def test_simple_scraping(session):
from mptracker.scraper.groups import GroupScraper
session.url_map.update({
STRUCTURE_URL:
PAGES_DIR / 'structura-index-modified',
STRUCTURE_URL + '?idg=0':
PAGES_DIR / 'structura-group0',
STRUCTURE_URL + '?idg=1':
PAGES_DIR / 'structura-group1',
STRUCTURE_URL + '?idg=4':
PAGES_DIR / 'structura-group4',
})
scraper = GroupScraper(session)
groups = list(scraper.fetch())
indep = groups[0]
assert indep.is_independent
assert len(indep.current_members) == 13
indep_2 = indep.current_members[2]
assert indep_2.mp_name == "Dumitru Ovidiu-Ioan"
assert indep_2.start_date is None
assert indep_2.end_date is None
psd = groups[1]
assert psd.name == "Grupul parlamentar al Partidului Social Democrat"
assert psd.short_name == "PSD"
assert not psd.is_independent
assert len(psd.current_members) == 165
assert len(psd.former_members) == 2
current_3 = psd.current_members[3]
assert current_3.mp_name == "Itu Cornel"
assert current_3.start_date is None
assert current_3.end_date is None
assert current_3.title == 'Vicelideri'
current_84 = psd.current_members[84]
assert current_84.start_date == date(2013, 9, 18)
former_0 = psd.former_members[0]
assert former_0.mp_name == "Cernea Remus-Florinel"
assert former_0.start_date is None
assert former_0.end_date == date(2013, 5, 21)
ppdd = groups[2]
current_2 = ppdd.current_members[2]
assert current_2.mp_name == "Ciuhodaru Tudor"
former_2 = ppdd.former_members[2]
assert former_2.mp_name == "Chebac Eugen"
assert former_2.start_date is None
assert former_2.end_date == date(2013, 9, 30)
former_4 = ppdd.former_members[4]
assert former_4.start_date == date(2013, 2, 11)
assert former_4.end_date == date(2013, 5, 7)
def test_simple_scraping(session):
from mptracker.scraper.groups import GroupScraper
session.url_map.update({
STRUCTURE_URL: PAGES_DIR / 'structura-index-modified',
STRUCTURE_URL + '?idg=0': PAGES_DIR / 'structura-group0',
STRUCTURE_URL + '?idg=1': PAGES_DIR / 'structura-group1',
STRUCTURE_URL + '?idg=4': PAGES_DIR / 'structura-group4',
})
scraper = GroupScraper(session)
groups = list(scraper.fetch())
indep = groups[0]
assert indep.is_independent
assert len(indep.current_members) == 13
indep_2 = indep.current_members[2]
assert indep_2.mp_name == "Dumitru Ovidiu-Ioan"
assert indep_2.start_date is None
assert indep_2.end_date is None
psd = groups[1]
assert psd.name == "Grupul parlamentar al Partidului Social Democrat"
assert psd.short_name == "PSD"
assert not psd.is_independent
assert len(psd.current_members) == 165
assert len(psd.former_members) == 2
current_3 = psd.current_members[3]
assert current_3.mp_name == "Itu Cornel"
assert current_3.start_date is None
assert current_3.end_date is None
assert current_3.title == 'Vicelideri'
current_84 = psd.current_members[84]
assert current_84.start_date == date(2013, 9, 18)
former_0 = psd.former_members[0]
assert former_0.mp_name == "Cernea Remus-Florinel"
assert former_0.start_date is None
assert former_0.end_date == date(2013, 5, 21)
ppdd = groups[2]
current_2 = ppdd.current_members[2]
assert current_2.mp_name == "Ciuhodaru Tudor"
former_2 = ppdd.former_members[2]
assert former_2.mp_name == "Chebac Eugen"
assert former_2.start_date is None
assert former_2.end_date == date(2013, 9, 30)
former_4 = ppdd.former_members[4]
assert former_4.start_date == date(2013, 2, 11)
assert former_4.end_date == date(2013, 5, 7)
def groups(
cache_name=None,
throttle=None,
no_commit=False,
):
from mptracker.scraper.groups import GroupScraper, Interval
http_session = create_session(cache_name=cache_name,
throttle=throttle and float(throttle))
group_scraper = GroupScraper(http_session)
mandate_lookup = models.MandateLookup()
mandate_intervals = defaultdict(list)
groups = list(group_scraper.fetch())
independents = groups[0]
assert independents.is_independent
for group in groups[1:] + [independents]:
for member in group.current_members + group.former_members:
(year, chamber, number) = member.mp_ident
assert chamber == 2
mandate = mandate_lookup.find(member.mp_name, year, number)
interval_list = mandate_intervals[mandate]
interval = member.get_interval()
if interval.start is None:
interval = interval._replace(start=TERM_2012_START)
if group.is_independent:
if interval_list:
start = interval_list[-1].end
interval = interval._replace(start=start)
interval_list.append(interval)
interval_list.sort(key=lambda i: i[0])
for mandate, interval_list in mandate_intervals.items():
# make sure interval_list are continuous
new_intervals = []
for interval_one, interval_two in \
zip(interval_list[:-1], interval_list[1:]):
assert interval_one.start < interval_one.end
if interval_one.end < interval_two.start:
interval = Interval(
