def _create_new_object(self, dataservice_vote):
return Vote.objects.create(
src_id=dataservice_vote.id,
title=u'{vote} - {sess}'.format(vote=dataservice_vote.item_dscr, sess=dataservice_vote.sess_item_dscr),
time_string = u'יום '+hebrew_strftime(dataservice_vote.datetime),
importance = 1,
time = dataservice_vote.datetime,
meeting_number = dataservice_vote.session_num,
vote_number = dataservice_vote.nbr_in_sess,
src_url = dataservice_vote.id
)
def _create_new_object(self, dataservice_meeting):
meeting = CommitteeMeeting.objects.create(
committee=get_committee(dataservice_meeting.committee_id),
date_string=hebrew_strftime(dataservice_meeting.datetime, u'%d/%m/%Y'),
date=dataservice_meeting.date,
topics=dataservice_meeting.agendum1,
datetime=dataservice_meeting.datetime,
knesset_id=dataservice_meeting.id,
src_url=dataservice_meeting.link,
)
meeting.reparse_protocol(mks=self.mks, mk_names=self.mk_names)
return meeting
class Command(BaseKnessetDataserviceCollectionCommand):
DATASERVICE_CLASS = DataserviceVote
DATASERVICE_MODEL_MAP = {
# model attribute name | dataservice attribute name, or lambda to get the value
'src_id':
'id',
'title':
lambda vote: u'{vote} - {sess}'.format(vote=vote.item_dscr,
sess=vote.sess_item_dscr),
'time_string':
lambda vote: u'יום %s' % hebrew_strftime(vote.datetime),
'importance':
lambda vote: 1,
'time':
'datetime',
'meeting_number':
"session_num",
'vote_number':
'nbr_in_sess',
'src_url':
lambda vote:
"http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s" %
vote.id
}
VALIDATE_FIELDS_TO_AUTOFIX = ['title', 'src_url']
option_list = BaseKnessetDataserviceCollectionCommand.option_list + (
make_option(
'--validate-votes-pages',
dest='validatevotepages',
help=
"validate votes between (and including) given page range\npages in this case are based on vote id ascending, so you'll have the same page number each time"
),
make_option(
'--validate-skip-to',
dest='validateskipto',
help=
"skip to the given vote id (for use with --validate-votes-pages)"),
make_option(
'--create-vote-src-id',
dest='createvotesrcid',
help=
"create the given vote/s from the comma-separated src ids (assuming they don't already exist in DB)"
),
make_option(
'--validate-output-file',
dest='validateoutputfile',
help="where to write the validation results to (defaults to stdout)"
),
make_option(
'--validate-fix',
dest='validatefix',
action='store_true',
help=
"try to fix some problems directly in DB which are safe to automatically fix"
))
help = "Scrape votes data from the knesset"
def _update_or_create_vote(self, dataservice_vote, oknesset_vote=None):
vote_kwargs = self._get_dataservice_model_kwargs(dataservice_vote)
if oknesset_vote:
[setattr(oknesset_vote, k, v) for k, v in vote_kwargs.iteritems()]
oknesset_vote.save()
else:
oknesset_vote = Vote.objects.create(**vote_kwargs)
self._add_vote_actions(dataservice_vote, oknesset_vote)
oknesset_vote.update_vote_properties()
SyncdataCommand().find_synced_protocol(oknesset_vote)
Link.objects.create(
title=u'ההצבעה באתר הכנסת',
url=
'http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s'
% oknesset_vote.src_id,
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=str(oknesset_vote.id))
return oknesset_vote
# if v.full_text_url != None:
# l = Link(title=u'מסמך הצעת החוק באתר הכנסת', url=v.full_text_url, content_type=ContentType.objects.get_for_model(v), object_pk=str(v.id))
# l.save()
def _add_vote_actions(self, dataservice_vote, oknesset_vote):
for member_id, vote_result_code in HtmlVote.get_from_vote_id(
dataservice_vote.id).member_votes:
member_qs = Member.objects.filter(pk=member_id)
if member_qs.exists():
member = member_qs.first()
vote_type = self._resolve_vote_type(vote_result_code)
vote_action, created = VoteAction.objects.get_or_create(
vote=oknesset_vote,
member=member,
defaults={
'type': vote_type,
'party': member.current_party
})
if created:
vote_action.save()
else:
raise VoteScraperException(
'vote %s: could not find member id %s' %
(dataservice_vote.id, member_id))
