def committees(
cache_name=None,
throttle=None,
no_commit=False,
):
from mptracker.scraper.committees import CommitteeScraper
patcher = TablePatcher(
models.MpCommittee,
models.db.session,
key_columns=['chamber_id', 'cdep_id'],
)
http_session = create_session(
cache_name=cache_name,
throttle=throttle and float(throttle),
)
scraper = CommitteeScraper(http_session)
with patcher.process(autoflush=1000, remove=True) as add:
for committee in scraper.fetch_committees():
add(committee.as_dict(['chamber_id', 'cdep_id', 'name']))
if no_commit:
logger.warn("Rolling back the transaction")
models.db.session.rollback()
else:
models.db.session.commit()
def load(name, include_columns=None, create=True, remove=False):
if include_columns:
include_columns = set(include_columns.split(','))
def filter_record(r):
return {k: r[k] for k in r if k in include_columns}
else:
filter_record = lambda r: r
loader = TableLoader(name)
patcher = TablePatcher(loader.model, db.session, key_columns=['id'])
records = (filter_record(loader.decode_dict(flask.json.loads(line)))
for line in sys.stdin)
patcher.update(records, create=create, remove=remove)
def committee_summaries(year=2013):
from mptracker.scraper.committee_summaries import SummaryScraper
patcher = TablePatcher(models.CommitteeSummary,
models.db.session,
key_columns=['pdf_url'])
summary_scraper = SummaryScraper(get_cached_session(),
get_cached_session('question-pdf'))
records = summary_scraper.fetch_summaries(year, get_pdf_text=True)
patcher.update(records)
def committee_summaries(year=2013):
from mptracker.scraper.committee_summaries import SummaryScraper
patcher = TablePatcher(models.CommitteeSummary,
models.db.session,
key_columns=['pdf_url'])
summary_scraper = SummaryScraper(get_cached_session(),
get_cached_session('question-pdf'))
records = summary_scraper.fetch_summaries(year, get_pdf_text=True)
patcher.update(records)
models.db.session.commit()
def load(name,
include_columns=None,
create=True,
remove=False,
_file=sys.stdin):
if include_columns:
include_columns = set(include_columns.split(','))
def filter_record(r):
return {k: r[k] for k in r if k in include_columns}
else:
filter_record = lambda r: r
loader = TableLoader(name)
patcher = TablePatcher(loader.model, db.session, key_columns=['id'])
records = (filter_record(loader.decode_dict(flask.json.loads(line)))
for line in _file)
patcher.update(records, create=create, remove=remove)
db.session.commit()
def questions(year='2013'):
from mptracker.scraper.questions import QuestionScraper
known_urls = set(q.url for q in models.Question.query)
def skip_question(url):
return url in known_urls
questions_scraper = QuestionScraper(session=create_session(throttle=0.5),
skip=skip_question)
mandate_lookup = models.MandateLookup()
question_patcher = TablePatcher(models.Question,
models.db.session,
key_columns=['number', 'date'])
with question_patcher.process() as add:
for question in questions_scraper.run(int(year)):
name, person_year, person_number = question.pop('person')
mandate = mandate_lookup.find(name, person_year, person_number)
question['mandate_id'] = mandate.id
question['addressee'] = '; '.join(question['addressee'])
add(question)
def people(year='2012'):
from mptracker.scraper.people import PersonScraper
patcher = TablePatcher(models.Person,
models.db.session,
key_columns=['cdep_id'])
def get_people():
person_scraper = PersonScraper(get_cached_session())
for row in person_scraper.fetch_people(year):
county_name = row.pop('county_name')
if county_name:
ok_name = fix_local_chars(county_name.title())
if ok_name == "Bistrița-Năsăud":
ok_name = "Bistrița Năsăud"
county = models.County.query.filter_by(name=ok_name).first()
if county is None:
logger.warn("Can't match county name %r", ok_name)
else:
row['county'] = county
yield row
patcher.update(get_people())
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 votes(
start=None,
days=1,
cache_name=None,
throttle=None,
no_commit=False,
autoanalyze=False,
):
from mptracker.scraper.votes import VoteScraper
if start is None:
start = models.db.session.execute(
'select date from voting_session '
'order by date desc limit 1').scalar() + ONE_DAY
else:
start = parse_date(start)
days = int(days)
http_session = create_session(cache_name=cache_name,
throttle=throttle and float(throttle))
vote_scraper = VoteScraper(http_session)
voting_session_patcher = TablePatcher(
models.VotingSession,
models.db.session,
key_columns=['cdeppk'],
)
vote_patcher = TablePatcher(
models.Vote,
models.db.session,
key_columns=['voting_session_id', 'mandate_id'],
)
proposal_ids = {p.cdeppk_cdep: p.id for p in models.Proposal.query}
mandate_lookup = models.MandateLookup()
new_voting_session_list = []
with voting_session_patcher.process() as add_voting_session:
with vote_patcher.process() as add_vote:
for delta in range(days):
the_date = start + ONE_DAY * delta
if the_date >= date.today():
