code fix

ispras · Sep 19, 2024 · 90fb752 · 90fb752
1 parent a4f12b9
commit 90fb752
Show file tree

Hide file tree

Showing 3 changed files with 21 additions and 18 deletions.
diff --git a/dedoc/readers/pdf_reader/pdf_base_reader.py b/dedoc/readers/pdf_reader/pdf_base_reader.py
@@ -114,21 +114,8 @@ 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 (isinstance(self, PdfImageReader) or isinstance(self, PdfTxtlayerReader)):
-            gost_analyzed_images = Parallel(n_jobs=self.config["n_jobs"])(delayed(self.gost_frame_recognizer.rec_and_clean_frame)(image) for image in images)
-            page_range = range(first_page, first_page + len(gost_analyzed_images))
-            gost_analyzed_images = dict(zip(page_range, gost_analyzed_images))
-            if isinstance(self, PdfImageReader):
-                result = Parallel(n_jobs=self.config["n_jobs"])(
-                    delayed(self._process_one_page)(image, parameters, page_number, path) for page_number, (image, box, original_image_shape) in
-                    gost_analyzed_images.items()
-                )
-            else:
-                self.gost_frame_boxes = dict(zip(page_range, [item[1] for item in gost_analyzed_images.values()]))
-                result = Parallel(n_jobs=self.config["n_jobs"])(
-                    delayed(self._process_one_page)(image, parameters, page_number, path) for page_number, (image, box, original_image_shape) in
-                    gost_analyzed_images.items()
-                )
+        if parameters.need_gost_frame_analysis and isinstance(self, (PdfImageReader, PdfTxtlayerReader)):
+            result, gost_analyzed_images = self._process_document_with_gost_frame(images=images, first_page=first_page, parameters=parameters, path=path)
         else:
             result = Parallel(n_jobs=self.config["n_jobs"])(
                 delayed(self._process_one_page)(image, parameters, page_number, path) for page_number, image in enumerate(images, start=first_page)
@@ -166,6 +153,22 @@ def _parse_document(self, path: str, parameters: ParametersForParseDoc) -> (
             metadata["rotated_page_angles"] = page_angles
         return all_lines_with_paragraphs, mp_tables, attachments, warnings, metadata
 
+    def _process_document_with_gost_frame(self, images: Iterator[np.ndarray], first_page: int, parameters: ParametersForParseDoc, path: str) -> \
+            Tuple[Tuple[List[LineWithLocation], List[ScanTable], List[PdfImageAttachment], List[float]], Dict[int, Tuple[np.ndarray, BBox, Tuple[int, ...]]]]:
+        from joblib import Parallel, delayed
+        from dedoc.readers.pdf_reader.pdf_txtlayer_reader.pdf_txtlayer_reader import PdfTxtlayerReader
+
+        gost_analyzed_images = Parallel(n_jobs=self.config["n_jobs"])(delayed(self.gost_frame_recognizer.rec_and_clean_frame)(image) for image in images)
+        page_range = range(first_page, first_page + len(gost_analyzed_images))
+        gost_analyzed_images = dict(zip(page_range, gost_analyzed_images))
+        if isinstance(self, PdfTxtlayerReader):
+            self.gost_frame_boxes = dict(zip(page_range, [item[1] for item in gost_analyzed_images.values()]))
+        result = Parallel(n_jobs=self.config["n_jobs"])(
+            delayed(self._process_one_page)(image, parameters, page_number, path) for page_number, (image, box, original_image_shape) in
+            gost_analyzed_images.items()
+        )
+        return result, gost_analyzed_images
+
     def _shift_all_contents(self, lines: List[LineWithMeta], unref_tables: List[ScanTable], attachments: List[PdfImageAttachment],
                             gost_analyzed_images: Dict[int, Tuple[np.ndarray, BBox, Tuple[int, ...]]]) -> None:
         # shift unref_tables

diff --git a/docs/source/dedoc_api_usage/api.rst b/docs/source/dedoc_api_usage/api.rst
@@ -228,7 +228,7 @@ Api parameters description
       - 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.
+        The GOST frame recognizer is used recognize and ignore GOST frame on images and PDF documents.
 
     * - language
       - rus, eng, rus+eng, fra, spa

diff --git a/docs/source/parameters/pdf_handling.rst b/docs/source/parameters/pdf_handling.rst
@@ -159,8 +159,8 @@ PDF and images handling
         * :meth:`dedoc.readers.ReaderComposition.read`
       - 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.
+        ignore the GOST frame on the document. It allows :class:`dedoc.readers.PdfImageReader` and :class:`dedoc.readers.PdfTxtlayerReader`
+        to properly process the content of the document containing GOST frame.
 
     * - orient_analysis_cells
       - True, False