Take article_url into account for articles, translated articles, and pages

.gitignore

@@ -1,2 +1,3 @@
 .pdm-python
 pdm.lock
+__pycache__/

pelican/plugins/sitemap/sitemap.py

@@ -130,7 +130,14 @@ def is_excluded(item):
                 or any(re.search(pattern, url) for pattern in excluded)
             )
 
-        page_queue = [(clean_url(to_url(path)), obj) for path, obj in self.page_queue]
+        # Use obj.url for articles/pages to respect custom URL settings
+        # (e.g., ARTICLE_URL). Fall back to to_url(path) for index pages
+        # (archives, tags, etc.) which don't have a .url property as they're
+        # not Article or Page objects.
+        page_queue = [
+            (clean_url(obj.url if obj else to_url(path)), obj)
+            for path, obj in self.page_queue
+        ]
         page_queue = [page for page in page_queue if not is_excluded(page)]
         page_queue.sort(key=lambda i: i[0])
 
@@ -174,12 +181,12 @@ def is_excluded(item):
                     )
                     priority = priorities[content_type]
 
+                # Use trans.url to respect custom URL settings for translations too
                 translations = "".join(
                     XML_TRANSLATION.format(
                         trans.lang,
                         siteurl,
-                        # save_as path is already output-relative
-                        clean_url(pathname2url(trans.save_as)),
+                        clean_url(trans.url),
                     )
                     for trans in getattr(obj, "translations", ())
                 )

pelican/plugins/sitemap/test_data/pages/about.md

@@ -0,0 +1,4 @@
+Title: About Us
+Status: published
+
+This is a test page for the about section.

pelican/plugins/sitemap/test_data/translated.md

@@ -0,0 +1,7 @@
+Title: Translated Post
+Date: 2023-07-12 13:00:00
+Category: test
+Lang: en
+Slug: translated-post
+
+This is an English article with a French translation.

pelican/plugins/sitemap/test_data/translated_fr.md

@@ -0,0 +1,7 @@
+Title: Article Traduit
+Date: 2023-07-12 13:00:00
+Category: test
+Lang: fr
+Slug: translated-post
+
+Ceci est un article français.

pelican/plugins/sitemap/test_sitemap.py

@@ -48,12 +48,15 @@ def test_txt(self):
 http://localhost/authors.html
 http://localhost/categories.html
 http://localhost/category/test.html
+http://localhost/pages/about-us.html
 http://localhost/tag/bar.html
 http://localhost/tag/foo.html
 http://localhost/tag/foobar.html
 http://localhost/tags.html
 http://localhost/test-post-daily.html
 http://localhost/test-post.html
+http://localhost/translated-post-fr.html
+http://localhost/translated-post.html
 """
         self.assertEqual(expected, contents)
 
@@ -80,3 +83,107 @@ def test_xml(self):
 </url>
 """
         self.assertIn(needle, contents)
+
+    def test_custom_article_url(self):
+        """Test that custom ARTICLE_URL settings are respected in sitemap."""
+        settings = read_settings(
+            override={
+                "PATH": TEST_DATA,
+                "CACHE_CONTENT": False,
+                "SITEURL": "http://localhost",
+                "OUTPUT_PATH": self.output_path,
+                "PLUGINS": [sitemap],
+                "SITEMAP": {
+                    "format": "txt",
+                },
+                # Custom URL settings like a user might have
+                "ARTICLE_URL": "blog/{slug}/",
+                "ARTICLE_SAVE_AS": "{slug}/index.html",
+            }
+        )
+        pelican = Pelican(settings=settings)
+        pelican.run()
+
+        with open(Path(self.output_path) / "sitemap.txt") as fd:
+            contents = fd.read()
+
+        # Articles should use ARTICLE_URL (blog/{slug}/),
+        # not ARTICLE_SAVE_AS ({slug}/index.html)
+        self.assertIn("http://localhost/blog/test-post/", contents)
+        self.assertIn("http://localhost/blog/test-post-daily/", contents)
+        # Should NOT contain the filesystem paths
+        self.assertNotIn("http://localhost/test-post/", contents)
+        self.assertNotIn("http://localhost/test-post-daily/", contents)
+
+    def test_translations_in_sitemap(self):
+        """Test that translation links use trans.url and appear in sitemap."""
+        self._run_pelican(sitemap_format="xml")
+        with open(Path(self.output_path) / "sitemap.xml") as fd:
+            contents = fd.read()
+
+        # Verify translation alternate links exist
+        self.assertIn('<xhtml:link rel="alternate"', contents)
+        # Verify they reference the correct translated article URLs
+        self.assertIn('hreflang="en"', contents)
+        self.assertIn('hreflang="fr"', contents)
+        # Verify translated articles are in the sitemap
+        self.assertIn("http://localhost/translated-post.html", contents)
+        self.assertIn("http://localhost/translated-post-fr.html", contents)
+
+    def test_translations_with_custom_article_url(self):
+        """Test translation links respect custom ARTICLE_URL settings."""
+        settings = read_settings(
+            override={
+                "PATH": TEST_DATA,
+                "CACHE_CONTENT": False,
+                "SITEURL": "http://localhost",
+                "OUTPUT_PATH": self.output_path,
+                "PLUGINS": [sitemap],
+                "SITEMAP": {
+                    "format": "xml",
+                },
+                # Custom URL settings
+                "ARTICLE_URL": "blog/{slug}/",
+                "ARTICLE_SAVE_AS": "{slug}/index.html",
+            }
+        )
+        pelican = Pelican(settings=settings)
+        pelican.run()
+
+        with open(Path(self.output_path) / "sitemap.xml") as fd:
+            contents = fd.read()
+
+        # Verify translation link for English uses custom ARTICLE_URL
+        self.assertIn('ref="http://localhost/blog/translated-post/"', contents)
+        # Verify the French translation is also in sitemap
+        # (French uses ARTICLE_LANG_URL which defaults differently)
+        self.assertIn("translated-post-fr", contents)
+
+    def test_custom_page_url(self):
+        """Test that custom PAGE_URL settings are respected in sitemap."""
+        settings = read_settings(
+            override={
+                "PATH": TEST_DATA,
+                "CACHE_CONTENT": False,
+                "SITEURL": "http://localhost",
+                "OUTPUT_PATH": self.output_path,
+                "PLUGINS": [sitemap],
+                "SITEMAP": {
+                    "format": "txt",
+                },
+                # Custom URL settings for pages
+                "PAGE_URL": "pages/{slug}/",
+                "PAGE_SAVE_AS": "p/{slug}/index.html",
+            }
+        )
+        pelican = Pelican(settings=settings)
+        pelican.run()
+
+        with open(Path(self.output_path) / "sitemap.txt") as fd:
+            contents = fd.read()
+
+        # Pages should use PAGE_URL (pages/{slug}/),
+        # not PAGE_SAVE_AS (p/{slug}/index.html)
+        self.assertIn("http://localhost/pages/about-us/", contents)
+        # Should NOT contain the filesystem path
+        self.assertNotIn("http://localhost/p/about-us/", contents)