create utils.es.ESIndex, successor to ESIndexer

biothings/hub/dataindex/indexer_task.py

@@ -1,11 +1,11 @@
 from collections import namedtuple
-import functools
 import logging
 from enum import Enum
 from functools import partial
 from types import SimpleNamespace
 
 from biothings.utils.loggers import get_logger
+from biothings.utils.es import ESIndex as BaseESIndex
 from elasticsearch import Elasticsearch, helpers
 from pymongo import MongoClient
 
@@ -20,27 +20,12 @@
 
 _IDExists = namedtuple("IDExists", ("id", "exists"))
 
-class ESIndex():
-    """ An Elasticsearch Index Wrapping A Client.
-    Counterpart for pymongo.collection.Collection """
-
-    # previously using biothings.utils.es.ESIndexer
-    # recently reimplemented here for better clarity.
+class ESIndex(BaseESIndex):
 
     def __init__(self, client, index_name, **bulk_index_args):
-        self.client = client
-        self.index_name = index_name  # MUST exist
+        super().__init__(client, index_name)
         self.bulk_index_args = bulk_index_args
 
-    @property
-    @functools.lru_cache()
-    def doc_type(self):
-        if int(self.client.info()['version']['number'].split('.')[0]) < 7:
-            mappings = self.client.indices.get_mapping(self.index_name)
-            mappings = mappings[self.index_name]["mappings"]
-            return next(iter(mappings.keys()))
-        return None
-
     # --------------------
     # bulk operations (m*)
     # --------------------
@@ -100,6 +85,14 @@ def _action(doc):
             self.client, map(_action, docs),
             **self.bulk_index_args)[0]
 
+    # NOTE
+    # Why doesn't "mget", "mexists", "mindex" belong to the base class?
+    # At this moment, their interfaces are too opinionated/customized
+    # for usage in this module. Unless we find them directly useful in
+    # another module in the future, proving their genericity, they should
+    # stay close in this module only.
+
+
 # Data Collection Client
 
 def _get_es_client(es_client_args, es_blk_args, es_idx_name):

biothings/utils/es.py

@@ -4,6 +4,7 @@
 import json
 import re
 import time
+import functools
 
 from elasticsearch import (Elasticsearch, NotFoundError, RequestError,
                            TransportError, ElasticsearchException)
@@ -73,6 +74,36 @@ def outter_fn(*args, **kwargs):
 class IndexerException(Exception):
     pass
 
+class ESIndex():
+    """ An Elasticsearch Index Wrapping A Client.
+    Counterpart for pymongo.collection.Collection """
+
+    # a new approach to biothings.utils.es.ESIndexer
+    # but not intended to be a replacement in features.
+
+    def __init__(self, client, index_name):
+        self.client = client  # synchronous
+        self.index_name = index_name  # MUST exist
+
+    @property
+    @functools.lru_cache()
+    def doc_type(self):
+        if int(self.client.info()['version']['number'].split('.')[0]) < 7:
+            mappings = self.client.indices.get_mapping(self.index_name)
+            mappings = mappings[self.index_name]["mappings"]
+            return next(iter(mappings.keys()))
+        return None
+
+    # SUBCLASS NOTE &&
+    # BEFORE YOU ADD A METHOD UNDER THIS CLASS:
+
+    # An object of this class refers to an existing ES index. All operations
+    # should target this index. Do not put uncommon methods here. They belong
+    # to the subclasses. This class is to provide a common framework to support
+    # Index-specific ES operations, an concept does not exist in the low-level
+    # ES library, thus only providing low-level common operations, like doc_type
+    # parsing across ES versions for biothings usecases. (single type per index)
+
 class ESIndexer():
     # RETIRING THIS CLASS