Put under feature flag

Author: stsewd

readthedocs/projects/models.py

@@ -1568,8 +1568,14 @@ def add_features(sender, **kwargs):
     SKIP_SYNC_VERSIONS = 'skip_sync_versions'
     CACHED_ENVIRONMENT = 'cached_environment'
     LIMIT_CONCURRENT_BUILDS = 'limit_concurrent_builds'
+
+    # Search related features
     DISABLE_SERVER_SIDE_SEARCH = 'disable_server_side_search'
     ENABLE_MKDOCS_SERVER_SIDE_SEARCH = 'enable_mkdocs_server_side_search'
+    DEFAULT_TO_FUZZY_SEARCH = 'default_to_fuzzy_search'
+    INDEX_FROM_HTML_FILES = 'index_from_html_files'
+    SEARCH_SUBPROJECTS_ON_DEFAULT_VERSION = 'search_subprojects_on_default_version'
+
     FORCE_SPHINX_FROM_VENV = 'force_sphinx_from_venv'
     LIST_PACKAGES_INSTALLED_ENV = 'list_packages_installed_env'
     VCS_REMOTE_LISTING = 'vcs_remote_listing'
@@ -1578,8 +1584,6 @@ def add_features(sender, **kwargs):
     USE_SPHINX_BUILDERS = 'use_sphinx_builders'
     DEDUPLICATE_BUILDS = 'deduplicate_builds'
     USE_SPHINX_RTD_EXT_LATEST = 'rtd_sphinx_ext_latest'
-    DEFAULT_TO_FUZZY_SEARCH = 'default_to_fuzzy_search'
-    INDEX_FROM_HTML_FILES = 'index_from_html_files'
     DONT_CREATE_INDEX = 'dont_create_index'
     USE_NEW_PIP_RESOLVER = 'use_new_pip_resolver'
     DONT_INSTALL_LATEST_PIP = 'dont_install_latest_pip'
@@ -1667,6 +1671,8 @@ def add_features(sender, **kwargs):
             LIMIT_CONCURRENT_BUILDS,
             _('Limit the amount of concurrent builds'),
         ),
+
+        # Search related features.
         (
             DISABLE_SERVER_SIDE_SEARCH,
             _('Disable server side search'),
@@ -1675,6 +1681,19 @@ def add_features(sender, **kwargs):
             ENABLE_MKDOCS_SERVER_SIDE_SEARCH,
             _('Enable server side search for MkDocs projects'),
         ),
+        (
+            DEFAULT_TO_FUZZY_SEARCH,
+            _('Default to fuzzy search for simple search queries'),
+        ),
+        (
+            INDEX_FROM_HTML_FILES,
+            _('Index content directly from html files instead or relying in other sources'),
+        ),
+        (
+            SEARCH_SUBPROJECTS_ON_DEFAULT_VERSION,
+            _('When searching subprojects default to its default version if it doesn\'t have the same version as the main project'),
+        ),
+
         (
             FORCE_SPHINX_FROM_VENV,
             _('Force to use Sphinx from the current virtual environment'),
@@ -1710,14 +1729,6 @@ def add_features(sender, **kwargs):
             USE_SPHINX_RTD_EXT_LATEST,
             _('Use latest version of the Read the Docs Sphinx extension'),
         ),
-        (
-            DEFAULT_TO_FUZZY_SEARCH,
-            _('Default to fuzzy search for simple search queries'),
-        ),
-        (
-            INDEX_FROM_HTML_FILES,
-            _('Index content directly from html files instead or relying in other sources'),
-        ),
         (
             DONT_CREATE_INDEX,
             _('Do not create index.md or README.rst if the project does not have one.'),

