feat: restore registry procurement collection

2026-07-19 17:46:15 +02:00
parent 59769be400
commit c344bef777
14 changed files with 1010 additions and 96 deletions
--- a/tests/apps/parsers/test_eis_registry_client.py
+++ b/tests/apps/parsers/test_eis_registry_client.py
@@ -0,0 +1,124 @@
+from __future__ import annotations
+
+from datetime import date
+from decimal import Decimal
+from urllib.parse import parse_qs, urlsplit
+
+from apps.parsers.clients.base import HTTPClientError
+from apps.parsers.clients.common.eis_registry import EisRegistryProcurementClient
+from apps.parsers.clients.common.schemas import GenericParserItem
+from django.test import SimpleTestCase
+
+
+def _record(*, external_id: str, inn: str, amount: str = "1") -> GenericParserItem:
+    return GenericParserItem(
+        source="procurements_44fz",
+        external_id=external_id,
+        inn=inn,
+        organisation_name="Наименование из ЕИС",
+        title="Закупка",
+        amount=Decimal(amount),
+        payload={"raw": external_id},
+    )
+
+
+class _FakeHttpClient:
+    def __init__(self, pages: dict[int, bytes], *, error_page: int | None = None):
+        self.pages = pages
+        self.error_page = error_page
+        self.urls: list[str] = []
+
+    def download_file(self, url: str, **_kwargs) -> bytes:
+        self.urls.append(url)
+        page = int(parse_qs(urlsplit(url).query)["pageNumber"][0])
+        if page == self.error_page:
+            raise HTTPClientError("page failed")
+        return self.pages[page]
+
+
+class _FakeStructuredClient:
+    def __init__(self, pages: dict[int, bytes], records: dict[bytes, list]):
+        self.http_client = _FakeHttpClient(pages)
+        self.records = records
+
+    def fetch_records(self, *, content: bytes, file_name: str):
+        assert file_name == "results.html"
+        return self.records[content]
+
+
+class EisRegistryProcurementClientTest(SimpleTestCase):
+    def test_fetches_all_pages_filters_inn_and_deduplicates_external_id(self):
+        page_1 = b'<a data-pagenumber="2"></a>'
+        page_2 = b'<a data-pagenumber="2"></a>'
+        structured = _FakeStructuredClient(
+            {1: page_1, 2: page_2},
+            {
+                page_1: [
+                    _record(external_id="same", inn="7701000001", amount="1"),
+                    _record(external_id="foreign", inn="7701000099"),
+                ],
+                page_2: [
+                    _record(external_id="same", inn="7701000001", amount="2"),
+                    _record(external_id="other", inn="7701000001"),
+                ],
+            },
+        )
+        client = EisRegistryProcurementClient(
+            source="procurements_44fz",
+            law="44",
+            _structured_client=structured,
+        )
+
+        records = client.fetch_for_customer(
+            inn="7701000001",
+            organization_name="ОПК Заказчик",
+            organization_id="org-1",
+            today=date(2026, 7, 19),
+        )
+
+        self.assertEqual([row.external_id for row in records], ["same", "other"])
+        self.assertEqual(records[0].amount, Decimal("2"))
+        self.assertTrue(all(row.inn == "7701000001" for row in records))
+        self.assertEqual(records[0].payload["organization_id"], "org-1")
+        query = parse_qs(urlsplit(structured.http_client.urls[0]).query)
+        self.assertEqual(query["fz44"], ["on"])
+        self.assertEqual(query["strictEqual"], ["true"])
+        self.assertEqual(query["publishDateFrom"], ["19.07.2025"])
+        self.assertEqual(query["publishDateTo"], ["19.07.2026"])
+
+    def test_builds_223fz_query_and_accepts_empty_page(self):
+        empty_page = b"<html></html>"
+        structured = _FakeStructuredClient({1: empty_page}, {empty_page: []})
+        client = EisRegistryProcurementClient(
+            source="procurements_223fz",
+            law="223",
+            _structured_client=structured,
+        )
+
+        records = client.fetch_for_customer(
+            inn="7701000001",
+            organization_name="ОПК Заказчик",
+            organization_id="org-1",
+        )
+
+        self.assertEqual(records, [])
+        query = parse_qs(urlsplit(structured.http_client.urls[0]).query)
+        self.assertEqual(query["fz223"], ["on"])
+        self.assertNotIn("fz44", query)
+
+    def test_page_error_is_not_silenced(self):
+        page_1 = b'<a data-pagenumber="2"></a>'
+        structured = _FakeStructuredClient({1: page_1}, {page_1: []})
+        structured.http_client.error_page = 2
+        client = EisRegistryProcurementClient(
+            source="procurements_44fz",
+            law="44",
+            _structured_client=structured,
+        )
+
+        with self.assertRaises(HTTPClientError):
+            client.fetch_for_customer(
+                inn="7701000001",
+                organization_name="ОПК Заказчик",
+                organization_id="org-1",
+            )