Merge pull request #7439 from readthedocs/revert-7429-revert-7408-update-es-dep

Revert "Revert ES: update dependencies"

Author: stsewd

readthedocs/search/documents.py

@@ -1,7 +1,7 @@
 import logging
 
 from django.conf import settings
-from django_elasticsearch_dsl import DocType, Index, fields
+from django_elasticsearch_dsl import Document, Index, fields
 from elasticsearch import Elasticsearch
 
 from readthedocs.projects.models import HTMLFile, Project
@@ -28,8 +28,8 @@ def update(self, *args, **kwargs):
         super().update(*args, **kwargs)
 
 
-@project_index.doc_type
-class ProjectDocument(RTDDocTypeMixin, DocType):
+@project_index.document
+class ProjectDocument(RTDDocTypeMixin, Document):
 
     # Metadata
     url = fields.TextField(attr='get_absolute_url')
@@ -43,14 +43,14 @@ class ProjectDocument(RTDDocTypeMixin, DocType):
 
     modified_model_field = 'modified_date'
 
-    class Meta:
+    class Django:
         model = Project
         fields = ('name', 'slug', 'description')
         ignore_signals = True
 
 
-@page_index.doc_type
-class PageDocument(RTDDocTypeMixin, DocType):
+@page_index.document
+class PageDocument(RTDDocTypeMixin, Document):
 
     # Metadata
     project = fields.KeywordField(attr='project.slug')
@@ -88,7 +88,7 @@ class PageDocument(RTDDocTypeMixin, DocType):
 
     modified_model_field = 'modified_date'
 
-    class Meta:
+    class Django:
         model = HTMLFile
         fields = ('commit', 'build')
         ignore_signals = True

readthedocs/search/faceted_search.py

@@ -148,7 +148,7 @@ def query(self, search, query):
 class ProjectSearchBase(RTDFacetedSearch):
     facets = {'language': TermsFacet(field='language')}
     doc_types = [ProjectDocument]
-    index = ProjectDocument._doc_type.index
+    index = ProjectDocument._index._name
     fields = ('name^10', 'slug^5', 'description')
     operators = ['and', 'or']
 
@@ -163,7 +163,7 @@ class PageSearchBase(RTDFacetedSearch):
         ),
     }
     doc_types = [PageDocument]
-    index = PageDocument._doc_type.index
+    index = PageDocument._index._name
 
     # boosting for these fields need to be close enough
     # to be re-boosted by the page rank.

readthedocs/search/management/commands/reindex_elasticsearch.py

@@ -1,15 +1,19 @@
 import datetime
 import logging
 
-from celery import chord, chain
+from celery import chain, chord
 from django.apps import apps
 from django.conf import settings
 from django.core.management import BaseCommand
 from django.utils import timezone
 from django_elasticsearch_dsl.registries import registry
 
-from ...tasks import (index_objects_to_es, switch_es_index, create_new_es_index,
-                      index_missing_objects)
+from ...tasks import (
+    create_new_es_index,
+    index_missing_objects,
+    index_objects_to_es,
+    switch_es_index,
+)
 
 log = logging.getLogger(__name__)
 
@@ -64,11 +68,11 @@ def _run_reindex_tasks(self, models, queue):
             app_label = queryset.model._meta.app_label
             model_name = queryset.model.__name__
 
-            index_name = doc._doc_type.index
+            index_name = doc._index._name
             new_index_name = "{}_{}".format(index_name, timestamp)
             # Set index temporarily for indexing,
             # this will only get set during the running of this command
-            doc._doc_type.index = new_index_name
+            doc._index._name = new_index_name
 
             pre_index_task = create_new_es_index.si(app_label=app_label,
                                                     model_name=model_name,

readthedocs/search/serializers.py

@@ -130,7 +130,7 @@ def get_blocks(self, obj):
 
         sorted_results = sorted(
             itertools.chain(sections, domains),
-            key=attrgetter('_score'),
+            key=attrgetter('meta.score'),
             reverse=True,
         )
         sorted_results = [
@@ -157,11 +157,11 @@ def get_content(self, obj):
 class DomainSearchSerializer(serializers.Serializer):
 
     type = serializers.CharField(default='domain', source=None, read_only=True)
-    role = serializers.CharField(source='_source.role_name')
-    name = serializers.CharField(source='_source.name')
-    id = serializers.CharField(source='_source.anchor')
-    content = serializers.CharField(source='_source.docstrings')
-    highlights = DomainHighlightSerializer(source='highlight', default=dict)
+    role = serializers.CharField(source='role_name')
+    name = serializers.CharField()
+    id = serializers.CharField(source='anchor')
+    content = serializers.CharField(source='docstrings')
+    highlights = DomainHighlightSerializer(source='meta.highlight', default=dict)
 
 
 class SectionHighlightSerializer(serializers.Serializer):
@@ -181,7 +181,7 @@ def get_content(self, obj):
 class SectionSearchSerializer(serializers.Serializer):
 
     type = serializers.CharField(default='section', source=None, read_only=True)
-    id = serializers.CharField(source='_source.id')
-    title = serializers.CharField(source='_source.title')
-    content = serializers.CharField(source='_source.content')
-    highlights = SectionHighlightSerializer(source='highlight', default=dict)
+    id = serializers.CharField()
+    title = serializers.CharField()
+    content = serializers.CharField()
+    highlights = SectionHighlightSerializer(source='meta.highlight', default=dict)

readthedocs/search/tasks.py

@@ -9,7 +9,8 @@
 from readthedocs.projects.models import Project
 from readthedocs.search.models import SearchQuery
 from readthedocs.worker import app
-from .utils import _get_index, _get_document
+
+from .utils import _get_document, _get_index
 
 log = logging.getLogger(__name__)
 
