def get_abstract_information(pubmed_id):
article_xml_tree = get_pubmed_article_xml_tree(pubmed_id)
title, abstract_xml_tree = parse_article_xml_tree(article_xml_tree)
if isinstance(title, unicode):
title = convert_unicode_to_ascii(title)
if abstract_xml_tree:
return (title, split_abstract(abstract_xml_tree))
return (title, [])
def split_abstract(abstract_xml_tree):
"""
Splits an abstract XML tree into individual chunks, if they exist.
Preserves the background/methods/etc format of some papers (eg pmid 24885308)
"""
abstract_chunks = []
for child in abstract_xml_tree.iter("AbstractText"):
section_name = child.get("Label")
text = child.text
if isinstance(text, unicode):
text = convert_unicode_to_ascii(text)
if section_name is not None:
text = "{0}: {1}".format(section_name, text)
abstract_chunks.append(text)
return abstract_chunks