def _has_existing_object(self, dataservice_vote):
qs = Vote.objects.filter(src_id=dataservice_vote.id)
return qs.exists()
def _create_new_object(self, dataservice_vote):
return self._update_or_create_vote(dataservice_vote)
def _resolve_vote_type(cls, vote_result_code):
return {
'voted for': u'for',
'voted against': u'against',
'abstain': u'abstain',
'did not vote': u'no-vote',
}[vote_result_code]
def recreate_objects(self, vote_ids):
recreated_votes = []
for vote_id in vote_ids:
oknesset_vote = Vote.objects.get(id=int(vote_id))
vote_src_id = oknesset_vote.src_id
dataservice_vote = self.DATASERVICE_CLASS.get(vote_src_id)
VoteAction.objects.filter(vote=oknesset_vote).delete()
Link.objects.filter(
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=oknesset_vote.id).delete()
recreated_votes.append(
self._update_or_create_vote(dataservice_vote, oknesset_vote))
return recreated_votes
def _get_validate_header_row(self):
return ['knesset vote id', 'open knesset vote id', 'error']
def _get_validate_error_row(self, dataservice_object, oknesset_object,
error):
return [dataservice_object.id, oknesset_object.id, error]
def _get_validate_first_object_title(self, dataservice_object):
return 'date: %s' % dataservice_object.datetime
def _validate_attr_actual_expected(self, attr_name, actual_value,
expected_value):
if attr_name == 'time_string':
# remove some unprintable artifacts which for some reason are in the old scraper's votes
actual_value = actual_value.replace(u"\u200f",
"").replace(u"\xa0", " ")
elif attr_name == 'title' and actual_value != expected_value:
# try a slightly different format which exists in DB in some cases
actual_value = actual_value.replace(u" - הצעת חוק", u" - חוק")
return super(Command, self)._validate_attr_actual_expected(
attr_name, actual_value, expected_value)
class Command(BaseKnessetDataserviceCollectionCommand):
DATASERVICE_CLASS = DataserviceVote
option_list = BaseKnessetDataserviceCollectionCommand.option_list + (
make_option(
'--validate-votes-pages',
dest='validatevotepages',
help=
"validate votes between (and including) given page range\npages in this case are based on vote id ascending, so you'll have the same page number each time"
),
make_option(
'--validate-skip-to',
dest='validateskipto',
help=
"skip to the given vote id (for use with --validate-votes-pages)"),
make_option(
'--create-vote-src-id',
dest='createvotesrcid',
help=
"create the given vote/s from the comma-separated src ids (assuming they don't already exist in DB)"
),
make_option(
'--validate-output-file',
dest='validateoutputfile',
help="where to write the validation results to (defaults to stdout)"
),
make_option(
'--validate-fix',
dest='validatefix',
action='store_true',
help=
"try to fix some problems directly in DB which are safe to automatically fix"
))
help = "Scrape votes data from the knesset"
dataservice_model_map = {
# model attribute name | dataservice attribute name, or lambda to get the value
'src_id':
'id',
'title':
lambda vote: u'{vote} - {sess}'.format(vote=vote.item_dscr,
sess=vote.sess_item_dscr),
'time_string':
lambda vote: u'יום %s' % hebrew_strftime(vote.datetime),
'importance':
lambda vote: 1,
'time':
'datetime',
'meeting_number':
"session_num",
'vote_number':
'nbr_in_sess',
'src_url':
lambda vote:
"http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s" %
vote.id
}
def _get_dataservice_model_kwargs(self, dataservice_vote):
return {
k: getattr(dataservice_vote, v)
if isinstance(v, str) else v(dataservice_vote)
for k, v in self.dataservice_model_map.iteritems()
}
def _update_or_create_vote(self, dataservice_vote, oknesset_vote=None):
vote_kwargs = self._get_dataservice_model_kwargs(dataservice_vote)
if oknesset_vote:
[setattr(oknesset_vote, k, v) for k, v in vote_kwargs.iteritems()]
oknesset_vote.save()
else:
oknesset_vote = Vote.objects.create(**vote_kwargs)
self._add_vote_actions(dataservice_vote, oknesset_vote)
oknesset_vote.update_vote_properties()
SyncdataCommand().find_synced_protocol(oknesset_vote)
