TLDR-483 fixed box extraction from cropped cells (#343)

ispras · Oct 2, 2023 · 840bed2 · 840bed2
1 parent 58ce902
commit 840bed2
Show file tree

Hide file tree

Showing 4 changed files with 5 additions and 5 deletions.
diff --git a/dedoc/data_structures/line_with_meta.py b/dedoc/data_structures/line_with_meta.py
@@ -7,10 +7,11 @@
 
 from dedoc.data_structures.annotation import Annotation
 from dedoc.data_structures.line_metadata import LineMetadata
+from dedoc.data_structures.serializable import Serializable
 from dedoc.utils.annotation_merger import AnnotationMerger
 
 
-class LineWithMeta(Sized):
+class LineWithMeta(Sized, Serializable):
     """
     Structural unit of document - line (or paragraph) of text and its metadata.
     One LineWithMeta should not contain text from different logical parts of the document

diff --git a/dedoc/readers/pdf_reader/pdf_image_reader/ocr/ocr_cell_extractor.py b/dedoc/readers/pdf_reader/pdf_image_reader/ocr/ocr_cell_extractor.py
@@ -54,8 +54,8 @@ def get_cells_text(self, page_image: np.ndarray, tree_nodes: List["TableTree"],
                     word.bbox.y_top_left -= chunk_boxes[chunk_index].y_top_left
                     word.bbox.x_top_left -= chunk_boxes[chunk_index].x_top_left
                     # do absolute coordinate on src_image (inside src_image)
-                    word.bbox.y_top_left += nodes_batch[chunk_index].cell_box.y_top_left
-                    word.bbox.x_top_left += nodes_batch[chunk_index].cell_box.x_top_left
+                    word.bbox.y_top_left += nodes_batch[chunk_index].crop_text_box.y_top_left
+                    word.bbox.x_top_left += nodes_batch[chunk_index].crop_text_box.x_top_left
 
                 originalbox_to_fastocrbox[nodes_batch[chunk_index].cell_box].append(line.words)
 

diff --git a/dedoc/scripts/test_words_bbox_extraction.py b/dedoc/scripts/test_words_bbox_extraction.py
@@ -159,7 +159,7 @@ def test_table_word_extraction(self):
         os.makedirs(output_path, exist_ok=True)
         file_names = ["tables/example_with_table5.png", "tables/example_with_table3.png", "tables/example_with_table4.jpg",
                       "tables/example_with_table6.png", "tables/example_with_table_horizontal_union.jpg",
-                      "scanned/orient_1.png"]
+                      "scanned/orient_1.png", "tables/rotated_table.png"]
 
         for file_name in file_names:
             result = self._send_request(file_name, data=dict())
@@ -173,4 +173,3 @@ def test_table_word_extraction(self):
 
             image = self.draw_word_annotations(image, word_annotations, angle=table_angle)
             cv2.imwrite(os.path.join(output_path, file_name.split('/')[-1]), image)
-
diff --git a/tests/data/tables/rotated_table.png b/tests/data/tables/rotated_table.png