace_same_agent_annotation(obj, annotation)
def get_annotation_list(nex_el, nexson_version):
ae_s_obj = find_nested_meta_first(nex_el, 'ot:annotationEvents', nexson_version)
if not ae_s_obj:
ae_s_obj = add_literal_meta(nex_el, 'ot:annotationEvents', {'annotation': []}, nexson_version)
# _LOG.debug('ae_s_obj = ' + str(ae_s_obj))
annotation_list = ae_s_obj.setdefault('annotation', [])
# _LOG.debug('annotation_list = ' + str(annotation_list))
return annotation_list
def get_annotation_list(nex_el, nexson_version):
ae_s_obj = find_nested_meta_first(nex_el, 'ot:annotationEvents',
nexson_version)
if not ae_s_obj:
ae_s_obj = add_literal_meta(nex_el, 'ot:annotationEvents',
{'annotation': []}, nexson_version)
#_LOG.debug('ae_s_obj = ' + str(ae_s_obj))
annotation_list = ae_s_obj.setdefault('annotation', [])
#_LOG.debug('annotation_list = ' + str(annotation_list))
return annotation_list
def _post_key_check_validate_tree(self,
tree_nex_id,
tree_obj,
vc,
otus_group_id=None):
node_list = tree_obj.get('node')
if isinstance(node_list, dict):
node_list = [node_list]
elif (not node_list) or (not isinstance(node_list, list)):
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=['node',])
return errorReturn('no "node" in "trees"')
edge_list = tree_obj.get('edge')
if isinstance(edge_list, dict):
edge_list = [edge_list]
elif (not edge_list) or (not isinstance(edge_list, list)):
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=['edge',])
return errorReturn('no "edge" in tree')
edge_id_list = [(i.get('@id'), i) for i in edge_list]
valid = self._validate_edge_list(edge_id_list, vc)
if not valid:
return False
node_id_obj_list = [(i.get('@id'), i) for i in node_list]
valid = self._validate_node_list(node_id_obj_list, vc)
if not valid:
return False
node_dict = {}
for i in node_id_obj_list:
nid, nd = i
node_dict[nid] = nd
missing_src = []
missing_target = []
for el in edge_id_list:
e = el[1]
sid = e.get('@source')
tid = e.get('@target')
if sid not in node_dict:
missing_src.append(sid)
if tid not in node_dict:
missing_target.append(tid)
if missing_src:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_ReferencedIDNotFoundWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=missing_src)
return errorReturn('no "@source" in edge')
if missing_target:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_ReferencedIDNotFoundWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=missing_target)
return errorReturn('no "@target" in edge')
if otus_group_id is None:
tree_group = vc.anc_list[-1][1]
otus_group_id = tree_group.get('@otus')
lowest_nodeid_set = set()
encountered_nodes = set()
edge_by_target = {}
edge_by_source = {}
multi_parent_node = []
for e in edge_list:
t = e.get('@target')
if t in edge_by_target:
multi_parent_node.append(t)
else:
edge_by_target[t] = e
#_LOG.debug('e=' + str(e))
sid = e['@source']
edge_by_source.setdefault(sid, []).append(e)
if multi_parent_node:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_NodeWithMultipleParents,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=multi_parent_node)
otuid2leaf = {}
unflagged_leaves = []
nonleaves_with_leaf_flags = []
with_at_root_prop = {}
first_lowest_node = None
for nd in node_list:
nid = nd.get('@id')
cycle_node, path_to_root = construct_path_to_root(nd, encountered_nodes, edge_by_target)
if cycle_node:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_TreeCycleWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
cycle_node=cycle_node)
return errorReturn('"@id" in node')
if path_to_root:
lowest_nodeid_set.add(path_to_root[-1])
if first_lowest_node is None:
first_lowest_node = path_to_root[-1]
is_flagged_as_leaf = self._find_first_literal_meta(nd, 'ot:isLeaf')
ch_list = edge_by_source.get(nid)
if ch_list is None:
otu_id = nd.get('@otu')
if otu_id is None:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
self._error_event(_NEXEL.NODE,
obj=nd,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=nid,
key_list=['@otu',])
return errorReturn('"@otu" in leaf')
finally:
vc.pop_context()
else:
