[bugfix] Handle missing SDC mediainfo

Test if mediainfo exists. Added get_data_for_new_entity() for creating new mediainfo.

Bug: T345038
Change-Id: Ifdd4d707807f2dc6f901f10e3ae40d9537e00fe7
---
M pywikibot/page/_collections.py
M pywikibot/page/_wikibase.py
M tests/file_tests.py
3 files changed, 71 insertions(+), 1 deletion(-)

diff --git a/pywikibot/page/_collections.py b/pywikibot/page/_collections.py
index a88ccf7..cea1f71 100644
--- a/pywikibot/page/_collections.py
+++ b/pywikibot/page/_collections.py
@@ -210,6 +210,8 @@
     def fromJSON(cls, data, repo):
         """Construct a new ClaimCollection from JSON."""
         this = cls(repo)
+        if data == []:  # workaround for T222159
+            return this
         for key, claims in data.items():
             this[key] = [pywikibot.page.Claim.fromJSON(repo, claim)
                          for claim in claims]
diff --git a/pywikibot/page/_wikibase.py b/pywikibot/page/_wikibase.py
index 959c0b1..5141e00 100644
--- a/pywikibot/page/_wikibase.py
+++ b/pywikibot/page/_wikibase.py
@@ -408,6 +408,12 @@
 
         return self._file
 
+    def get_data_for_new_entity(self) -> dict:
+        """Return data required for creation of a new mediainfo."""
+        self.id = 'M' + str(self.file.pageid)
+        self._content = {}
+        return super().get()
+
     def get(self, force: bool = False) -> dict:
         """Fetch all MediaInfo entity data and cache it.
 
@@ -427,6 +433,8 @@
                     data = self.file.latest_revision.slots['mediainfo']['*']
                 except NoPageError as exc:
                     raise NoWikibaseEntityError(self) from exc
+                except KeyError as exc:
+                    raise NoWikibaseEntityError(self) from exc
 
                 self._content = jsonlib.loads(data)
                 self.id = self._content['id']
diff --git a/tests/file_tests.py b/tests/file_tests.py
index fb3bdf0..911c595 100755
--- a/tests/file_tests.py
+++ b/tests/file_tests.py
@@ -17,6 +17,7 @@
     NoWikibaseEntityError,
     PageRelatedError,
 )
+from pywikibot import pagegenerators
 from tests import join_images_path
 from tests.aspects import TestCase
 
@@ -370,14 +371,61 @@
         del item._file
         self.assertEqual(page, item.file)
 
-    def test_data_item_not_existing(self):
+    def test_data_item_when_no_file_or_data_item(self):
         """Test data item associated to file that does not exist."""
         page = pywikibot.FilePage(self.site,
                                   'File:Notexisting_Albert Einstein.jpg')
+        self.assertFalse(page.exists())
         item = page.data_item()
+        self.assertIsInstance(item, pywikibot.MediaInfo)
+
         with self.assertRaises(NoWikibaseEntityError):
             item.get()
 
+    def test_data_item_when_file_exist_but_without_item(self):
+        """Test if data item is missing from file."""
+        # Get latest uploaded files.
+        gen = pagegenerators.RecentChangesPageGenerator(
+            site=self.site,
+            namespaces=[6],  # File namespace
+            changetype='new',
+            total=100
+        )
+
+        # Seek to first page without mediainfo.
+        for page in gen:
+            if 'mediainfo' not in page.latest_revision.slots:
+                item = page.data_item()
+                self.assertIsInstance(item, pywikibot.MediaInfo)
+
+                # Get fails as there is no mediainfo.
+                with self.assertRaises(NoWikibaseEntityError):
+                    item.get()
+
+                # Create new empty mediainfo.
+                item.get_data_for_new_entity()
+                self.assertIsInstance(
+                    item.labels, pywikibot.page._collections.LanguageDict)
+                self.assertIsInstance(
+                    item.statements,
+                    pywikibot.page._collections.ClaimCollection)
+
+                # break the loop after checking first file
+                break
+
+    def test_data_list_to_dict_workaround(self):
+        """Test that T222159 workaround converts [] to {}."""
+        page = pywikibot.FilePage(self.site, 'File:Albert Einstein.jpg')
+        item = page.data_item()
+        item.get(force=True)
+        item._content['labels'] = []
+        item._content['statements'] = []
+        item.get()
+        self.assertIsInstance(
+            item.labels, pywikibot.page._collections.LanguageDict)
+        self.assertIsInstance(
+            item.statements, pywikibot.page._collections.ClaimCollection)
+
 
 if __name__ == '__main__':  # pragma: no cover
     with suppress(SystemExit):

