LGVISIUM-102: LayerIdentifierColumn extends from DepthColumn

pyproject.toml

@@ -12,7 +12,6 @@ dependencies = [
     "boto3",
     "pandas",
     "levenshtein",
-    "pathlib",
     "python-dotenv",
     "setuptools",
     "tqdm",

src/stratigraphy/annotations/draw.py

@@ -9,7 +9,7 @@
 from dotenv import load_dotenv
 from stratigraphy.data_extractor.data_extractor import FeatureOnPage
 from stratigraphy.depthcolumn.depthcolumn import DepthColumn
-from stratigraphy.depths_materials_column_pairs.depths_materials_column_pairs import DepthsMaterialsColumnPairs
+from stratigraphy.depths_materials_column_pairs.depths_materials_column_pairs import DepthsMaterialsColumnPair
 from stratigraphy.groundwater.groundwater_extraction import Groundwater
 from stratigraphy.layer.layer import Layer
 from stratigraphy.metadata.coordinate_extraction import Coordinate
@@ -98,7 +98,7 @@ def draw_predictions(
                     draw_depth_columns_and_material_rect(
                         shape,
                         page.derotation_matrix,
-                        [pair for pair in depths_materials_column_pairs if pair.page == page_number],
+                        [pair for pair in depths_materials_column_pairs if pair.depth_column.page == page_number],
                     )
                     draw_material_descriptions(
                         shape,
@@ -245,7 +245,7 @@ def draw_material_descriptions(shape: fitz.Shape, derotation_matrix: fitz.Matrix
 
 
 def draw_depth_columns_and_material_rect(
-    shape: fitz.Shape, derotation_matrix: fitz.Matrix, depths_materials_column_pairs: list[DepthsMaterialsColumnPairs]
+    shape: fitz.Shape, derotation_matrix: fitz.Matrix, depths_materials_column_pairs: list[DepthsMaterialsColumnPair]
 ):
     """Draw depth columns as well as the material rects on a pdf page.
 

src/stratigraphy/depthcolumn/boundarydepthcolumnvalidator.py

@@ -63,27 +63,26 @@ def is_valid(self, column: BoundaryDepthColumn, corr_coef_threshold: float = 0.9
 
         return corr_coef and corr_coef > corr_coef_threshold
 
-    def reduce_until_valid(self, column: BoundaryDepthColumn, page_number: int) -> BoundaryDepthColumn:
+    def reduce_until_valid(self, column: BoundaryDepthColumn) -> BoundaryDepthColumn:
         """Removes entries from the depth column until it fulfills the is_valid condition.
 
         is_valid checks whether there is too much noise (i.e. other text) in the column and whether the entries are
         linearly correlated with their vertical position.
 
         Args:
             column (BoundaryDepthColumn): The depth column to validate
-            page_number (int): The page number of the depth column
         Returns:
             BoundaryDepthColumn: The current depth column with entries removed until it is valid.
         """
         while column:
             if self.is_valid(column):
                 return column
-            elif self.correct_OCR_mistakes(column, page_number) is not None:
-                return self.correct_OCR_mistakes(column, page_number)
+            elif self.correct_OCR_mistakes(column) is not None:
+                return self.correct_OCR_mistakes(column)
             else:
                 column = column.remove_entry_by_correlation_gradient()
 
-    def correct_OCR_mistakes(self, column: BoundaryDepthColumn, page_number: int) -> BoundaryDepthColumn | None:
+    def correct_OCR_mistakes(self, column: BoundaryDepthColumn) -> BoundaryDepthColumn | None:
         """Corrects OCR mistakes in the depth column entries.
 
         Loops through all values and corrects common OCR mistakes for the given entry. Then, the column with the
@@ -102,15 +101,14 @@ def correct_OCR_mistakes(self, column: BoundaryDepthColumn, page_number: int) ->
 
         Args:
             column (BoundaryDepthColumn): The depth column to validate
-            page_number (int): The page number of the depth column
 
         Returns:
             BoundaryDepthColumn | None: The corrected depth column, or None if no correction was possible.
         """
-        new_columns = [BoundaryDepthColumn()]
+        new_columns = [BoundaryDepthColumn(entries=[], page=column.page)]
         for entry in column.entries:
             new_columns = [
-                BoundaryDepthColumn([*column.entries, DepthColumnEntry(entry.rect, new_value, page_number)])
+                BoundaryDepthColumn([*column.entries, DepthColumnEntry(entry.rect, new_value)], page=column.page)
                 for column in new_columns
                 for new_value in _value_alternatives(entry.value)
             ]