Normalize URLs to GitHub project repos

In particular we want to remove any trailing `.git` from the project
URL, since this results in broken links when we contruct the "Fork"
and "Open Issues" links in the plugin page.

(While `https://github.com/_owner_/_project_.git` will work to get to the
repo, `https://github.com/_owner_/_project_.git/fork` is a 404.)

Author: dairiki

packages/project-data/lektor_project_data.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 import cgi
+import re
 
 import readme_renderer.markdown
 import readme_renderer.rst
@@ -17,6 +18,24 @@
 }
 
 
+def normalize_url(url):
+    """Normalize project home page URLs."""
+    # Normalize any URLS to GitHub project repos.
+    m = re.match(
+        r"""
+            https?://(?:www\.)?github\.com
+            / (?P<owner>[^/]+)
+            / (?P<project>[^/]+?) (?:\.git)
+            /? \Z
+        """,
+        url,
+        flags=re.VERBOSE
+    )
+    if m:
+        return "https://github.com/{owner}/{project}".format(**m.groupdict())
+    return url
+
+
 class ProjectDataPlugin(Plugin):
     name = 'Project Data'
     description = u'Retrieve project information from PyPI.'
@@ -71,6 +90,8 @@ def package_data(self, name, entry_point=None):
             self.data['description'], self.data['description_content_type'])
         if not self.data.get('home_page'):
             self.data['home_page'] = f'https://pypi.org/project/{name}/'
+        else:
+            self.data['home_page'] = normalize_url(self.data['home_page'])
 
     def github_data(self, owner=None, repo=None):
         url = 'https://api.github.com/repos/{}/{}'.format(owner, repo)