feat: expand platform APIs, sources, and test coverage

tests/apps/parsers/test_clients.py

@@ -13,7 +13,12 @@ from apps.parsers.clients.base import (
 )
 from apps.parsers.clients.minpromtorg.industrial import IndustrialProductionClient
 from apps.parsers.clients.minpromtorg.manufactures import ManufacturesClient
-from apps.parsers.clients.minpromtorg.schemas import IndustrialCertificate, Manufacturer
+from apps.parsers.clients.minpromtorg.products import IndustrialProductsClient
+from apps.parsers.clients.minpromtorg.schemas import (
+    IndustrialCertificate,
+    IndustrialProduct,
+    Manufacturer,
+)
 from apps.parsers.clients.proverki import ProverkiClient
 from apps.parsers.clients.proverki.schemas import Inspection
 from django.test import TestCase, tag
@@ -23,6 +28,7 @@ from tests.utils import Response, TestHTTPServer
 from tests.utils.fixtures import (
     build_minpromtorg_certificates_excel,
     build_minpromtorg_manufacturers_excel,
+    build_minpromtorg_products_excel,
     build_proverki_xml,
     fake,
 )
@@ -43,6 +49,10 @@ def _proxy_address() -> str:
     return f"http://{fake.ipv4()}:{fake.port_number()}"
 
 
+def _digits(length: int) -> str:
+    return "".join(str(fake.random_int(0, 9)) for _ in range(length))
+
+
 class _RaisingAdapter(BaseAdapter):
     def __init__(self, exc: Exception) -> None:
         super().__init__()
@@ -440,6 +450,126 @@ class ManufacturesClientTest(TestCase):
         self.assertEqual(result.address, "")
 
 
+class IndustrialProductsClientTest(TestCase):
+    """Tests for IndustrialProductsClient."""
+
+    def test_client_initialization(self):
+        client = IndustrialProductsClient()
+        self.assertIsNone(client.proxies)
+        self.assertEqual(client.host, "minpromtorg.gov.ru")
+
+    def test_fetch_products_success(self):
+        excel_bytes, rows = build_minpromtorg_products_excel(count=5)
+        date_str = fake.date_between(start_date="-30d", end_date="today").strftime(
+            "%Y%m%d"
+        )
+        file_name = f"industrial_products_{date_str}.xlsx"
+
+        with TestHTTPServer() as server:
+            server.add_json(
+                "/api/kss-document-preview",
+                {
+                    "data": [
+                        {
+                            "name": IndustrialProductsClient().query,
+                            "files": [
+                                {"name": file_name, "url": f"/files/{file_name}"}
+                            ],
+                        }
+                    ]
+                },
+            )
+            server.add_bytes(
+                f"/files/{file_name}",
+                excel_bytes,
+                content_type=(
+                    "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+                ),
+            )
+
+            client = IndustrialProductsClient(
+                host=_host_from_base_url(server.base_url),
+                scheme="http",
+                http_adapter=server.adapter,
+            )
+            products = client.fetch_products()
+
+        self.assertEqual(len(products), len(rows))
+        self.assertIsInstance(products[0], IndustrialProduct)
+        self.assertSetEqual(
+            {product.registry_number for product in products},
+            {row.registry_number for row in rows},
+        )
+
+    def test_fetch_products_no_files(self):
+        with TestHTTPServer() as server:
+            server.add_json("/api/kss-document-preview", {"data": []})
+            client = IndustrialProductsClient(
+                host=_host_from_base_url(server.base_url),
+                scheme="http",
+                http_adapter=server.adapter,
+            )
+            products = client.fetch_products()
+
+        self.assertEqual(products, [])
+
+    def test_get_latest_file_url_falls_back_to_excel_file(self):
+        client = IndustrialProductsClient()
+        files = [
+            {"name": "readme.txt", "url": "/files/readme.txt"},
+            {"name": "registry.xlsx", "url": "/files/registry.xlsx"},
+        ]
+
+        url = client._get_latest_file_url(files)
+        self.assertEqual(url, "https://minpromtorg.gov.ru/files/registry.xlsx")
+
+    def test_parse_row_valid(self):
+        client = IndustrialProductsClient()
+        header_map = {
+            "full_organisation_name": 0,
+            "ogrn": 1,
+            "inn": 2,
+            "registry_number": 3,
+            "product_name": 4,
+            "product_model": 5,
+            "okpd2_code": 6,
+            "tnved_code": 7,
+            "regulatory_document": 8,
+        }
+        row = (
+            fake.company(),
+            _digits(13),
+            _digits(10),
+            f"MPP-{_digits(8)}",
+            fake.sentence(nb_words=4),
+            fake.bothify(text="MODEL-###"),
+            "25.11",
+            _digits(10),
+            fake.sentence(nb_words=5),
+        )
+
+        result = client._parse_row(row, header_map)
+
+        self.assertIsInstance(result, IndustrialProduct)
+        self.assertEqual(result.registry_number, row[3])
+        self.assertEqual(result.product_name, row[4])
+
+    def test_parse_row_without_required_fields(self):
+        client = IndustrialProductsClient()
+        header_map = {
+            "full_organisation_name": 0,
+            "ogrn": 1,
+            "inn": 2,
+            "registry_number": 3,
+            "product_name": 4,
+        }
+
+        result = client._parse_row(
+            (fake.company(), _digits(13), _digits(10), "", ""), header_map
+        )
+        self.assertIsNone(result)
+
+
 @tag("integration", "slow")
 class IndustrialProductionClientIntegrationTest(TestCase):
     """Integration test using local HTTP server instead of external API."""
@@ -516,6 +646,44 @@ class ManufacturesClientIntegrationTest(TestCase):
         self.assertEqual(len(manufacturers), len(rows))
 
