Search: stop relying on the DB when indexing

docs/dev/server-side-search.rst

@@ -62,7 +62,7 @@ You can fix this by deleting the page index and :ref:`re-indexing <server-side-s
 .. prompt:: bash
 
    inv docker.manage 'search_index --delete'
-   inv docker.manage reindex_elasticsearch
+   inv docker.manage 'reindex_elasticsearch --queue web'
 
 How we index documentations
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~

readthedocs/builds/admin.py

@@ -12,8 +12,7 @@
 )
 from readthedocs.core.utils import trigger_build
 from readthedocs.core.utils.admin import pretty_json_field
-from readthedocs.projects.models import HTMLFile
-from readthedocs.search.utils import _indexing_helper
+from readthedocs.projects.tasks.search import reindex_version
 
 
 class BuildCommandResultInline(admin.TabularInline):
@@ -89,7 +88,7 @@ class VersionAdmin(admin.ModelAdmin):
     list_filter = ("type", "privacy_level", "active", "built")
     search_fields = ("slug", "project__slug")
     raw_id_fields = ("project",)
-    actions = ["build_version", "reindex_version", "wipe_version_indexes"]
+    actions = ["build_version", "reindex_version"]
 
     def project_slug(self, obj):
         return obj.project.slug
@@ -117,41 +116,11 @@ def build_version(self, request, queryset):
     @admin.action(description="Reindex version to ES")
     def reindex_version(self, request, queryset):
         """Reindexes all selected versions to ES."""
-        html_objs_qs = []
-        for version in queryset.iterator():
-            html_objs = HTMLFile.objects.filter(
-                project=version.project, version=version
-            )
-
-            if html_objs.exists():
-                html_objs_qs.append(html_objs)
-
-        if html_objs_qs:
-            _indexing_helper(html_objs_qs, wipe=False)
+        for version_id in queryset.values_list("id", flat=True).iterator():
+            reindex_version.delay(version_id)
 
         self.message_user(request, "Task initiated successfully.", messages.SUCCESS)
 
-    @admin.action(description="Wipe version from ES")
-    def wipe_version_indexes(self, request, queryset):
-        """Wipe selected versions from ES."""
-        html_objs_qs = []
-        for version in queryset.iterator():
-            html_objs = HTMLFile.objects.filter(
-                project=version.project, version=version
-            )
-
-            if html_objs.exists():
-                html_objs_qs.append(html_objs)
-
-        if html_objs_qs:
-            _indexing_helper(html_objs_qs, wipe=True)
-
-        self.message_user(
-            request,
-            "Task initiated successfully",
-            messages.SUCCESS,
-        )
-
 
 @admin.register(RegexAutomationRule)
 class RegexAutomationRuleAdmin(PolymorphicChildModelAdmin, admin.ModelAdmin):

readthedocs/builds/models.py

@@ -320,7 +320,7 @@ def config(self):
         :rtype: dict
         """
         last_build = (
-            self.builds(manager=INTERNAL).filter(
+            self.builds.filter(
                 state=BUILD_STATE_FINISHED,
                 success=True,
             ).order_by('-date')

readthedocs/builds/querysets.py

@@ -118,6 +118,28 @@ def public(
     def api(self, user=None):
         return self.public(user, only_active=False)
 
+    def for_reindex(self):
+        """
+        Get all versions that can be reindexed.
+
+        A version can be reindexed if:
+
+        - It's active and has been built at least once successfully.
+          Since that means that it has files to be indexed.
+        - Its project is not delisted or marked as spam.
+        """
+        return (
+            self.filter(
+                active=True,
+                built=True,
+                builds__state=BUILD_STATE_FINISHED,
+                builds__success=True,
+            )
+            .exclude(project__delisted=True)
+            .exclude(project__is_spam=True)
+            .distinct()
+        )
+
 
 class VersionQuerySet(SettingsOverrideObject):
     _default_class = VersionQuerySetBase

