From abef3522d6f3f50993ee60414649724aa007121f Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Wed, 13 May 2020 20:44:25 +0200
Subject: [PATCH 01/30] PageType.get_AllRegions to list all kinds of regions

---
 .../ocrd_models/ocrd_page_generateds.py       | 18 ++++++++++++++--
 ocrd_models/ocrd_page_user_methods.py         | 21 +++++++++++++++++++
 tests/model/test_ocrd_page.py                 | 11 ++++++++++
 3 files changed, 48 insertions(+), 2 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 941851c35b..0295dc8ef0 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/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):
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index 927c2298b1..8b87138692 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/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,
     )
 
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 978d2a47e5..605ee4e483 100644
--- a/tests/model/test_ocrd_page.py
+++ b/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()

From 3445f87f027970d57227170115116cc440e624af Mon Sep 17 00:00:00 2001
From: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
Date: Thu, 14 May 2020 14:12:44 +0200
Subject: [PATCH 02/30] Update ocrd_models/ocrd_page_user_methods.py

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/ocrd_page_user_methods.py | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index 8b87138692..da77ad7f5e 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -94,13 +94,32 @@ def get_AllRegions(self, regions=None, reading_order=False):
         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 :(")
+            reading_order = self.get_ReadingOrder()
         if not regions:
-            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Table', 'Text']
+            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
         ret = []
         for region in regions:
             ret += getattr(self, 'get_{}Region'.format(region))()
-        return ret
+        if reading_order:
+            reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
+        if reading_order:
+            def get_recursive_reading_order(rogroup):
+                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
+                    elements = rogroup.get_AllIndexed()
+                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+                regionrefs = list()
+                for elem in elements:
+                    regionrefs.append(elem.get_regionRef())
+                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                        regionrefs.extend(get_recursive_reading_order(elem))
+                return regionrefs
+            reading_order = get_recursive_reading_order(reading_order)
+        if reading_order:
+            ret = dict([(region.id, region) for region in ret])
+            return [ret[region_id] for region_id in reading_order if region_id in ret]
+        else:
+            return ret
     ''', class_names=r'^(PageType)$')
 
 #

From a48b8c1a80653f106c48e57b260e6048e46235f7 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Thu, 14 May 2020 15:40:10 +0200
Subject: [PATCH 03/30] update generateds page, add region filter if using
 reading order, wip

---
 .../ocrd_models/ocrd_page_generateds.py       | 31 +++++++++++++++----
 ocrd_models/ocrd_page_user_methods.py         |  6 ++--
 tests/model/test_ocrd_page.py                 |  8 +++--
 3 files changed, 33 insertions(+), 12 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 0295dc8ef0..cd0e078bce 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,8 +2,8 @@
 # -*- coding: utf-8 -*-
 
 #
-# 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]
+# Generated Thu May 14 15:35:20 2020 by generateDS.py version 2.35.20.
+# Python 3.6.6 (default, Jul 24 2018, 16:39:20)  [GCC 4.9.2]
 #
 # Command line options:
 #   ('-f', '')
@@ -16,7 +16,7 @@
 #   repo/assets/data/schema/data/2019.xsd
 #
 # Command line:
-#   /home/kba/miniconda3/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
+#   /data/monorepo/venv3.6/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
 #
 # Current working directory (os.getcwd()):
 #   core
@@ -2857,13 +2857,32 @@ def get_AllRegions(self, regions=None, reading_order=False):
         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 :(")
+            reading_order = self.get_ReadingOrder()
         if not regions:
-            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Table', 'Text']
+            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
         ret = []
         for region in regions:
             ret += getattr(self, 'get_{}Region'.format(region))()
-        return ret
+        if reading_order:
+            reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
+        if reading_order:
+            def get_recursive_reading_order(rogroup):
+                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
+                    elements = rogroup.get_AllIndexed()
+                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+                regionrefs = list()
+                for elem in elements:
+                    regionrefs.append(elem.get_regionRef())
+                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                        regionrefs.extend(get_recursive_reading_order(elem))
+                return regionrefs
+            reading_order = get_recursive_reading_order(reading_order)
+        if reading_order:
+            ret = dict([(region.id, region) for region in ret])
+            return [ret[region_id] for region_id in reading_order if region_id in ret]
+        else:
+            return ret
     # end class PageType
 
 
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index da77ad7f5e..bf4261022e 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -117,9 +117,9 @@ def get_recursive_reading_order(rogroup):
             reading_order = get_recursive_reading_order(reading_order)
         if reading_order:
             ret = dict([(region.id, region) for region in ret])
-            return [ret[region_id] for region_id in reading_order if region_id in ret]
-        else:
-            return ret
+            ret = [ret[region_id] for region_id in reading_order if region_id in ret]
+        ret = [r in ret if r.__class__.__name__.replace('RegionType', '') in regions
+        return ret
     ''', class_names=r'^(PageType)$')
 
 #
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 605ee4e483..b24d1e45ff 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -184,14 +184,16 @@ def test_empty_groups_to_regionrefindexed(self):
             children = og.get_AllIndexed()
             self.assertTrue(isinstance(children[1], RegionRefIndexedType))
 
-    def test_all_regions(self):
+    def test_all_regions_without_reading_order(self):
         """
-        Corrolary See https://github.com/OCR-D/core/issues/475
+        https://github.com/OCR-D/core/pull/479
+        https://github.com/OCR-D/core/issues/240#issuecomment-493135797
         """
         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()), 45)
+            self.assertEqual(len(pg.get_AllRegions(['Separator'])), 25)
             self.assertEqual(len(pg.get_AllRegions(['Table'])), 3)
             self.assertEqual(len(pg.get_AllRegions(['Text'])), 17)
 

From d2a01bb799b32d89d8d04b6f695e37b9e83cd8a6 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Fri, 15 May 2020 16:17:48 +0200
Subject: [PATCH 04/30] refactoring: move generateDS methods to their own files

---
 .../ocrd_models/ocrd_page_generateds.py       | 131 +++++++----------
 ocrd_models/ocrd_page_user_methods.py         | 139 +++---------------
 .../ocrd_page_user_methods/__hash__.py        |   2 +
 .../ocrd_page_user_methods/add_AllIndexed.py  |  12 ++
 .../clear_AllIndexed.py                       |   7 +
 .../ocrd_page_user_methods/exportChildren.py  |  20 +++
 .../ocrd_page_user_methods/get_AllIndexed.py  |   3 +
 .../ocrd_page_user_methods/get_AllRegions.py  |  32 ++++
 8 files changed, 143 insertions(+), 203 deletions(-)
 create mode 100644 ocrd_models/ocrd_page_user_methods/__hash__.py
 create mode 100644 ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
 create mode 100644 ocrd_models/ocrd_page_user_methods/clear_AllIndexed.py
 create mode 100644 ocrd_models/ocrd_page_user_methods/exportChildren.py
 create mode 100644 ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
 create mode 100644 ocrd_models/ocrd_page_user_methods/get_AllRegions.py

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index cd0e078bce..8a3705a3e4 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,8 +2,8 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 14 15:35:20 2020 by generateDS.py version 2.35.20.
-# Python 3.6.6 (default, Jul 24 2018, 16:39:20)  [GCC 4.9.2]
+# Generated Fri May 15 16:15:27 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:
 #   ('-f', '')
@@ -16,7 +16,7 @@
 #   repo/assets/data/schema/data/2019.xsd
 #
 # Command line:
-#   /data/monorepo/venv3.6/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
+#   /home/kba/miniconda3/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
 #
 # Current working directory (os.getcwd()):
 #   core
@@ -2850,40 +2850,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'CustomRegion'
     def __hash__(self):
         return hash(self.id)
-
-    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:
-            reading_order = self.get_ReadingOrder()
-        if not regions:
-            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
-        ret = []
-        for region in regions:
-            ret += getattr(self, 'get_{}Region'.format(region))()
-        if reading_order:
-            reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
-        if reading_order:
-            def get_recursive_reading_order(rogroup):
-                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
-                    elements = rogroup.get_AllIndexed()
-                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
-                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
-                regionrefs = list()
-                for elem in elements:
-                    regionrefs.append(elem.get_regionRef())
-                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
-                        regionrefs.extend(get_recursive_reading_order(elem))
-                return regionrefs
-            reading_order = get_recursive_reading_order(reading_order)
-        if reading_order:
-            ret = dict([(region.id, region) for region in ret])
-            return [ret[region_id] for region_id in reading_order if region_id in ret]
-        else:
-            return ret
-    # end class PageType
+# end class PageType
 
 
 class CoordsType(GeneratedsSuper):
@@ -5380,7 +5347,28 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroupIndexed'
     def __hash__(self):
         return hash(self.id)
-
+    def get_AllIndexed(self):
+        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+    
+    def clear_AllIndexed(self):
+        ret = self.get_AllIndexed()
+        self.set_RegionRefIndexed([])
+        self.set_OrderedGroupIndexed([])
+        self.set_UnorderedGroupIndexed([])
+        return ret
+    
+    def add_AllIndexed(self, elements):
+        if not isinstance(elements, list):
+            elements = [elements]
+        for element in sorted(elements, key=lambda x : x.index):
+            if isinstance(element, RegionRefIndexedType):
+                self.add_RegionRefIndexed(element)
+            elif isinstance(element, OrderedGroupIndexedType):
+                self.add_OrderedGroupIndexed(element)
+            elif isinstance(element, UnorderedGroupIndexedType):
+                self.add_UnorderedGroupIndexed(element)
+        return self.get_AllIndexed()
+    
     def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
         eol_ = '\n' if pretty_print else ''
         namespaceprefix_ = 'pc:'
@@ -5400,27 +5388,7 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
                 cleaned.append(entry)
         for entry in cleaned:
             entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-
-    def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index) 
-    def add_AllIndexed(self, elements):
-        if not isinstance(elements, list):
-            elements = [elements]
-        for element in sorted(elements, key=lambda x : x.index):
-            if isinstance(element, RegionRefIndexedType):
-                self.add_RegionRefIndexed(element)
-            elif isinstance(element, OrderedGroupIndexedType):
-                self.add_OrderedGroupIndexed(element)
-            elif isinstance(element, UnorderedGroupIndexedType):
-                self.add_UnorderedGroupIndexed(element)
-        return self.get_AllIndexed()
-
-    def clear_AllIndexed(self):
-        ret = self.get_AllIndexed()
-        self.set_RegionRefIndexed([])
-        self.set_OrderedGroupIndexed([])
-        self.set_UnorderedGroupIndexed([])
-        return ret
+    
 # end class OrderedGroupIndexedType
 
 
@@ -6169,7 +6137,28 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroupIndexed'
     def __hash__(self):
         return hash(self.id)
-
+    def get_AllIndexed(self):
+        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+    
+    def clear_AllIndexed(self):
+        ret = self.get_AllIndexed()
+        self.set_RegionRefIndexed([])
+        self.set_OrderedGroupIndexed([])
+        self.set_UnorderedGroupIndexed([])
+        return ret
+    
+    def add_AllIndexed(self, elements):
+        if not isinstance(elements, list):
+            elements = [elements]
+        for element in sorted(elements, key=lambda x : x.index):
+            if isinstance(element, RegionRefIndexedType):
+                self.add_RegionRefIndexed(element)
+            elif isinstance(element, OrderedGroupIndexedType):
+                self.add_OrderedGroupIndexed(element)
+            elif isinstance(element, UnorderedGroupIndexedType):
+                self.add_UnorderedGroupIndexed(element)
+        return self.get_AllIndexed()
+    
     def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
         eol_ = '\n' if pretty_print else ''
         namespaceprefix_ = 'pc:'
@@ -6189,27 +6178,7 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
                 cleaned.append(entry)
         for entry in cleaned:
             entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-
-    def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index) 
-    def add_AllIndexed(self, elements):
-        if not isinstance(elements, list):
-            elements = [elements]
-        for element in sorted(elements, key=lambda x : x.index):
-            if isinstance(element, RegionRefIndexedType):
-                self.add_RegionRefIndexed(element)
-            elif isinstance(element, OrderedGroupIndexedType):
-                self.add_OrderedGroupIndexed(element)
-            elif isinstance(element, UnorderedGroupIndexedType):
-                self.add_UnorderedGroupIndexed(element)
-        return self.get_AllIndexed()
-
-    def clear_AllIndexed(self):
-        ret = self.get_AllIndexed()
-        self.set_RegionRefIndexed([])
-        self.set_OrderedGroupIndexed([])
-        self.set_UnorderedGroupIndexed([])
-        return ret
+    
 # end class OrderedGroupType
 
 
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index bf4261022e..7f93323794 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -3,6 +3,8 @@
 # source: https://bitbucket.org/dkuhlman/generateds/src/default/gends_user_methods.py
 
 import re
+import codecs
+from os.path import dirname, join
 
 #
 # You must include the following class definition at the top of
@@ -80,134 +82,27 @@ def show(self):
 #   generated superclass file and also section "User Methods" in
 #   the documentation, as well as the examples below.
 
