Merge pull request #2406 from mosra/metadata-intrasite-links-process-only-once

Don't process metadata links before all files are present

pelican/contents.py

@@ -142,9 +142,9 @@ class Content(object):
         if not hasattr(self, 'status'):
             self.status = getattr(self, 'default_status', None)
 
-        if (len(self._context.get('generated_content', [])) > 0 or
-                len(self._context.get('static_content', [])) > 0):
-            self.refresh_metadata_intersite_links()
+        # store the summary metadata if it is set
+        if 'summary' in metadata:
+            self._summary = metadata['summary']
 
         signals.content_object_init.send(self)
 
@@ -469,7 +469,7 @@ class Content(object):
 
     def refresh_metadata_intersite_links(self):
         for key in self.settings['FORMATTED_FIELDS']:
-            if key in self.metadata:
+            if key in self.metadata and key != 'summary':
                 value = self._update_content(
                     self.metadata[key],
                     self.get_siteurl()
@@ -477,6 +477,16 @@ class Content(object):
                 self.metadata[key] = value
                 setattr(self, key.lower(), value)
 
+        # _summary is an internal variable that some plugins may be writing to,
+        # so ensure changes to it are picked up
+        if ('summary' in self.settings['FORMATTED_FIELDS'] and
+                'summary' in self.metadata):
+            self._summary = self._update_content(
+                self._summary,
+                self.get_siteurl()
+            )
+            self.metadata['summary'] = self._summary
+
 
 class Page(Content):
     mandatory_properties = ('title',)

pelican/tests/cyclic_intersite_links/first-article.rst

@@ -0,0 +1,7 @@
+First article
+#############
+
+:date: 2018-11-10
+:summary: Here's the `second <{filename}/second-article.rst>`_,
+    `third <{filename}/third-article.rst>`_ and a
+    `nonexistent article <{filename}/nonexistent.rst>`_.

pelican/tests/cyclic_intersite_links/second-article.rst

@@ -0,0 +1,7 @@
+Second article
+##############
+
+:date: 2018-11-10
+:summary: Here's the `first <{filename}/first-article.rst>`_,
+    `third <{filename}/third-article.rst>`_ and a
+    `nonexistent article <{filename}/nonexistent.rst>`_.

pelican/tests/cyclic_intersite_links/third-article.rst

@@ -0,0 +1,7 @@
+Third article
+#############
+
+:date: 2018-11-10
+:summary: Here's the `first <{filename}/first-article.rst>`_,
+    `second <{filename}/second-article.rst>`_ and a
+    `nonexistent article <{filename}/nonexistent.rst>`_.

pelican/tests/test_contents.py

@@ -333,6 +333,10 @@ class TestPage(LoggedTestCase):
         args['metadata']['custom'] = parsed
         args['context']['localsiteurl'] = 'http://notmyidea.org'
         p = Page(**args)
+        # This is called implicitly from all generators and Pelican.run() once
+        # all files are processed. Here we process just one page so it needs
+        # to be called explicitly.
+        p.refresh_metadata_intersite_links()
         self.assertEqual(p.summary, linked)
         self.assertEqual(p.custom, linked)
 

pelican/tests/test_pelican.py

@@ -211,6 +211,29 @@ class TestPelican(LoggedTestCase):
             msg="Writing .*",
             level=logging.INFO)
 
+    def test_cyclic_intersite_links_no_warnings(self):
+        settings = read_settings(path=None, override={
+            'PATH': os.path.join(CURRENT_DIR, 'cyclic_intersite_links'),
+            'OUTPUT_PATH': self.temp_path,
+            'CACHE_PATH': self.temp_cache,
+        })
+        pelican = Pelican(settings=settings)
+        mute(True)(pelican.run)()
+        # There are four different intersite links:
+        # - one pointing to the second article from first and third
+        # - one pointing to the first article from second and third
+        # - one pointing to the third article from first and second
+        # - one pointing to a nonexistent from each
+        # If everything goes well, only the warning about the nonexistent
+        # article should be printed. Only two articles are not sufficient,
+        # since the first will always have _context['generated_content'] empty
+        # (thus skipping the link resolving) and the second will always have it
+        # non-empty, containing the first, thus always succeeding.
+        self.assertLogCountEqual(
+            count=1,
+            msg="Unable to find '.*\\.rst', skipping url replacement.",
+            level=logging.WARNING)
+
     def test_md_extensions_deprecation(self):
         """Test that a warning is issued if MD_EXTENSIONS is used"""
         settings = read_settings(path=None, override={