Perform redirects at DB level

These queries a little more complex but they allow us to rely on the
database to perform the filter and check for all the redirect in a
single query than fetching all the instances of the Redirect objects
and iterating over them.

This will perform only one query to the DB and the result will be the
redirect we need to apply or no redirects at all.

Author: humitos

readthedocs/redirects/querysets.py

@@ -1,9 +1,9 @@
 """Queryset for the redirects app."""
 
 from django.db import models
+from django.db.models import Value, CharField, Q, F
 
 from readthedocs.core.utils.extend import SettingsOverrideObject
-from readthedocs.projects import constants
 
 
 class RedirectQuerySetBase(models.QuerySet):
@@ -25,8 +25,55 @@ def api(self, user=None, detail=True):
             queryset = self._add_user_repos(queryset, user)
         return queryset
 
-    def get_redirect_path_with_status(self, path, language=None, version_slug=None):
-        for redirect in self.select_related('project'):
+    def get_redirect_path_with_status(self, path, full_path=None, language=None, version_slug=None):
+        # add extra fields with the ``path`` and ``full_path`` to perform a
+        # filter at db level instead with Python
+        queryset = self.annotate(
+            path=Value(
+                path,
+                output_field=CharField(),
+            ),
+            full_path=Value(
+                full_path,
+                output_field=CharField(),
+            ),
+        )
+        prefix = Q(
+            redirect_type='prefix',
+            path__startswith=F('from_url'),
+        )
+        page = Q(
+            redirect_type='page',
+            path__iexact=F('from_url'),
+        )
+        exact = (
+            Q(
+                redirect_type='exact',
+                from_url__endswith='$rest',  # Python implementation does "in"
+            ) | Q(
+                redirect_type='exact',
+                full_path__iexact=F('from_url'),
+            )
+        )
+        sphinx_html = (
+            Q(
+                redirect_type='sphinx_html',
+                path__endswith='/',
+            ) | Q(
+                redirect_type='sphinx_html',
+                path__endswith='/index.html',
+            )
+        )
+        sphinx_htmldir = Q(
+            redirect_type='sphinx_html',
+            path__endswith='.html',
+        )
+
+        # There should be one and only one redirect returned by this query. I
+        # can't think in a case where there can be more at this point. I'm
+        # leaving the loop just in case for now
+        queryset = queryset.filter(prefix | page | exact | sphinx_html | sphinx_htmldir)
+        for redirect in queryset.select_related('project'):
             new_path = redirect.get_redirect_path(
                 path=path,
                 language=language,

readthedocs/redirects/utils.py

@@ -84,7 +84,7 @@ def get_redirect_response(request, full_path):
         language, version_slug, path = language_and_version_from_path(path)
 
     path, http_status = project.redirects.get_redirect_path_with_status(
-        path=path, language=language, version_slug=version_slug
+        path=path, full_path=full_path, language=language, version_slug=version_slug
     )
 
     if path is None: