Merge pull request #78 from patrickfournier/sitename-and-summary

Enhance OpenGraph tags

Author: MaevaBrunelles

README.md

@@ -269,6 +269,11 @@ Image: https://www.example.com/article-image.jpg
 
 Based on [Open Graph protocol](https://ogp.me), the SEO plugin implements required properties and some aditionnals ones:
 
+```
+<meta property="og:site_name" content=":sitename:" />
+```
+`:sitename:`: The `SITENAME` from the Pelican settings.
+
 ```
 <meta property="og:url" content=":fileurl:" />
 ```
@@ -293,6 +298,7 @@ og_image: https://www.example.com/og-image.jpg
 ```
 
 If these metadata are not declared, `:title:`, `:description:`, `:image:` will be filled by the default `Title`, `Description` (Pelican metadata) and `Image` (plugin metadata) if they exist.
+If `Description` is not defined, a plain text version of `Summary` will be used instead.
 
 ```
 <meta property="og:locale" content=":language:">

pelican/plugins/seo/seo_enhancer/html_enhancer/__init__.py

@@ -1,5 +1,7 @@
 """HTML Enhancer : get instances of HTML enhancements."""
 
+from bs4 import BeautifulSoup
+
 from pelican.contents import Article, Page
 
 from .article_schema_creator import ArticleSchemaCreator
@@ -9,6 +11,17 @@
 from .twitter_cards import TwitterCards
 
 
+def _get_plain_text_summary(metadata):
+    """Get content from summary, without HTML tags."""
+
+    soup = BeautifulSoup(
+        metadata.get("summary", ""),
+        "html.parser",
+    )
+    text_summary = soup.get_text().strip()
+    return text_summary
+
+
 class HTMLEnhancer:
     """HTML Enhancer : get instances of HTML enhancements."""
 
@@ -63,12 +76,14 @@ def __init__(self, file, output_path, path, open_graph=False, twitter_cards=Fals
 
         if open_graph:
             self.open_graph = OpenGraph(
+                sitename=_settings.get("SITENAME"),
                 siteurl=_settings.get("SITEURL"),
                 fileurl=_fileurl,
                 file_type=_file_type,
                 title=_metadata.get("og_title") or _title,
                 description=_metadata.get("og_description")
-                or _metadata.get("description"),
+                or _metadata.get("description")
+                or _get_plain_text_summary(_metadata),
                 image=_metadata.get("og_image") or _image,
                 locale=_settings.get("LOCALE"),
             )

pelican/plugins/seo/seo_enhancer/html_enhancer/open_graph.py

@@ -9,8 +9,9 @@ class OpenGraph:
     """
 
     def __init__(
-        self, siteurl, fileurl, file_type, title, description, image, locale
+        self, sitename, siteurl, fileurl, file_type, title, description, image, locale
     ) -> None:
+        self.sitename = sitename
         self.siteurl = siteurl
         self.fileurl = fileurl
         self.type = file_type
@@ -57,6 +58,9 @@ def create_tags(self) -> dict:
         """
         open_graph_tags = {}
 
+        if self.sitename:
+            open_graph_tags["site_name"] = self.sitename
+
         open_graph_tags["url"] = self._create_absolute_fileurl()
         open_graph_tags["type"] = self.type
 

pelican/plugins/seo/seo_report/seo_analyzer/__init__.py

@@ -11,7 +11,9 @@ class SEOAnalyzer:
 
     def __init__(self, article):
         self._title = getattr(article, "title", None)
-        self._description = getattr(article, "description", None)
+        self._description = getattr(article, "description", None) or getattr(
+            article, "summary", None
+        )
         self._content = getattr(article, "content", None)
         self._settings = getattr(article, "settings", None)
 

pelican/plugins/seo/tests/conftest.py

@@ -75,6 +75,7 @@ def fake_article():
         "og_description": "OG Description",
         "og_image": "https://www.fakesite.com/og-image.jpg",
         "tw_account": "@TestTWCards",
+        "summary": "Fake summary",
     }
     title = "Fake Title"
     description = "Fake description"

