fixed code according to comments

tests/api_tests/content_checker.py

@@ -77,7 +77,7 @@ def __check_metadata(self, metadata: dict) -> None:
         if "other_fields" in metadata:
             self.assertIsInstance(metadata["other_fields"], dict)
 
-    def check_english_doc(self, result: dict) -> None:
+    def _check_english_doc(self, result: dict) -> None:
         content = result["content"]
         structure = content["structure"]
         self._check_tree_sanity(structure)

tests/api_tests/test_api_doctype_law.py

@@ -616,16 +616,6 @@ def test_foiv_html(self) -> None:
         self.assertEqual("4.2.1.", node["text"].strip())
         self.assertEqual("item", node['metadata']['paragraph_type'])
 
-    @unittest.skip("TODO fix this")
-    def test_number_not_part(self) -> None:
-        file_name = "31(1).txt"
-        result = self._send_request(file_name, dict(document_type="law"), expected_code=200)
-        document_tree = result["content"]["structure"]
-        self.__test_law_tree_sanity(document_tree)
-        node = self._get_by_tree_path(document_tree, "0.0.3.5.0.0")
-        self.assertTrue(node["text"].strip().endswith("2 настоящей статьи."))
-        self.assertEqual("raw_text", node['metadata']['paragraph_type'])
-
     def test_html_invisible_table(self) -> None:
         file_name = "invisibly_table4.html"
         result = self._send_request(file_name, dict(document_type="law"), expected_code=200)

tests/api_tests/test_api_doctype_tz.py

@@ -6,7 +6,7 @@
 from tests.test_utils import tree2linear
 
 
-class TestLawApiDocReader(AbstractTestApiDocReader):
+class TestTZApiDocReader(AbstractTestApiDocReader):
 
     def test_doc_tz(self) -> None:
         file_name = "alpaca_tz.doc"

tests/api_tests/test_api_format_archives.py

@@ -12,7 +12,7 @@ def _check_archive_with_english_doc(self, file_name: str) -> None:
         result = self._send_request(file_name, dict(with_attachments="True"))
         self.assertEqual(len(result['attachments']), 4)
         english_doc = [doc for doc in result['attachments'] if doc["metadata"]["file_name"].startswith("english_doc")][0]
-        self.check_english_doc(english_doc)
+        self._check_english_doc(english_doc)
 
     def test_zip(self) -> None:
         file_name = "arch_with_attachs.zip"
@@ -77,4 +77,4 @@ def test_broken_archive(self) -> None:
         result = self._send_request(file_name, dict(with_attachments="True"))
         self.assertEqual(len(result['attachments']), 7)
         english_doc = [doc for doc in result['attachments'] if doc["metadata"]["file_name"].startswith("english_doc")][0]
-        self.check_english_doc(english_doc)
+        self._check_english_doc(english_doc)

tests/api_tests/test_api_format_docx.py

@@ -102,18 +102,12 @@ def test_tricky_doc(self) -> None:
         file_name = "doc.docx"
         _ = self._send_request(file_name)
 
-    def test_broken_docx(self) -> None:
-        self._send_request("broken.docx", expected_code=415)
-
     def test_not_stripped_xml(self) -> None:
         self._send_request("not_stripped_xml.docx", expected_code=200)
 
     def test_docx_with_comments(self) -> None:
         _ = self._send_request("with_comments.docx", expected_code=200)
 
-    def test_send_wo_file(self) -> None:
-        self._send_request_wo_file(expected_code=422)
-
     def test_return_html(self) -> None:
         file_name = "example.doc"
         result = self._send_request(file_name, data={"structure_type": "tree", "return_format": "html"})

tests/api_tests/test_api_format_json.py

@@ -5,7 +5,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiCSVReader(AbstractTestApiDocReader):
+class TestApiJSONReader(AbstractTestApiDocReader):
 
     def _get_abs_path(self, file_name: str) -> str:
         return os.path.abspath(os.path.join(self.data_directory_path, "json", file_name))

