def update_trustees(session,
charity_collection,
lower_limit,
upper_limit,
batch_size=10000):
logging.info("Updating trustees")
while lower_limit <= upper_limit:
q = session\
.query(ExtractMainCharity.regno, ExtractTrustee)\
.join(ExtractTrustee, ExtractTrustee.regno==ExtractMainCharity.regno)\
.filter(ExtractMainCharity.regno >= lower_limit)\
.filter(ExtractMainCharity.regno < lower_limit + batch_size)
trustees = {}
for i, x in enumerate(q):
regno = x.ExtractTrustee.regno
if regno not in trustees:
trustees[regno] = []
trustees[regno].append(
stripped_or_none(x.ExtractTrustee.trustee, 'title'))
requests = [
UpdateOne({'ids.GB-CHC': i},
{'$set': {
'trustees.names': trustees[i]
}}) for i in trustees.keys()
]
logging.info(lower_limit)
lower_limit += batch_size
if len(requests) == 0:
continue
try:
result = charity_collection.bulk_write(requests)
# print(lower_limit, lower_limit + batch_size, result.matched_count)
except BulkWriteError as bwe:
logging.error(bwe.details)
def update_beneficiaries(session, charity_collection, lower_limit, upper_limit, batch_size=10000):
logging.info("Updating beneficiaries.")
while lower_limit <= upper_limit:
q = session\
.query(ExtractMainCharity.regno, ExtractClass.id, ExtractClassRef)\
.join(ExtractClass, ExtractClass.regno==ExtractMainCharity.regno)\
.join(ExtractClassRef, ExtractClassRef.classno==ExtractClass.classno)\
.filter(ExtractClass.classno >= 200)\
.filter(ExtractClass.classno < 300)\
.filter(ExtractMainCharity.regno >= lower_limit)\
.filter(ExtractMainCharity.regno < lower_limit + batch_size)
beneficiaries = {}
for i, x in enumerate(q):
regno = x.regno
if regno not in beneficiaries:
beneficiaries[regno] = []
beneficiaries[regno].append({
'id': int(x.ExtractClassRef.classno),
'name': stripped_or_none(x.ExtractClassRef.classtext, 'sentence'),
})
requests = [UpdateOne(
{'ids.GB-CHC': i},
{'$set': {'beneficiaries': beneficiaries[i]}}
) for i in beneficiaries.keys()]
logging.info(lower_limit)
lower_limit += batch_size
if len(requests) == 0:
continue
try:
result = charity_collection.bulk_write(requests)
# print(lower_limit, lower_limit + batch_size, result.matched_count)
except BulkWriteError as bwe:
logging.error(bwe.details)
def create_subsidiary(charity):
c = {}
c['id'] = charity.subno
c['name'] = stripped_or_none(charity.name, 'title')
c['isRegistered'] = charity.orgtype and charity.orgtype.strip() == 'R'
c['governingDoc'] = stripped_or_none(charity.gd)
c['areaOfBenefit'] = stripped_or_none(charity.aob)
c['contact'] = {
'email':
None,
'person':
stripped_or_none(charity.corr, 'title'),
'phone':
stripped_or_none(charity.phone),
'postcode':
stripped_or_none(charity.postcode),
'address': [
stripped_or_none(getattr(charity, add), 'title')
for add in ['add1', 'add2', 'add3', 'add4', 'add5']
if getattr(charity, add) != None
],
}
return c
def create_charity(charity, main_charity):
c = {}
c['regulator'] = 'GB-CHC'
c['ids'] = {
'charityId': 'GB-CHC-%d' % charity.regno,
'GB-CHC': charity.regno,
}
c['name'] = stripped_or_none(charity.name, 'title')
c['isRegistered'] = charity.orgtype and charity.orgtype.strip() == 'R'
c['governingDoc'] = stripped_or_none(charity.gd)
c['areaOfBenefit'] = stripped_or_none(charity.aob, 'title')
c['contact'] = {
'email': stripped_or_none(main_charity.email, 'lower'),
'person': stripped_or_none(charity.corr, 'title'),
'phone': stripped_or_none(charity.phone),
'postcode': stripped_or_none(charity.postcode),
'address': [stripped_or_none(getattr(charity, add), 'title') for add in ['add1', 'add2', 'add3', 'add4', 'add5'] if getattr(charity, add) != None]
}
c['isWelsh'] = boolean_on_value(main_charity.welsh, 'T')
c['trustees'] = {
'incorporated': boolean_on_value(main_charity.trustees, 'T'),
'names': []
}
c['website'] = stripped_or_none(main_charity.web, 'lower')
c['isSchool'] = boolean_on_value(main_charity.grouptype, 'SCH')
c['income'] = {
'latest': {
'date': main_charity.incomedate,
'total': int(main_charity.income) if main_charity.income != None else None,
},
'annual': []
}
c['fyend'] = stripped_or_none(main_charity.fyend)
c['companiesHouseNumber'] = stripped_or_none(main_charity.coyno)
c['areasOfOperation'] = []
c['causes'] = []
c['beneficiaries'] = []
c['operations'] = []
c['subsidiaries'] = []
c['alternativeNames'] = []
c['activities'] = None
return c
def update_aoo(session,
charity_collection,
lower_limit,
upper_limit,
batch_size=10000):
logging.info("Updating areas of operation.")
while lower_limit <= upper_limit:
q = session\
.query(ExtractMainCharity.regno, ExtractCharityAoo.id, ExtractAooRef)\
.join(ExtractCharityAoo, ExtractCharityAoo.regno==ExtractMainCharity.regno)\
.join(ExtractAooRef, and_(ExtractAooRef.aootype==ExtractCharityAoo.aootype, ExtractAooRef.aookey==ExtractCharityAoo.aookey))\
.filter(ExtractMainCharity.regno >= lower_limit)\
.filter(ExtractMainCharity.regno < lower_limit + batch_size)
aoo = {}
for i, x in enumerate(q):
regno = x.regno
if regno not in aoo:
aoo[regno] = []
aootype = stripped_or_none(x.ExtractAooRef.aootype)
scale = {
'A': 'UK Division',
'B': 'Local Authority',
'C': 'Metropolitan County',
'D': 'Country',
'E': 'Continent',
}
parent_aoo_id = None
if aootype in [
'B', 'D'
] and x.ExtractAooRef.master and x.ExtractAooRef.master > 0:
parent_aoo_id = '%s-%d' % (chr(ord(aootype) + 1),
x.ExtractAooRef.master)
aoo[regno].append({
'id':
'%s-%d' % (x.ExtractAooRef.aootype, x.ExtractAooRef.aookey),
'parentId':
parent_aoo_id,
'name':
stripped_or_none(x.ExtractAooRef.aooname, 'title'),
'alternativeName':
stripped_or_none(x.ExtractAooRef.aoosort, 'title'),
'locationType':
scale[aootype],
'isWelsh':
boolean_on_value(x.ExtractAooRef.welsh, 'Y'),
})
requests = [
UpdateOne({'ids.GB-CHC': i},
{'$set': {
'areasOfOperation': aoo[i]
}}) for i in aoo.keys()
]
logging.info(lower_limit)
lower_limit += batch_size
if len(requests) == 0:
continue
try:
result = charity_collection.bulk_write(requests)
# print(lower_limit, lower_limit + batch_size, result.matched_count)
except BulkWriteError as bwe:
logging.error(bwe.details)