def set_journal(self, journal):
node = self._fetch_node("titles")
cleanj = clean_line(journal)
self._set_item(cleanj, self.journaltag, node)
#and do the abbreviation
node = self._fetch_node("periodical")
abbrev = JournalCleanup.abbreviate(journal)
self._set_item(abbrev, self.abbrevtag, node)
def process_authors(entries):
authors = []
for entry in entries:
if len(entry) == 1: #hmm, chinese name
entry = map(lambda x: x.strip(), entry[0].split())
last, first = entry[:2]
#check to see if we have stupidness
first_first = first.split()[0]
match = re.compile("[A-Z]{2}").search(first_first)
if match: #I f*****g hate you papers
initials = []
for entry in first_first:
initials.append(entry)
first = " ".join(initials)
else:
first = clean_line(first)
#capitalize last name
last = capitalize_word(last)
name = "%s, %s" % (last, first)
authors.append(name)
return authors