# don't scrape today, maybe voting is not done yet!
break
logger.info("Scraping votes from %s", the_date)
for voting_session in vote_scraper.scrape_day(the_date):
record = model_to_dict(
voting_session,
['cdeppk', 'subject', 'subject_html'],
)
record['date'] = the_date
proposal_cdeppk = voting_session.proposal_cdeppk
record['proposal_id'] = (proposal_ids.get(proposal_cdeppk)
if proposal_cdeppk else None)
record['final'] = bool("vot final" in
record['subject'].lower())
vs = add_voting_session(record).row
if vs.id is None:
models.db.session.flush()
new_voting_session_list.append(vs.id)
for vote in voting_session.votes:
record = model_to_dict(vote, ['choice'])
record['voting_session_id'] = vs.id
mandate = mandate_lookup.find(
vote.mandate_name,
vote.mandate_year,
vote.mandate_number,
)
record['mandate_id'] = mandate.id
add_vote(record)
if no_commit:
logger.warn("Rolling back the transaction")
models.db.session.rollback()
else:
models.db.session.commit()
if autoanalyze:
from mptracker.votes import calculate_voting_session_loyalty
logger.info("Scheduling %d jobs", len(new_voting_session_list))
for voting_session_id in new_voting_session_list:
calculate_voting_session_loyalty.delay(voting_session_id)
def import_person_xls(xls_path):
from mptracker.scraper.person_xls import read_person_xls
mandate_lookup = models.MandateLookup()
people_data = []
committees = {}
committee_memberships = []
groups = {}
group_memberships = []
mandate_patcher = TablePatcher(models.Mandate,
models.db.session,
key_columns=['year', 'cdep_number'])
with mandate_patcher.process() as add:
for record in read_person_xls(xls_path):
mandate = mandate_lookup.find(record.pop('name'), record['year'],
record['cdep_number'])
person_data = record.pop('person_data')
person_data['id'] = mandate.person_id
people_data.append(person_data)
mandate_committees = record.pop('committees')
mp_group = record.pop('mp_group')
mandate = add(record).row
for data in mandate_committees:
committees[data['name']] = None
committee_memberships.append(
(mandate.id, data['name'], data['role']))
groups[mp_group['short_name']] = None
group_memberships.append(
(mandate.id, mp_group['short_name'], mp_group['role']))
person_patcher = TablePatcher(models.Person,
models.db.session,
key_columns=['id'])
with person_patcher.process() as add:
for person_data in people_data:
add(person_data)
committee_patcher = TablePatcher(models.MpCommittee,
models.db.session,
key_columns=['name'])
with committee_patcher.process() as add:
for name in list(committees):
mp_committee = add({'name': name}).row
committees[name] = mp_committee.id
committee_membership_patcher = TablePatcher(models.MpCommitteeMembership,
models.db.session, key_columns=['mandate_id', 'mp_committee_id'])
with committee_membership_patcher.process() as add:
for mandate_id, name, role in committee_memberships:
add({
'mandate_id': mandate_id,
'mp_committee_id': committees[name],
'role': role,
})
mp_group_patcher = TablePatcher(models.MpGroup,
models.db.session,
key_columns=['short_name'])
with mp_group_patcher.process() as add:
for short_name in list(groups):
mp_group = add({'short_name': short_name}).row
groups[short_name] = mp_group.id
mp_group_membership_patcher = TablePatcher(models.MpGroupMembership,
models.db.session, key_columns=['mandate_id', 'mp_group_id'])
with mp_group_membership_patcher.process() as add:
for mandate_id, name, role in group_memberships:
add({
'mandate_id': mandate_id,
'mp_group_id': groups[name],
'role': role,
})
models.db.session.commit()
def transcripts(start=None, n_sessions=1, cache_name=None, throttle=None):
from mptracker.scraper.transcripts import TranscriptScraper
if start is None:
max_serial = models.db.session.execute(
'select serial from transcript_chapter '
'order by serial desc limit 1').scalar()
start = int(max_serial.split('/')[0]) + 1
cdeppk = int(start) - 1
n_sessions = int(n_sessions)
transcript_scraper = TranscriptScraper(
session=create_session(cache_name=cache_name,