-#
-# 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:
-            reading_order = self.get_ReadingOrder()
-        if not regions:
-            regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
-        ret = []
-        for region in regions:
-            ret += getattr(self, 'get_{}Region'.format(region))()
-        if reading_order:
-            reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
-        if reading_order:
-            def get_recursive_reading_order(rogroup):
-                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
-                    elements = rogroup.get_AllIndexed()
-                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
-                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
-                regionrefs = list()
-                for elem in elements:
-                    regionrefs.append(elem.get_regionRef())
-                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
-                        regionrefs.extend(get_recursive_reading_order(elem))
-                return regionrefs
-            reading_order = get_recursive_reading_order(reading_order)
-        if reading_order:
-            ret = dict([(region.id, region) for region in ret])
-            ret = [ret[region_id] for region_id in reading_order if region_id in ret]
-        ret = [r in ret if r.__class__.__name__.replace('RegionType', '') in regions
-        return ret
-    ''', class_names=r'^(PageType)$')
-
-#
-# List all *Indexed children sorted by @index
-#
-get_AllIndexed = MethodSpec(name='get_AllIndexed',
-    source=r'''
-    def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index) ''', class_names=r'^(OrderedGroupType|OrderedGroupIndexedType)$')
-
-#
-# Clear all *Indexed children sorted by @index
-#
-clear_AllIndexed = MethodSpec(name='clear_AllIndexed',
-    source=r'''
-    def clear_AllIndexed(self):
-        ret = self.get_AllIndexed()
-        self.set_RegionRefIndexed([])
-        self.set_OrderedGroupIndexed([])
-        self.set_UnorderedGroupIndexed([])
-        return ret
-''', class_names=r'^(OrderedGroupType|OrderedGroupIndexedType)$')
-
-#
-# Add all *Indexed children sorted by @index
-#
-add_AllIndexed = MethodSpec(name='add_AllIndexed',
-    source=r'''
-    def add_AllIndexed(self, elements):
-        if not isinstance(elements, list):
-            elements = [elements]
-        for element in sorted(elements, key=lambda x : x.index):
-            if isinstance(element, RegionRefIndexedType):
-                self.add_RegionRefIndexed(element)
-            elif isinstance(element, OrderedGroupIndexedType):
-                self.add_OrderedGroupIndexed(element)
-            elif isinstance(element, UnorderedGroupIndexedType):
-                self.add_UnorderedGroupIndexed(element)
-        return self.get_AllIndexed()
-''', class_names=r'^(OrderedGroupType|OrderedGroupIndexedType)$')
-
+def _add_method(class_re, method_name):
+    """
+    Loads a file ./ocrd_page_user_methods/{{ method_name }}.py and defines a MethodSpec applying to class_re
+    """
+    source = []
+    with codecs.open(join(dirname(__file__), 'ocrd_page_user_methods', '%s.py' % method_name)) as f:
+        for line in f.readlines():
+            source.append('    %s' % line if line else line)
+    return MethodSpec(name=method_name, class_names=class_re, source=''.join(source))
 
-#
-# export children sorted by index of the childelement
-#
-exportChildren = MethodSpec(name='exportChildren',
-    source=r'''
-    def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
-        eol_ = '\n' if pretty_print else ''
-        namespaceprefix_ = 'pc:'
-        if self.UserDefined is not None:
-            self.UserDefined.export(outfile, level, namespaceprefix_, namespacedef_='', name_='UserDefined', pretty_print=pretty_print)
-        for Labels_ in self.Labels:
-            Labels_.export(outfile, level, namespaceprefix_, namespacedef_='', name_='Labels', pretty_print=pretty_print)
-        cleaned = []
-        # remove emtpy groups and replace with RegionRefIndexedType
-        for entry in self.get_AllIndexed():
-            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed():
-                rri = RegionRefIndexedType.factory(parent_object_=self)
-                rri.index = entry.index
-                rri.regionRef = entry.regionRef
-                cleaned.append(rri)
-            else:
-                cleaned.append(entry)
-        for entry in cleaned:
-            entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-''', class_names=r'^(OrderedGroupType|OrderedGroupIndexedType)$')
-#
-# Hash by memory adress/id()
-#
-hash_by_id = MethodSpec(name='hash',
-    source='''\
-    def __hash__(self):
-        return hash(self.id)
-''',
-    class_names=r'^.*$',
-    )
 #
 # Provide a list of your method specifications.
 #   This list of specifications must be named METHOD_SPECS.
 #
 METHOD_SPECS = (
-    hash_by_id,
-    exportChildren,
-    get_AllIndexed,
-    add_AllIndexed,
-    get_AllRegions,
-    clear_AllIndexed,
+    _add_method(r'^.*$', '__hash__'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'get_AllIndexed'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'clear_AllIndexed'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'add_AllIndexed'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'exportChildren'),
+    _add_method(r'^(PageType)$', 'get_AllRegions'),
     )
 
 
diff --git a/ocrd_models/ocrd_page_user_methods/__hash__.py b/ocrd_models/ocrd_page_user_methods/__hash__.py
new file mode 100644
index 0000000000..dc34db5932
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/__hash__.py
@@ -0,0 +1,2 @@
+def __hash__(self):
+    return hash(self.id)
diff --git a/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
new file mode 100644
index 0000000000..36e90ea9f6
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
@@ -0,0 +1,12 @@
+def add_AllIndexed(self, elements):
+    if not isinstance(elements, list):
+        elements = [elements]
+    for element in sorted(elements, key=lambda x : x.index):
+        if isinstance(element, RegionRefIndexedType):
+            self.add_RegionRefIndexed(element)
+        elif isinstance(element, OrderedGroupIndexedType):
+            self.add_OrderedGroupIndexed(element)
+        elif isinstance(element, UnorderedGroupIndexedType):
+            self.add_UnorderedGroupIndexed(element)
+    return self.get_AllIndexed()
+
diff --git a/ocrd_models/ocrd_page_user_methods/clear_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/clear_AllIndexed.py
new file mode 100644
index 0000000000..a213e53852
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/clear_AllIndexed.py
@@ -0,0 +1,7 @@
+def clear_AllIndexed(self):
+    ret = self.get_AllIndexed()
+    self.set_RegionRefIndexed([])
+    self.set_OrderedGroupIndexed([])
+    self.set_UnorderedGroupIndexed([])
+    return ret
+
diff --git a/ocrd_models/ocrd_page_user_methods/exportChildren.py b/ocrd_models/ocrd_page_user_methods/exportChildren.py
new file mode 100644
index 0000000000..9f53f66dde
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/exportChildren.py
@@ -0,0 +1,20 @@
+def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
+    eol_ = '\n' if pretty_print else ''
+    namespaceprefix_ = 'pc:'
+    if self.UserDefined is not None:
+        self.UserDefined.export(outfile, level, namespaceprefix_, namespacedef_='', name_='UserDefined', pretty_print=pretty_print)
+    for Labels_ in self.Labels:
+        Labels_.export(outfile, level, namespaceprefix_, namespacedef_='', name_='Labels', pretty_print=pretty_print)
+    cleaned = []
+    # remove emtpy groups and replace with RegionRefIndexedType
+    for entry in self.get_AllIndexed():
+        if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed():
+            rri = RegionRefIndexedType.factory(parent_object_=self)
+            rri.index = entry.index
+            rri.regionRef = entry.regionRef
+            cleaned.append(rri)
+        else:
+            cleaned.append(entry)
+    for entry in cleaned:
+        entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
+
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
new file mode 100644
index 0000000000..4d79632353
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
@@ -0,0 +1,3 @@
+def get_AllIndexed(self):
+    return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
new file mode 100644
index 0000000000..5fb645e350
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -0,0 +1,32 @@
+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:
+        reading_order = self.get_ReadingOrder()
+    if not regions:
+        regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
+    ret = []
+    for region in regions:
+        ret += getattr(self, 'get_{}Region'.format(region))()
+    if reading_order:
+        reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
+    if reading_order:
+        def get_recursive_reading_order(rogroup):
+            if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
+                elements = rogroup.get_AllIndexed()
+            if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+            regionrefs = list()
+            for elem in elements:
+                regionrefs.append(elem.get_regionRef())
+                if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                    regionrefs.extend(get_recursive_reading_order(elem))
+            return regionrefs
+        reading_order = get_recursive_reading_order(reading_order)
+    if reading_order:
+        ret = dict([(region.id, region) for region in ret])
+        ret = [ret[region_id] for region_id in reading_order if region_id in ret]
+    ret = [r in ret if r.__class__.__name__.replace('RegionType', '') in regions
+    return ret

From be7f026a8c1e8e213e53b1e3f562596fc876fa06 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Fri, 15 May 2020 16:50:55 +0200
Subject: [PATCH 05/30] get_AllRegions: adapt to signature proposed in #240,
 test with order=reading-order

---
 .../ocrd_models/ocrd_page_generateds.py       | 39 +++++++++++-
 .../ocrd_page_user_methods/get_AllRegions.py  | 59 ++++++++++---------
 tests/model/test_ocrd_page.py                 | 21 ++++++-
 3 files changed, 88 insertions(+), 31 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 8a3705a3e4..f6152110b1 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 15 16:15:27 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 15 16:48:56 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:
@@ -2850,6 +2850,43 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'CustomRegion'
     def __hash__(self):
         return hash(self.id)
+    def get_AllRegions(self, classes=None, order='document'):
+        """
+        Get all the *Region element or only those provided by ``classes``.
+        Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+        Arguments:
+            classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
+            order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
+        """
+        if order not in ['document', 'reading-order']:
+            raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
+        if not classes:
+            classes = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
+        def region_class(x):
+            return x.__class__.__name__.replace('RegionType', '')
+        ret = []
+        for region in classes:
+            ret += getattr(self, 'get_{}Region'.format(region))()
+        if order == 'reading-order':
+            reading_order = self.get_ReadingOrder().get_OrderedGroup() or reading_order.get_UnorderedGroup()
+            if reading_order:
+                def get_recursive_reading_order(rogroup):
+                    if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
+                        elements = rogroup.get_AllIndexed()
+                    if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                        elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+                    regionrefs = list()
+                    for elem in elements:
+                        regionrefs.append(elem.get_regionRef())
+                        if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                            regionrefs.extend(get_recursive_reading_order(elem))
+                    return regionrefs
+                reading_order = get_recursive_reading_order(reading_order)
+            if reading_order:
+                id2region = dict([(region.id, region) for region in ret])
+                ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+        ret = [r for r in ret if region_class(r) in classes]
+        return ret
 # end class PageType
 
 
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 5fb645e350..d8f8cf8c2d 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -1,32 +1,37 @@
-def get_AllRegions(self, regions=None, reading_order=False):
-    """"
-    Get all the *Region element or only those provided by ``regions``.
+def get_AllRegions(self, classes=None, order='document'):
+    """
+    Get all the *Region element or only those provided by ``classes``.
     Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+    Arguments:
+        classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
+        order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
     """
-    if reading_order:
-        reading_order = self.get_ReadingOrder()
-    if not regions:
-        regions = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
+    if order not in ['document', 'reading-order']:
+        raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
+    if not classes:
+        classes = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
+    def region_class(x):
+        return x.__class__.__name__.replace('RegionType', '')
     ret = []
-    for region in regions:
+    for region in classes:
         ret += getattr(self, 'get_{}Region'.format(region))()
-    if reading_order:
-        reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
-    if reading_order:
-        def get_recursive_reading_order(rogroup):
-            if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
-                elements = rogroup.get_AllIndexed()
-            if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
-                elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
-            regionrefs = list()
-            for elem in elements:
-                regionrefs.append(elem.get_regionRef())
-                if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
-                    regionrefs.extend(get_recursive_reading_order(elem))
-            return regionrefs
-        reading_order = get_recursive_reading_order(reading_order)
-    if reading_order:
-        ret = dict([(region.id, region) for region in ret])
-        ret = [ret[region_id] for region_id in reading_order if region_id in ret]
-    ret = [r in ret if r.__class__.__name__.replace('RegionType', '') in regions
+    if order == 'reading-order':
+        reading_order = self.get_ReadingOrder().get_OrderedGroup() or reading_order.get_UnorderedGroup()
+        if reading_order:
+            def get_recursive_reading_order(rogroup):
+                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
+                    elements = rogroup.get_AllIndexed()
+                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+                regionrefs = list()
+                for elem in elements:
+                    regionrefs.append(elem.get_regionRef())
+                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                        regionrefs.extend(get_recursive_reading_order(elem))
+                return regionrefs
+            reading_order = get_recursive_reading_order(reading_order)
+        if reading_order:
+            id2region = dict([(region.id, region) for region in ret])
+            ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+    ret = [r for r in ret if region_class(r) in classes]
     return ret
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index b24d1e45ff..7ee58e421f 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -193,9 +193,24 @@ def test_all_regions_without_reading_order(self):
             pcgts = parseString(f.read().encode('utf8'), silence=True)
             pg = pcgts.get_Page()
             self.assertEqual(len(pg.get_AllRegions()), 45)
-            self.assertEqual(len(pg.get_AllRegions(['Separator'])), 25)
-            self.assertEqual(len(pg.get_AllRegions(['Table'])), 3)
-            self.assertEqual(len(pg.get_AllRegions(['Text'])), 17)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Separator'])), 25)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Table'])), 3)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'])), 17)
+
+    def test_all_regions_with_reading_order(self):
+        """
+        https://github.com/OCR-D/core/pull/479
+        https://github.com/OCR-D/core/issues/240#issuecomment-493135797
+        """
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            pcgts = parseString(f.read().encode('utf8'), silence=True)
+            pg = pcgts.get_Page()
+            with self.assertRaisesRegex(Exception, "Argument 'order' must be either 'document' or 'reading-order', not 'random'"):
+                pg.get_AllRegions(order='random')
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order')), 20)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 17)
 
 if __name__ == '__main__':
     main()

From e1740f78ac80d5ce48de6f8d0d8905fd737feb44 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <unixprog@gmail.com>
Date: Fri, 15 May 2020 16:58:19 +0200
Subject: [PATCH 06/30] README: explain how to add user methods to PAGE API

---
 ocrd_models/README.md | 27 +++++++++++++++++++++++++++
 1 file changed, 27 insertions(+)

diff --git a/ocrd_models/README.md b/ocrd_models/README.md
index 9564566c9f..c98d69b9b8 100644
--- a/ocrd_models/README.md
+++ b/ocrd_models/README.md
@@ -3,3 +3,30 @@
 > OCR-D framework - file format APIs and schemas
 