Link.objects.create(
title=u'ההצבעה באתר הכנסת',
url=
'http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s'
% oknesset_vote.src_id,
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=str(oknesset_vote.id))
return oknesset_vote
# if v.full_text_url != None:
# l = Link(title=u'מסמך הצעת החוק באתר הכנסת', url=v.full_text_url, content_type=ContentType.objects.get_for_model(v), object_pk=str(v.id))
# l.save()
def _add_vote_actions(self, dataservice_vote, oknesset_vote):
for member_id, vote_result_code in HtmlVote.get_from_vote_id(
dataservice_vote.id).member_votes:
member_qs = Member.objects.filter(pk=member_id)
if member_qs.exists():
member = member_qs.first()
vote_type = self._resolve_vote_type(vote_result_code)
vote_action, created = VoteAction.objects.get_or_create(
vote=oknesset_vote,
member=member,
defaults={
'type': vote_type,
'party': member.current_party
})
if created:
vote_action.save()
else:
raise VoteScraperException(
'vote %s: could not find member id %s' %
(dataservice_vote.id, member_id))
def _has_existing_object(self, dataservice_vote):
qs = Vote.objects.filter(src_id=dataservice_vote.id)
return qs.exists()
def _create_new_object(self, dataservice_vote):
return self._update_or_create_vote(dataservice_vote)
def _resolve_vote_type(cls, vote_result_code):
return {
'voted for': u'for',
'voted against': u'against',
'abstain': u'abstain',
'did not vote': u'no-vote',
}[vote_result_code]
def recreate_objects(self, vote_ids):
recreated_votes = []
for vote_id in vote_ids:
oknesset_vote = Vote.objects.get(id=int(vote_id))
vote_src_id = oknesset_vote.src_id
dataservice_vote = self.DATASERVICE_CLASS.get(vote_src_id)
VoteAction.objects.filter(vote=oknesset_vote).delete()
Link.objects.filter(
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=oknesset_vote.id).delete()
recreated_votes.append(
self._update_or_create_vote(dataservice_vote, oknesset_vote))
return recreated_votes
def _validate_vote(self, dataservice_vote, csv_writer, fix=False):
# check the basic metadata
qs = Vote.objects.filter(src_id=dataservice_vote.id)
if qs.count() != 1:
if fix:
self._log_info(
'could not find corresponding vote in DB, creating it now')
self._create_new_object(dataservice_vote)
else:
error = 'could not find corresponding vote in DB (qs.count=%s)' % (
qs.count(), )
self._log_warn(error)
csv_writer.writerow(
[dataservice_vote.id, '',
error.encode('utf-8')])
else:
oknesset_vote = qs.first()
for attr_name, expected_value in self._get_dataservice_model_kwargs(
dataservice_vote).iteritems():
actual_value = getattr(oknesset_vote, attr_name)
if attr_name == 'time_string':
# remove some unprintable artifacts which for some reason are in the old scraper's votes
actual_value = actual_value.replace(u"\u200f", "").replace(
u"\xa0", " ")
if attr_name == 'title' and actual_value != expected_value:
# try a slightly different format which exists in DB in some cases
actual_value = actual_value.replace(
u" - הצעת חוק", u" - חוק")
if actual_value != expected_value:
if fix and attr_name in ['title', 'src_url']:
self._log_info('fixing mismatch in %s attribute' %
(attr_name, ))
setattr(oknesset_vote, attr_name, expected_value)
oknesset_vote.save()
else:
error = 'value mismatch for %s (expected="%s", actual="%s")' % (
attr_name, expected_value, actual_value)
self._log_warn(error)
csv_writer.writerow([
dataservice_vote.id, oknesset_vote.id,
error.encode('utf-8')
])
# validate the vote counts
for type_title, oknesset_count, dataservice_count in zip(
('for', 'against', 'abstain'), [
oknesset_vote.actions.filter(type=t).count()
for t in 'for', 'against', 'abstain'
], [
int(getattr(dataservice_vote, t))
for t in 'total_for', 'total_against', 'total_abstain'
]):
if oknesset_count != dataservice_count:
error = 'mismatch in %s count (expected=%s, actual=%s)' % (
type_title, dataservice_count, oknesset_count)
self._log_warn(error)
csv_writer.writerow([
dataservice_vote.id, oknesset_vote.id,
error.encode('utf-8')
])
class Command(BaseKnessetDataserviceCollectionCommand):
