tests/workflow/test_export_to_model.py

@@ -4,6 +4,7 @@ import time
 import zipfile
 
 import pytest
+from django.utils.encoding import force_bytes
 from PIL import Image
 from pyzbar.pyzbar import ZBarSymbol
 from pyzbar.pyzbar import decode as zbar_decode_qrcode
@@ -299,10 +300,6 @@ def test_export_to_model_django_template(pub):
 
 
 def test_export_to_model_xml(pub):
-    LoggedError = pub.loggederror_class
-    if LoggedError:
-        LoggedError.wipe()
-
     formdef = FormDef()
     formdef.name = 'foo-export-to-template-with-django'
     formdef.fields = [
@@ -320,9 +317,10 @@ def test_export_to_model_xml(pub):
     item.attach_to_history = True
 
     def run(template, filename='/foo/template.xml', content_type='application/xml'):
+        pub.loggederror_class.wipe()
         upload = QuixoteUpload(filename, content_type=content_type)
         upload.fp = io.BytesIO()
-        upload.fp.write(template.encode())
+        upload.fp.write(force_bytes(template))
         upload.fp.seek(0)
         item.model_file = UploadedFile(pub.app_dir, None, upload)
         item.convert_to_pdf = False
@@ -340,18 +338,23 @@ def test_export_to_model_xml(pub):
     assert run(template='<a>{{ form_var_string }}</a>', filename='/foo/template.svg') == '<a>écho</a>'
 
     # unknown file format
-    with pytest.raises(UploadValidationError):
+    with pytest.raises(UploadValidationError) as e:
         run(
             template='<a>{{ form_var_string }}</a>',
             filename='/foo/template.txt',
             content_type='application/octet-stream',
         )
+    assert str(e.value) == 'Only OpenDocument and XML files can be used.'
+
+    # invalid UTF-8
+    with pytest.raises(UploadValidationError) as e:
+        assert run(template=b'<name>test \xE0 {{form_var_string}}</name>') == ''
+    assert str(e.value) == 'XML model files must be UTF-8.'
 
     # malformed XML
-    assert not LoggedError or LoggedError.count() == 0
     assert run(template='<a>{{ form_var_string }}<a>') == '<a>écho<a>'
     # on error in the XML correctness no exception is raised but an error is logged
-    assert not LoggedError or LoggedError.count() == 1
+    assert pub.loggederror_class.count() == 1
 
 
 @pytest.mark.parametrize('filename', ['template-form-details.odt', 'template-form-details-no-styles.odt'])

wcs/wf/export_to_model.py

@@ -361,13 +361,18 @@ class ExportToModel(WorkflowStatusItem):
 
         # XML
         fp.seek(0)
-        xml_prefix = fp.read(1)
+        xml_content = fp.read()
         fp.seek(0)
         if (upload.content_type and upload.content_type in ('text/xml', 'application/xml')) or (
-            upload.base_filename and upload.base_filename.endswith('.xml') and xml_prefix == b'<'
+            upload.base_filename and upload.base_filename.endswith('.xml') and xml_content.startswith(b'<')
         ):
+            # check XML content is valid UTF-8
+            try:
+                xml_content.decode('utf-8')
+            except UnicodeDecodeError:
+                raise UploadValidationError(_('XML model files must be UTF-8.'))
             return 'xml'
-        raise UploadValidationError(_('Only OpenDocument and XML files can be used'))
+        raise UploadValidationError(_('Only OpenDocument and XML files can be used.'))
 
     def get_parameters(self):
         parameters = ('model_file',)