 
+@tag("integration", "slow")
+class IndustrialProductsClientIntegrationTest(TestCase):
+    """Integration test using local HTTP server instead of external API."""
+
+    def test_fetch_products_local_server(self):
+        excel_bytes, rows = build_minpromtorg_products_excel(count=3)
+        date_str = fake.date_between(start_date="-30d", end_date="today").strftime(
+            "%Y%m%d"
+        )
+        file_name = f"industrial_products_{date_str}.xlsx"
+
+        with TestHTTPServer() as server:
+            server.add_json(
+                "/api/kss-document-preview",
+                {
+                    "data": [
+                        {
+                            "name": IndustrialProductsClient().query,
+                            "files": [
+                                {"name": file_name, "url": f"/files/{file_name}"}
+                            ],
+                        }
+                    ]
+                },
+            )
+            server.add_bytes(f"/files/{file_name}", excel_bytes)
+
+            client = IndustrialProductsClient(
+                host=_host_from_base_url(server.base_url),
+                scheme="http",
+                timeout=30,
+                http_adapter=server.adapter,
+            )
+            products = client.fetch_products()
+
+        self.assertEqual(len(products), len(rows))
+
+
 class ProverkiClientTest(TestCase):
     """Tests for ProverkiClient."""
 
@@ -574,7 +742,7 @@ class ProverkiClientTest(TestCase):
         self.assertEqual(inspections[0].control_authority, authority)
 
     def test_parse_xml_record_with_attributes(self):
-        from xml.etree import ElementTree as ET
+        from defusedxml import ElementTree as ET
 
         row_inn = "".join(str(fake.random_int(0, 9)) for _ in range(10))
         reg_num = "".join(str(fake.random_int(0, 9)) for _ in range(12))
@@ -590,7 +758,7 @@ class ProverkiClientTest(TestCase):
         self.assertEqual(result.registration_number, reg_num)
 
     def test_parse_xml_record_invalid(self):
-        from xml.etree import ElementTree as ET
+        from defusedxml import ElementTree as ET
 
         element = ET.fromstring("<empty_record></empty_record>")  # noqa: S314
         client = ProverkiClient()