 See https://github.com/OCR-D/core
+
+## Adding user methods to the generated PAGE API
+
+Let's say you want to add a method `get_FirstTextRegion` on the pc:Page element:
+
+1. Create a file `ocrd_models/ocrd_page_user_methods/get_FirstTextRegion.py`
+
+```python
+def getFirstTextRegion(self)
+  return self.get_TextRegion[0]
+```
+
+2. Edit `ocrd_models/ocrd_page_user_methods.py` and append to the `METHOD_SPECS` list:
+
+```python
+METHOD_SPECS = (
+  # ...
+  _add_method(r'^PageType$', 'get_TextRegion')
+  # ...
+)
+```
+
+3. Regenerate the PAGE API:
+
+```sh
+make generate-page
+```

From 6f9163ee8e64c70c1f937994af9f8e611beb8c71 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Thu, 28 May 2020 13:39:39 +0200
Subject: [PATCH 07/30] Update ocrd_models/README.md

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocrd_models/README.md b/ocrd_models/README.md
index c98d69b9b8..47901e654c 100644
--- a/ocrd_models/README.md
+++ b/ocrd_models/README.md
@@ -20,7 +20,7 @@ def getFirstTextRegion(self)
 ```python
 METHOD_SPECS = (
   # ...
-  _add_method(r'^PageType$', 'get_TextRegion')
+  _add_method(r'^PageType$', 'get_FirstTextRegion')
   # ...
 )
 ```

From 0c73b3e7f56639c0d7e545dbb890117e2e67f76f Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Thu, 28 May 2020 13:40:14 +0200
Subject: [PATCH 08/30] Update ocrd_models/README.md

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/README.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocrd_models/README.md b/ocrd_models/README.md
index 47901e654c..44b6e95a65 100644
--- a/ocrd_models/README.md
+++ b/ocrd_models/README.md
@@ -11,7 +11,7 @@ Let's say you want to add a method `get_FirstTextRegion` on the pc:Page element:
 1. Create a file `ocrd_models/ocrd_page_user_methods/get_FirstTextRegion.py`
 
 ```python
-def getFirstTextRegion(self)
+def get_FirstTextRegion(self):
   return self.get_TextRegion[0]
 ```
 

From 5c2f3a8028e54322fdccfe7a9578319a8582b785 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Thu, 28 May 2020 13:40:59 +0200
Subject: [PATCH 09/30] Update ocrd_models/README.md

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/README.md | 1 +
 1 file changed, 1 insertion(+)

diff --git a/ocrd_models/README.md b/ocrd_models/README.md
index 44b6e95a65..7e6ec6216b 100644
--- a/ocrd_models/README.md
+++ b/ocrd_models/README.md
@@ -15,6 +15,7 @@ def get_FirstTextRegion(self):
   return self.get_TextRegion[0]
 ```
 
+   (Note that the method name and file name must be identical.)
 2. Edit `ocrd_models/ocrd_page_user_methods.py` and append to the `METHOD_SPECS` list:
 
 ```python

From 6a575061b383edb213e580beb37d82349b2c4e99 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Thu, 28 May 2020 13:51:45 +0200
Subject: [PATCH 10/30] recursion (with both finite or arbitrary depth) for
 get_AllRegions

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 .../ocrd_page_user_methods/get_AllRegions.py  | 31 ++++++++++++++-----
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index d8f8cf8c2d..bda88180e2 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -8,15 +8,33 @@ def get_AllRegions(self, classes=None, order='document'):
     """
     if order not in ['document', 'reading-order']:
         raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-    if not classes:
-        classes = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
     def region_class(x):
         return x.__class__.__name__.replace('RegionType', '')
-    ret = []
-    for region in classes:
-        ret += getattr(self, 'get_{}Region'.format(region))()
+    def get_recursive_regions(regions, level):
+        if level == 1:
+            # stop recursion, filter classes
+            if classes:
+                return [r for r in regions if region_class(r) in classes]
+            else:
+                return regions
+        # find more regions recursively
+        more_regions = []
+        for region in regions:
+            more_regions.append([])
+            for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']:
+                if class_ == 'Map' and not isinstance(region, PageType):
+                    # 'Map' is not recursive in 2019 schema
+                    continue
+                 more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
+        if not any(more_regions):
+            return get_recursive_regions(regions, 1)
+        regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
+        return get_recursive_regions(regions, level - 1 if level else 0)
+    ret = get_recursive_regions([self], depth + 1 if depth else 0)
     if order == 'reading-order':
-        reading_order = self.get_ReadingOrder().get_OrderedGroup() or reading_order.get_UnorderedGroup()
+        reading_order = self.get_ReadingOrder()
+        if reading_order:
+            reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
         if reading_order:
             def get_recursive_reading_order(rogroup):
                 if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
@@ -33,5 +51,4 @@ def get_recursive_reading_order(rogroup):
         if reading_order:
             id2region = dict([(region.id, region) for region in ret])
             ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
-    ret = [r for r in ret if region_class(r) in classes]
     return ret

From a9072c8a92dfee62f72a1c04afd3413839e128e9 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 14:26:15 +0200
Subject: [PATCH 11/30] regenerate PAGE API

---
 .../ocrd_models/ocrd_page_generateds.py       | 39 +++++++++++++------
 .../ocrd_page_user_methods/get_AllRegions.py  |  2 +-
 2 files changed, 29 insertions(+), 12 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index f6152110b1..0d6d27940c 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,8 +2,8 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 15 16:48:56 2020 by generateDS.py version 2.35.20.
-# Python 3.7.6 (default, Jan  8 2020, 19:59:22)  [GCC 7.3.0]
+# Generated Thu May 28 14:25:40 2020 by generateDS.py version 2.35.20.
+# Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
 #   ('-f', '')
@@ -16,7 +16,7 @@
 #   repo/assets/data/schema/data/2019.xsd
 #
 # Command line:
-#   /home/kba/miniconda3/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
+#   /home/kba/ocrd_all/venv/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" repo/assets/data/schema/data/2019.xsd
 #
 # Current working directory (os.getcwd()):
 #   core
@@ -2853,22 +2853,40 @@ def __hash__(self):
     def get_AllRegions(self, classes=None, order='document'):
         """
         Get all the *Region element or only those provided by ``classes``.
-        Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+        Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
         Arguments:
             classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
             order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
         """
         if order not in ['document', 'reading-order']:
             raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-        if not classes:
-            classes = ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']
         def region_class(x):
             return x.__class__.__name__.replace('RegionType', '')
-        ret = []
-        for region in classes:
-            ret += getattr(self, 'get_{}Region'.format(region))()
+        def get_recursive_regions(regions, level):
+            if level == 1:
+                # stop recursion, filter classes
+                if classes:
+                    return [r for r in regions if region_class(r) in classes]
+                else:
+                    return regions
+            # find more regions recursively
+            more_regions = []
+            for region in regions:
+                more_regions.append([])
+                for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']:
+                    if class_ == 'Map' and not isinstance(region, PageType):
+                        # 'Map' is not recursive in 2019 schema
+                        continue
+                     more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
+            if not any(more_regions):
+                return get_recursive_regions(regions, 1)
+            regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
+            return get_recursive_regions(regions, level - 1 if level else 0)
+        ret = get_recursive_regions([self], depth + 1 if depth else 0)
         if order == 'reading-order':
-            reading_order = self.get_ReadingOrder().get_OrderedGroup() or reading_order.get_UnorderedGroup()
+            reading_order = self.get_ReadingOrder()
+            if reading_order:
+                reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
             if reading_order:
                 def get_recursive_reading_order(rogroup):
                     if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
@@ -2885,7 +2903,6 @@ def get_recursive_reading_order(rogroup):
             if reading_order:
                 id2region = dict([(region.id, region) for region in ret])
                 ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
-        ret = [r for r in ret if region_class(r) in classes]
         return ret
 # end class PageType
 
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index bda88180e2..62dbe74e35 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -1,7 +1,7 @@
 def get_AllRegions(self, classes=None, order='document'):
     """
     Get all the *Region element or only those provided by ``classes``.
-    Returned in random order unless ``reading_order`` is set (NOT CURRENTLY IMPLEMENTED)
+    Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
     Arguments:
         classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
         order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order

From ac62b857476967b44f7d81115424ca9c51fdbf93 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 15:59:05 +0200
Subject: [PATCH 12/30] get_AllRegions: clean-up merge artifacts and reorganize

---
 .../ocrd_models/ocrd_page_generateds.py       | 88 +++++++++++--------
 .../ocrd_page_user_methods/get_AllRegions.py  | 86 ++++++++++--------
 2 files changed, 97 insertions(+), 77 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 0d6d27940c..aac3997445 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 14:25:40 2020 by generateDS.py version 2.35.20.
+# Generated Thu May 28 15:57:42 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2850,58 +2850,68 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'CustomRegion'
     def __hash__(self):
         return hash(self.id)
-    def get_AllRegions(self, classes=None, order='document'):
+    # pylint: disable=line-too-long,invalid-name,protected-access,missing-module-docstring
+    def _region_class(self, x): # pylint: disable=unused-argument
+        return x.__class__.__name__.replace('RegionType', '')
+    
+    def _region_id(self, x): # pylint: disable=unused-argument
+        return x.id if hasattr(x, 'id') else x.pcGtsId
+    
+    def _get_recursive_regions(self, regions, level, classes=None):
+        if level == 1:
+            # stop recursion, filter classes
+            if classes:
+                return [r for r in regions if self._region_class(r) in classes]
+            # remove the first element (PageType)
+            return regions[1:]
+        # find more regions recursively
+        more_regions = []
+        for region in regions:
+            more_regions.append([])
+            for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image',
+                           'LineDrawing', 'Map', 'Maths', 'Music', 'Noise',
+                           'Separator', 'Table', 'Text', 'Unknown']:
+                if class_ == 'Map' and not isinstance(region, PageType): # pylint: disable=undefined-variable
+                    # 'Map' is not recursive in 2019 schema
+                    continue
+                more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
+        if not any(more_regions):
+            return _get_recursive_regions(regions, 1, classes)
+        regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
+        return self._get_recursive_regions(regions, level - 1 if level else 0, classes)
+    
+    def _get_recursive_reading_order(self, rogroup):
+        if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)): # pylint: disable=undefined-variable
+            elements = rogroup.get_AllIndexed()
+        if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)): # pylint: disable=undefined-variable
+            elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+        regionrefs = list()
+        for elem in elements:
+            regionrefs.append(elem.get_regionRef())
+            if not isinstance(elem, (RegionRefType, RegionRefIndexedType)): # pylint: disable=undefined-variable
+                regionrefs.extend(self._get_recursive_reading_order(elem))
+        return regionrefs
+    
+    def get_AllRegions(self, classes=None, order='document', depth=1):
         """
         Get all the *Region element or only those provided by ``classes``.
         Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
         Arguments:
-            classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
+            classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
             order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
+            depth (int) Recursive depth to look for regions at. Default: 1
         """
         if order not in ['document', 'reading-order']:
             raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-        def region_class(x):
-            return x.__class__.__name__.replace('RegionType', '')
-        def get_recursive_regions(regions, level):
-            if level == 1:
-                # stop recursion, filter classes
-                if classes:
-                    return [r for r in regions if region_class(r) in classes]
-                else:
-                    return regions
-            # find more regions recursively
-            more_regions = []
-            for region in regions:
-                more_regions.append([])
-                for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']:
-                    if class_ == 'Map' and not isinstance(region, PageType):
-                        # 'Map' is not recursive in 2019 schema
-                        continue
-                     more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
-            if not any(more_regions):
-                return get_recursive_regions(regions, 1)
-            regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
-            return get_recursive_regions(regions, level - 1 if level else 0)
-        ret = get_recursive_regions([self], depth + 1 if depth else 0)
+        ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
         if order == 'reading-order':
             reading_order = self.get_ReadingOrder()
             if reading_order:
                 reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
             if reading_order:
-                def get_recursive_reading_order(rogroup):
-                    if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
-                        elements = rogroup.get_AllIndexed()
-                    if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
-                        elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
-                    regionrefs = list()
-                    for elem in elements:
-                        regionrefs.append(elem.get_regionRef())
-                        if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
-                            regionrefs.extend(get_recursive_reading_order(elem))
-                    return regionrefs
-                reading_order = get_recursive_reading_order(reading_order)
+                reading_order = self._get_recursive_reading_order(reading_order)
             if reading_order:
-                id2region = dict([(region.id, region) for region in ret])
+                id2region = dict([(self._region_id(region), region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
                 ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
         return ret
 # end class PageType
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 62dbe74e35..6f3b2342a9 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -1,54 +1,64 @@
-def get_AllRegions(self, classes=None, order='document'):
+# pylint: disable=line-too-long,invalid-name,protected-access,missing-module-docstring
+def _region_class(self, x): # pylint: disable=unused-argument
+    return x.__class__.__name__.replace('RegionType', '')
+
+def _region_id(self, x): # pylint: disable=unused-argument
+    return x.id if hasattr(x, 'id') else x.pcGtsId
+
+def _get_recursive_regions(self, regions, level, classes=None):
+    if level == 1:
+        # stop recursion, filter classes
+        if classes:
+            return [r for r in regions if self._region_class(r) in classes]
+        # remove the first element (PageType)
+        return regions[1:]
+    # find more regions recursively
+    more_regions = []
+    for region in regions:
+        more_regions.append([])
+        for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image',
+                       'LineDrawing', 'Map', 'Maths', 'Music', 'Noise',
+                       'Separator', 'Table', 'Text', 'Unknown']:
+            if class_ == 'Map' and not isinstance(region, PageType): # pylint: disable=undefined-variable
+                # 'Map' is not recursive in 2019 schema
+                continue
+            more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
+    if not any(more_regions):
+        return _get_recursive_regions(regions, 1, classes)
+    regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
+    return self._get_recursive_regions(regions, level - 1 if level else 0, classes)
+
+def _get_recursive_reading_order(self, rogroup):
+    if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)): # pylint: disable=undefined-variable
+        elements = rogroup.get_AllIndexed()
+    if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)): # pylint: disable=undefined-variable
+        elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
+    regionrefs = list()
+    for elem in elements:
+        regionrefs.append(elem.get_regionRef())
+        if not isinstance(elem, (RegionRefType, RegionRefIndexedType)): # pylint: disable=undefined-variable
+            regionrefs.extend(self._get_recursive_reading_order(elem))
+    return regionrefs
+
+def get_AllRegions(self, classes=None, order='document', depth=1):
     """
     Get all the *Region element or only those provided by ``classes``.
     Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
     Arguments:
-        classes (list) Classes of regions that shall be returned, e.g. ['Text', 'Image']
+        classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
         order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
+        depth (int) Recursive depth to look for regions at. Default: 1
     """
     if order not in ['document', 'reading-order']:
         raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-    def region_class(x):
-        return x.__class__.__name__.replace('RegionType', '')
-    def get_recursive_regions(regions, level):
-        if level == 1:
-            # stop recursion, filter classes
-            if classes:
-                return [r for r in regions if region_class(r) in classes]
-            else:
-                return regions
-        # find more regions recursively
-        more_regions = []
-        for region in regions:
-            more_regions.append([])
-            for class_ in ['Advert', 'Chart', 'Chem', 'Custom', 'Graphic', 'Image', 'LineDrawing', 'Map', 'Maths', 'Music', 'Noise', 'Separator', 'Table', 'Text', 'Unknown']:
-                if class_ == 'Map' and not isinstance(region, PageType):
-                    # 'Map' is not recursive in 2019 schema
-                    continue
-                 more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
-        if not any(more_regions):
-            return get_recursive_regions(regions, 1)
-        regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
-        return get_recursive_regions(regions, level - 1 if level else 0)
-    ret = get_recursive_regions([self], depth + 1 if depth else 0)
+    ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
     if order == 'reading-order':
         reading_order = self.get_ReadingOrder()
         if reading_order:
             reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
         if reading_order:
-            def get_recursive_reading_order(rogroup):
-                if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)):
-                    elements = rogroup.get_AllIndexed()
-                if isinstance(rogroup, (UnorderedGroupType, UnorderedGroupIndexedType)):
-                    elements = (rogroup.get_RegionRef() + rogroup.get_OrderedGroup() + rogroup.get_UnorderedGroup())
-                regionrefs = list()
-                for elem in elements:
-                    regionrefs.append(elem.get_regionRef())
-                    if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
-                        regionrefs.extend(get_recursive_reading_order(elem))
-                return regionrefs
-            reading_order = get_recursive_reading_order(reading_order)
+            reading_order = self._get_recursive_reading_order(reading_order)
         if reading_order:
-            id2region = dict([(region.id, region) for region in ret])
+            id2region = dict([(self._region_id(region), region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
             ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
     return ret

