FilePage: raise ValueError when title doesn't have a valid file exten…

…sion

Bug: T345786
Change-Id: Iba951506619f1f14f75e7bfb33c7afc99a240a5a

pywikibot/page/_filepage.py

@@ -44,6 +44,13 @@ def __init__(self, source, title: str = '') -> None:
         super().__init__(source, title, 6)
         if self.namespace() != 6:
             raise ValueError(f"'{self.title()}' is not in the file namespace!")
+        title = self.title(with_section=False)
+        extension = title.rpartition('.')[2].lower()
+        if extension not in self.site.file_extensions:
+            raise ValueError(
+                f'{title!r} does not have a valid extension '
+                f'({", ".join(self.site.file_extensions)}).'
+            )
 
     def _load_file_revisions(self, imageinfo) -> None:
         for file_rev in imageinfo:

pywikibot/site/_apisite.py

@@ -471,6 +471,11 @@ def logout(self) -> None:
         # Clear also cookies for site's second level domain (T224712)
         api._invalidate_superior_cookies(self.family)
 
+    @property
+    def file_extensions(self) -> List[str]:
+        """File extensions enabled on the wiki."""
+        return sorted(e['ext'] for e in self.siteinfo.get('fileextensions'))
+
     @property
     def maxlimit(self) -> int:
         """Get the maximum limit of pages to be retrieved.

tests/file_tests.py

@@ -164,24 +164,24 @@ class TestFilePage(TestCase):
     def test_file_info_with_no_page(self):
         """FilePage:latest_file_info raises NoPageError for missing pages."""
         site = self.get_site()
-        image = pywikibot.FilePage(site, 'File:NoPage')
+        image = pywikibot.FilePage(site, 'File:NoPage.jpg')
         self.assertFalse(image.exists())
 
         with self.assertRaisesRegex(
                 NoPageError,
-                (r'Page \[\[(wikipedia\:|)test:File:NoPage\]\] '
+                (r'Page \[\[(wikipedia\:|)test:File:NoPage\.jpg\]\] '
                  r"doesn't exist\.")):
             image = image.latest_file_info
 
     def test_file_info_with_no_file(self):
         """FilePage:latest_file_info raises PagerelatedError if no file."""
         site = self.get_site()
-        image = pywikibot.FilePage(site, 'File:Test with no image')
+        image = pywikibot.FilePage(site, 'File:Test with no image.png')
         self.assertTrue(image.exists())
         with self.assertRaisesRegex(
                 PageRelatedError,
                 (r'loadimageinfo: Query on '
-                 r'\[\[(wikipedia\:|)test:File:Test with no image\]\]'
+                 r'\[\[(wikipedia\:|)test:File:Test with no image\.png\]\]'
                  r' returned no imageinfo')):
             image = image.latest_file_info
 
@@ -322,13 +322,13 @@ def test_changed_title(self):
     def test_not_existing_download(self):
         """Test not existing download."""
         page = pywikibot.FilePage(self.site,
-                                  'File:Albert Einstein.jpg_notexisting')
+                                  'File:notexisting_Albert Einstein.jpg')
         filename = join_images_path('Albert Einstein.jpg')
 
         with self.assertRaisesRegex(
                 NoPageError,
-                re.escape('Page [[commons:File:Albert Einstein.jpg '
-                          "notexisting]] doesn't exist.")):
+                re.escape('Page [[commons:File:Notexisting Albert '
+                          "Einstein.jpg]] doesn't exist.")):
             page.download(filename)
 
 
@@ -373,7 +373,7 @@ def test_data_item(self):
     def test_data_item_not_existing(self):
         """Test data item associated to file that does not exist."""
         page = pywikibot.FilePage(self.site,
-                                  'File:Albert Einstein.jpg_notexisting')
+                                  'File:Notexisting_Albert Einstein.jpg')
         item = page.data_item()
         with self.assertRaises(NoWikibaseEntityError):
             item.get()

tests/page_tests.py

@@ -263,6 +263,14 @@ def testFileTitle(self):
             p2.title(as_link=True, textlink=True,
                      with_ns=False, insite=site),
             '[[:File:Jean-Léon Gérôme 003.jpg|Jean-Léon Gérôme 003.jpg]]')
+        for title in (
+            'Help:Example',  # wrong namespace
+            'File:Example',  # no file extension
+            'File:Example #3.jpg',  # file extension in section
+        ):
+            with self.subTest(title=title):
+                with self.assertRaises(ValueError):
+                    pywikibot.FilePage(site, title)
 
     def testImageAndDataRepository(self):
         """Test image_repository and data_repository page attributes."""