start=interval_one.end,
end=interval_two.start,
group=independents,
)
new_intervals.append(interval)
elif interval_one.end > interval_two.start:
raise RuntimeError("Overlapping intervals")
interval_list.extend(new_intervals)
interval_list.sort()
mandate_end = mandate.interval.upper
if mandate_end == date.max:
mandate_end = None
if interval_list[-1].end != mandate_end:
logger.warn("Mandate %s ends at %s",
mandate, interval_list[-1].end)
group_patcher = TablePatcher(
models.MpGroup,
models.db.session,
key_columns=['short_name'],
)
with group_patcher.process(remove=True) as add_group:
for group in groups:
record = group.as_dict(['name', 'short_name'])
group.row = add_group(record).row
models.db.session.flush()
membership_patcher = TablePatcher(
models.MpGroupMembership,
models.db.session,
key_columns=['mandate_id', 'mp_group_id', 'interval'],
)
with membership_patcher.process(
autoflush=1000,
remove=True,
) as add_membership:
for mandate, interval_list in mandate_intervals.items():
for interval in interval_list:
row = add_membership({
'mandate_id': mandate.id,
'mp_group_id': interval.group.row.id,
'interval': DateRange(
interval.start or date.min,
interval.end or date.max,
),
}).row
if no_commit:
logger.warn("Rolling back the transaction")
models.db.session.rollback()
else:
models.db.session.commit()
def groups(
cache_name=None,
throttle=None,
no_commit=False,
year='2012',
):
year = int(year)
from mptracker.scraper.groups import GroupScraper, Interval
http_session = create_session(cache_name=cache_name,
throttle=throttle and float(throttle))
group_scraper = GroupScraper(http_session)
mandate_lookup = models.MandateLookup()
mandate_intervals = defaultdict(list)
term_interval = TERM_INTERVAL[year]
groups = list(group_scraper.fetch(year))
independents = groups[0]
assert independents.is_independent
for group in groups[1:] + [independents]:
for member in group.current_members + group.former_members:
(myear, chamber, number) = member.mp_ident
assert chamber == 2
mandate = mandate_lookup.find(member.mp_name, myear, number)
interval_list = mandate_intervals[mandate]
interval = member.get_interval()
if interval.start is None:
interval = interval._replace(start=term_interval.lower)
if interval.end is None:
interval = interval._replace(end=term_interval.upper)
if group.is_independent:
if interval_list:
start = interval_list[-1].end
interval = interval._replace(start=start)
interval_list.append(interval)
interval_list.sort(key=lambda i: i[0])
for mandate, interval_list in mandate_intervals.items():
# make sure interval_list are continuous
new_intervals = []
for interval_one, interval_two in \
zip(interval_list[:-1], interval_list[1:]):
assert interval_one.start < interval_one.end
if interval_one.end < interval_two.start:
interval = Interval(
start=interval_one.end,
end=interval_two.start,
group=independents,
)
new_intervals.append(interval)
elif interval_one.end > interval_two.start:
import pdb; pdb.set_trace()
raise RuntimeError("Overlapping intervals")
interval_list.extend(new_intervals)
interval_list.sort()
mandate_end = mandate.interval.upper
if mandate_end == date.max:
mandate_end = None
if interval_list[-1].end != mandate_end:
logger.warn("Mandate %s ends at %s",
mandate, interval_list[-1].end)
group_patcher = TablePatcher(
models.MpGroup,
models.db.session,
key_columns=['short_name', 'year'],
)
with group_patcher.process(remove=True, filter={'year': year}) as add_group:
for group in groups:
record = group.as_dict(['name', 'short_name', 'year'])
group.row = add_group(record).row
models.db.session.flush()
membership_patcher = TablePatcher(
models.MpGroupMembership,
models.db.session,
key_columns=['mandate_id', 'mp_group_id', 'interval'],
)
current_membership_query = (
models.db.session.query(models.MpGroupMembership.id)
.join(models.MpGroupMembership.mandate)
.filter_by(year=year)
)
remove_membership_ids = set(row.id for row in current_membership_query)
with membership_patcher.process(autoflush=1000) as add_membership:
for mandate, interval_list in mandate_intervals.items():
for interval in interval_list:
res = add_membership({
'mandate_id': mandate.id,
'mp_group_id': interval.group.row.id,
'interval': DateRange(
interval.start or date.min,
interval.end or date.max,
),
})
if not res.is_new:
remove_membership_ids.remove(res.row.id)
if remove_membership_ids:
unseen_items = (
models.MpGroupMembership.query
.filter(models.MpGroupMembership.id.in_(remove_membership_ids))
)
unseen_items.delete(synchronize_session=False)
logger.info("Deleted %d stale memberships", len(remove_membership_ids))
if no_commit:
logger.warn("Rolling back the transaction")
models.db.session.rollback()
else:
models.db.session.commit()