From fd6d54541248ca605f849abf72395eb657658c59 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Thu, 28 May 2020 16:09:05 +0200
Subject: [PATCH 13/30] Update
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 6f3b2342a9..b7eb4e68bb 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -43,7 +43,7 @@ def _get_recursive_reading_order(self, rogroup):
 def get_AllRegions(self, classes=None, order='document', depth=1):
     """
     Get all the *Region element or only those provided by ``classes``.
-    Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
+    Returned in document order unless ``order`` is ``reading-order``
     Arguments:
         classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
         order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order

From 86a7133579a11dfd3dac100d1ce38ca659d601bb Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 16:10:17 +0200
Subject: [PATCH 14/30] get_AllRegions: _region_id method unneccessary now

---
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)

diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 6f3b2342a9..ace12ff622 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -2,9 +2,6 @@
 def _region_class(self, x): # pylint: disable=unused-argument
     return x.__class__.__name__.replace('RegionType', '')
 
-def _region_id(self, x): # pylint: disable=unused-argument
-    return x.id if hasattr(x, 'id') else x.pcGtsId
-
 def _get_recursive_regions(self, regions, level, classes=None):
     if level == 1:
         # stop recursion, filter classes
@@ -59,6 +56,6 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
         if reading_order:
             reading_order = self._get_recursive_reading_order(reading_order)
         if reading_order:
-            id2region = dict([(self._region_id(region), region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
+            id2region = dict([(region.id, region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
             ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
     return ret

From 5c8d89bc8a9ed6815202a6e3281ad97b81e87f20 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 16:11:16 +0200
Subject: [PATCH 15/30] regenerate PAGE API

---
 ocrd_models/ocrd_models/ocrd_page_generateds.py | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index aac3997445..5725af42b5 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 15:57:42 2020 by generateDS.py version 2.35.20.
+# Generated Thu May 28 16:10:37 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2854,9 +2854,6 @@ def __hash__(self):
     def _region_class(self, x): # pylint: disable=unused-argument
         return x.__class__.__name__.replace('RegionType', '')
     
-    def _region_id(self, x): # pylint: disable=unused-argument
-        return x.id if hasattr(x, 'id') else x.pcGtsId
-    
     def _get_recursive_regions(self, regions, level, classes=None):
         if level == 1:
             # stop recursion, filter classes
@@ -2895,7 +2892,7 @@ def _get_recursive_reading_order(self, rogroup):
     def get_AllRegions(self, classes=None, order='document', depth=1):
         """
         Get all the *Region element or only those provided by ``classes``.
-        Returned in document order unless ``order`` is ``reading-order`` is set (NOT CURRENTLY IMPLEMENTED)
+        Returned in document order unless ``order`` is ``reading-order``
         Arguments:
             classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
             order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
@@ -2911,7 +2908,7 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
             if reading_order:
                 reading_order = self._get_recursive_reading_order(reading_order)
             if reading_order:
-                id2region = dict([(self._region_id(region), region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
+                id2region = dict([(region.id, region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
                 ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
         return ret
 # end class PageType

From f6e3da5fb6162f9ecae31cd76c04d406e849e3ff Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 17:03:19 +0200
Subject: [PATCH 16/30] :art: pylint

---
 .../ocrd_models/ocrd_page_generateds.py       | 38 +++++++++----------
 .../ocrd_page_user_methods/add_AllIndexed.py  |  9 +++--
 .../ocrd_page_user_methods/exportChildren.py  |  9 ++---
 3 files changed, 28 insertions(+), 28 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 5725af42b5..a9c0484dc7 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 16:10:37 2020 by generateDS.py version 2.35.20.
+# Generated Thu May 28 17:02:23 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -5418,20 +5418,21 @@ def clear_AllIndexed(self):
         self.set_UnorderedGroupIndexed([])
         return ret
     
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
     def add_AllIndexed(self, elements):
         if not isinstance(elements, list):
             elements = [elements]
-        for element in sorted(elements, key=lambda x : x.index):
-            if isinstance(element, RegionRefIndexedType):
+        for element in sorted(elements, key=lambda x: x.index):
+            if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
                 self.add_RegionRefIndexed(element)
-            elif isinstance(element, OrderedGroupIndexedType):
+            elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
                 self.add_OrderedGroupIndexed(element)
-            elif isinstance(element, UnorderedGroupIndexedType):
+            elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
                 self.add_UnorderedGroupIndexed(element)
         return self.get_AllIndexed()
     
-    def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
-        eol_ = '\n' if pretty_print else ''
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+    def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True): # pylint: disable=unused-argument,too-many-arguments
         namespaceprefix_ = 'pc:'
         if self.UserDefined is not None:
             self.UserDefined.export(outfile, level, namespaceprefix_, namespacedef_='', name_='UserDefined', pretty_print=pretty_print)
@@ -5440,8 +5441,8 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
         cleaned = []
         # remove emtpy groups and replace with RegionRefIndexedType
         for entry in self.get_AllIndexed():
-            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed():
-                rri = RegionRefIndexedType.factory(parent_object_=self)
+            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
+                rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
                 rri.index = entry.index
                 rri.regionRef = entry.regionRef
                 cleaned.append(rri)
@@ -5449,7 +5450,6 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
                 cleaned.append(entry)
         for entry in cleaned:
             entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-    
 # end class OrderedGroupIndexedType
 
 
@@ -6208,20 +6208,21 @@ def clear_AllIndexed(self):
         self.set_UnorderedGroupIndexed([])
         return ret
     
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
     def add_AllIndexed(self, elements):
         if not isinstance(elements, list):
             elements = [elements]
-        for element in sorted(elements, key=lambda x : x.index):
-            if isinstance(element, RegionRefIndexedType):
+        for element in sorted(elements, key=lambda x: x.index):
+            if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
                 self.add_RegionRefIndexed(element)
-            elif isinstance(element, OrderedGroupIndexedType):
+            elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
                 self.add_OrderedGroupIndexed(element)
-            elif isinstance(element, UnorderedGroupIndexedType):
+            elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
                 self.add_UnorderedGroupIndexed(element)
         return self.get_AllIndexed()
     
-    def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
-        eol_ = '\n' if pretty_print else ''
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+    def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True): # pylint: disable=unused-argument,too-many-arguments
         namespaceprefix_ = 'pc:'
         if self.UserDefined is not None:
             self.UserDefined.export(outfile, level, namespaceprefix_, namespacedef_='', name_='UserDefined', pretty_print=pretty_print)
@@ -6230,8 +6231,8 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
         cleaned = []
         # remove emtpy groups and replace with RegionRefIndexedType
         for entry in self.get_AllIndexed():
-            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed():
-                rri = RegionRefIndexedType.factory(parent_object_=self)
+            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
+                rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
                 rri.index = entry.index
                 rri.regionRef = entry.regionRef
                 cleaned.append(rri)
@@ -6239,7 +6240,6 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
                 cleaned.append(entry)
         for entry in cleaned:
             entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-    
 # end class OrderedGroupType
 
 
diff --git a/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
index 36e90ea9f6..45d73eaacc 100644
--- a/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
@@ -1,12 +1,13 @@
+# pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
 def add_AllIndexed(self, elements):
     if not isinstance(elements, list):
         elements = [elements]
-    for element in sorted(elements, key=lambda x : x.index):
-        if isinstance(element, RegionRefIndexedType):
+    for element in sorted(elements, key=lambda x: x.index):
+        if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
             self.add_RegionRefIndexed(element)
-        elif isinstance(element, OrderedGroupIndexedType):
+        elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
             self.add_OrderedGroupIndexed(element)
-        elif isinstance(element, UnorderedGroupIndexedType):
+        elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
             self.add_UnorderedGroupIndexed(element)
     return self.get_AllIndexed()
 
diff --git a/ocrd_models/ocrd_page_user_methods/exportChildren.py b/ocrd_models/ocrd_page_user_methods/exportChildren.py
index 9f53f66dde..85c7f660c2 100644
--- a/ocrd_models/ocrd_page_user_methods/exportChildren.py
+++ b/ocrd_models/ocrd_page_user_methods/exportChildren.py
@@ -1,5 +1,5 @@
-def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True):
-    eol_ = '\n' if pretty_print else ''
+# pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True): # pylint: disable=unused-argument,too-many-arguments
     namespaceprefix_ = 'pc:'
     if self.UserDefined is not None:
         self.UserDefined.export(outfile, level, namespaceprefix_, namespacedef_='', name_='UserDefined', pretty_print=pretty_print)
@@ -8,8 +8,8 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
     cleaned = []
     # remove emtpy groups and replace with RegionRefIndexedType
     for entry in self.get_AllIndexed():
-        if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed():
-            rri = RegionRefIndexedType.factory(parent_object_=self)
+        if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
+            rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
             rri.index = entry.index
             rri.regionRef = entry.regionRef
             cleaned.append(rri)
@@ -17,4 +17,3 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
             cleaned.append(entry)
     for entry in cleaned:
         entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
-

From 8351056a3a40ab39fa4bb9e0a7079e3ba7f6a5dc Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 17:10:56 +0200
Subject: [PATCH 17/30] add_AllIndexed -> extend_AllIndexed

---
 ocrd_models/ocrd_models/ocrd_page_generateds.py             | 6 +++---
 ocrd_models/ocrd_page_user_methods.py                       | 2 +-
 .../{add_AllIndexed.py => extend_AllIndexed.py}             | 2 +-
 3 files changed, 5 insertions(+), 5 deletions(-)
 rename ocrd_models/ocrd_page_user_methods/{add_AllIndexed.py => extend_AllIndexed.py} (94%)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index a9c0484dc7..ed8697d9fa 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 17:02:23 2020 by generateDS.py version 2.35.20.
+# Generated Thu May 28 17:07:14 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -5419,7 +5419,7 @@ def clear_AllIndexed(self):
         return ret
     
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
-    def add_AllIndexed(self, elements):
+    def extend_AllIndexed(self, elements):
         if not isinstance(elements, list):
             elements = [elements]
         for element in sorted(elements, key=lambda x: x.index):
@@ -6209,7 +6209,7 @@ def clear_AllIndexed(self):
         return ret
     
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
-    def add_AllIndexed(self, elements):
+    def extend_AllIndexed(self, elements):
         if not isinstance(elements, list):
             elements = [elements]
         for element in sorted(elements, key=lambda x: x.index):
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index 7f93323794..42f5bfa481 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -100,7 +100,7 @@ def _add_method(class_re, method_name):
     _add_method(r'^.*$', '__hash__'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'get_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'clear_AllIndexed'),
-    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'add_AllIndexed'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'extend_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'exportChildren'),
     _add_method(r'^(PageType)$', 'get_AllRegions'),
     )
diff --git a/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
similarity index 94%
rename from ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
rename to ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
index 45d73eaacc..fb129f5b5f 100644
--- a/ocrd_models/ocrd_page_user_methods/add_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
@@ -1,5 +1,5 @@
 # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
-def add_AllIndexed(self, elements):
+def extend_AllIndexed(self, elements):
     if not isinstance(elements, list):
         elements = [elements]
     for element in sorted(elements, key=lambda x: x.index):

From f2022057c16fe2e3c49f571aee75b21681ebdc75 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Thu, 28 May 2020 20:30:41 +0200
Subject: [PATCH 18/30] get_AllRegions: differentiate
 "reading-order"/"reading-order-only"