throttle=throttle and float(throttle)))
mandate_lookup = models.MandateLookup()
transcript_patcher = TablePatcher(models.Transcript,
models.db.session,
key_columns=['serial'])
with transcript_patcher.process() as add:
while n_sessions > 0:
n_sessions -= 1
cdeppk += 1
logger.info("Fetching session %s", cdeppk)
session_data = transcript_scraper.fetch_session(cdeppk)
if session_data is None:
logger.info("No content")
continue
for chapter in session_data.chapters:
chapter_row = (models.TranscriptChapter.query
.filter_by(serial=chapter.serial)
.first())
if chapter_row is None:
chapter_row = models.TranscriptChapter(
serial=chapter.serial)
models.db.session.add(chapter_row)
models.db.session.flush()
chapter_row.date = session_data.date
chapter_row.headline = chapter.headline
for paragraph in chapter.paragraphs:
if paragraph['mandate_chamber'] != 2:
continue
try:
mandate = mandate_lookup.find(
paragraph['speaker_name'],
paragraph['mandate_year'],
paragraph['mandate_number'])
except models.LookupError as e:
logger.warn("at %s %s", paragraph['serial'], e)
continue
transcript_data = {
'chapter_id': chapter_row.id,
'text': paragraph['text'],
'serial': paragraph['serial'],
'mandate_id': mandate.id,
}
add(transcript_data)
models.db.session.commit()
def proposals(
cache_name=None,
throttle=None,
autoanalyze=False,
):
from mptracker.scraper.proposals import ProposalScraper
from mptracker.proposals import ocr_proposal
from mptracker.policy import calculate_proposal
proposal_scraper = ProposalScraper(create_session(
cache_name=cache_name,
throttle=float(throttle) if throttle else None))
def cdep_id(mandate):
return (mandate.year, mandate.cdep_number)
by_cdep_id = {cdep_id(m): m
for m in models.Mandate.query
if m.year == 2012}
id_cdeppk_cdep = {}
id_cdeppk_senate = {}
for proposal in models.Proposal.query:
if proposal.cdeppk_cdep:
id_cdeppk_cdep[proposal.cdeppk_cdep] = proposal.id
if proposal.cdeppk_senate:
id_cdeppk_senate[proposal.cdeppk_senate] = proposal.id
chamber_by_slug = {c.slug: c for c in models.Chamber.query}
proposals = proposal_scraper.fetch_from_mp_pages(set(by_cdep_id.keys()))
all_activity = defaultdict(list)
for item in models.ProposalActivityItem.query:
all_activity[item.proposal_id].append(item)
proposal_patcher = TablePatcher(models.Proposal,
models.db.session,
key_columns=['id'])
activity_patcher = TablePatcher(models.ProposalActivityItem,
models.db.session,
key_columns=['id'])
sp_updates = sp_added = sp_removed = 0
changed = []
seen = []
with proposal_patcher.process(autoflush=1000, remove=True) as add_proposal:
with activity_patcher.process(autoflush=1000, remove=True) \
as add_activity:
for prop in proposals:
record = model_to_dict(prop, ['cdeppk_cdep', 'cdeppk_senate',
'decision_chamber', 'url', 'title', 'date', 'number_bpi',
'number_cdep', 'number_senate', 'proposal_type',
'pdf_url'])
slug = prop.decision_chamber
if slug:
record['decision_chamber'] = chamber_by_slug[slug]
idc = id_cdeppk_cdep.get(prop.cdeppk_cdep)
ids = id_cdeppk_senate.get(prop.cdeppk_senate)
if idc and ids and idc != ids:
logger.warn("Two different records for the same proposal: "
"(%s, %s). Removing the 2nd.", idc, ids)
models.db.session.delete(models.Proposal.query.get(ids))
ids = None
record['id'] = idc or ids or models.random_uuid()
result = add_proposal(record)
row = result.row
if result.is_changed:
changed.append(row)
seen.append(row)
new_people = set(by_cdep_id[ci] for ci in prop.sponsorships)
existing_sponsorships = {sp.mandate: sp
for sp in row.sponsorships}
to_remove = set(existing_sponsorships) - set(new_people)
to_add = set(new_people) - set(existing_sponsorships)
if to_remove:
logger.info("Removing sponsors %s: %r", row.id,
[cdep_id(m) for m in to_remove])
sp_removed += 1
for m in to_remove:
sp = existing_sponsorships[m]
models.db.session.delete(sp)
if to_add:
logger.info("Adding sponsors %s: %r", row.id,
[cdep_id(m) for m in to_add])
sp_added += 1
for m in to_add:
row.sponsorships.append(models.Sponsorship(mandate=m))
if to_remove or to_add:
sp_updates += 1
db_activity = all_activity[row.id]
db_activity.sort(key=lambda a: a.order)
act_fields = lambda r: (r.date, r.location)
if ([act_fields(r) for r in db_activity] !=
[act_fields(r) for r in prop.activity[:len(db_activity)]]):
logger.warn("History doesn't match for %s, "
"%d items will be removed",
row.id,len(db_activity))
db_activity = []
for n, ac in enumerate(prop.activity):
record = model_to_dict(ac, ['date', 'location', 'html'])
record['proposal_id'] = row.id
record['order'] = n
if n < len(db_activity):
item = db_activity[n]
record['id'] = item.id
assert item.date == record['date']
assert item.location == record['location']
assert item.order == record['order']
else:
record['id'] = models.random_uuid()
add_activity(record)
models.db.session.commit()
logger.info("Updated sponsorship for %d proposals (+%d, -%d)",
sp_updates, sp_added, sp_removed)
if autoanalyze:
logger.info("Scheduling analysis jobs for %d proposals", len(changed))
for proposal in changed:
if proposal.pdf_url:
ocr_proposal.delay(proposal.id, autoanalyze=True)
logger.info("Scheduling policy jobs for %d proposals", len(seen))
for proposal in seen:
if proposal.policy_domain_id is None:
calculate_proposal.delay(proposal.id)
def proposals(dry_run=False):
from mptracker.scraper.proposals import ProposalScraper
proposal_scraper = ProposalScraper(create_session(cache_name='page-cache',
throttle=0.5))
def cdep_id(mandate):
return (mandate.year, mandate.cdep_number)
by_cdep_id = {cdep_id(m): m
for m in models.Mandate.query
if m.year == 2012}
chamber_by_slug = {c.slug: c for c in models.Chamber.query}
proposals = proposal_scraper.fetch_from_mp_pages(set(by_cdep_id.keys()))
proposal_patcher = TablePatcher(models.Proposal,
models.db.session,
key_columns=['combined_id'])
sp_updates = sp_added = sp_removed = 0
with proposal_patcher.process(autoflush=1000, remove=True) as add:
for record in proposals:
if 'decision_chamber' in record:
slug = record.pop('decision_chamber')
record['decision_chamber'] = chamber_by_slug[slug]
sponsorships = record.pop('_sponsorships')
url = record['url']
result = add(record)
row = result.row
new_people = set(by_cdep_id[ci] for ci in sponsorships)
existing_sponsorships = {sp.mandate: sp for sp in row.sponsorships}
to_remove = set(existing_sponsorships) - set(new_people)
to_add = set(new_people) - set(existing_sponsorships)
if to_remove:
logger.info("Removing sponsors %s: %r", row.combined_id,
[cdep_id(m) for m in to_remove])
sp_removed += 1
for m in to_remove:
sp = existing_sponsorships[m]
models.db.session.delete(sp)
if to_add:
logger.info("Adding sponsors %s: %r", row.combined_id,
[cdep_id(m) for m in to_add])
sp_added += 1
for m in to_add:
row.sponsorships.append(models.Sponsorship(mandate=m))
if to_remove or to_add:
sp_updates += 1
if dry_run:
models.db.session.rollback()
logger.info("Updated sponsorship for %d proposals (+%d, -%d)",
sp_updates, sp_added, sp_removed)
def patcher_twin_key(db_app):
from mptracker.common import TablePatcher
return TablePatcher(Thing, db.session, key_columns=['code', 'number'])
def questions(
year='2013',
reimport_existing=False,
cache_name=None,
throttle=None,
autoanalyze=False,
):
from mptracker.scraper.questions import QuestionScraper
from mptracker.questions import ocr_question
from mptracker.policy import calculate_question
if reimport_existing:
known_urls = set()
else:
known_urls = set(q.url for q in models.Question.query)
def skip_question(url):
return url in known_urls
http_session = create_session(cache_name=cache_name,
throttle=throttle and float(throttle),
counters=True)
questions_scraper = QuestionScraper(session=http_session,
skip=skip_question)
mandate_lookup = models.MandateLookup()
question_patcher = TablePatcher(models.Question,
models.db.session,
key_columns=['number', 'date'])
new_ask_rows = 0
changed = []
with question_patcher.process() as add:
for question in questions_scraper.run(int(year)):
person_list = question.pop('person')
