jenkins-bot submitted this change.

View Change

[fix] raise SectionError if a section does not exists on a page

Also add some tests and update documentation.

Bug: T107141
Change-Id: Ib515e94657e7100695b69d77934d448ce424e232
---
M pywikibot/page/_basepage.py
M tests/page_tests.py
2 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/pywikibot/page/_basepage.py b/pywikibot/page/_basepage.py
index 8eaeeb0..1c51f20 100644
--- a/pywikibot/page/_basepage.py
+++ b/pywikibot/page/_basepage.py
@@ -292,11 +292,10 @@
         return title
 
     def section(self) -> str | None:
-        """
-        Return the name of the section this Page refers to.
+        """Return the name of the section this Page refers to.
 
-        The section is the part of the title following a '#' character, if
-        any. If no section is present, return None.
+        The section is the part of the title following a ``#`` character,
+        if any. If no section is present, return None.
         """
         try:
             section = self._link.section
@@ -373,12 +372,14 @@
          ...
         pywikibot.exceptions.IsRedirectPageError: ... is a redirect page.
 
+        .. versionchanged:: 9.2
+           :exc:`exceptions.SectionError` is raised if the
+           :meth:`section` does not exists
         .. seealso:: :attr:`text` property
 
         :param force: reload all page attributes, including errors.
         :param get_redirect: return the redirect text, do not follow the
             redirect, do not raise an exception.
-
         :raises NoPageError: The page does not exist.
         :raises IsRedirectPageError: The page is a redirect.
         :raises SectionError: The section does not exist on a page with
@@ -394,7 +395,18 @@
             if not get_redirect:
                 raise
 
-        return self.latest_revision.text  # type: ignore[attr-defined]
+        text = self.latest_revision.text
+
+        # check for valid section in title
+        page_section = self.section()
+        if page_section:
+            content = textlib.extract_sections(text, self.site)
+            headings = {section.heading for section in content.sections}
+            if page_section not in headings:
+                raise SectionError(f'{page_section!r} is not a valid section '
+                                   f'of {self.title(with_section=False)}')
+
+        return text
 
     def has_content(self) -> bool:
         """
diff --git a/tests/page_tests.py b/tests/page_tests.py
index b56cf4b..ea4dfd8 100755
--- a/tests/page_tests.py
+++ b/tests/page_tests.py
@@ -22,6 +22,7 @@
     IsNotRedirectPageError,
     IsRedirectPageError,
     NoPageError,
+    SectionError,
     TimeoutError,
     UnknownExtensionError,
 )
@@ -974,6 +975,7 @@
 
     def testPageGet(self):
         """Test ``Page.get()`` on different types of pages."""
+        fail_msg = '{page!r}.get() raised {error!r} unexpectedly!'
         site = self.get_site('en')
         p1 = pywikibot.Page(site, 'User:Legoktm/R2')
         p2 = pywikibot.Page(site, 'User:Legoktm/R1')
@@ -986,9 +988,27 @@
                                     r'{} is a redirect page\.'
                                     .format(re.escape(str(p2)))):
             p2.get()
+
+        try:
+            p2.get(get_redirect=True)
+        except (IsRedirectPageError, NoPageError, SectionError) as e:
+            self.fail(fail_msg.format(page=p2, error=e))
+
         with self.assertRaisesRegex(NoPageError, NO_PAGE_RE):
             p3.get()
 
+        page = pywikibot.Page(site, 'User:Legoktm/R2#Section')
+        with self.assertRaisesRegex(SectionError,
+                                    "'Section' is not a valid section"):
+            page.get()
+
+        site = pywikibot.Site('mediawiki')
+        page = pywikibot.Page(site, 'Manual:Pywikibot/2.0 #See_also')
+        try:
+            page.get()
+        except (IsRedirectPageError, NoPageError, SectionError) as e:
+            self.fail(fail_msg.format(page=page, error=e))
+
     def test_set_redirect_target(self):
         """Test set_redirect_target method."""
         # R1 redirects to R2 and R3 doesn't exist.

To view, visit change 1030049. To unsubscribe, or for help writing mail filters, visit settings.