---
 ocrd_models/README.md                         |  5 +++-
 .../ocrd_models/ocrd_page_generateds.py       | 29 ++++++++++++++-----
 .../ocrd_page_user_methods/get_AllRegions.py  | 27 ++++++++++++-----
 tests/model/test_ocrd_page.py                 | 10 +++++--
 4 files changed, 53 insertions(+), 18 deletions(-)

diff --git a/ocrd_models/README.md b/ocrd_models/README.md
index 7e6ec6216b..36ebf381d0 100644
--- a/ocrd_models/README.md
+++ b/ocrd_models/README.md
@@ -15,7 +15,10 @@ def get_FirstTextRegion(self):
   return self.get_TextRegion[0]
 ```
 
-   (Note that the method name and file name must be identical.)
+ **NOTE** The method name and file name must be identical.
+
+ **NOTE** Do not use Python's `%` string interpolation operator, it will break generateDS. Use `"".format(...)` instead.
+
 2. Edit `ocrd_models/ocrd_page_user_methods.py` and append to the `METHOD_SPECS` list:
 
 ```python
diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index ed8697d9fa..1bc34f8f4c 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 17:07:14 2020 by generateDS.py version 2.35.20.
+# Generated Thu May 28 20:28:48 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2860,7 +2860,7 @@ def _get_recursive_regions(self, regions, level, classes=None):
             if classes:
                 return [r for r in regions if self._region_class(r) in classes]
             # remove the first element (PageType)
-            return regions[1:]
+            return list(set(regions[1:]))
         # find more regions recursively
         more_regions = []
         for region in regions:
@@ -2895,21 +2895,34 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
         Returned in document order unless ``order`` is ``reading-order``
         Arguments:
             classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
-            order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
+            order ("document"|"reading-order"|"reading-order-only") Whether to
+                return regions sorted by document order (``document``, default) or by
+                reading order with regions not in the reading order at the end of the
+                returned list (``reading-order``) or regions not in the reading order
+                omitted (``reading-order-only``)
             depth (int) Recursive depth to look for regions at. Default: 1
         """
-        if order not in ['document', 'reading-order']:
+        if order not in ['document', 'reading-order', 'reading-order-only']:
             raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-        ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
-        if order == 'reading-order':
+        ret = self._get_recursive_regions([self], depth + 1, classes)
+        if order.startswith('reading-order'):
             reading_order = self.get_ReadingOrder()
             if reading_order:
                 reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
             if reading_order:
                 reading_order = self._get_recursive_reading_order(reading_order)
             if reading_order:
-                id2region = dict([(region.id, region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
-                ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+                id2region = {region.id: region for region in ret}
+                in_reading_order = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+                #  print("ret: {} / in_ro: {} / not-in-ro: {}".format(
+                #      len(ret),
+                #      len([id2region[region_id] for region_id in reading_order if region_id in id2region]),
+                #      len([r for r in ret if r not in in_reading_order])
+                #      ))
+                if order == 'reading-order-only':
+                    ret = in_reading_order
+                else:
+                    ret = in_reading_order + [r for r in ret if r not in in_reading_order]
         return ret
 # end class PageType
 
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 98a539c90d..8dbf7de135 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -8,7 +8,7 @@ def _get_recursive_regions(self, regions, level, classes=None):
         if classes:
             return [r for r in regions if self._region_class(r) in classes]
         # remove the first element (PageType)
-        return regions[1:]
+        return list(set(regions[1:]))
     # find more regions recursively
     more_regions = []
     for region in regions:
@@ -43,19 +43,32 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
     Returned in document order unless ``order`` is ``reading-order``
     Arguments:
         classes (list) Classes of regions that shall be returned, e.g. ``['Text', 'Image']``
-        order ("document"|"reading-order") Whether to return regions sorted by document order (default) or by reading order
+        order ("document"|"reading-order"|"reading-order-only") Whether to
+            return regions sorted by document order (``document``, default) or by
+            reading order with regions not in the reading order at the end of the
+            returned list (``reading-order``) or regions not in the reading order
+            omitted (``reading-order-only``)
         depth (int) Recursive depth to look for regions at. Default: 1
     """
-    if order not in ['document', 'reading-order']:
+    if order not in ['document', 'reading-order', 'reading-order-only']:
         raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
-    ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
-    if order == 'reading-order':
+    ret = self._get_recursive_regions([self], depth + 1, classes)
+    if order.startswith('reading-order'):
         reading_order = self.get_ReadingOrder()
         if reading_order:
             reading_order = reading_order.get_OrderedGroup() or reading_order.get_UnorderedGroup()
         if reading_order:
             reading_order = self._get_recursive_reading_order(reading_order)
         if reading_order:
-            id2region = dict([(region.id, region) for region in ret]) # pylint: disable=consider-using-dict-comprehension
-            ret = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+            id2region = {region.id: region for region in ret}
+            in_reading_order = [id2region[region_id] for region_id in reading_order if region_id in id2region]
+            #  print("ret: {} / in_ro: {} / not-in-ro: {}".format(
+            #      len(ret),
+            #      len([id2region[region_id] for region_id in reading_order if region_id in id2region]),
+            #      len([r for r in ret if r not in in_reading_order])
+            #      ))
+            if order == 'reading-order-only':
+                ret = in_reading_order
+            else:
+                ret = in_reading_order + [r for r in ret if r not in in_reading_order]
     return ret
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 7ee58e421f..ce36433443 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -193,9 +193,13 @@ def test_all_regions_without_reading_order(self):
             pcgts = parseString(f.read().encode('utf8'), silence=True)
             pg = pcgts.get_Page()
             self.assertEqual(len(pg.get_AllRegions()), 45)
+            self.assertEqual(len(pg.get_AllRegions(depth=1)), 45)
+            self.assertEqual(len(pg.get_AllRegions(depth=2)), 65)
+            self.assertEqual(len(pg.get_AllRegions(depth=3)), 65)
             self.assertEqual(len(pg.get_AllRegions(classes=['Separator'])), 25)
             self.assertEqual(len(pg.get_AllRegions(classes=['Table'])), 3)
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'])), 17)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=2)), 54)
 
     def test_all_regions_with_reading_order(self):
         """
@@ -207,8 +211,10 @@ def test_all_regions_with_reading_order(self):
             pg = pcgts.get_Page()
             with self.assertRaisesRegex(Exception, "Argument 'order' must be either 'document' or 'reading-order', not 'random'"):
                 pg.get_AllRegions(order='random')
-            self.assertEqual(len(pg.get_AllRegions(order='reading-order')), 20)
-            self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order-only')), 20)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order-only', depth=2)), 40)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=1)), 45)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=2)), 65)
             self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 17)
 

From ffba6f937dcb6418104025fad8fe8dfaa10ce25c Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 13:40:22 +0200
Subject: [PATCH 19/30] get_AllRegions: catch negative depth, test depth==0

---
 ocrd_models/ocrd_models/ocrd_page_generateds.py      | 6 ++++--
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py | 4 +++-
 tests/model/test_ocrd_page.py                        | 9 ++++++---
 3 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 1bc34f8f4c..a1aa2f976c 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Thu May 28 20:28:48 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 13:39:11 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2903,7 +2903,9 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
             depth (int) Recursive depth to look for regions at. Default: 1
         """
         if order not in ['document', 'reading-order', 'reading-order-only']:
-            raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
+            raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))
+        if depth < 0:
+            raise Exception("Argument 'depth' must be an integer greater-or-equal 0, not '{}'".format(depth))
         ret = self._get_recursive_regions([self], depth + 1, classes)
         if order.startswith('reading-order'):
             reading_order = self.get_ReadingOrder()
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 8dbf7de135..90d4a193f6 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -51,7 +51,9 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
         depth (int) Recursive depth to look for regions at. Default: 1
     """
     if order not in ['document', 'reading-order', 'reading-order-only']:
-        raise Exception("Argument 'order' must be either 'document' or 'reading-order', not '{}'".format(order))
+        raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))
+    if depth < 0:
+        raise Exception("Argument 'depth' must be an integer greater-or-equal 0, not '{}'".format(depth))
     ret = self._get_recursive_regions([self], depth + 1, classes)
     if order.startswith('reading-order'):
         reading_order = self.get_ReadingOrder()
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index ce36433443..0e32cbeaff 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -193,6 +193,7 @@ def test_all_regions_without_reading_order(self):
             pcgts = parseString(f.read().encode('utf8'), silence=True)
             pg = pcgts.get_Page()
             self.assertEqual(len(pg.get_AllRegions()), 45)
+            self.assertEqual(len(pg.get_AllRegions(depth=0)), 0)
             self.assertEqual(len(pg.get_AllRegions(depth=1)), 45)
             self.assertEqual(len(pg.get_AllRegions(depth=2)), 65)
             self.assertEqual(len(pg.get_AllRegions(depth=3)), 65)
@@ -207,12 +208,14 @@ def test_all_regions_with_reading_order(self):
         https://github.com/OCR-D/core/issues/240#issuecomment-493135797
         """
         with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
-            pcgts = parseString(f.read().encode('utf8'), silence=True)
-            pg = pcgts.get_Page()
-            with self.assertRaisesRegex(Exception, "Argument 'order' must be either 'document' or 'reading-order', not 'random'"):
+            pg = parseString(f.read().encode('utf8'), silence=True).get_Page()
+            with self.assertRaisesRegex(Exception, "Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not 'random'"):
                 pg.get_AllRegions(order='random')
+            with self.assertRaisesRegex(Exception, "Argument 'depth' must be an integer greater-or-equal 0, not '-1'"):
+                pg.get_AllRegions(depth=-1)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order-only')), 20)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order-only', depth=2)), 40)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=0)), 0)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=1)), 45)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=2)), 65)
             self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)

From 207f396c2fe4cf3d32f5456f74a01c64e7a5ac17 Mon Sep 17 00:00:00 2001
From: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
Date: Fri, 29 May 2020 14:46:43 +0200
Subject: [PATCH 20/30] :memo: get_AllRegions: document example

---
 ocrd_models/ocrd_models/ocrd_page_generateds.py      | 8 +++++++-
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py | 6 ++++++
 2 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index a1aa2f976c..2bf8827c3d 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 13:39:11 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 14:45:59 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2901,6 +2901,12 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
                 returned list (``reading-order``) or regions not in the reading order
                 omitted (``reading-order-only``)
             depth (int) Recursive depth to look for regions at. Default: 1
+    
+        For example, to get all text anywhere on the page in reading order, use:
+        ::
+            '\n'.join(line.get_TextEquiv()[0].Unicode
+                      for region in page.get_AllRegions(classes='Text', depth=0, order='reading-order')
+                      for line in region.get_TextLine())
         """
         if order not in ['document', 'reading-order', 'reading-order-only']:
             raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 90d4a193f6..4a4dacbf9a 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -49,6 +49,12 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
             returned list (``reading-order``) or regions not in the reading order
             omitted (``reading-order-only``)
         depth (int) Recursive depth to look for regions at. Default: 1
+
+    For example, to get all text anywhere on the page in reading order, use:
+    ::
+        '\n'.join(line.get_TextEquiv()[0].Unicode
+                  for region in page.get_AllRegions(classes='Text', depth=0, order='reading-order')
+                  for line in region.get_TextLine())
     """
     if order not in ['document', 'reading-order', 'reading-order-only']:
         raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))

From 9ced31514599a87bfabf8c2a0accf28e26747b21 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 16:34:54 +0200
Subject: [PATCH 21/30] get_AllRegions: fix recursion

---
 .../ocrd_models/ocrd_page_generateds.py       | 22 +++++++++++--------
 .../ocrd_page_user_methods/get_AllRegions.py  | 20 ++++++++++-------
 tests/model/test_ocrd_page.py                 | 17 ++++++++------
 3 files changed, 35 insertions(+), 24 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 2bf8827c3d..ab581a4ab4 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 14:45:59 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 16:34:32 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2859,8 +2859,9 @@ def _get_recursive_regions(self, regions, level, classes=None):
             # stop recursion, filter classes
             if classes:
                 return [r for r in regions if self._region_class(r) in classes]
-            # remove the first element (PageType)
-            return list(set(regions[1:]))
+            if regions and regions[0].__class__.__name__ == 'PageType':
+                regions = regions[1:]
+            return regions
         # find more regions recursively
         more_regions = []
         for region in regions:
@@ -2873,9 +2874,12 @@ def _get_recursive_regions(self, regions, level, classes=None):
                     continue
                 more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
         if not any(more_regions):
-            return _get_recursive_regions(regions, 1, classes)
-        regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
-        return self._get_recursive_regions(regions, level - 1 if level else 0, classes)
+            return self._get_recursive_regions(regions, 1, classes)
+        ret = []
+        for r, more in zip(regions, more_regions):
+            ret.append(r)
+            ret += self._get_recursive_regions(more, level - 1 if level else 0, classes)
+        return self._get_recursive_regions(ret, 1, classes)
     
     def _get_recursive_reading_order(self, rogroup):
         if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)): # pylint: disable=undefined-variable
@@ -2889,7 +2893,7 @@ def _get_recursive_reading_order(self, rogroup):
                 regionrefs.extend(self._get_recursive_reading_order(elem))
         return regionrefs
     
-    def get_AllRegions(self, classes=None, order='document', depth=1):
+    def get_AllRegions(self, classes=None, order='document', depth=0):
         """
         Get all the *Region element or only those provided by ``classes``.
         Returned in document order unless ``order`` is ``reading-order``
@@ -2900,7 +2904,7 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
                 reading order with regions not in the reading order at the end of the
                 returned list (``reading-order``) or regions not in the reading order
                 omitted (``reading-order-only``)
-            depth (int) Recursive depth to look for regions at. Default: 1
+            depth (int) Recursive depth to look for regions at, set to `0` for all regions at any depth. Default: 0
     