tests/api_tests/test_api_format_pdf_auto_text_layer.py

@@ -44,21 +44,21 @@ def test_auto_pdf_with_text_layer(self) -> None:
         file_name = os.path.join("..", "pdf_with_text_layer", "english_doc.pdf")
         result = self._send_request(file_name, dict(pdf_with_text_layer="auto"))
         self.assertIn("Assume document has a correct textual layer", result["warnings"])
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     def test_auto_pdf_with_wrong_text_layer(self) -> None:
         file_name = "english_doc_bad_text.pdf"
         result = self._send_request(file_name, dict(pdf_with_text_layer="auto"))
         self.assertIn("Assume document has incorrect textual layer", result["warnings"])
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     def test_auto_document_mixed(self) -> None:
         file_name = "mixed_pdf.pdf"
         for pdf_with_text_layer in "auto", "auto_tabby":
             result = self._send_request(file_name, dict(pdf_with_text_layer=pdf_with_text_layer))
             self.assertIn("Assume document has a correct textual layer", result["warnings"])
             self.assertIn("Assume the first page hasn't a textual layer", result["warnings"])
-            self.check_english_doc(result)
+            self._check_english_doc(result)
             structure = result["content"]["structure"]
             list_items = structure["subparagraphs"][1]["subparagraphs"]
             self.assertEqual("3) продолжаем список\n", list_items[2]["text"])

tests/api_tests/test_api_format_pdf_page_limit.py

@@ -3,7 +3,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiPdfReader(AbstractTestApiDocReader):
+class TestApiPdfPageLimit(AbstractTestApiDocReader):
 
     def _get_abs_path(self, file_name: str) -> str:
         return os.path.join(self.data_directory_path, "pdf_with_text_layer", file_name)

tests/api_tests/test_api_format_pdf_tabby_reader.py

@@ -16,7 +16,7 @@ def __filter_by_name(self, annotations: List[dict], name: str) -> List[dict]:
     def test_example_file(self) -> None:
         file_name = "english_doc.pdf"
         result = self._send_request(file_name, data=dict(pdf_with_text_layer="tabby"))
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     @unittest.skip("TODO: add two layers output order support, e.g footnotes and main text.")
     def test_former_txt_file(self) -> None:
@@ -231,93 +231,6 @@ def test_pdf_with_tables(self) -> None:
         self.assertEqual("", table[1][0])
         self.assertEqual("Прогноз", table[1][1])
         self.assertEqual("Прогноз бюджета", table[1][2])
-        self.assertEqual("Прогноз бюджета", table[1][3])
-        self.assertEqual("Прогноз бюджета", table[1][4])
-        self.assertEqual("Расходы", table[2][0])
-        self.assertEqual("19,8", table[2][1])
-        self.assertEqual("18,6", table[2][2])
-        self.assertEqual("17,3", table[2][3])
-        self.assertEqual("16,1", table[2][4])
-        self.assertEqual("Доходы", table[3][0])
-        self.assertEqual("16,1", table[3][1])
-        self.assertEqual("15,4", table[3][2])
-        self.assertEqual("15,1", table[3][3])
-        self.assertEqual("15,0", table[3][4])
-        self.assertEqual("Нефтегазовые\nдоходы", table[4][0])
-        self.assertEqual("5,8", table[4][1])
-        self.assertEqual("5,8", table[4][2])
-        self.assertEqual("5,5", table[4][3])
-        self.assertEqual("5,4", table[4][4])
-        self.assertEqual("Ненефтегазов\nые доходы", table[5][0])
-        self.assertEqual("10,4", table[5][1])
-        self.assertEqual("9,6", table[5][2])
-        self.assertEqual("9,6", table[5][3])
-        self.assertEqual("9,6", table[5][4])
-        self.assertEqual("Сальдо\nбюджета", table[6][0])
-        self.assertEqual("-3,7", table[6][1])
-        self.assertEqual("-3,2", table[6][2])
-        self.assertEqual("-2,2", table[6][3])
-        self.assertEqual("-1,2", table[6][4])
-        self.assertEqual("", table[7][0])
-        self.assertEqual("2016", table[7][1])
-        self.assertEqual("2017", table[7][2])
-        self.assertEqual("2018", table[7][3])
-        self.assertEqual("2019", table[7][4])
-        self.assertEqual("", table[8][0])
-        self.assertEqual("Прогноз", table[8][1])
-        self.assertEqual("Прогноз бюджета", table[8][2])
-        self.assertEqual("Прогноз бюджета", table[8][3])
-        self.assertEqual("Прогноз бюджета", table[8][4])
-        self.assertEqual("Расходы", table[9][0])
-        self.assertEqual("19,8", table[9][1])
-        self.assertEqual("18,6", table[9][2])
-        self.assertEqual("17,3", table[9][3])
-        self.assertEqual("16,1", table[9][4])
-        self.assertEqual("Доходы", table[10][0])
-        self.assertEqual("16,1", table[10][1])
-        self.assertEqual("15,4", table[10][2])
-        self.assertEqual("15,1", table[10][3])
-        self.assertEqual("15,0", table[10][4])
-        self.assertEqual("Нефтегазовые\nдоходы", table[11][0])
-        self.assertEqual("5,8", table[11][1])
-        self.assertEqual("5,8", table[11][2])
-        self.assertEqual("5,5", table[11][3])
-        self.assertEqual("5,4", table[11][4])
-        self.assertEqual("Ненефтегазов\nые доходы", table[12][0])
-        self.assertEqual("10,4", table[12][1])
-        self.assertEqual("9,6", table[12][2])
-        self.assertEqual("9,6", table[12][3])
-        self.assertEqual("9,6", table[12][4])
-        self.assertEqual("Сальдо\nбюджета", table[13][0])
-        self.assertEqual("-3,7", table[13][1])
-        self.assertEqual("-3,2", table[13][2])
-        self.assertEqual("-2,2", table[13][3])
-        self.assertEqual("-1,2", table[13][4])
-        self.assertEqual("", table[14][0])
-        self.assertEqual("2016", table[14][1])
-        self.assertEqual("2017", table[14][2])
-        self.assertEqual("2018", table[14][3])
-        self.assertEqual("2019", table[14][4])
-        self.assertEqual("", table[15][0])
-        self.assertEqual("Прогноз", table[15][1])
-        self.assertEqual("Прогноз бюджета", table[15][2])
-        self.assertEqual("Прогноз бюджета", table[15][3])
-        self.assertEqual("Прогноз бюджета", table[15][4])
-        self.assertEqual("Расходы", table[16][0])
-        self.assertEqual("19,8", table[16][1])
-        self.assertEqual("18,6", table[16][2])
-        self.assertEqual("17,3", table[16][3])
-        self.assertEqual("16,1", table[16][4])
-        self.assertEqual("Доходы", table[17][0])
-        self.assertEqual("16,1", table[17][1])
-        self.assertEqual("15,4", table[17][2])
-        self.assertEqual("15,1", table[17][3])
-        self.assertEqual("15,0", table[17][4])
-        self.assertEqual("Нефтегазовые\nдоходы", table[18][0])
-        self.assertEqual("5,8", table[18][1])
-        self.assertEqual("5,8", table[18][2])
-        self.assertEqual("5,5", table[18][3])
-        self.assertEqual("5,4", table[18][4])
         self.assertEqual("Ненефтегазов\nые доходы", table[19][0])
         self.assertEqual("10,4", table[19][1])
         self.assertEqual("9,6", table[19][2])