readthedocs/search/api.py

@@ -232,8 +232,13 @@ def _get_all_projects_data(self):
                 version_slug=main_version.slug,
                 subproject=project,
             )
+
             # Fallback to the default version of the subproject.
-            if not version and project.default_version:
+            if (
+                not version
+                and main_project.has_feature(Feature.SEARCH_SUBPROJECTS_ON_DEFAULT_VERSION)
+                and project.default_version
+            ):
                 version = self._get_subproject_version(
                     version_slug=project.default_version,
                     subproject=project,
@@ -297,24 +302,40 @@ def get_queryset(self):
            calling ``search.execute().hits``. This is why an DSL search object
            is compatible with DRF's paginator.
         """
-        projects = {
-            project: version.slug
-            for project, version in self._get_all_projects_data().items()
-        }
-
-        # Check to avoid searching all projects in case it's empty.
-        if not projects:
-            log.info('Unable to find a version to search')
-            return []
+        main_project = self._get_project()
+        main_version = self._get_version()
+        projects = {}
+        filters = {}
+
+        if main_project.has_feature(Feature.SEARCH_SUBPROJECTS_ON_DEFAULT_VERSION):
+            projects = {
+                project: version.slug
+                for project, version in self._get_all_projects_data().items()
+            }
+            # Check to avoid searching all projects in case it's empty.
+            if not projects:
+                log.info('Unable to find a version to search')
+                return []
+        else:
+            filters['project'] = list(self._get_all_projects_data().keys())
+            filters['version'] = main_version.slug
+            # Check to avoid searching all projects in case these filters are empty.
+            if not filters['project']:
+                log.info('Unable to find a project to search')
+                return []
+            if not filters['version']:
+                log.info('Unable to find a version to search')
+                return []
 
         query = self.request.query_params['q']
         queryset = PageSearch(
             query=query,
             projects=projects,
+            filters=filters,
             user=self.request.user,
             # We use a permission class to control authorization
             filter_by_user=False,
-            use_advanced_query=not self._get_project().has_feature(Feature.DEFAULT_TO_FUZZY_SEARCH),
+            use_advanced_query=not main_project.has_feature(Feature.DEFAULT_TO_FUZZY_SEARCH),
         )
         return queryset
 

readthedocs/search/tests/test_api.py

@@ -264,11 +264,62 @@ def test_doc_search_subprojects(self, api_client, all_projects):
         # First result should be the subproject
         first_result = data[0]
         assert first_result['project'] == subproject.slug
+        # The result is from the same version as the main project.
+        assert first_result['version'] == version.slug
         # Check the link is the subproject document link
         document_link = subproject.get_docs_url(version_slug=version.slug)
         link = first_result['domain'] + first_result['path']
         assert document_link in link
 
+    def test_doc_search_subprojects_default_version(self, api_client, all_projects):
+        """Return results from subprojects that match the version from the main project or fallback to its default version."""
+        project = all_projects[0]
+        version = project.versions.all()[0]
+        feature, _ = Feature.objects.get_or_create(
+            feature_id=Feature.SEARCH_SUBPROJECTS_ON_DEFAULT_VERSION,
+        )
+        project.feature_set.add(feature)
+
+        subproject = all_projects[1]
+        subproject_version = subproject.versions.all()[0]
+
+        # Change the name of the version, and make it default.
+        subproject_version.slug = 'different'
+        subproject_version.save()
+        subproject.default_version = subproject_version.slug
+        subproject.save()
+        subproject.versions.filter(slug=version.slug).delete()
+
+        # Refresh index
+        version_files = HTMLFile.objects.all().filter(version=subproject_version)
+        for f in version_files:
+            PageDocument().update(f)
+
+        # Add another project as subproject of the project
+        project.add_subproject(subproject)
+
+        # Now search with subproject content but explicitly filter by the parent project
+        query = get_search_query_from_project_file(project_slug=subproject.slug)
+        search_params = {
+            'q': query,
+            'project': project.slug,
+            'version': version.slug
+        }
+        resp = self.get_search(api_client, search_params)
+        assert resp.status_code == 200
+
+        data = resp.data['results']
+        assert len(data) >= 1  # there may be results from another projects
+
+        # First result should be the subproject
+        first_result = data[0]
+        assert first_result['project'] == subproject.slug
+        assert first_result['version'] == 'different'
+        # Check the link is the subproject document link
+        document_link = subproject.get_docs_url(version_slug=subproject_version.slug)
+        link = first_result['domain'] + first_result['path']
+        assert document_link in link
+
     def test_doc_search_unexisting_project(self, api_client):
         project = 'notfound'
         assert not Project.objects.filter(slug=project).exists()