readthedocs/projects/admin.py

@@ -11,8 +11,8 @@
 from readthedocs.core.history import ExtraSimpleHistoryAdmin, set_change_reason
 from readthedocs.core.utils import trigger_build
 from readthedocs.notifications.views import SendNotificationView
+from readthedocs.projects.tasks.search import reindex_version
 from readthedocs.redirects.models import Redirect
-from readthedocs.search.utils import _indexing_helper
 
 from .forms import FeatureForm
 from .models import (
@@ -264,7 +264,6 @@ class ProjectAdmin(ExtraSimpleHistoryAdmin):
         "run_spam_rule_checks",
         "build_default_version",
         "reindex_active_versions",
-        "wipe_all_versions",
         "import_tags_from_vcs",
     ]
 
@@ -362,66 +361,26 @@ def reindex_active_versions(self, request, queryset):
         """Reindex all active versions of the selected projects to ES."""
         qs_iterator = queryset.iterator()
         for project in qs_iterator:
-            version_qs = Version.internal.filter(project=project)
-            active_versions = version_qs.filter(active=True)
+            versions_id_to_reindex = project.versions.for_reindex().values_list(
+                "pk", flat=True
+            )
 
-            if not active_versions.exists():
+            if not versions_id_to_reindex.exists():
                 self.message_user(
                     request,
-                    "No active versions of project {}".format(project),
+                    "No versions to be re-indexed for project {}".format(project),
                     messages.ERROR,
                 )
             else:
-                html_objs_qs = []
-                for version in active_versions.iterator():
-                    html_objs = HTMLFile.objects.filter(
-                        project=project, version=version
-                    )
-
-                    if html_objs.exists():
-                        html_objs_qs.append(html_objs)
-
-                if html_objs_qs:
-                    _indexing_helper(html_objs_qs, wipe=False)
+                for version_id in versions_id_to_reindex.iterator():
+                    reindex_version.delay(version_id)
 
                 self.message_user(
                     request,
                     "Task initiated successfully for {}".format(project),
                     messages.SUCCESS,
                 )
 
-    # TODO: rename method to mention "indexes" on its name
-    @admin.action(description="Wipe all versions from ES")
-    def wipe_all_versions(self, request, queryset):
-        """Wipe indexes of all versions of selected projects."""
-        qs_iterator = queryset.iterator()
-        for project in qs_iterator:
-            version_qs = Version.internal.filter(project=project)
-            if not version_qs.exists():
-                self.message_user(
-                    request,
-                    "No active versions of project {}.".format(project),
-                    messages.ERROR,
-                )
-            else:
-                html_objs_qs = []
-                for version in version_qs.iterator():
-                    html_objs = HTMLFile.objects.filter(
-                        project=project, version=version
-                    )
-
-                    if html_objs.exists():
-                        html_objs_qs.append(html_objs)
-
-                if html_objs_qs:
-                    _indexing_helper(html_objs_qs, wipe=True)
-
-                self.message_user(
-                    request,
-                    "Task initiated successfully for {}.".format(project),
-                    messages.SUCCESS,
-                )
-
     @admin.action(description="Import tags from the version control API")
     def import_tags_from_vcs(self, request, queryset):
         for project in queryset.iterator():

readthedocs/projects/tasks/builds.py

@@ -67,7 +67,7 @@
 from ..models import APIProject, WebHookEvent
 from ..signals import before_vcs
 from .mixins import SyncRepositoryMixin
-from .search import fileify
+from .search import index_build
 from .utils import (
     BuildRequest,
     clean_build,
@@ -653,13 +653,7 @@ def on_success(self, retval, task_id, args, kwargs):
                 )
 
         # Index search data
-        fileify.delay(
-            version_pk=self.data.version.pk,
-            commit=self.data.build['commit'],
-            build=self.data.build['id'],
-            search_ranking=self.data.config.search.ranking,
-            search_ignore=self.data.config.search.ignore,
-        )
+        index_build.delay(build_id=self.data.build["id"])
 
         if not self.data.project.has_valid_clone:
             self.set_valid_clone()