         For example, to get all text anywhere on the page in reading order, use:
         ::
@@ -2912,7 +2916,7 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
             raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))
         if depth < 0:
             raise Exception("Argument 'depth' must be an integer greater-or-equal 0, not '{}'".format(depth))
-        ret = self._get_recursive_regions([self], depth + 1, classes)
+        ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
         if order.startswith('reading-order'):
             reading_order = self.get_ReadingOrder()
             if reading_order:
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 4a4dacbf9a..25c1d2c69e 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -7,8 +7,9 @@ def _get_recursive_regions(self, regions, level, classes=None):
         # stop recursion, filter classes
         if classes:
             return [r for r in regions if self._region_class(r) in classes]
-        # remove the first element (PageType)
-        return list(set(regions[1:]))
+        if regions and regions[0].__class__.__name__ == 'PageType':
+            regions = regions[1:]
+        return regions
     # find more regions recursively
     more_regions = []
     for region in regions:
@@ -21,9 +22,12 @@ def _get_recursive_regions(self, regions, level, classes=None):
                 continue
             more_regions[-1] += getattr(region, 'get_{}Region'.format(class_))()
     if not any(more_regions):
-        return _get_recursive_regions(regions, 1, classes)
-    regions = [region for r, more in zip(regions, more_regions) for region in [r] + more]
-    return self._get_recursive_regions(regions, level - 1 if level else 0, classes)
+        return self._get_recursive_regions(regions, 1, classes)
+    ret = []
+    for r, more in zip(regions, more_regions):
+        ret.append(r)
+        ret += self._get_recursive_regions(more, level - 1 if level else 0, classes)
+    return self._get_recursive_regions(ret, 1, classes)
 
 def _get_recursive_reading_order(self, rogroup):
     if isinstance(rogroup, (OrderedGroupType, OrderedGroupIndexedType)): # pylint: disable=undefined-variable
@@ -37,7 +41,7 @@ def _get_recursive_reading_order(self, rogroup):
             regionrefs.extend(self._get_recursive_reading_order(elem))
     return regionrefs
 
-def get_AllRegions(self, classes=None, order='document', depth=1):
+def get_AllRegions(self, classes=None, order='document', depth=0):
     """
     Get all the *Region element or only those provided by ``classes``.
     Returned in document order unless ``order`` is ``reading-order``
@@ -48,7 +52,7 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
             reading order with regions not in the reading order at the end of the
             returned list (``reading-order``) or regions not in the reading order
             omitted (``reading-order-only``)
-        depth (int) Recursive depth to look for regions at. Default: 1
+        depth (int) Recursive depth to look for regions at, set to `0` for all regions at any depth. Default: 0
 
     For example, to get all text anywhere on the page in reading order, use:
     ::
@@ -60,7 +64,7 @@ def get_AllRegions(self, classes=None, order='document', depth=1):
         raise Exception("Argument 'order' must be either 'document', 'reading-order' or 'reading-order-only', not '{}'".format(order))
     if depth < 0:
         raise Exception("Argument 'depth' must be an integer greater-or-equal 0, not '{}'".format(depth))
-    ret = self._get_recursive_regions([self], depth + 1, classes)
+    ret = self._get_recursive_regions([self], depth + 1 if depth else 0, classes)
     if order.startswith('reading-order'):
         reading_order = self.get_ReadingOrder()
         if reading_order:
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 0e32cbeaff..5a950d5149 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -192,15 +192,16 @@ def test_all_regions_without_reading_order(self):
         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()), 45)
-            self.assertEqual(len(pg.get_AllRegions(depth=0)), 0)
+            self.assertEqual(len(pg.get_AllRegions()), 65)
+            self.assertEqual(len(pg.get_AllRegions(depth=0)), 65)
             self.assertEqual(len(pg.get_AllRegions(depth=1)), 45)
             self.assertEqual(len(pg.get_AllRegions(depth=2)), 65)
             self.assertEqual(len(pg.get_AllRegions(depth=3)), 65)
             self.assertEqual(len(pg.get_AllRegions(classes=['Separator'])), 25)
             self.assertEqual(len(pg.get_AllRegions(classes=['Table'])), 3)
-            self.assertEqual(len(pg.get_AllRegions(classes=['Text'])), 17)
-            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=2)), 54)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'])), 37)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=1)), 17)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], depth=2)), 37)
 
     def test_all_regions_with_reading_order(self):
         """
@@ -213,13 +214,15 @@ def test_all_regions_with_reading_order(self):
                 pg.get_AllRegions(order='random')
             with self.assertRaisesRegex(Exception, "Argument 'depth' must be an integer greater-or-equal 0, not '-1'"):
                 pg.get_AllRegions(depth=-1)
-            self.assertEqual(len(pg.get_AllRegions(order='reading-order-only')), 20)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order-only')), 40)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order-only', depth=1)), 20)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order-only', depth=2)), 40)
-            self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=0)), 0)
+            self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=0)), 65)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=1)), 45)
             self.assertEqual(len(pg.get_AllRegions(order='reading-order', depth=2)), 65)
             self.assertEqual(len(pg.get_AllRegions(classes=['Table'], order='reading-order')), 3)
-            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 17)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 37)
+            self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order', depth=1)), 17)
 
 if __name__ == '__main__':
     main()

From 629f38d2ce45bac6887c3ac3a8db47cc85a53b15 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Fri, 29 May 2020 19:58:28 +0200
Subject: [PATCH 22/30] get_AllRegions: Update example

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/ocrd_page_user_methods/get_AllRegions.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
index 25c1d2c69e..bae63bef66 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllRegions.py
@@ -56,8 +56,8 @@ def get_AllRegions(self, classes=None, order='document', depth=0):
 
     For example, to get all text anywhere on the page in reading order, use:
     ::
-        '\n'.join(line.get_TextEquiv()[0].Unicode
-                  for region in page.get_AllRegions(classes='Text', depth=0, order='reading-order')
+        '\\n'.join(line.get_TextEquiv()[0].Unicode
+                  for region in page.get_AllRegions(classes=['Text'], depth=0, order='reading-order')
                   for line in region.get_TextLine())
     """
     if order not in ['document', 'reading-order', 'reading-order-only']:

From e958559fbdf702347a0ed870e9df7b625b59a5e0 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 21:54:54 +0200
Subject: [PATCH 23/30] wip

---
 .../ocrd_models/ocrd_page_generateds.py       | 54 +++++++++++++++----
 .../extend_AllIndexed.py                      |  5 +-
 .../ocrd_page_user_methods/get_AllIndexed.py  | 19 ++++++-
 tests/model/TEMP1_Gutachten2-2.xml            |  3 ++
 tests/model/test_ocrd_page.py                 |  9 ++++
 5 files changed, 78 insertions(+), 12 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index ab581a4ab4..1bc292bc15 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 16:34:32 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 21:46:01 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2908,8 +2908,8 @@ def get_AllRegions(self, classes=None, order='document', depth=0):
     
         For example, to get all text anywhere on the page in reading order, use:
         ::
-            '\n'.join(line.get_TextEquiv()[0].Unicode
-                      for region in page.get_AllRegions(classes='Text', depth=0, order='reading-order')
+            '\\n'.join(line.get_TextEquiv()[0].Unicode
+                      for region in page.get_AllRegions(classes=['Text'], depth=0, order='reading-order')
                       for line in region.get_TextLine())
         """
         if order not in ['document', 'reading-order', 'reading-order-only']:
@@ -5433,9 +5433,24 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroupIndexed'
     def __hash__(self):
         return hash(self.id)
-    def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+    # pylint: disable=invalid-name,missing-module-docstring,line-too-long
+    def get_AllIndexed(self, classes=None):
+        """
+        Get all indexed children sorted by their ``@index``.
     
+        Arguments:
+            classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+        """
+        if not classes:
+            classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+        ret = []
+        if 'RegionRef' in classes:
+            ret += self.get_RegionRefIndexed()
+        if 'OrderedGroup' in classes:
+            ret += self.get_OrderedGroupIndexed()
+        if 'UnorderedGroup' in classes:
+            ret += self.get_UnorderedGroupIndexed()
+        return sorted(ret, key=lambda x: x.index)
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
         self.set_RegionRefIndexed([])
@@ -5443,8 +5458,11 @@ def clear_AllIndexed(self):
         self.set_UnorderedGroupIndexed([])
         return ret
     
-    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring
     def extend_AllIndexed(self, elements):
+        """
+        Add all elements in list ``elements``, respecting ``@index`` order.
+        """
         if not isinstance(elements, list):
             elements = [elements]
         for element in sorted(elements, key=lambda x: x.index):
@@ -6223,9 +6241,24 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroupIndexed'
     def __hash__(self):
         return hash(self.id)
-    def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+    # pylint: disable=invalid-name,missing-module-docstring,line-too-long
+    def get_AllIndexed(self, classes=None):
+        """
+        Get all indexed children sorted by their ``@index``.
     
+        Arguments:
+            classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+        """
+        if not classes:
+            classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+        ret = []
+        if 'RegionRef' in classes:
+            ret += self.get_RegionRefIndexed()
+        if 'OrderedGroup' in classes:
+            ret += self.get_OrderedGroupIndexed()
+        if 'UnorderedGroup' in classes:
+            ret += self.get_UnorderedGroupIndexed()
+        return sorted(ret, key=lambda x: x.index)
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
         self.set_RegionRefIndexed([])
@@ -6233,8 +6266,11 @@ def clear_AllIndexed(self):
         self.set_UnorderedGroupIndexed([])
         return ret
     
-    # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring
     def extend_AllIndexed(self, elements):
+        """
+        Add all elements in list ``elements``, respecting ``@index`` order.
+        """
         if not isinstance(elements, list):
             elements = [elements]
         for element in sorted(elements, key=lambda x: x.index):
diff --git a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
index fb129f5b5f..e364a8cc11 100644
--- a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
@@ -1,5 +1,8 @@
-# pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
+# pylint: disable=line-too-long,invalid-name,missing-module-docstring
 def extend_AllIndexed(self, elements):
+    """
+    Add all elements in list ``elements``, respecting ``@index`` order.
+    """
     if not isinstance(elements, list):
         elements = [elements]
     for element in sorted(elements, key=lambda x: x.index):
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
index 4d79632353..6618b6bf3e 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
@@ -1,3 +1,18 @@
-def get_AllIndexed(self):
-    return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+# pylint: disable=invalid-name,missing-module-docstring,line-too-long
+def get_AllIndexed(self, classes=None):
+    """
+    Get all indexed children sorted by their ``@index``.
 
+    Arguments:
+        classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+    """
+    if not classes:
+        classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+    ret = []
+    if 'RegionRef' in classes:
+        ret += self.get_RegionRefIndexed()
+    if 'OrderedGroup' in classes:
+        ret += self.get_OrderedGroupIndexed()
+    if 'UnorderedGroup' in classes:
+        ret += self.get_UnorderedGroupIndexed()
+    return sorted(ret, key=lambda x: x.index)
diff --git a/tests/model/TEMP1_Gutachten2-2.xml b/tests/model/TEMP1_Gutachten2-2.xml
index f0d7cbd33b..581af560d9 100644
--- a/tests/model/TEMP1_Gutachten2-2.xml
+++ b/tests/model/TEMP1_Gutachten2-2.xml
@@ -117,6 +117,9 @@
                 </pc:OrderedGroupIndexed>
                 <pc:RegionRefIndexed index="18" regionRef="Gutachten2-2_region0016"/>
                 <pc:RegionRefIndexed index="19" regionRef="Gutachten2-2_region0017"/>
+                <pc:UnorderedGroupIndexed id="unordered-group-for-testing_group" regionRef="unordered-group-for-testing" index="18">
+                    <pc:RegionRef regionRef="unordered-group-for-testing_region0001"/>
+                </pc:UnorderedGroupIndexed>
             </pc:OrderedGroup>
         </pc:ReadingOrder>
         <pc:TextRegion id="Gutachten2-2_region0001">
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 5a950d5149..d795aa7c0b 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -224,5 +224,14 @@ def test_all_regions_with_reading_order(self):
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 37)
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order', depth=1)), 17)
 
+    def test_get_AllIndexed_classes(self):
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
+            self.assertEqual(len(og.get_AllIndexed(classes=['RegionRef'])), 17)
+            self.assertEqual(len(og.get_AllIndexed(classes=['OrderedGroup'])), 3)
+            self.assertEqual(len(og.get_AllIndexed(classes=['UnorderedGroup'])), 1)
+
+
+
 if __name__ == '__main__':
     main()

From 196456386e8e656ad6c463379ba8085d42ac3c8b Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 22:35:46 +0200
Subject: [PATCH 24/30] reading order test sample: add unorderedgroups for
 testing

---
 tests/model/TEMP1_Gutachten2-2.xml | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/tests/model/TEMP1_Gutachten2-2.xml b/tests/model/TEMP1_Gutachten2-2.xml
index f0d7cbd33b..1f54b4db9a 100644
--- a/tests/model/TEMP1_Gutachten2-2.xml
+++ b/tests/model/TEMP1_Gutachten2-2.xml
@@ -117,6 +117,11 @@
                 </pc:OrderedGroupIndexed>
                 <pc:RegionRefIndexed index="18" regionRef="Gutachten2-2_region0016"/>
                 <pc:RegionRefIndexed index="19" regionRef="Gutachten2-2_region0017"/>
+                <pc:UnorderedGroupIndexed id="unordered-group-for-testing_group" regionRef="unordered-group-for-testing" index="20">
+                    <pc:RegionRef regionRef="unordered-group-for-testing_region0001"/>
+                </pc:UnorderedGroupIndexed>
+                <pc:UnorderedGroupIndexed id="empty-group-for-testing_group" regionRef="empty-group-for-testing" index="21">
+                </pc:UnorderedGroupIndexed>
             </pc:OrderedGroup>
         </pc:ReadingOrder>
         <pc:TextRegion id="Gutachten2-2_region0001">

