PageType.get_AllRegions to list all kinds of regions

ocrd_models/ocrd_models/ocrd_page_generateds.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Wed May 13 16:09:07 2020 by generateDS.py version 2.35.20.
+# Generated Wed May 13 20:41:21 2020 by generateDS.py version 2.35.20.
 # Python 3.7.6 (default, Jan  8 2020, 19:59:22)  [GCC 7.3.0]
 #
 # Command line options:
@@ -2850,7 +2850,21 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'CustomRegion'
     def __hash__(self):
         return hash(self.id)
-# end class PageType
+
+    def get_AllRegions(self, regions=None, reading_order=False):
+        """"
+        Get all the *Region element or only those provided by ``regions``.
+        Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+        """
+        if reading_order:
+            raise NotImplementedError("Ordering of regions by Reading Order not currently Implemented :(")
+        if not regions:
+            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Table', 'Text']
+        ret = []
+        for region in regions:
+            ret += getattr(self, 'get_{}Region'.format(region))()
+        return ret
+    # end class PageType
 
 
 class CoordsType(GeneratedsSuper):

ocrd_models/ocrd_page_user_methods.py

@@ -83,6 +83,26 @@ def show(self):
 #
 # Replace the following method specifications with your own.
 
+#
+# List all *Regions on the PAGE
+#
+get_AllRegions = MethodSpec(name='get_AllRegions',
+    source=r'''
+    def get_AllRegions(self, regions=None, reading_order=False):
+        """"
+        Get all the *Region element or only those provided by ``regions``.
+        Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+        """
+        if reading_order:
+            raise NotImplementedError("Ordering of regions by Reading Order not currently Implemented :(")
+        if not regions:
+            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Table', 'Text']
+        ret = []
+        for region in regions:
+            ret += getattr(self, 'get_{}Region'.format(region))()
+        return ret
+    ''', class_names=r'^(PageType)$')
+
 #
 # List all *Indexed children sorted by @index
 #
@@ -167,6 +187,7 @@ def __hash__(self):
     exportChildren,
     get_AllIndexed,
     add_AllIndexed,
+    get_AllRegions,
     clear_AllIndexed,
     )
 

tests/model/test_ocrd_page.py

@@ -184,5 +184,16 @@ def test_empty_groups_to_regionrefindexed(self):
             children = og.get_AllIndexed()
             self.assertTrue(isinstance(children[1], RegionRefIndexedType))
 
+    def test_all_regions(self):
+        """
+        Corrolary See https://github.com/OCR-D/core/issues/475
+        """
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            pcgts = parseString(f.read().encode('utf8'), silence=True)
+            pg = pcgts.get_Page()
+            self.assertEqual(len(pg.get_AllRegions()), 20)
+            self.assertEqual(len(pg.get_AllRegions(['Table'])), 3)
+            self.assertEqual(len(pg.get_AllRegions(['Text'])), 17)
+
 if __name__ == '__main__':
     main()