prepare for segmenter use and add logs

Author: jnswidmer

flow_preprocessor/preprocessing_logic/parse_textlines.py

@@ -13,7 +13,6 @@
 
 from flow_preprocessor.exceptions.exceptions import ParseTextLinesException
 from flow_preprocessor.utils.logging.preprocessing_logger import logger
-# from flow_preprocessor.preprocessing_logic.segmentation import SegmenterYOLO
 
 
 # ===============================================================================
@@ -284,29 +283,35 @@ def parse_xml_file(self, xml_file: str, segment: bool = False) -> None:
             self.namespace_uri = self.root.tag.split('}')[0][1:]
             self.namespace = {'prefix': self.namespace_uri}
             self.xmlns = {'ns': self.namespace_uri}
-            image_filename = self.get_image_file_name()
+            # image_filename = self.get_image_file_name()
 
-            """
             if segment:
                 existing_segmentation = self.check_segmentation()
                 if existing_segmentation == 'ground_truth':
-                    pass
-                else:
-                    segmenter = SegmenterYOLO(
-                        models=['Riksarkivet/yolov9-regions-1', 'Riksarkivet/yolov9-lines-within-regions-1'],
-                        batch_sizes=4,
-                        order_lines=True,
+                    # If the XML file already contains ground truth segmentation, do not segment again.
+                    logger.info(
+                        '%s - XML file %s already contains ground truth segmentation, skipping segmentation.',
+                        self.__class__.__name__,
+                        xml_file,
                     )
-                    self.tree = segmenter.segment(self.tree, image_filename)
-                    self.root = self.tree.getroot()
-                
+                    pass
                 elif existing_segmentation == 'segmented':
-                    segmenter = Segmenter('linemasks')
-                    self.root = segmenter.segment(self.root)
+                    # If the XML file is already segmented, do not segment again - maybe linemasks recognition?
+                    logger.info(
+                        '%s - XML file %s is already segmented, skipping segmentation.',
+                        self.__class__.__name__,
+                        xml_file,
+                    )
+                    pass
                 else:
-                    segmenter = Segmenter('yolo')
-                    self.root = segmenter.segment(self.root)
-                """
+                    # If the XML file is not segmented, segment it.
+                    logger.info(
+                        '%s - XML file %s is not segmented, processing lines.',
+                        self.__class__.__name__,
+                        xml_file,
+                    )
+                    pass
+
 
         except (et.XMLSyntaxError, et.ParseError) as e:
             self.failed_processing.append(xml_file)

flow_preprocessor/preprocessing_logic/preprocess.py

@@ -45,7 +45,7 @@ def __init__(
             abbrev: bool = False,
             stop_on_fail: bool = True,
             minwidth: Union[int, float] = None,
-            # segment: bool = False,
+            segment: bool = False,
             **kwargs,
     ) -> None:
         """
@@ -86,12 +86,13 @@ def __init__(
         else:
             self.minwidth = None
         self.kwargs = kwargs
-        # TODO: Change as soon as Segmenter is implemented
-        # self.segment = segment
-        if 'segment' in kwargs:
-            self.segment = kwargs['segment']
-            del kwargs['segment']
+        if segment and 'segmentation_models' in kwargs:
+            self.segmentation_models = kwargs['segmentation_models']
+            del kwargs['segmentation_models']
         else:
+            self.segmentation_models = None
+
+        if not segment:
             self.segment = False
 
         self.image_processor = ImageProcessor()
@@ -215,7 +216,7 @@ def preprocess_single_xml_file(self, xml_file: str) -> None:
             page = page_parser.get_page()
             gt_dict = {}
         except ParseTextLinesException as e:
-            logger.error("Error parsing XML file %s: %s", xml_file, exc_info=True)
+            logger.error("Error parsing XML file %s: %s", xml_file, e, exc_info=True)
             if self.stop_on_fail:
                 raise
             return