Merge pull request #17 from alonisser/master

Faster committee protocol speaker id parsing

knesset_data_django/committees/protocol_part_builder.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*
-
+from functools32 import lru_cache
 from knesset_data.protocols.protocol_header_parser import ProtocolHeaderParser
 from collections import namedtuple
 
@@ -9,19 +9,27 @@
 ProtocolPartResult = namedtuple('ProtocolPartResult', ['meeting', 'order', 'header', 'body', 'speaker'])
 
 
-def create_person_name_mk_id_map(mks):
-    persons = Person.objects.filter(mk__in=mks).select_related()
+@lru_cache(maxsize=1000)
+def create_person_name_mk_id_map(mks_ids):
+    persons = Person.objects.filter(mk_id__in=mks_ids).select_related()
     aliases = PersonAlias.objects.filter(person__in=persons).select_related()
     persons_map = dict()
     for person in persons:
-        persons_map[person.name] = person.mk.id
+        persons_map[person.name] = person.id
 
     for alias in aliases:
-        persons_map[alias.name] = alias.person.mk.id
+        persons_map[alias.name] = alias.person.id
 
     return persons_map
 
 
+@lru_cache(maxsize=1000)
+def parse_header(mks_ids, header):
+    mapped_names_ids = create_person_name_mk_id_map(mks_ids)
+    header_text, speaker_id = ProtocolHeaderParser(header, mapped_names_ids).parse()
+    return speaker_id
+
+
 class CommitteeProtocolPartBuilder(object):
     def __init__(self, meeting, order, header, body, mks):
         self.mks = mks
@@ -32,10 +40,9 @@ def __init__(self, meeting, order, header, body, mks):
         self.body = body
 
     def build(self):
-        # mapped_names_ids = create_person_name_mk_id_map(self.mks)
-        # header_text, speaker_id = ProtocolHeaderParser(self.header, mapped_names_ids).parse()
+        speaker_id = parse_header(tuple([mk.id for mk in self.mks]), self.header)
         return ProtocolPart(meeting=self.committee_meeting,
                             order=self.order,
                             body=self.body,
-                            # speaker_id=speaker_id,
+                            speaker_id=speaker_id,
                             header=self.header)

knesset_data_django/committees/tests/testapp.py

@@ -29,11 +29,7 @@ def given_party_exists_in_knesset(cls, party_name, knesset):
     @classmethod
     def given_member_exists_in_knesset(cls, member_name, party, start_date=ten_days_ago.date(), end_date=None):
         member, create = Member.objects.get_or_create(name=member_name, start_date=ten_days_ago.date())
-        # membership, create = Membership.objects.get_or_create(member=member, party=party,
-        #                                                       start_date=party.knesset.start_date)
-        # if end_date:
-        #     membership.end_date = end_date
-        #     membership.save()
+
         return member
 
     @classmethod

knesset_data_django/settings.py

@@ -2,6 +2,10 @@
 # Django settings for knesset-data-django testing app
 import os, sys
 from knesset_data_django import KNESSET_DATA_DJANGO_APPS
+import logging
+logging.basicConfig(level=logging.INFO, stream=sys.stdout,
+                    format="%(asctime)s:%(name)s:%(lineno)d:%(levelname)s:%(message)s")
+
 
 DEBUG = True
 TEMPLATE_DEBUG = DEBUG