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 +1,134 @@
"""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)