pelican/plugins/seo/tests/test_open_graph.py

@@ -17,6 +17,7 @@ def test_create_absolute_fileurl(self, fake_article):
         """
 
         og = OpenGraph(
+            sitename=None,
             siteurl=fake_article.settings["SITEURL"],
             fileurl=fake_article.url,
             file_type=None,
@@ -39,6 +40,7 @@ def test_create_absolute_fileurl_with_site_url_with_path(self, site_url, file_ur
         slash and a file URL with or without a leading slash properly.
         """
         og = OpenGraph(
+            sitename=None,
             siteurl=site_url,
             fileurl=file_url,
             file_type=None,
@@ -67,6 +69,7 @@ def test_get_locale(self, locale, expected_result):
         """
 
         og = OpenGraph(
+            sitename=None,
             siteurl=None,
             fileurl=None,
             file_type=None,
@@ -86,6 +89,7 @@ def test_create_tags(self, fake_article):
         """
 
         og = OpenGraph(
+            sitename=fake_article.settings["SITENAME"],
             siteurl=fake_article.settings["SITEURL"],
             fileurl=fake_article.url,
             file_type="article",
@@ -97,6 +101,7 @@ def test_create_tags(self, fake_article):
 
         og_tags = og.create_tags()
 
+        assert og_tags["site_name"] == "Fake Site Name"
         assert og_tags["url"] == "https://www.fakesite.com/fake-title.html"
         assert og_tags["type"] == "article"
         assert og_tags["title"] == "OG Title"
@@ -111,6 +116,7 @@ def test_create_tags_missing_elements(self, fake_article_missing_elements):
         """
 
         og = OpenGraph(
+            sitename=fake_article_missing_elements.settings["SITENAME"],
             siteurl=fake_article_missing_elements.settings["SITEURL"],
             fileurl=fake_article_missing_elements.url,
             file_type="article",
@@ -122,6 +128,7 @@ def test_create_tags_missing_elements(self, fake_article_missing_elements):
 
         og_tags = og.create_tags()
 
+        assert "site_name" not in og_tags
         assert "title" not in og_tags
         assert "description" not in og_tags
         assert "image" not in og_tags

