def clean_date(date):
splits = [
"a)",
"b)",
"c)",
"d)",
"e)",
"f)",
"g)",
"h)",
"i)",
" or ",
" to ",
" and ",
"alt DOB:",
"alt DOB",
";",
">>",
]
dates = set()
if isinstance(date, float):
date = str(int(date))
if isinstance(date, datetime):
date = date.date().isoformat()
date = remove_bracketed(date)
if date is None:
return dates
date = date.replace("\n", " ")
for part in multi_split(date, splits):
part = part.strip().strip(",")
if not len(part):
continue
dates.update(h.parse_date(part, FORMATS))
return dates
def parse_entry(context: Context, entry):
entity = context.make("LegalEntity")
if entry.findtext("./type-entry") == "2":
entity = context.make("Person")
entry_id = entry.findtext("number-entry")
entity.id = context.make_slug(entry_id)
sanction = h.make_sanction(context, entity)
sanction.add("program", entry.findtext("./program-entry"))
date_entry = entry.findtext("./date-entry")
if date_entry:
date = datetime.strptime(date_entry, "%Y%m%d")
entity.add("createdAt", date.date())
sanction.add("listingDate", date.date())
sanction.add("startDate", date.date())
for aka in entry.findall("./aka-list"):
h.apply_name(
entity,
name1=aka.findtext("./aka-name1"),
name2=aka.findtext("./aka-name2"),
name3=aka.findtext("./aka-name3"),
tail_name=aka.findtext("./aka-name4"),
alias=aka.findtext("type-aka") != "N",
is_weak=aka.findtext("./quality-aka") == "2",
quiet=True,
)
for node in entry.findall("./title-list"):
entity.add("title", node.text, quiet=True)
for doc in entry.findall("./document-list"):
reg = doc.findtext("./document-reg")
number = doc.findtext("./document-id")
country = doc.findtext("./document-country")
passport = context.make("Passport")
passport.id = context.make_id("Passport", entity.id, reg, number, country)
passport.add("holder", entity)
passport.add("passportNumber", number)
passport.add("summary", reg)
passport.add("country", country)
context.emit(passport)
for doc in entry.findall("./id-number-list"):
entity.add("idNumber", doc.text)
for node in entry.findall("./address-list"):
address = h.make_address(context, full=node.findtext("./address"))
h.apply_address(context, entity, address)
for pob in entry.findall("./place-of-birth-list"):
entity.add_cast("Person", "birthPlace", pob.text)
for dob in entry.findall("./date-of-birth-list"):
date = parse_date(dob.text)
entity.add_cast("Person", "birthDate", date)
for nat in entry.findall("./nationality-list"):
for country in multi_split(nat.text, [";", ","]):
country = remove_bracketed(country)
entity.add("nationality", country, quiet=True)
entity.add("topics", "sanction")
context.emit(entity, target=True)
context.emit(sanction)
def parse_countries(text):
countries = set()
for country in multi_split(text, COUNTRY_SPLIT):
country = remove_bracketed(country)
countries.add(country)
return countries
def parse_entry(context, entry):
entity = context.make("LegalEntity")
if entry.findtext("./type-entry") == "2":
entity = context.make("Person")
entry_id = entry.findtext("number-entry")
entity.id = context.make_slug(entry_id)
sanction = h.make_sanction(context, entity)
sanction.add("program", entry.findtext("./program-entry"))
date_entry = entry.findtext("./date-entry")
if date_entry:
date = datetime.strptime(date_entry, "%Y%m%d")
entity.context["created_at"] = date.isoformat()
sanction.add("startDate", date.date())
for aka in entry.findall("./aka-list"):
first_name = aka.findtext("./aka-name1")
entity.add("firstName", first_name, quiet=True)
second_name = aka.findtext("./aka-name2")
entity.add("secondName", second_name, quiet=True)
third_name = aka.findtext("./aka-name3")
entity.add("middleName", third_name, quiet=True)
last_name = aka.findtext("./aka-name4")
entity.add("lastName", last_name, quiet=True)
name = jointext(first_name, second_name, third_name, last_name)
if aka.findtext("type-aka") == "N":
entity.add("name", name)
else:
if aka.findtext("./quality-aka") == "2":
entity.add("weakAlias", name)
else:
entity.add("alias", name)
for node in entry.findall("./title-list"):
entity.add("title", node.text, quiet=True)
for doc in entry.findall("./document-list"):
reg = doc.findtext("./document-reg")
number = doc.findtext("./document-id")
country = doc.findtext("./document-country")
passport = context.make("Passport")
passport.id = context.make_id("Passport", entity.id, reg, number,
country)
passport.add("holder", entity)
passport.add("passportNumber", number)
passport.add("summary", reg)
passport.add("country", country)
context.emit(passport)
for doc in entry.findall("./id-number-list"):
entity.add("idNumber", doc.text)
for node in entry.findall("./address-list"):
address = h.make_address(context, full=node.findtext("./address"))
h.apply_address(context, entity, address)
for pob in entry.findall("./place-of-birth-list"):
entity.add_cast("Person", "birthPlace", pob.text)
for dob in entry.findall("./date-of-birth-list"):
date = parse_date(dob.text)
entity.add_cast("Person", "birthDate", date)
for nat in entry.findall("./nationality-list"):
for country in multi_split(nat.text, [";", ","]):
country = remove_bracketed(country)
entity.add("nationality", country, quiet=True)
entity.add("topics", "sanction")
context.emit(entity, target=True, unique=True)
context.emit(sanction)