tests/api_tests/test_api_format_pdf_with_text.py

@@ -5,7 +5,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiPdfReader(AbstractTestApiDocReader):
+class TestApiPdfWithText(AbstractTestApiDocReader):
 
     def _get_abs_path(self, file_name: str) -> str:
         return os.path.join(self.data_directory_path, "pdf_with_text_layer", file_name)

tests/api_tests/test_api_format_pptx.py

@@ -3,7 +3,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiExcelReader(AbstractTestApiDocReader):
+class TestApiPPTXReader(AbstractTestApiDocReader):
 
     data_directory_path = os.path.join(AbstractTestApiDocReader.data_directory_path, "pptx")
 

tests/api_tests/test_api_misc_language.py

@@ -3,26 +3,26 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiDocReader(AbstractTestApiDocReader):
+class TestLanguage(AbstractTestApiDocReader):
 
     data_directory_path = os.path.join(AbstractTestApiDocReader.data_directory_path, "docx")
 
     def test_en_doc(self) -> None:
         file_name = "english_doc.doc"
         result = self._send_request(file_name, dict(language="eng", structure_type="tree"))
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     def test_en_docx(self) -> None:
         file_name = "english_doc.docx"
         result = self._send_request(file_name, dict(language="eng", structure_type="tree"))
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     def test_en_odt(self) -> None:
         file_name = "english_doc.odt"
         result = self._send_request(file_name, dict(language="eng", structure_type="tree"))
-        self.check_english_doc(result)
+        self._check_english_doc(result)
 
     def test_en_pdf(self) -> None:
         file_name = "../pdf_with_text_layer/english_doc.pdf"
         result = self._send_request(file_name, dict(language="eng"))
-        self.check_english_doc(result)
+        self._check_english_doc(result)

tests/api_tests/test_api_misc_list_patching.py

@@ -1,7 +1,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiDocReader(AbstractTestApiDocReader):
+class TestListPatching(AbstractTestApiDocReader):
 
     def test_list_patching(self) -> None:
         file_name = "docx/13_moloko_1_polug.docx"

tests/api_tests/test_api_misc_main.py

@@ -4,9 +4,6 @@
 
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
-# test_structure
-# тесты на ошибки вынести в отдельный файл
-# вести файлик с темами тестов
 
 class TestApi(AbstractTestApiDocReader):
 
