def find_related(pmid):
"""find_related(pmid) -> ids
Search PubMed for a list of citations related to pmid. pmid can
be a PubMed ID, a MEDLINE UID, or a list of those.
"""
class ResultParser(sgmllib.SGMLParser):
# Parse the ID's out of the HTML-formatted page that PubMed
# returns. The format of the page is:
# [...]
# <Link>
# <Id>######</Id>
# <Score>######</Score>
# [...]
# </Link>
# [...]
def __init__(self):
sgmllib.SGMLParser.__init__(self)
self.ids = []
self.in_link = 0
self.in_id = 0
def start_id(self, attributes):
self.in_id = 1
def end_id(self):
self.in_id = 0
def start_link(self, attributes):
self.in_link = 1
def end_link(self):
self.in_link = 0
_not_pmid_re = re.compile(r'\D')
def handle_data(self, data):
if not self.in_link or not self.in_id:
return
# Everything here should be a PMID. Check and make sure
# data really is one. A PMID should be a string consisting
# of only integers. Should I check to make sure it
# meets a certain minimum length?
if self._not_pmid_re.search(data):
raise ValueError, \
"I expected an ID, but '%s' doesn't look like one." % \
repr(data)
self.ids.append(data)
parser = ResultParser()
if type(pmid) is type([]):
pmid = string.join(pmid, ',')
h = NCBI.elink(dbfrom='pubmed', id=pmid)
parser.feed(h.read())
return parser.ids