question['addressee'] = '; '.join(question['addressee'])
result = add(question)
q = result.row
old_asked = {ask.mandate_id: ask for ask in q.asked}
for name, person_year, person_number in person_list:
mandate = mandate_lookup.find(name, person_year, person_number)
if mandate.id in old_asked:
old_asked.pop(mandate.id)
else:
ask = models.Ask(mandate=mandate)
q.asked.append(ask)
ask.set_meta('new', True)
logger.info("Adding ask for %s: %s", q, mandate)
new_ask_rows += 1
if result.is_changed:
changed.append(q)
assert not old_asked
models.db.session.commit()
if new_ask_rows:
logger.info("Added %d ask records", new_ask_rows)
counters = http_session.counters
logger.info("HTTP: %d kb in %s requests, %.2f seconds",
counters['bytes'] / 1024, counters['requests'],
counters['download_time'].total_seconds())
if autoanalyze:
logger.info("Scheduling jobs for %d questions", len(changed))
for question in changed:
if question.pdf_url:
ocr_question.delay(question.id, autoanalyze=True)
if question.policy_domain_id is None:
calculate_question.delay(question.id)
def people(
year='2012',
cache_name=None,
throttle=None,
no_commit=False,
):
from mptracker.scraper.people import MandateScraper
http_session = create_session(
cache_name=cache_name,
throttle=throttle and float(throttle),
)
mandate_scraper = MandateScraper(http_session)
mandate_patcher = TablePatcher(
models.Mandate,
models.db.session,
key_columns=['year', 'cdep_number'],
)
with mandate_patcher.process() as add_mandate:
for mandate in mandate_scraper.fetch(year):
row = mandate.as_dict([
'year',
'cdep_number',
'minority',
'college',
'constituency',
])
if year == '2012':
end_date = mandate.end_date or date.max
row['interval'] = DateRange(TERM_2012_START, end_date)
person = (
models.Person.query
.filter_by(name=mandate.person_name)
.first())
if person is None:
raise RuntimeError("Can't find person named %r"
% mandate.person_name)
row['person_id'] = person.id
if not mandate.minority:
county = (
models.County.query
.filter_by(name=mandate.county_name)
.first())
if county is None:
raise RuntimeError("Can't match county name %r"
% mandate.county_name)
row['county'] = county
add_mandate(row)
if no_commit:
logger.warn("Rolling back the transaction")
models.db.session.rollback()
else:
models.db.session.commit()
def votes(
start=None,
n_days=1,
cache_name=None,
throttle=None,
):
from mptracker.scraper.votes import VoteScraper
if start is None:
start = models.db.session.execute(
'select date from voting_session '
'order by date desc limit 1').scalar() + ONE_DAY
else:
start = parse_date(start)
n_days = int(n_days)
http_session = create_session(cache_name=cache_name,
throttle=throttle and float(throttle))
vote_scraper = VoteScraper(http_session)
voting_session_patcher = TablePatcher(
models.VotingSession,
models.db.session,
key_columns=['cdeppk'],
)
vote_patcher = TablePatcher(
models.Vote,
models.db.session,
key_columns=['voting_session_id', 'mandate_id'],
)
proposal_ids = {p.cdeppk_cdep: p.id for p in models.Proposal.query}
mandate_lookup = models.MandateLookup()
with voting_session_patcher.process() as add_voting_session:
with vote_patcher.process() as add_vote:
for delta in range(n_days):
the_date = start + ONE_DAY * delta
if the_date >= date.today():
# don't scrape today, maybe voting is not done yet!
break
logger.info("Scraping votes from %s", the_date)
for voting_session in vote_scraper.scrape_day(the_date):
record = model_to_dict(
voting_session,
['cdeppk', 'subject', 'subject_html'],
)
record['date'] = the_date
proposal_cdeppk = voting_session.proposal_cdeppk
record['proposal_id'] = (proposal_ids.get(proposal_cdeppk)
if proposal_cdeppk else None)
vs = add_voting_session(record).row
if vs.id is None:
models.db.session.flush()
for vote in voting_session.votes:
record = model_to_dict(vote, ['choice'])
record['voting_session_id'] = vs.id
mandate = mandate_lookup.find(
vote.mandate_name,
vote.mandate_year,
vote.mandate_number,
)
record['mandate_id'] = mandate.id
add_vote(record)
models.db.session.commit()