@@ -45,6 +42,3 @@ def test_text(self) -> None:
         self.assertEqual(content["subparagraphs"][1]["subparagraphs"][0]["text"].strip(), '1. Элемент нумерованного списка')
         self.assertEqual(content["subparagraphs"][1]["subparagraphs"][0]["metadata"]['paragraph_type'], 'list_item')
         self._check_metainfo(result['metadata'], 'text/plain', file_name)
-
-    def test_bin_file(self) -> None:
-        self._send_request("file.bin", expected_code=415)

tests/api_tests/test_api_misc_multipage_table.py

@@ -4,7 +4,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestRecognizedTable(AbstractTestApiDocReader):
+class TestMultipageTable(AbstractTestApiDocReader):
 
     def _get_abs_path(self, file_name: str) -> str:
         return os.path.join(self.data_directory_path, "tables", file_name)

tests/api_tests/test_api_misc_nesting_list.py

@@ -1,7 +1,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiDocReader(AbstractTestApiDocReader):
+class TestNestingList(AbstractTestApiDocReader):
 
     def test_list_nesting_content(self) -> None:
         file_name = "docx/pr14tz_v5_2007_03_01.docx"

tests/api_tests/test_api_misc_structure.py

@@ -3,7 +3,7 @@
 from tests.api_tests.abstract_api_test import AbstractTestApiDocReader
 
 
-class TestApiDocReader(AbstractTestApiDocReader):
+class TestStructure(AbstractTestApiDocReader):
 
     data_directory_path = os.path.join(AbstractTestApiDocReader.data_directory_path, "docx")
 

tests/api_tests/test_api_misc_with_attachments.py

@@ -11,18 +11,6 @@ class TestApiAttachmentsReader(AbstractTestApiDocReader):
 
     data_directory_path = AbstractTestApiDocReader.data_directory_path
 
-    def check_pdf_1(self, pdf: dict) -> None:
-        content = pdf["content"]['structure']
-        self.assertEqual("Глава 543\n", content["subparagraphs"][0]["text"])
-        self.assertEqual("Какой-то текст.\n", content["subparagraphs"][0]["subparagraphs"][0]["text"])
-        self.assertEqual(content["subparagraphs"][0]["subparagraphs"][1]["subparagraphs"][0]['text'], '1.\n')
-        self.assertEqual(content["subparagraphs"][0]["subparagraphs"][1]["subparagraphs"][1]['text'], '2.\n')
-        self.assertEqual(content["subparagraphs"][0]["subparagraphs"][1]["subparagraphs"][2]['text'], '3.\n')
-
-    def check_pdf_2(self, pdf: dict) -> None:
-        content = pdf["content"]
-        self.assertEqual("Пример документа\n", content['structure']['subparagraphs'][0]["text"])
-
     def _check_attachments(self, attachments: List[dict]) -> None:
         for attachment in attachments:
             self.assertTrue(attachment["attachments"] is not None)
@@ -172,7 +160,7 @@ def test_docx_images_base64(self) -> None:
             with open(path, "wb") as file_out:
                 file_out.write(base64.decodebytes(base64_encode.encode()))
             result_english = self._send_request(file_name=path, data={})
-            self.check_english_doc(result_english)
+            self._check_english_doc(result_english)
 
     def test_docx_images_no_base64(self) -> None:
         metadata = self.__check_base64(False)

tests/unit_tests/abstract_converter_test.py

@@ -7,9 +7,6 @@
 
 
 class AbstractConverterTest(TestCase):
-    """
-    Class for testing abstract converter
-    """
     path = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "data"))
 
     def setUp(self) -> None:
@@ -27,9 +24,6 @@ def tearDown(self) -> None:
         self.tmp_dir.cleanup()
 
     def _convert(self, filename: str, extension: str, converter: AbstractConverter) -> None:
-        """
-        Method for converting file into another extension
-        """
         filename_with_extension = filename + extension
         file = os.path.join(self.path, filename_with_extension)
         tmp_file = os.path.join(self.tmp_dir.name, filename_with_extension)