if otu_id in otuid2leaf:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
self._error_event(_NEXEL.NODE,
obj=nd,
err_type=gen_RepeatedOTUWarning,
anc=vc.anc_list,
obj_nex_id=nid,
key_list=[otu_id])
return errorReturn('repeated "@otu" in leaves')
finally:
vc.pop_context()
otuid2leaf[otu_id] = nd
if not is_flagged_as_leaf:
if not self._logger.retain_deprecated:
add_literal_meta(nd, 'ot:isLeaf', True, self._syntax_version)
else:
unflagged_leaves.append(nid)
elif is_flagged_as_leaf:
if not self._logger.retain_deprecated:
delete_first_literal_meta(nd, 'ot:isLeaf', self._syntax_version)
else:
nonleaves_with_leaf_flags.append(nid)
if nd.get('@root'):
with_at_root_prop[nid] = nd
if unflagged_leaves:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
#_LOG.debug('unflagged_leaves="{f}"'.format(f=unflagged_leaves))
self._error_event(_NEXEL.NODE,
obj=tree_obj,
err_type=gen_MissingMandatoryKeyWarning,
anc=vc.anc_list,
obj_nex_id=unflagged_leaves,
key_list=['ot:isLeaf'])
finally:
vc.pop_context()
if nonleaves_with_leaf_flags:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_InvalidKeyWarning,
anc=vc.anc_list,
obj_nex_id=nonleaves_with_leaf_flags,
key_list=['ot:isLeaf'])
return errorReturn('"ot:isLeaf" for internal')
self._detect_multilabelled_tree(otus_group_id,
tree_nex_id,
otuid2leaf)
if len(lowest_nodeid_set) > 1:
lowest_nodeid_set = list(lowest_nodeid_set)
lowest_nodeid_set.sort()
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=lowest_nodeid_set)
return errorReturn('multiple roots in a tree')
root_node_id = first_lowest_node
if root_node_id not in with_at_root_prop:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=list(with_at_root_prop.keys()) + [root_node_id])
return errorReturn('root without "@root"')
elif len(with_at_root_prop) > 1:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=list(with_at_root_prop.keys()))
return errorReturn('Multiple nodes with "@root"')
elif len(with_at_root_prop) == 0:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_NoRootWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id)
return errorReturn('no node with "@root"')
og = set([i['@id'] for i in self._otu_group_by_id[otus_group_id]['otu']])
nli = [(i['@id'], i) for i in node_list]
return self._validate_otu_key_if_present(nli, og, vc)
def _post_key_check_validate_tree(self,
tree_nex_id,
tree_obj,
vc,
otus_group_id=None):
node_list = tree_obj.get('node')
if isinstance(node_list, dict):
node_list = [node_list]
elif (not node_list) or (not isinstance(node_list, list)):
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=[
'node',
])
return errorReturn('no "node" in "trees"')
edge_list = tree_obj.get('edge')
if isinstance(edge_list, dict):
edge_list = [edge_list]
elif (not edge_list) or (not isinstance(edge_list, list)):
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=[
'edge',
])
return errorReturn('no "edge" in tree')
edge_id_list = [(i.get('@id'), i) for i in edge_list]
vc.push_context(_NEXEL.EDGE, (tree_obj, tree_nex_id))
try:
valid = self._validate_edge_list(edge_id_list, vc)
if not valid:
return False
finally:
vc.pop_context()
node_id_obj_list = [(i.get('@id'), i) for i in node_list]
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
valid = self._validate_node_list(node_id_obj_list, vc)
if not valid:
return False
finally:
vc.pop_context()
node_dict = {}
for i in node_id_obj_list:
nid, nd = i
node_dict[nid] = nd
missing_src = []
missing_target = []
for el in edge_id_list:
e = el[1]
sid = e.get('@source')
tid = e.get('@target')
if sid not in node_dict:
missing_src.append(sid)
if tid not in node_dict:
missing_target.append(tid)
if missing_src:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_ReferencedIDNotFoundWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=missing_src)
return errorReturn('no "@source" in edge')
if missing_target:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_ReferencedIDNotFoundWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
key_list=missing_target)
return errorReturn('no "@target" in edge')
