fixed code according to mr comments

dedoc/readers/pdf_reader/pdf_base_reader.py

@@ -113,7 +113,7 @@ def _parse_document(self, path: str, parameters: ParametersForParseDoc) -> (
         last_page = math.inf if parameters.last_page is None else parameters.last_page
         images = self._get_images(path, first_page, last_page)
 
-        if parameters.need_gost_frame_analysis and parameters.pdf_with_txt_layer == "false":
+        if parameters.need_gost_frame_analysis and type(self).__name__ == "PdfImageReader":
             gost_analyzed_images = Parallel(n_jobs=self.config["n_jobs"])(delayed(self.gost_frame_recognizer.rec_and_clean_frame)(image) for image in images)
             result = Parallel(n_jobs=self.config["n_jobs"])(
                 delayed(self._process_one_page)(image, parameters, page_number, path) for page_number, (image, box) in
@@ -152,7 +152,7 @@ def _parse_document(self, path: str, parameters: ParametersForParseDoc) -> (
         all_lines_with_paragraphs = self.paragraph_extractor.extract(all_lines_with_links)
         if page_angles:
             metadata["rotated_page_angles"] = page_angles
-        if parameters.need_gost_frame_analysis and parameters.pdf_with_txt_layer == "false":
+        if parameters.need_gost_frame_analysis and type(self).__name__ == "PdfImageReader":
             self._shift_all_contents(lines=all_lines_with_paragraphs, mp_tables=mp_tables, attachments=attachments, gost_analyzed_images=gost_analyzed_images)
         return all_lines_with_paragraphs, mp_tables, attachments, warnings, metadata
 

docs/source/dedoc_api_usage/api.rst

@@ -225,12 +225,10 @@ Api parameters description
         * **false** -- use the textual layer classifier to detect textual layer and prove its correctness.
 
     * - need_gost_frame_analysis
-      - True, False
-      - False
-      - This option is used to enable GOST(Russian government standard) frame recognition for PDF documents or images.
-        The GOST frame recognizer is used in :meth:`dedoc.readers.PdfBaseReader.read`. Its main function is to recognize and
-        ignore the GOST frame on the document. It allows :class:`dedoc.readers.PdfImageReader` to properly process the content
-        of the document containing GOST frame. Currently works only when ``pdf_with_text_layer="false"``.
+      - true, false
+      - false
+      - This option is used to enable GOST (Russian government standard) frame recognition for PDF documents or images.
+        The GOST frame recognizer is used recognize and ignore GOST frame on images and PDF documents without correct textual layer.
 
     * - language
       - rus, eng, rus+eng, fra, spa

docs/source/parameters/pdf_handling.rst

@@ -157,10 +157,10 @@ PDF and images handling
       - * :meth:`dedoc.DedocManager.parse`
         * :meth:`dedoc.readers.PdfAutoReader.read`, :meth:`dedoc.readers.PdfBaseReader.read`
         * :meth:`dedoc.readers.ReaderComposition.read`
-      - This option is used to enable GOST(Russian government standard) frame recognition for PDF documents or images.
+      - This option is used to enable GOST (Russian government standard) frame recognition for PDF documents or images.
         The GOST frame recognizer is used in :meth:`dedoc.readers.PdfBaseReader.read`. Its main function is to recognize and
         ignore the GOST frame on the document. It allows :class:`dedoc.readers.PdfImageReader` to properly process the content
-        of the document containing GOST frame. Currently works only when ``pdf_with_text_layer="false"``.
+        of the document containing GOST frame.
 
     * - orient_analysis_cells
       - True, False

tests/api_tests/test_api_module_table_recognizer.py

@@ -213,3 +213,13 @@ def test_detect_small_table(self) -> None:
         result = self._send_request(file_name, data={"language": "rus"})
         tables = result["content"]["tables"]
         self.assertEqual(2, len(tables))
+
+    def test_multipage_gost_table(self):
+        file_name = "gost_multipage_table.pdf"
+        result = self._send_request(file_name, data={"need_gost_frame_analysis": "True"})  # don't pass pdf_with_text_layer to check condition in PDFBaseReader
+        self.assertTrue(len(result['content']['tables'][0]['cells']) > 35)
+        self.assertTrue('KR13' in result['content']['tables'][0]['cells'][-1][0]['lines'][0]['text'])  # check the last row of multipage table
+        self.assertTrue('R13.1' in result['content']['tables'][0]['cells'][-1][1]['lines'][0]['text'])  # check that it belongs to first and only table
+        self.assertTrue('Испытание по проверке' in result['content']['tables'][0]['cells'][-1][2]['lines'][0]['text'])
+        self.assertTrue('3.6' in result['content']['tables'][0]['cells'][-1][3]['lines'][0]['text'])
+        self.assertTrue('7.4.9' in result['content']['tables'][0]['cells'][-1][4]['lines'][0]['text'])