From 27e256f52165763b4a4d02d93bac6ac0a61f619c Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 22:36:48 +0200
Subject: [PATCH 25/30] add get_UnorderedGroupChildren, let get_AllIndexed
 handle UnorderedGroupIndexed properly

---
 .../ocrd_models/ocrd_page_generateds.py       | 54 +++++++++++++------
 ocrd_models/ocrd_page_user_methods.py         |  1 +
 .../ocrd_page_user_methods/exportChildren.py  | 15 ++++--
 .../ocrd_page_user_methods/get_AllIndexed.py  |  2 +-
 .../get_UnorderedGroupChildren.py             |  7 +++
 tests/model/test_ocrd_page.py                 | 20 +++++--
 6 files changed, 73 insertions(+), 26 deletions(-)
 create mode 100644 ocrd_models/ocrd_page_user_methods/get_UnorderedGroupChildren.py

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index ab581a4ab4..4878390c5a 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 16:34:32 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 22:34:55 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -2908,8 +2908,8 @@ def get_AllRegions(self, classes=None, order='document', depth=0):
     
         For example, to get all text anywhere on the page in reading order, use:
         ::
-            '\n'.join(line.get_TextEquiv()[0].Unicode
-                      for region in page.get_AllRegions(classes='Text', depth=0, order='reading-order')
+            '\\n'.join(line.get_TextEquiv()[0].Unicode
+                      for region in page.get_AllRegions(classes=['Text'], depth=0, order='reading-order')
                       for line in region.get_TextLine())
         """
         if order not in ['document', 'reading-order', 'reading-order-only']:
@@ -5434,7 +5434,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
     def __hash__(self):
         return hash(self.id)
     def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x: x.index)
     
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
@@ -5464,13 +5464,18 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
         for Labels_ in self.Labels:
             Labels_.export(outfile, level, namespaceprefix_, namespacedef_='', name_='Labels', pretty_print=pretty_print)
         cleaned = []
+        def replaceWithRRI(group):
+            rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
+            rri.index = group.index
+            rri.regionRef = group.regionRef
+            cleaned.append(rri)
         # remove emtpy groups and replace with RegionRefIndexedType
         for entry in self.get_AllIndexed():
-            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
-                rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
-                rri.index = entry.index
-                rri.regionRef = entry.regionRef
-                cleaned.append(rri)
+            # pylint: disable=undefined-variable
+            if isinstance(entry, (OrderedGroupIndexedType)) and not entry.get_AllIndexed():
+                replaceWithRRI(entry)
+            elif isinstance(entry, UnorderedGroupIndexedType) and not entry.get_UnorderedGroupChildren():
+                replaceWithRRI(entry)
             else:
                 cleaned.append(entry)
         for entry in cleaned:
@@ -5811,6 +5816,13 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroup'
     def __hash__(self):
         return hash(self.id)
+    def get_UnorderedGroupChildren(self):
+        """
+        List all non-metadata children of an UnorderedGroup
+        """
+        # TODO: should not change order
+        return self.get_RegionRef() + self.get_OrderedGroup() + self.get_UnorderedGroup()
+    
 # end class UnorderedGroupIndexedType
 
 
@@ -6224,7 +6236,7 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
     def __hash__(self):
         return hash(self.id)
     def get_AllIndexed(self):
-        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+        return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x: x.index)
     
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
@@ -6254,13 +6266,18 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
         for Labels_ in self.Labels:
             Labels_.export(outfile, level, namespaceprefix_, namespacedef_='', name_='Labels', pretty_print=pretty_print)
         cleaned = []
+        def replaceWithRRI(group):
+            rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
+            rri.index = group.index
+            rri.regionRef = group.regionRef
+            cleaned.append(rri)
         # remove emtpy groups and replace with RegionRefIndexedType
         for entry in self.get_AllIndexed():
-            if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
-                rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
-                rri.index = entry.index
-                rri.regionRef = entry.regionRef
-                cleaned.append(rri)
+            # pylint: disable=undefined-variable
+            if isinstance(entry, (OrderedGroupIndexedType)) and not entry.get_AllIndexed():
+                replaceWithRRI(entry)
+            elif isinstance(entry, UnorderedGroupIndexedType) and not entry.get_UnorderedGroupChildren():
+                replaceWithRRI(entry)
             else:
                 cleaned.append(entry)
         for entry in cleaned:
@@ -6585,6 +6602,13 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
             obj_.original_tagname_ = 'UnorderedGroup'
     def __hash__(self):
         return hash(self.id)
+    def get_UnorderedGroupChildren(self):
+        """
+        List all non-metadata children of an UnorderedGroup
+        """
+        # TODO: should not change order
+        return self.get_RegionRef() + self.get_OrderedGroup() + self.get_UnorderedGroup()
+    
 # end class UnorderedGroupType
 
 
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index 42f5bfa481..59ca85b7e0 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -102,6 +102,7 @@ def _add_method(class_re, method_name):
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'clear_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'extend_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'exportChildren'),
+    _add_method(r'^(UnorderedGroupType|UnorderedGroupIndexedType)$', 'get_UnorderedGroupChildren'),
     _add_method(r'^(PageType)$', 'get_AllRegions'),
     )
 
diff --git a/ocrd_models/ocrd_page_user_methods/exportChildren.py b/ocrd_models/ocrd_page_user_methods/exportChildren.py
index 85c7f660c2..8f118b71be 100644
--- a/ocrd_models/ocrd_page_user_methods/exportChildren.py
+++ b/ocrd_models/ocrd_page_user_methods/exportChildren.py
@@ -6,13 +6,18 @@ def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xml
     for Labels_ in self.Labels:
         Labels_.export(outfile, level, namespaceprefix_, namespacedef_='', name_='Labels', pretty_print=pretty_print)
     cleaned = []
+    def replaceWithRRI(group):
+        rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
+        rri.index = group.index
+        rri.regionRef = group.regionRef
+        cleaned.append(rri)
     # remove emtpy groups and replace with RegionRefIndexedType
     for entry in self.get_AllIndexed():
-        if isinstance(entry, (UnorderedGroupIndexedType, OrderedGroupIndexedType)) and not entry.get_AllIndexed(): # pylint: disable=undefined-variable
-            rri = RegionRefIndexedType.factory(parent_object_=self) # pylint: disable=undefined-variable
-            rri.index = entry.index
-            rri.regionRef = entry.regionRef
-            cleaned.append(rri)
+        # pylint: disable=undefined-variable
+        if isinstance(entry, (OrderedGroupIndexedType)) and not entry.get_AllIndexed():
+            replaceWithRRI(entry)
+        elif isinstance(entry, UnorderedGroupIndexedType) and not entry.get_UnorderedGroupChildren():
+            replaceWithRRI(entry)
         else:
             cleaned.append(entry)
     for entry in cleaned:
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
index 4d79632353..7f2a5434b4 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
@@ -1,3 +1,3 @@
 def get_AllIndexed(self):
-    return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x : x.index)
+    return sorted(self.get_RegionRefIndexed() + self.get_OrderedGroupIndexed() + self.get_UnorderedGroupIndexed(), key=lambda x: x.index)
 
diff --git a/ocrd_models/ocrd_page_user_methods/get_UnorderedGroupChildren.py b/ocrd_models/ocrd_page_user_methods/get_UnorderedGroupChildren.py
new file mode 100644
index 0000000000..61c1d9450f
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/get_UnorderedGroupChildren.py
@@ -0,0 +1,7 @@
+def get_UnorderedGroupChildren(self):
+    """
+    List all non-metadata children of an UnorderedGroup
+    """
+    # TODO: should not change order
+    return self.get_RegionRef() + self.get_OrderedGroup() + self.get_UnorderedGroup()
+
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 5a950d5149..34d79b5f4a 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -8,6 +8,7 @@
     TextRegionType,
     TextLineType,
     OrderedGroupIndexedType,
+    UnorderedGroupIndexedType,
     RegionRefIndexedType,
     WordType,
     GlyphType,
@@ -151,19 +152,19 @@ def test_orderedgroup_export_order(self):
             og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
             xml_before = to_xml(og)
             children = og.get_AllIndexed()
-            self.assertEqual(len(children), 20)
-            self.assertEqual([c.index for c in children], list(range(0, 20)))
+            self.assertEqual(len(children), 22)
+            self.assertEqual([c.index for c in children], list(range(0, 22)))
             # mix up the indexes
             children[0].index = 11
             children[11].index = 3
             children[3].index = 0
-            self.assertEqual([c.index for c in children], [11, 1, 2, 0, 4, 5, 6, 7, 8, 9, 10, 3, 12, 13, 14, 15, 16, 17, 18, 19])
-            self.assertEqual([c.index for c in og.get_AllIndexed()], list(range(0, 20)))
+            self.assertEqual([c.index for c in children], [11, 1, 2, 0, 4, 5, 6, 7, 8, 9, 10, 3, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21])
+            self.assertEqual([c.index for c in og.get_AllIndexed()], list(range(0, 22)))
             self.assertEqual(og.get_AllIndexed()[1].__class__, OrderedGroupIndexedType)
             # serialize and make sure the correct order was serialized
             new_pcgts = parseString(to_xml(pcgts).encode('utf8'), silence=True)
             new_og = new_pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
-            self.assertEqual([c.index for c in new_og.get_AllIndexed()], list(range(0, 20)))
+            self.assertEqual([c.index for c in new_og.get_AllIndexed()], list(range(0, 22)))
             # xml_after = to_xml(new_og)
             # self.assertEqual(xml_after, xml_before)
 
@@ -176,6 +177,7 @@ def test_empty_groups_to_regionrefindexed(self):
             og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
             children = og.get_AllIndexed()
             self.assertTrue(isinstance(children[1], OrderedGroupIndexedType))
+            self.assertTrue(isinstance(children[21], UnorderedGroupIndexedType))
             # empty all the elements in the first orederdGroupIndexed
             children[1].set_RegionRefIndexed([])
             # serialize apnd parse to see empty group converted
@@ -183,6 +185,7 @@ def test_empty_groups_to_regionrefindexed(self):
             og = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup()
             children = og.get_AllIndexed()
             self.assertTrue(isinstance(children[1], RegionRefIndexedType))
+            self.assertTrue(isinstance(children[21], RegionRefIndexedType))
 
     def test_all_regions_without_reading_order(self):
         """
@@ -224,5 +227,12 @@ def test_all_regions_with_reading_order(self):
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order')), 37)
             self.assertEqual(len(pg.get_AllRegions(classes=['Text'], order='reading-order', depth=1)), 17)
 
+    def test_get_UnorderdGroupChildren(self):
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            pcgts = parseString(f.read().encode('utf8'), silence=True)
+            ug = pcgts.get_Page().get_ReadingOrder().get_OrderedGroup().get_UnorderedGroupIndexed()[0]
+            self.assertEqual(len(ug.get_UnorderedGroupChildren()), 1)
+
+
 if __name__ == '__main__':
     main()

From ae613cfb757319e954839c74cef709359c49dcf0 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 23:04:44 +0200
Subject: [PATCH 26/30] get_AllIndexed: index_sort parameter to enable/disable
 sorting

---
 ocrd_models/ocrd_models/ocrd_page_generateds.py  | 16 +++++++++++-----
 .../ocrd_page_user_methods/get_AllIndexed.py     |  7 +++++--
 tests/model/test_ocrd_page.py                    | 10 ++++++++++
 3 files changed, 26 insertions(+), 7 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index ca2209571c..7ed0d71494 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 22:47:44 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 23:03:43 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -5434,19 +5434,22 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
     def __hash__(self):
         return hash(self.id)
     # pylint: disable=invalid-name,missing-module-docstring,line-too-long
-    def get_AllIndexed(self, classes=None):
+    def get_AllIndexed(self, classes=None, index_sort=True):
         """
         Get all indexed children sorted by their ``@index``.
     
         Arguments:
             classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+            index_sort (boolean): Whether to sort by ``@index``
         """
         if not classes:
             classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
         ret = []
         for class_ in classes:
             ret += getattr(self, 'get_{}Indexed'.format(class_))()
-        return sorted(ret, key=lambda x: x.index)
+        if index_sort:
+            return sorted(ret, key=lambda x: x.index)
+        return ret
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
         self.set_RegionRefIndexed([])
@@ -6250,19 +6253,22 @@ def buildChildren(self, child_, node, nodeName_, fromsubclass_=False, gds_collec
     def __hash__(self):
         return hash(self.id)
     # pylint: disable=invalid-name,missing-module-docstring,line-too-long
-    def get_AllIndexed(self, classes=None):
+    def get_AllIndexed(self, classes=None, index_sort=True):
         """
         Get all indexed children sorted by their ``@index``.
     
         Arguments:
             classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+            index_sort (boolean): Whether to sort by ``@index``
         """
         if not classes:
             classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
         ret = []
         for class_ in classes:
             ret += getattr(self, 'get_{}Indexed'.format(class_))()
-        return sorted(ret, key=lambda x: x.index)
+        if index_sort:
+            return sorted(ret, key=lambda x: x.index)
+        return ret
     def clear_AllIndexed(self):
         ret = self.get_AllIndexed()
         self.set_RegionRefIndexed([])
diff --git a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
index 6fa157cdfd..da660d9893 100644
--- a/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/get_AllIndexed.py
@@ -1,14 +1,17 @@
 # pylint: disable=invalid-name,missing-module-docstring,line-too-long
-def get_AllIndexed(self, classes=None):
+def get_AllIndexed(self, classes=None, index_sort=True):
     """
     Get all indexed children sorted by their ``@index``.
 