if otus_group_id is None:
tree_group = vc.anc_list[-1][1]
otus_group_id = tree_group.get('@otus')
lowest_nodeid_set = set()
encountered_nodes = set()
edge_by_target = {}
edge_by_source = {}
multi_parent_node = []
for e in edge_list:
t = e.get('@target')
if t in edge_by_target:
multi_parent_node.append(t)
else:
edge_by_target[t] = e
#_LOG.debug('e=' + str(e))
sid = e['@source']
edge_by_source.setdefault(sid, []).append(e)
if multi_parent_node:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_NodeWithMultipleParents,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=multi_parent_node)
otuid2leaf = {}
unflagged_leaves = []
nonleaves_with_leaf_flags = []
with_at_root_prop = {}
first_lowest_node = None
for nd in node_list:
nid = nd.get('@id')
cycle_node, path_to_root = construct_path_to_root(
nd, encountered_nodes, edge_by_target)
if cycle_node:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_TreeCycleWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
cycle_node=cycle_node)
return errorReturn('"@id" in node')
if path_to_root:
lowest_nodeid_set.add(path_to_root[-1])
if first_lowest_node is None:
first_lowest_node = path_to_root[-1]
is_flagged_as_leaf = self._find_first_literal_meta(nd, 'ot:isLeaf')
ch_list = edge_by_source.get(nid)
if ch_list is None:
otu_id = nd.get('@otu')
if otu_id is None:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
self._error_event(
_NEXEL.NODE,
obj=nd,
err_type=gen_MissingCrucialContentWarning,
anc=vc.anc_list,
obj_nex_id=nid,
key_list=[
'@otu',
])
return errorReturn('"@otu" in leaf')
finally:
vc.pop_context()
else:
if otu_id in otuid2leaf:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
self._error_event(_NEXEL.NODE,
obj=nd,
err_type=gen_RepeatedOTUWarning,
anc=vc.anc_list,
obj_nex_id=nid,
key_list=[otu_id])
return errorReturn('repeated "@otu" in leaves')
finally:
vc.pop_context()
otuid2leaf[otu_id] = nd
if not is_flagged_as_leaf:
if not self._logger.retain_deprecated:
add_literal_meta(nd, 'ot:isLeaf', True,
self._syntax_version)
else:
unflagged_leaves.append(nid)
elif is_flagged_as_leaf:
if not self._logger.retain_deprecated:
delete_first_literal_meta(nd, 'ot:isLeaf',
self._syntax_version)
else:
nonleaves_with_leaf_flags.append(nid)
if nd.get('@root'):
with_at_root_prop[nid] = nd
if unflagged_leaves:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
#_LOG.debug('unflagged_leaves="{f}"'.format(f=unflagged_leaves))
self._error_event(_NEXEL.NODE,
obj=tree_obj,
err_type=gen_MissingMandatoryKeyWarning,
anc=vc.anc_list,
obj_nex_id=unflagged_leaves,
key_list=['ot:isLeaf'])
finally:
vc.pop_context()
if nonleaves_with_leaf_flags:
vc.push_context(_NEXEL.NODE, (tree_obj, tree_nex_id))
try:
self._error_event(_NEXEL.NODE,
obj=tree_obj,
err_type=gen_InvalidKeyWarning,
anc=vc.anc_list,
obj_nex_id=nonleaves_with_leaf_flags,
key_list=['ot:isLeaf'])
return errorReturn('"ot:isLeaf" for internal')
finally:
vc.pop_context()
self._detect_multilabelled_tree(otus_group_id, tree_nex_id, otuid2leaf)
if len(lowest_nodeid_set) > 1:
lowest_nodeid_set = list(lowest_nodeid_set)
lowest_nodeid_set.sort()
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=lowest_nodeid_set)
return errorReturn('multiple roots in a tree')
root_node_id = first_lowest_node
if root_node_id not in with_at_root_prop:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=list(with_at_root_prop.keys()) +
[root_node_id])
return errorReturn('root without "@root"')
elif len(with_at_root_prop) > 1:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_MultipleRootsWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id,
node_id_list=list(with_at_root_prop.keys()))
return errorReturn('Multiple nodes with "@root"')
elif len(with_at_root_prop) == 0:
self._error_event(_NEXEL.TREE,
obj=tree_obj,
err_type=gen_NoRootWarning,
anc=vc.anc_list,
obj_nex_id=tree_nex_id)
return errorReturn('no node with "@root"')
og = set(
[i['@id'] for i in self._otu_group_by_id[otus_group_id]['otu']])
nli = [(i['@id'], i) for i in node_list]
return self._validate_otu_key_if_present(nli, og, vc)