OCR-D · kba · Nov 15, 2023 · Nov 13, 2023
diff --git a/ocrd/ocrd/cli/workspace.py b/ocrd/ocrd/cli/workspace.py
@@ -629,6 +629,20 @@ def set_id(ctx, id):   # pylint: disable=redefined-builtin
     workspace.mets.unique_identifier = id
     workspace.save_mets()
 
+@workspace_cli.command('update-page')
+@click.option('--order', help="@ORDER attribute for this mets:div", metavar='ORDER')
+@click.option('--orderlabel', help="@ORDERLABEL attribute for this mets:div", metavar='ORDERLABEL')
+@click.option('--contentids', help="@CONTENTIDS attribute for this mets:div", metavar='ORDERLABEL')
+@click.argument('PAGE_ID')
+@pass_workspace
+def update_page(ctx, order, orderlabel, contentids, page_id):
+    """
+    Update the @ORDER, @ORDERLABEL o @CONTENTIDS attributes of the mets:div with @ID=PAGE_ID
+    """
+    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace.mets.update_physical_page_attributes(page_id, order=order, orderlabel=orderlabel, contentids=contentids)
+    workspace.save_mets()
+
 # ----------------------------------------------------------------------
 # ocrd workspace merge
 # ----------------------------------------------------------------------

diff --git a/ocrd_models/ocrd_models/ocrd_mets.py b/ocrd_models/ocrd_models/ocrd_mets.py
@@ -6,6 +6,7 @@
 import typing
 from lxml import etree as ET
 from copy import deepcopy
+from warnings import warn
 
 from ocrd_utils import (
     getLogger,
@@ -665,6 +666,22 @@ def set_physical_page_for_file(self, pageId, ocrd_file, order=None, orderlabel=N
             # Assign the ocrd fileID to the pageId in the cache
             self._fptr_cache[el_pagediv.get('ID')].update({ocrd_file.ID : el_fptr})
 
+    def update_physical_page_attributes(self, page_id, **kwargs):
+        mets_div = None
+        if self._cache_flag:
+            if page_id in self._page_cache.keys():
+                mets_div = [self._page_cache[page_id]]
+        else:
+            mets_div = self._tree.getroot().xpath(
+                'mets:structMap[@TYPE="PHYSICAL"]/mets:div[@TYPE="physSequence"]/mets:div[@TYPE="page"][@ID="%s"]' % page_id,
+                namespaces=NS)
+        if mets_div:
+            for attr_name, attr_value in kwargs.items():
+                if attr_value:
+                    mets_div[0].set(attr_name.upper(), attr_value)
+        else:
+            warn("Could not find mets:div[@ID={page_id}]")
+
     def get_physical_page_for_file(self, ocrd_file):
         """
         Get the physical page ID (``@ID`` of the physical ``mets:structMap`` ``mets:div`` entry)

diff --git a/tests/model/test_ocrd_mets.py b/tests/model/test_ocrd_mets.py
@@ -379,5 +379,21 @@ def test_envvar():
         assert not OcrdMets(filename=assets.url_of('SBB0000F29300010000/data/mets.xml'), cache_flag=True)._cache_flag
         assert not OcrdMets(filename=assets.url_of('SBB0000F29300010000/data/mets.xml'), cache_flag=False)._cache_flag
 
+def test_update_physical_page_attributes(sbb_directory_ocrd_mets):
+    m = sbb_directory_ocrd_mets
+    m.remove_file()
+    assert len(m.physical_pages) == 0
+    m.add_file('FOO', pageId='new page', ID='foo1', mimetype='foo/bar')
+    m.add_file('FOO', pageId='new page', ID='foo2', mimetype='foo/bar')
+    m.add_file('FOO', pageId='new page', ID='foo3', mimetype='foo/bar')
+    m.add_file('FOO', pageId='new page', ID='foo4', mimetype='foo/bar')
+    assert len(m.physical_pages) == 1
+    assert b'ORDER' not in m.to_xml()
+    assert b'ORDERLABEL' not in m.to_xml()
+    m.update_physical_page_attributes('new page', order='foo', orderlabel='bar')
+    assert b'ORDER' in m.to_xml()
+    assert b'ORDERLABEL' in m.to_xml()
+
+
 if __name__ == '__main__':
     main(__file__)