feat: normalize streaming FSSP CSV records

This commit is contained in:
Aleksandr Meshchryakov
2026-10-05 15:20:47 +02:00
parent 524e06230f
commit 92290511c2
2 changed files with 346 additions and 0 deletions

View File

@@ -0,0 +1,212 @@
"""Streaming normalization of the two verified FSSP legal-entity CSV schemas."""
from __future__ import annotations
import csv
import re
from collections.abc import Iterator
from datetime import date, datetime
from decimal import Decimal, InvalidOperation
from typing import TextIO
from apps.parsers.registry_snapshots import SnapshotValidationError
from django.utils import timezone
from organizations.source_ingestion import SourceRecordInput
FSSP_SOURCE = "fssp_enforcement_proceedings"
GENERAL_DATASET = "7709576929-iplegallist"
ENDED_DATASET = "7709576929-iplegallistcomplete"
FSSP_DATASETS = (GENERAL_DATASET, ENDED_DATASET)
FSSP_HEADERS = {
"The name of the debtor": "debtor_name",
"Address of the debtor organization": "debtor_legal_address",
"Actual address of the debtor organization": "debtor_actual_address",
"The number of enforcement proceeding": "proceeding_number",
"Date of the institution of proceeding": "initiated_on",
"The number of the consolidated enforcement proceedings": "consolidated_proceeding_number",
"The executive document type": "document_type",
"Date of the executive document": "document_date",
"The number of executive document": "document_number",
"The object of executive documents": "document_subject_raw",
"The object of the execution": "subject",
"Departments of bailiffs": "department_name",
"Address of departments of bailiff": "department_address",
"Debtor TIN": "debtor_inn",
"TIN of the organization collector": "creditor_inn",
}
def valid_legal_inn(value: str) -> bool:
"""Validate a ten-digit legal-entity INN without dropping leading zeroes."""
if not re.fullmatch(r"[0-9]{10}", value):
return False
weights = (2, 4, 10, 3, 5, 9, 4, 6, 8)
return sum(
int(digit) * weight for digit, weight in zip(value, weights)
) % 11 % 10 == int(value[-1])
def _date(value: str | None, warnings: set[str]) -> str | None:
if not value:
return None
for pattern in ("%d.%m.%Y", "%Y-%m-%d"):
try:
return datetime.strptime(value, pattern).date().isoformat()
except ValueError:
continue
warnings.add("unparsed_date")
return None
def _money(value: str | None, warnings: set[str], *, signed: bool) -> Decimal | None:
if not value:
return None
normalized = re.sub(r"\s", "", value).replace(",", ".")
pattern = r"-?[0-9]+(?:\.[0-9]{1,2})?" if signed else r"[0-9]+(?:\.[0-9]{1,2})?"
try:
if not re.fullmatch(pattern, normalized):
raise InvalidOperation
result = Decimal(normalized)
if abs(result) >= Decimal("1000000000000000000"):
raise InvalidOperation
return result
except InvalidOperation:
warnings.add("unparsed_amount")
return None
def _join(*values: str | None) -> str | None:
return "\n".join(value for value in values if value) or None
def _termination(payload: dict, warnings: set[str]) -> None:
raw = payload["termination_raw"]
if not raw:
warnings.add("unparsed_termination")
return
match = re.search(r"[0-9]{2}\.[0-9]{2}\.[0-9]{4}", raw)
if match:
payload["terminated_on"] = _date(match.group(), warnings)
for label, field in (
("ст", "termination_article"),
("ч", "termination_part"),
("п", "termination_point"),
):
match = re.search(rf"\b{label}\.\s*([0-9]+)", raw)
if match:
payload[field] = match.group(1)
if not payload["terminated_on"] or not payload["termination_article"]:
warnings.add("unparsed_termination")
def iter_fssp_records(
handle: TextIO,
*,
dataset: str,
