Add NamedBytesIO to work around pyxform change…

in XLSForm/pyxform#712

kobo/apps/openrosa/apps/viewer/models/data_dictionary.py

@@ -21,6 +21,7 @@
 from kobo.apps.openrosa.libs.utils.model_tools import queryset_iterator, set_uuid
 from kpi.constants import DEFAULT_SURVEY_NAME
 from kpi.utils.mongo_helper import MongoHelper
+from kpi.utils.pyxform_compatibility import NamedBytesIO
 
 
 class ColumnRename(models.Model):
@@ -157,8 +158,9 @@ def add_instances(self):
 
     def save(self, *args, **kwargs):
         if self.xls:
+            xls_io = NamedBytesIO.fromfieldfile(self.xls)
             survey = create_survey_from_xls(
-                self.xls, default_name=DEFAULT_SURVEY_NAME
+                xls_io, default_name=DEFAULT_SURVEY_NAME
             )
             if survey.name == DEFAULT_SURVEY_NAME:
                 survey.name = survey.id_string

kpi/utils/pyxform_compatibility.py

@@ -1,5 +1,7 @@
+from io import BytesIO
 from pyxform.constants import ALLOW_CHOICE_DUPLICATES
 
+
 def allow_choice_duplicates(content: dict) -> None:
     """
     Modify `content` to include `allow_choice_duplicates=Yes` in the settings
@@ -13,3 +15,39 @@ def allow_choice_duplicates(content: dict) -> None:
     settings = content.setdefault('settings', {})
     if ALLOW_CHOICE_DUPLICATES not in settings:
         settings[ALLOW_CHOICE_DUPLICATES] = 'yes'
+
+
+class NamedBytesIO(BytesIO):
+    """
+    Changes in XLSForm/pyxform#718 prevent
+    `pyxform.builder.create_survey_from_xls()` from accepting a
+    `django.db.models.fields.files.FieldFile`. Only instances of
+    `bytes | BytesIO | IOBase` are now accepted for treatment as file-like
+    objects, and furthermore, anything that is not already a `BytesIO` will
+    have its contents placed inside a newly instantiated one.
+
+    Problem: `BytesIO`s do not have `name`s, and the constructor for
+    `pyxform.xls2json.SurveyReader` fails because of that.
+
+    Workaround: a `BytesIO` with a `name` 🙃
+
+    For more details, see
+    https://github.com/kobotoolbox/kpi/pull/5126#discussion_r1829763316
+    """
+
+    def __init__(self, *args, name=None, **kwargs):
+        if name is None:
+            raise NotImplementedError('Use `BytesIO` if no `name` is needed')
+        super().__init__(*args, **kwargs)
+        self.name = name
+
+    @classmethod
+    def fromfieldfile(cls, django_fieldfile):
+        """
+        Given a Django `FieldFile`, return an instance of `NamedBytesIO`
+
+        à la `datetime.datetime.fromtimestamp()`
+        """
+        new_instance = cls(django_fieldfile.read(), name=django_fieldfile.name)
+        django_fieldfile.seek(0)  # Be kind: rewind
+        return new_instance