pelican/plugins/seo/tests/test_seo_enhancer.py

@@ -233,12 +233,77 @@ def test_add_html_enhancements_to_file_with_open_graph(
                     <link href="https://www.fakesite.com/fake-title.html" rel="canonical"/>
 <script type="application/ld+json">{"@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [{"@type": "ListItem", "position": 1, "name": "Fake Site Name", "item": "https://www.fakesite.com"}, {"@type": "ListItem", "position": 2, "name": "Fake_file", "item": "https://www.fakesite.com/fake_file.html"}]}</script>
 <script type="application/ld+json">{"@context": "https://schema.org", "@type": "Article", "author": {"@type": "Person", "name": "Fake author"}, "publisher": {"@type": "Organization", "name": "Fake Site Name", "logo": {"@type": "ImageObject", "url": "https://www.fakesite.com/fake-logo.jpg"}}, "headline": "Fake Title", "about": "Fake category", "datePublished": "2019-04-03 23:49"}</script>
+<meta content="Fake Site Name" property="og:site_name"/>
 <meta content="https://www.fakesite.com/fake-title.html" property="og:url"/>
 <meta content="website" property="og:type"/>
 <meta content="OG Title" property="og:title"/>
 <meta content="OG Description" property="og:description"/>
 <meta content="https://www.fakesite.com/og-image.jpg" property="og:image"/>
 <meta content="fr_FR" property="og:locale"/>
+</head>
+                    <body>
+                        <h1>Fake content title</h1>
+                        <p>Fake content 🙃</p>
+                        <a href="https://www.fakesite.com">Fake internal link</a>
+                        <p>Fake content with <code>inline code</code></p>
+                        <p>Fake content with "<a href="https://www.fakesite.com">Fake inline internal link</a>"</p>
+                    </body>
+                </html>"""
+            )
+
+    def test_add_html_enhancements_to_file_with_open_graph_using_summary_for_description(
+        self, fake_article, fake_seo_enhancer
+    ):
+        """
+        Test if add_html_to_file with open_graph setting
+        adds Open Graph tags to HTML files.
+        """
+
+        # Remove higher priority values for the og:description tag to force the use of
+        # the summary from article description.
+        del fake_article.metadata["og_description"]
+        del fake_article.description
+
+        path = "fake_output/fake_file.html"
+        fake_html_enhancements = fake_seo_enhancer.launch_html_enhancer(
+            file=fake_article,
+            output_path="fake_output",
+            path=path,
+            open_graph=True,
+        )
+
+        with patch(
+            "seo.seo_enhancer.open", mock_open(read_data=fake_article.content)
+        ) as mocked_open:
+            mocked_file_handle = mocked_open.return_value
+
+            fake_seo_enhancer.add_html_to_file(
+                enhancements=fake_html_enhancements, path=path
+            )
+            assert len(mocked_open.call_args_list) == 2
+            mocked_file_handle.read.assert_called_once()
+            mocked_file_handle.write.assert_called_once()
+
+            write_args, _ = mocked_file_handle.write.call_args_list[0]
+            fake_html_content = write_args[0]
+
+            # The og:description tag should now contain "Fake summary".
+            assert (
+                fake_html_content
+                == """<html>
+                    <head>
+                        <title>Fake Title</title>
+                        <meta content="Fake description" name="description"/>
+                    <link href="https://www.fakesite.com/fake-title.html" rel="canonical"/>
+<script type="application/ld+json">{"@context": "https://schema.org", "@type": "BreadcrumbList", "itemListElement": [{"@type": "ListItem", "position": 1, "name": "Fake Site Name", "item": "https://www.fakesite.com"}, {"@type": "ListItem", "position": 2, "name": "Fake_file", "item": "https://www.fakesite.com/fake_file.html"}]}</script>
+<script type="application/ld+json">{"@context": "https://schema.org", "@type": "Article", "author": {"@type": "Person", "name": "Fake author"}, "publisher": {"@type": "Organization", "name": "Fake Site Name", "logo": {"@type": "ImageObject", "url": "https://www.fakesite.com/fake-logo.jpg"}}, "headline": "Fake Title", "about": "Fake category", "datePublished": "2019-04-03 23:49"}</script>
+<meta content="Fake Site Name" property="og:site_name"/>
+<meta content="https://www.fakesite.com/fake-title.html" property="og:url"/>
+<meta content="website" property="og:type"/>
+<meta content="OG Title" property="og:title"/>
+<meta content="Fake summary" property="og:description"/>
+<meta content="https://www.fakesite.com/og-image.jpg" property="og:image"/>
+<meta content="fr_FR" property="og:locale"/>
 </head>
                     <body>
                         <h1>Fake content title</h1>
@@ -295,6 +360,7 @@ def test_add_html_enhancements_to_file_with_twitter_cards(
 <script type="application/ld+json">{"@context": "https://schema.org", "@type": "Article", "author": {"@type": "Person", "name": "Fake author"}, "publisher": {"@type": "Organization", "name": "Fake Site Name", "logo": {"@type": "ImageObject", "url": "https://www.fakesite.com/fake-logo.jpg"}}, "headline": "Fake Title", "about": "Fake category", "datePublished": "2019-04-03 23:49"}</script>
 <meta content="summary" name="twitter:card"/>
 <meta content="@TestTWCards" name="twitter:site"/>
+<meta content="Fake Site Name" property="og:site_name"/>
 <meta content="https://www.fakesite.com/fake-title.html" property="og:url"/>
 <meta content="website" property="og:type"/>
 <meta content="OG Title" property="og:title"/>