     Arguments:
         classes (list): Type of children to return. Default: ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
+        index_sort (boolean): Whether to sort by ``@index``
     """
     if not classes:
         classes = ['RegionRef', 'OrderedGroup', 'UnorderedGroup']
     ret = []
     for class_ in classes:
         ret += getattr(self, 'get_{}Indexed'.format(class_))()
-    return sorted(ret, key=lambda x: x.index)
+    if index_sort:
+        return sorted(ret, key=lambda x: x.index)
+    return ret
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 9362f58d9b..7aaef47eda 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -240,5 +240,15 @@ def test_get_AllIndexed_classes(self):
             self.assertEqual(len(og.get_AllIndexed(classes=['OrderedGroup'])), 3)
             self.assertEqual(len(og.get_AllIndexed(classes=['UnorderedGroup'])), 2)
 
+    def test_get_AllIndexed_index_sort(self):
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
+            unogs = og.get_UnorderedGroupIndexed()
+            self.assertEqual([x.index for x in unogs], [20, 21])
+            unogs[0].index = 21
+            unogs[1].index = 20
+            self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=True)], [20, 21])
+            self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)], [21, 20])
+
 if __name__ == '__main__':
     main()

From b1df95fccf079a9ee957e82dca84bd94162f2797 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 23:08:27 +0200
Subject: [PATCH 27/30] add sort_AllIndexed to sort in-place

---
 .../ocrd_models/ocrd_page_generateds.py       | 34 ++++++++++++++++++-
 ocrd_models/ocrd_page_user_methods.py         |  1 +
 .../ocrd_page_user_methods/sort_AllIndexed.py | 16 +++++++++
 tests/model/test_ocrd_page.py                 |  2 ++
 4 files changed, 52 insertions(+), 1 deletion(-)
 create mode 100644 ocrd_models/ocrd_page_user_methods/sort_AllIndexed.py

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 7ed0d71494..1e27dd5768 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 23:03:43 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 23:08:05 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -5473,6 +5473,22 @@ def extend_AllIndexed(self, elements):
                 self.add_UnorderedGroupIndexed(element)
         return self.get_AllIndexed()
     
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring
+    def sort_AllIndexed(self, validate_uniqueness=True):
+        """
+        Sort all indexed children in-place.
+        """
+        elements = self.get_AllIndexed(index_sort=True)
+        self.clear_AllIndexed()
+        for element in elements:
+            if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
+                self.add_RegionRefIndexed(element)
+            elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
+                self.add_OrderedGroupIndexed(element)
+            elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
+                self.add_UnorderedGroupIndexed(element)
+        return self.get_AllIndexed()
+    
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
     def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True): # pylint: disable=unused-argument,too-many-arguments
         namespaceprefix_ = 'pc:'
@@ -6292,6 +6308,22 @@ def extend_AllIndexed(self, elements):
                 self.add_UnorderedGroupIndexed(element)
         return self.get_AllIndexed()
     
+    # pylint: disable=line-too-long,invalid-name,missing-module-docstring
+    def sort_AllIndexed(self, validate_uniqueness=True):
+        """
+        Sort all indexed children in-place.
+        """
+        elements = self.get_AllIndexed(index_sort=True)
+        self.clear_AllIndexed()
+        for element in elements:
+            if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
+                self.add_RegionRefIndexed(element)
+            elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
+                self.add_OrderedGroupIndexed(element)
+            elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
+                self.add_UnorderedGroupIndexed(element)
+        return self.get_AllIndexed()
+    
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring,missing-function-docstring
     def exportChildren(self, outfile, level, namespaceprefix_='', namespacedef_='xmlns:pc="http://schema.primaresearch.org/PAGE/gts/pagecontent/2019-07-15"', name_='OrderedGroupType', fromsubclass_=False, pretty_print=True): # pylint: disable=unused-argument,too-many-arguments
         namespaceprefix_ = 'pc:'
diff --git a/ocrd_models/ocrd_page_user_methods.py b/ocrd_models/ocrd_page_user_methods.py
index 59ca85b7e0..27714c408a 100644
--- a/ocrd_models/ocrd_page_user_methods.py
+++ b/ocrd_models/ocrd_page_user_methods.py
@@ -101,6 +101,7 @@ def _add_method(class_re, method_name):
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'get_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'clear_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'extend_AllIndexed'),
+    _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'sort_AllIndexed'),
     _add_method(r'^(OrderedGroupType|OrderedGroupIndexedType)$', 'exportChildren'),
     _add_method(r'^(UnorderedGroupType|UnorderedGroupIndexedType)$', 'get_UnorderedGroupChildren'),
     _add_method(r'^(PageType)$', 'get_AllRegions'),
diff --git a/ocrd_models/ocrd_page_user_methods/sort_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/sort_AllIndexed.py
new file mode 100644
index 0000000000..eeeaf32a7c
--- /dev/null
+++ b/ocrd_models/ocrd_page_user_methods/sort_AllIndexed.py
@@ -0,0 +1,16 @@
+# pylint: disable=line-too-long,invalid-name,missing-module-docstring
+def sort_AllIndexed(self, validate_uniqueness=True):
+    """
+    Sort all indexed children in-place.
+    """
+    elements = self.get_AllIndexed(index_sort=True)
+    self.clear_AllIndexed()
+    for element in elements:
+        if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
+            self.add_RegionRefIndexed(element)
+        elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
+            self.add_OrderedGroupIndexed(element)
+        elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
+            self.add_UnorderedGroupIndexed(element)
+    return self.get_AllIndexed()
+
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index 7aaef47eda..ecd094e58c 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -249,6 +249,8 @@ def test_get_AllIndexed_index_sort(self):
             unogs[1].index = 20
             self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=True)], [20, 21])
             self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)], [21, 20])
+            og.sort_AllIndexed()
+            self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)], [20, 21])
 
 if __name__ == '__main__':
     main()

From fd9dc833017c464ce962481d52a07b81d414cbec Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 23:30:34 +0200
Subject: [PATCH 28/30] extend_AllIndexed: increment @index when adding
 elements

---
 .../ocrd_models/ocrd_page_generateds.py       | 32 ++++++++++++++++---
 .../extend_AllIndexed.py                      | 15 +++++++--
 tests/model/test_ocrd_page.py                 | 21 ++++++++++++
 3 files changed, 61 insertions(+), 7 deletions(-)

diff --git a/ocrd_models/ocrd_models/ocrd_page_generateds.py b/ocrd_models/ocrd_models/ocrd_page_generateds.py
index 1e27dd5768..46af7ebd28 100644
--- a/ocrd_models/ocrd_models/ocrd_page_generateds.py
+++ b/ocrd_models/ocrd_models/ocrd_page_generateds.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Fri May 29 23:08:05 2020 by generateDS.py version 2.35.20.
+# Generated Fri May 29 23:29:23 2020 by generateDS.py version 2.35.20.
 # Python 3.6.9 (default, Apr 18 2020, 01:56:04)  [GCC 8.4.0]
 #
 # Command line options:
@@ -5458,13 +5458,24 @@ def clear_AllIndexed(self):
         return ret
     
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring
-    def extend_AllIndexed(self, elements):
+    def extend_AllIndexed(self, elements, validate_continuity=False):
         """
         Add all elements in list ``elements``, respecting ``@index`` order.
         """
         if not isinstance(elements, list):
             elements = [elements]
-        for element in sorted(elements, key=lambda x: x.index):
+        siblings = self.get_AllIndexed()
+        highest_sibling_index = siblings[-1].index if siblings else -1
+        if validate_continuity:
+            elements = sorted(elements, key=lambda x: x.index)
+            lowest_element_index = elements[0].index
+            if lowest_element_index <= highest_sibling_index:
+                raise Exception("@index already used: {}".format(lowest_element_index))
+        else:
+            for element in elements:
+                highest_sibling_index += 1
+                element.index = highest_sibling_index
+        for element in elements:
             if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
                 self.add_RegionRefIndexed(element)
             elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
@@ -6293,13 +6304,24 @@ def clear_AllIndexed(self):
         return ret
     
     # pylint: disable=line-too-long,invalid-name,missing-module-docstring
-    def extend_AllIndexed(self, elements):
+    def extend_AllIndexed(self, elements, validate_continuity=False):
         """
         Add all elements in list ``elements``, respecting ``@index`` order.
         """
         if not isinstance(elements, list):
             elements = [elements]
-        for element in sorted(elements, key=lambda x: x.index):
+        siblings = self.get_AllIndexed()
+        highest_sibling_index = siblings[-1].index if siblings else -1
+        if validate_continuity:
+            elements = sorted(elements, key=lambda x: x.index)
+            lowest_element_index = elements[0].index
+            if lowest_element_index <= highest_sibling_index:
+                raise Exception("@index already used: {}".format(lowest_element_index))
+        else:
+            for element in elements:
+                highest_sibling_index += 1
+                element.index = highest_sibling_index
+        for element in elements:
             if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
                 self.add_RegionRefIndexed(element)
             elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
diff --git a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
index e364a8cc11..0eba30d1ea 100644
--- a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
@@ -1,11 +1,22 @@
 # pylint: disable=line-too-long,invalid-name,missing-module-docstring
-def extend_AllIndexed(self, elements):
+def extend_AllIndexed(self, elements, validate_continuity=False):
     """
     Add all elements in list ``elements``, respecting ``@index`` order.
     """
     if not isinstance(elements, list):
         elements = [elements]
-    for element in sorted(elements, key=lambda x: x.index):
+    siblings = self.get_AllIndexed()
+    highest_sibling_index = siblings[-1].index if siblings else -1
+    if validate_continuity:
+        elements = sorted(elements, key=lambda x: x.index)
+        lowest_element_index = elements[0].index
+        if lowest_element_index <= highest_sibling_index:
+            raise Exception("@index already used: {}".format(lowest_element_index))
+    else:
+        for element in elements:
+            highest_sibling_index += 1
+            element.index = highest_sibling_index
+    for element in elements:
         if isinstance(element, RegionRefIndexedType): # pylint: disable=undefined-variable
             self.add_RegionRefIndexed(element)
         elif isinstance(element, OrderedGroupIndexedType): # pylint: disable=undefined-variable
diff --git a/tests/model/test_ocrd_page.py b/tests/model/test_ocrd_page.py
index ecd094e58c..2f6f59031c 100644
--- a/tests/model/test_ocrd_page.py
+++ b/tests/model/test_ocrd_page.py
@@ -252,5 +252,26 @@ def test_get_AllIndexed_index_sort(self):
             og.sort_AllIndexed()
             self.assertEqual([x.index for x in og.get_AllIndexed(classes=['UnorderedGroup'], index_sort=False)], [20, 21])
 
+    def test_extend_AllIndexed_no_validation(self):
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
+            og.extend_AllIndexed([
+                RegionRefIndexedType(index=3, id='r3'),
+                RegionRefIndexedType(index=2, id='r2'),
+                RegionRefIndexedType(index=1, id='r1'),
+            ])
+            rrs = og.get_RegionRefIndexed()
+            self.assertEqual([x.index for x in rrs][-3:], [22, 23, 24])
+
+    def test_extend_AllIndexed_validate_continuity(self):
+        with open('tests/model/TEMP1_Gutachten2-2.xml', 'r') as f:
+            og = parseString(f.read().encode('utf8'), silence=True).get_Page().get_ReadingOrder().get_OrderedGroup()
+            with self.assertRaisesRegex(Exception, "@index already used: 1"):
+                og.extend_AllIndexed([
+                    RegionRefIndexedType(index=3, id='r3'),
+                    RegionRefIndexedType(index=2, id='r2'),
+                    RegionRefIndexedType(index=1, id='r1'),
+                ], validate_continuity=True)
+
 if __name__ == '__main__':
     main()

From 84f1d33ce1e61da323868ba3cfbbe73fef08f7d9 Mon Sep 17 00:00:00 2001
From: kba <unixprog@gmail.com>
Date: Fri, 29 May 2020 23:47:01 +0200
Subject: [PATCH 29/30] :memo: changelog

---
 CHANGELOG.md | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 7ae554911a..0758732177 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,6 +5,22 @@ Versioned according to [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
 
+Added:
+
+  * OcrdPage: `get_AllRegions`: retrieve all regions, sorted by document or reading order, #479
+  * OcrdPage: `sort_AllIndexed`: sort all children by `@index`  in-place
+  * OcrdPage: `clear_AllIndexed`: clear all `@index` children
+  * OcrdPage: `extend_AllIndexed`: Add elements with incrementing `@index`
+  * OcrdPage: Replace empty reading order groups with equivalent `RegionRef` on export
+  * OcrdPage: `get_UnorderedGroupChildren`: get reading order elements of an `UnorderedGroup`
+
+
+Changed:
+
+  * OcrdPage: `get_AllIndexed`: allow filtering by child type
+  * OcrdPage: `get_AllIndexed`: index_sort parameter to enable/disable sorting
+
+
 ## [2.7.1] - 2020-05-27
 
 Fixed:

From 0e146334ced3ca86f85a1ff9dff3af6ec622c291 Mon Sep 17 00:00:00 2001
From: Konstantin Baierer <kba@users.noreply.github.com>
Date: Wed, 3 Jun 2020 14:08:55 +0200
Subject: [PATCH 30/30] Document extend_AllIndexed validate_contiunuity param

Co-authored-by: Robert Sachunsky <38561704+bertsky@users.noreply.github.com>
---
 ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
index 0eba30d1ea..73645316c0 100644
--- a/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
+++ b/ocrd_models/ocrd_page_user_methods/extend_AllIndexed.py
@@ -2,6 +2,9 @@
 def extend_AllIndexed(self, elements, validate_continuity=False):
     """
     Add all elements in list ``elements``, respecting ``@index`` order.
+    With ``validate_continuity``, check that all new elements come after all old elements
+    (or raise an exception). 
+    Otherwise, ensure this condition silently (by increasing ``@index`` accordingly).
     """
     if not isinstance(elements, list):
         elements = [elements]
@@ -24,4 +27,3 @@ def extend_AllIndexed(self, elements, validate_continuity=False):
         elif isinstance(element, UnorderedGroupIndexedType): # pylint: disable=undefined-variable
             self.add_UnorderedGroupIndexed(element)
     return self.get_AllIndexed()
-