def _yield_items_from_stream(self, stream):
xml_doc = xmlprocessing.XmlDocument(
file_obj=stream, subelement_factory=self._subelement_factory)
self._namespace_registry = xml_doc.namespace_registry
xml_root = xml_doc.root
self._parse_taxon_namespaces(xml_root)
tree_parser = nexmlreader._NexmlTreeParser(
id_taxon_map=self._id_taxon_map,
annotations_processor_fn=self._parse_annotations,
)
for trees_idx, trees_element in enumerate(xml_root.iter_trees()):
trees_id = trees_element.get('id', "Trees" + str(trees_idx))
trees_label = trees_element.get('label', None)
otus_id = trees_element.get('otus', None)
if otus_id is None:
raise Exception(
"Taxa block not specified for trees block '{}'".format(
otus_id))
taxon_namespace = self._id_taxon_namespace_map.get(otus_id, None)
if not taxon_namespace:
raise Exception(
"Tree block '{}': Taxa block '{}' not found".format(
trees_id, otus_id))
for tree_element in trees_element.findall_tree():
tree_obj = self.tree_factory()
tree_parser.build_tree(tree_obj, tree_element, otus_id)
yield tree_obj
def parse_list_keys(self, stream):
keys = []
xml_doc = xmlprocessing.XmlDocument(file_obj=stream,
subelement_factory=GbifXmlElement)
xml_root = xml_doc.root
for txml in xml_root.iter_taxon_occurrence():
keys.append(txml.get("gbifKey"))
return keys
def parse_from_stream(stream):
xml_doc = xmlprocessing.XmlDocument(file_obj=stream,
subelement_factory=GbifXmlElement)
gb_recs = []
for txo in xml_doc.root.iter_taxon_occurrence():
gbo = GbifOccurrenceRecord()
gbo.parse_taxon_occurrence_xml(txo)
gb_recs.append(gbo)
return gb_recs
def _read(self,
stream,
taxon_namespace_factory=None,
tree_list_factory=None,
char_matrix_factory=None,
state_alphabet_factory=None,
global_annotations_target=None):
xml_doc = xmlprocessing.XmlDocument(
file_obj=stream, subelement_factory=self._subelement_factory)
self._namespace_registry = xml_doc.namespace_registry
self._taxon_namespace_factory = taxon_namespace_factory
self._tree_list_factory = tree_list_factory
self._char_matrix_factory = char_matrix_factory
self._state_alphabet_factory = state_alphabet_factory
self._global_annotations_target = global_annotations_target
self._parse_document(xml_doc)
self._product = self.Product(taxon_namespaces=self._taxon_namespaces,
tree_lists=self._tree_lists,
char_matrices=self._char_matrices)
return self._product