snapshot_date: str | None = None,
fetched_at: str | None = None,
) -> Iterator[SourceRecordInput]:
"""Read one logical row at a time; retain original columns for provenance."""
if dataset not in FSSP_DATASETS:
raise SnapshotValidationError("unknown_fssp_dataset")
fetched_at = fetched_at or timezone.now().isoformat()
comparison_date = date.fromisoformat(snapshot_date or fetched_at[:10])
expected = set(FSSP_HEADERS)
if dataset == GENERAL_DATASET:
expected.update(("Amount due", "Debt remaining balance"))
else:
expected.add("Date, complete IP reason")
reader = csv.reader(handle, strict=True)
try:
headers = next(reader, [])
if headers:
headers[0] = headers[0].lstrip("\ufeff")
if len(headers) != len(expected) or set(headers) != expected:
raise SnapshotValidationError("unknown_fssp_csv_headers")
for row_number, values in enumerate(reader, 1):
if len(values) != len(headers):
raise SnapshotValidationError("invalid_fssp_csv_row_width")
raw = dict(zip(headers, values))
payload = {
field: raw[key].strip() or None for key, field in FSSP_HEADERS.items()
}
warnings: set[str] = set()
for field in ("initiated_on", "document_date"):
payload[field] = _date(payload[field], warnings)
if (
payload["document_date"]
and date.fromisoformat(payload["document_date"]) > comparison_date
):
warnings.add("future_document_date")
debt = _money(raw.get("Debt remaining balance"), warnings, signed=True)
claim = _money(raw.get("Amount due"), warnings, signed=False)
if debt is not None and debt < 0:
warnings.add("negative_debt_balance")
if debt is not None and claim is not None and debt > claim:
warnings.add("debt_exceeds_claim_amount")
payload.update(
{
"origin_kind": "open_data_csv",
"queried_inn": None,
"debt_amount": format(debt, ".2f") if debt is not None else None,
"claim_amount": format(claim, ".2f") if claim is not None else None,
"debt_currency": "RUB"
if debt is not None or claim is not None
else None,
"document_issuer": None,
"termination_raw": raw.get("Date, complete IP reason", "").strip()
or None,
"terminated_on": None,
"termination_reason": None,
"termination_article": None,
"termination_part": None,
"termination_point": None,
"service_text": None,
"service_links": [],
"bailiff_raw": None,
"bailiff_name": None,
"bailiff_phones": [],
"source_url": f"https://opendata.fssp.gov.ru/{dataset}",
"fetched_at": fetched_at,
"raw_columns": raw,
"dataset": dataset,
"source_row_number": row_number,
}
)
payload["debtor_raw"] = _join(
payload["debtor_name"],
payload["debtor_inn"],
payload["debtor_legal_address"],
)
payload["proceeding_raw"] = _join(
payload["proceeding_number"], payload["initiated_on"]
)
payload["document_raw"] = _join(
payload["document_type"],
payload["document_date"],
payload["document_number"],
payload["document_subject_raw"],
)
payload["subject_raw"] = _join(payload["subject"], payload["debt_amount"])
payload["department_raw"] = _join(
payload["department_name"], payload["department_address"]
)
if dataset == ENDED_DATASET:
_termination(payload, warnings)
payload["parse_warnings"] = sorted(warnings)
inn = payload["debtor_inn"] or ""
number = payload["proceeding_number"] or ""
yield SourceRecordInput(
external_id=f"{inn}:{number}",
title=number[:255],
organization_name=payload["debtor_name"] or "",
record_type="enforcement_proceeding",
inn=inn,
record_date=payload["initiated_on"] or "",
amount=debt,
status="ended" if dataset == ENDED_DATASET else "unknown",
url=payload["source_url"],
payload=payload,
)
except (csv.Error, UnicodeError) as exc:
raise SnapshotValidationError("invalid_fssp_csv") from exc