tests/unit_tests/test_module_gost_frame_recognizer.py

@@ -8,6 +8,7 @@
 import dedoc.utils.parameter_utils as param_utils
 from dedoc.readers.pdf_reader.pdf_base_reader import ParametersForParseDoc
 from dedoc.readers.pdf_reader.pdf_image_reader.pdf_image_reader import PdfImageReader
+from dedoc.readers.pdf_reader.pdf_auto_reader.pdf_auto_reader import PdfAutoReader
 from dedoc.readers.pdf_reader.pdf_image_reader.table_recognizer.gost_frame_recognizer import GOSTFrameRecognizer
 from tests.test_utils import get_test_config
 
@@ -17,6 +18,7 @@ class TestGOSTFrameRecognizer(unittest.TestCase):
     gost_frame_recognizer = GOSTFrameRecognizer(config=get_test_config())
     test_data_folder = os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "data", "tables"))
     pdf_image_reader = PdfImageReader(config=get_test_config())
+    pdf_auto_reader = PdfAutoReader(config=get_test_config())
 
     def _get_params_for_parse(self, parameters: Optional[dict], file_path: Optional[str]) -> ParametersForParseDoc:
         parameters = parameters if parameters else {}
@@ -43,7 +45,7 @@ def _get_params_for_parse(self, parameters: Optional[dict], file_path: Optional[
 
     def test_gost_frame_recognition(self) -> None:
         image_names = [
-            "gost_frame_1.png", "gost_frame_2.png", "gost_frame_3.png", "example_with_table6.png", "example_with_table5.png", "example_with_table3.png"
+            "gost_frame_1.jpg", "gost_frame_2.png", "gost_frame_3.jpg", "example_with_table6.png", "example_with_table5.png", "example_with_table3.png"
         ]
         gt = [True, True, True, False, False, False]
         for index, image_name in enumerate(image_names):
@@ -52,9 +54,18 @@ def test_gost_frame_recognition(self) -> None:
             result_image, result_bbox = self.gost_frame_recognizer.rec_and_clean_frame(image)
             self.assertEqual(not np.array_equal(result_image, image), gt[index])  # check if we cut something from original image or not
 
+    def test_not_gost_frame(self) -> None:
+        path_image = os.path.join(self.test_data_folder, "not_gost_frame.jpg")
+        image = cv2.imread(path_image)
+        result_image, result_bbox = self.gost_frame_recognizer.rec_and_clean_frame(image)
+        self.assertTrue(abs(result_bbox.x_top_left - 26) < 10)
+        self.assertTrue(abs(result_bbox.y_top_left - 26) < 10)
+        self.assertTrue(abs(result_bbox.width - 722) < 10)
+        self.assertTrue(abs(result_bbox.height - 969) < 10)
+
     def test_coordinates_shift(self) -> None:
         file_path = os.path.join(self.test_data_folder, "gost_frame_2.png")
-        parameters = {"need_gost_frame_analysis": "True", "pdf_with_text_layer": "false"}
+        parameters = {"need_gost_frame_analysis": "True"}
         params_for_parse = self._get_params_for_parse(parameters=parameters, file_path=file_path)
         result = self.pdf_image_reader._parse_document(path=file_path, parameters=params_for_parse)
         self.assertTrue(len(result[0]) > 0)
@@ -65,3 +76,14 @@ def test_coordinates_shift(self) -> None:
         self.assertTrue(len(result[1]) > 0)
         self.assertTrue(abs(result[1][0].location.bbox.x_top_left - 81) < 10)
         self.assertTrue(abs(result[1][0].location.bbox.y_top_left - 49) < 10)
+
+    def test_pdf_auto_reader(self) -> None:
+        file_path = os.path.join(self.test_data_folder, "gost_frame_2.png")
+        parameters = {"need_gost_frame_analysis": "True"}
+        result = self.pdf_auto_reader.read(file_path=file_path, parameters=parameters)
+        self.assertTrue(len(result.tables) == 1)
+        self.assertEqual(result.tables[0].cells[0][1].get_text(), "Колонка 2")
+        self.assertEqual(result.tables[0].cells[0][2].get_text(), "Колонка 3")
+        self.assertEqual(len(result.tables[0].cells), 22)
+        self.assertTrue("Названне таблицы (продолженне)" in result.lines[0].line)
+