@@ -44,17 +45,19 @@ def index_objects_to_es(
 
     if index_name:
         # Hack the index name temporarily for reindexing tasks
-        old_index_name = document._doc_type.index
-        document._doc_type.index = index_name
+        old_index_name = document._index._name
+        document._index._name = index_name
         log.info('Replacing index name %s with %s', old_index_name, index_name)
 
     log.info("Indexing model: %s, '%s' objects", model.__name__, queryset.count())
     doc_obj.update(queryset.iterator())
 
     if index_name:
-        log.info('Undoing index replacement, settings %s with %s',
-                 document._doc_type.index, old_index_name)
-        document._doc_type.index = old_index_name
+        log.info(
+            'Undoing index replacement, settings %s with %s',
+            document._index._name, old_index_name,
+        )
+        document._index._name = old_index_name
 
 
 @app.task(queue='web')

readthedocs/search/utils.py

@@ -89,7 +89,7 @@ def _get_index(indices, index_name):
     :return: DED Index
     """
     for index in indices:
-        if str(index) == index_name:
+        if index._name == index_name:
             return index
 
 
@@ -116,7 +116,10 @@ def _indexing_helper(html_objs_qs, wipe=False):
     else, html_objs are indexed.
     """
     from readthedocs.search.documents import PageDocument
-    from readthedocs.search.tasks import index_objects_to_es, delete_objects_in_es
+    from readthedocs.search.tasks import (
+        delete_objects_in_es,
+        index_objects_to_es,
+    )
 
     if html_objs_qs:
         obj_ids = []
@@ -148,7 +151,7 @@ def _get_sorted_results(results, source_key='_source'):
             source_key: hit._source.to_dict(),
             'highlight': hit.highlight.to_dict() if hasattr(hit, 'highlight') else {}
         }
-        for hit in sorted(results, key=attrgetter('_score'), reverse=True)
+        for hit in sorted(results, key=attrgetter('meta.score'), reverse=True)
     ]
 
     return sorted_results

readthedocs/settings/base.py

@@ -581,16 +581,17 @@ def DOCKER_LIMITS(self):
     ES_INDEXES = {
         'project': {
             'name': 'project_index',
-            'settings': {'number_of_shards': 1,
-                         'number_of_replicas': 1
-                         }
+            'settings': {
+                'number_of_shards': 1,
+                'number_of_replicas': 1
+            },
         },
         'page': {
             'name': 'page_index',
             'settings': {
                 'number_of_shards': 1,
                 'number_of_replicas': 1,
-            }
+            },
         },
     }
 

requirements/pip.txt

@@ -47,23 +47,8 @@ GitPython==3.1.8
 
 # Search
 elasticsearch==6.8.1  # pyup: <7.0.0
-
-
-# elasticsearch-dsl==6.3.1 produces this error
-#   File "/home/travis/build/rtfd/readthedocs.org/.tox/py36/lib/python3.6/site-packages/django_elasticsearch_dsl/documents.py", line 8, in <module>
-#     from elasticsearch_dsl.document import DocTypeMeta as DSLDocTypeMeta
-# ImportError: cannot import name 'DocTypeMeta'
-#
-# Commit 97e3f75 adds the NestedFacet
-git+https://github.com/elastic/elasticsearch-dsl-py@97e3f756a8cacd1c863d3ced3d17abcafbb0f85e#egg=elasticsearch-dsl==6.1.1
-
-# django-elasticsearch-dsl==6.4.1 produces this error
-#   File "/home/travis/build/readthedocs/readthedocs.org/.tox/py36/lib/python3.6/site-packages/django_elasticsearch_dsl/__init__.py", line 3, in <module>
-#     from .documents import DocType  # noqa
-#   File "/home/travis/build/readthedocs/readthedocs.org/.tox/py36/lib/python3.6/site-packages/django_elasticsearch_dsl/documents.py", line 7, in <module>
-#     from elasticsearch_dsl import Document as DSLDocument
-# ImportError: cannot import name 'Document'
-django-elasticsearch-dsl==0.5.1  # pyup: ignore
+elasticsearch-dsl==6.4.0  # pyup: <7.0
+django-elasticsearch-dsl==6.4.2  # pyup: <7.0
 selectolax==0.2.7
 
 # NOTE: this dep can be removed in python 3.7 in favor of ``date.fromisoformat``