View File

@@ -0,0 +1,134 @@
"""FSSP CSV normalization without network or business data."""
import csv
import importlib
from decimal import Decimal
from io import StringIO
from pathlib import Path
import pytest
from apps.parsers.registry_snapshots import SnapshotValidationError
FIXTURES = Path(__file__).parents[2] / "fixtures" / "fssp"
GENERAL = "7709576929-iplegallist"
ENDED = "7709576929-iplegallistcomplete"
def records(text, dataset=GENERAL):
try:
module = importlib.import_module("apps.parsers.fssp_csv")
except ModuleNotFoundError:
pytest.fail("iter_fssp_records is not implemented")
return module.iter_fssp_records(
StringIO(text),
dataset=dataset,
snapshot_date="2026-10-05",
fetched_at="2026-10-05T08:00:00+00:00",
)
def fixture_text(name="general.csv", **changes):
rows = list(csv.reader(StringIO((FIXTURES / name).read_text())))
for key, value in changes.items():
rows[1][rows[0].index(key)] = value
handle = StringIO()
csv.writer(handle, lineterminator="\n").writerows(rows)
return handle.getvalue()
def test_signed_debt_and_multiline_preserved():
row = next(records(fixture_text()))
assert row.amount == Decimal("-12.50")
assert row.payload["debt_amount"] == "-12.50"
assert row.payload["claim_amount"] == "250.00"
assert row.payload["debtor_actual_address"] == "Москва, офис 2\nэтаж 3"
assert row.payload["origin_kind"] == "open_data_csv"
assert row.payload["queried_inn"] is None
assert row.payload["document_issuer"] is None
assert row.payload["document_subject_raw"] == "Взыскание сбора"
assert len(row.payload["raw_columns"]) == 17
assert row.record_date == "2026-10-01"
@pytest.mark.parametrize(
"value,want", [("0", "0.00"), ("", None), ("abc", None), ("1 234,50", "1234.50")]
)
def test_money_preserves_zero_and_missing(value, want):
row = next(records(fixture_text(**{"Debt remaining balance": value})))
assert row.payload["debt_amount"] == want
assert row.amount == (Decimal(want) if want is not None else None)
assert ("unparsed_amount" in row.payload["parse_warnings"]) == (value == "abc")
def test_debt_exceeds_claim_warning():
row = next(records(fixture_text(**{"Debt remaining balance": "251"})))
assert row.amount == Decimal("251.00")
assert "debt_exceeds_claim_amount" in row.payload["parse_warnings"]
def test_future_document_date_is_preserved():
row = next(
records(fixture_text(**{"Date of the executive document": "01.01.2099"}))
)
assert row.payload["document_date"] == "2099-01-01"
assert "future_document_date" in row.payload["parse_warnings"]
@pytest.mark.parametrize("raw,want", [("", None), ("bad", None), ("31.02.2026", None)])
def test_missing_and_invalid_dates(raw, want):
row = next(records(fixture_text(**{"Date of the institution of proceeding": raw})))
assert row.payload["initiated_on"] is want
assert row.record_date == ""
assert ("unparsed_date" in row.payload["parse_warnings"]) == bool(raw)
def test_reordered_headers_preserve_mapping():
rows = list(csv.reader(StringIO(fixture_text())))
handle = StringIO()
csv.writer(handle).writerows([list(reversed(row)) for row in rows])
row = next(records(handle.getvalue()))
assert row.inn == "7707083893"
assert row.payload["proceeding_number"] == "123/26/77001-ИП"
assert row.amount == Decimal("-12.50")
@pytest.mark.parametrize(
"kind", ["unknown", "duplicate", "missing", "truncated", "unclosed"]
)
def test_unknown_or_corrupt_structure_rejected(kind):
rows = list(csv.reader(StringIO(fixture_text())))
if kind == "unknown":
rows[0][0] = "new unknown column"
elif kind == "duplicate":
rows[0][0] = rows[0][1]
elif kind == "missing":
rows = [row[:-1] for row in rows]
elif kind == "truncated":
rows[1] = rows[1][:-1]
handle = StringIO()
csv.writer(handle).writerows(rows)
text = handle.getvalue() if kind != "unclosed" else handle.getvalue() + '"unclosed'
with pytest.raises(SnapshotValidationError):
list(records(text))
def test_ended_dataset_has_distinct_mapping_and_no_fake_debt():
row = next(records(fixture_text("ended.csv"), dataset=ENDED))
assert row.amount is None
assert row.payload["claim_amount"] is None
assert row.status == "ended"
assert row.payload["terminated_on"] == "2026-10-02"
assert row.payload["termination_article"] == "46"
assert row.payload["termination_part"] == "1"
assert row.payload["termination_point"] == "3"
assert len(row.payload["raw_columns"]) == 16
def test_iterator_does_not_read_next_row_until_requested():
rows = list(csv.reader(StringIO(fixture_text())))
handle = StringIO()
csv.writer(handle).writerows(rows)
iterator = records(handle.getvalue() + '"unclosed')
assert next(iterator).inn == "7707083893"
with pytest.raises(SnapshotValidationError):
next(iterator)