DATASERVICE_CLASS = DataserviceVote
DATASERVICE_MODEL_MAP = {
# model attribute name | dataservice attribute name, or lambda to get the value
'src_id':
'id',
'title':
lambda vote: u'{vote} - {sess}'.format(vote=vote.item_dscr,
sess=vote.sess_item_dscr),
'time_string':
lambda vote: u'יום %s' % hebrew_strftime(vote.datetime),
'importance':
lambda vote: 1,
'time':
'datetime',
'meeting_number':
"session_num",
'vote_number':
'nbr_in_sess',
'src_url':
lambda vote:
"http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s" %
vote.id
}
VALIDATE_FIELDS_TO_AUTOFIX = ['title', 'src_url']
help = "Scrape votes data from the knesset"
def _update_or_create_vote(self, dataservice_vote, oknesset_vote=None):
vote_kwargs = self._get_dataservice_model_kwargs(dataservice_vote)
if oknesset_vote:
[setattr(oknesset_vote, k, v) for k, v in vote_kwargs.iteritems()]
oknesset_vote.save()
else:
oknesset_vote = Vote.objects.create(**vote_kwargs)
self._add_vote_actions(dataservice_vote, oknesset_vote)
oknesset_vote.update_vote_properties()
SyncdataCommand().find_synced_protocol(oknesset_vote)
Link.objects.create(
title=u'ההצבעה באתר הכנסת',
url=
'http://www.knesset.gov.il/vote/heb/Vote_Res_Map.asp?vote_id_t=%s'
% oknesset_vote.src_id,
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=str(oknesset_vote.id))
return oknesset_vote
# if v.full_text_url != None:
# l = Link(title=u'מסמך הצעת החוק באתר הכנסת', url=v.full_text_url, content_type=ContentType.objects.get_for_model(v), object_pk=str(v.id))
# l.save()
def _add_vote_actions(self, dataservice_vote, oknesset_vote):
for member_id, vote_result_code in HtmlVote.get_from_vote_id(
dataservice_vote.id).member_votes:
member_qs = Member.objects.filter(pk=member_id)
if member_qs.exists():
member = member_qs.first()
vote_type = self._resolve_vote_type(vote_result_code)
vote_action, created = VoteAction.objects.get_or_create(
vote=oknesset_vote,
member=member,
defaults={
'type': vote_type,
'party': member.current_party
})
if created:
vote_action.save()
else:
raise VoteScraperException(
'vote %s: could not find member id %s' %
(dataservice_vote.id, member_id))
def _has_existing_object(self, dataservice_vote):
qs = Vote.objects.filter(src_id=dataservice_vote.id)
return qs.exists()
def _get_existing_object(self, dataservice_object):
return Vote.objects.get(src_id=dataservice_object.id)
def _create_new_object(self, dataservice_vote):
return self._update_or_create_vote(dataservice_vote)
def _resolve_vote_type(cls, vote_result_code):
return {
'voted for': u'for',
'voted against': u'against',
'abstain': u'abstain',
'did not vote': u'no-vote',
}[vote_result_code]
def recreate_objects(self, vote_ids):
recreated_votes = []
for vote_id in vote_ids:
oknesset_vote = Vote.objects.get(id=int(vote_id))
vote_src_id = oknesset_vote.src_id
dataservice_vote = self.DATASERVICE_CLASS.get(vote_src_id)
VoteAction.objects.filter(vote=oknesset_vote).delete()
Link.objects.filter(
content_type=ContentType.objects.get_for_model(oknesset_vote),
object_pk=oknesset_vote.id).delete()
recreated_votes.append(
self._update_or_create_vote(dataservice_vote, oknesset_vote))
return recreated_votes
def _get_validate_first_object_title(self, dataservice_object):
return 'date: %s' % dataservice_object.datetime
def _validate_attr_actual_expected(self, attr_name, actual_value,
expected_value):
if attr_name == 'time_string':
# remove some unprintable artifacts which for some reason are in the old scraper's votes
actual_value = actual_value.replace(u"\u200f",
"").replace(u"\xa0", " ")
elif attr_name == 'title' and actual_value != expected_value:
# try a slightly different format which exists in DB in some cases
actual_value = actual_value.replace(u" - הצעת חוק", u" - חוק")
return super(Command, self)._validate_attr_actual_expected(
attr_name, actual_value, expected_value)
def _validate_dataservice_oknesset_object(self, dataservice_object,
oknesset_object, writer, fix):
oknesset_object.update_from_knesset_data()
return None
