def get_unique_speaker_id(tidied_speaker, on_date):
ids = memberList.match_whole_speaker(tidied_speaker,str(on_date))
if ids is None:
# This special return value (None) indicates that the speaker
# is something we know about, but not an MSP (e.g Lord
# Advocate)
return None
else:
if len(ids) == 0:
log_speaker(tidied_speaker,str(on_date),"missing")
return None
elif len(ids) == 1:
Speech.speakers_so_far.append(ids[0])
return ids[0]
else:
final_id = None
# If there's an ambiguity there our best bet is to go
# back through the previous IDs used today, and pick
# the most recent one that's in the list we just got
# back...
for i in range(len(Speech.speakers_so_far) - 1, -1, -1):
older_id = Speech.speakers_so_far[i]
if older_id in ids:
final_id = older_id
break
if final_id:
Speech.speakers_so_far.append(final_id)
return final_id
else:
log_speaker(tidied_speaker,str(on_date),"genuine ambiguity")
def valid_speaker(self,speaker_name):
tidied_speaker = speaker_name
if not tidied_speaker:
return None
if self.odd_unknowns.has_key(speaker_name):
return "unknown"
tidied_speaker = re.sub("((on behalf of the )?Scottish Parliamentary Corporate Body)",'',tidied_speaker)
ids = memberList.match_whole_speaker(tidied_speaker,str(parser.date))
if ids == None:
return "unknown"
if not ids: # i.e. it's the empty list...
if verbose: print "No match for speaker: "+tidied_speaker+" on date "+str(parser.date)
return None
elif len(ids) > 1:
if verbose: print "Too many matches found for speaker: "+tidied_speaker+" on date "+str(parser.date)
return None
else:
return ids[0]
