fix: fix broken protoidx for #234

jina-ai · Apr 14, 2020 · a357e25 · a357e25
1 parent 6266c4a
commit a357e25
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 7 deletions.
diff --git a/jina/drivers/index.py b/jina/drivers/index.py
@@ -36,7 +36,7 @@ class DocPbIndexDriver(BaseIndexDriver):
 
     def __call__(self, *args, **kwargs):
         from google.protobuf.json_format import MessageToJson
-        content = {'d%d' % d.doc_id: MessageToJson(d) for d in self.req.docs}
+        content = {f'd{d.doc_id}': MessageToJson(d) for d in self.req.docs}
         if content:
             self.exec_fn(content)
 
@@ -48,7 +48,6 @@ class ChunkPbIndexDriver(BaseIndexDriver):
 
     def __call__(self, *args, **kwargs):
         from google.protobuf.json_format import MessageToJson
-        content = {'c%d' % c.chunk_id: MessageToJson(c) for d in self.req.docs for c in d.chunks}
-        print('write {}'.format(','.join([k for k in content.keys()])))
+        content = {f'c{c.chunk_id}': MessageToJson(c) for d in self.req.docs for c in d.chunks}
         if content:
             self.exec_fn(content)
diff --git a/jina/drivers/search.py b/jina/drivers/search.py
@@ -12,24 +12,30 @@ def __init__(self, executor: str = None, method: str = 'query', *args, **kwargs)
 class DocPbSearchDriver(BaseSearchDriver):
     """Fill in the doc-level top-k results using the :class:`jina.executors.indexers.meta.BasePbIndexer`
 
+    .. warning::
+        This driver loops over all doc's top-K results, each step fires a query.
+        This may not be very efficient, as the total number of quires is D x K
     """
 
     def __call__(self, *args, **kwargs):
         for d in self.req.docs:
             for tk in d.topk_results:
-                tk.match_doc.CopyFrom(self.exec_fn('d%d' % tk.match_doc.doc_id))
+                tk.match_doc.CopyFrom(self.exec_fn(f'd{tk.match_doc.doc_id}'))
 
 
 class ChunkPbSearchDriver(BaseSearchDriver):
     """Fill in the chunk-level top-k results using the :class:`jina.executors.indexers.meta.BasePbIndexer`
 
+    .. warning::
+        This driver loops over all chunk's top-K results, each step fires a query.
+        This may not be very efficient, as the total number of quires is D x C x K
     """
 
     def __call__(self, *args, **kwargs):
         for d in self.req.docs:
             for c in d.chunks:
                 for k in c.topk_results:
-                    k.match_chunk.CopyFrom(self.exec_fn('c%d' % k.match_chunk.chunk_id))
+                    k.match_chunk.CopyFrom(self.exec_fn(f'c{k.match_chunk.chunk_id}'))
 
 
 class ChunkSearchDriver(BaseSearchDriver):

diff --git a/jina/executors/indexers/keyvalue/proto.py b/jina/executors/indexers/keyvalue/proto.py
@@ -3,7 +3,6 @@
 from typing import Union
 
 from google.protobuf.json_format import Parse
-
 from jina.executors.indexers import BaseKVIndexer
 from jina.proto import jina_pb2
 
@@ -21,7 +20,7 @@ def get_query_handler(self):
                     tmp = json.loads(l)
                     for k, v in tmp.items():
                         _parser = jina_pb2.Chunk if k[0] == 'c' else jina_pb2.Document
-                        r[int(k[1:])] = Parse(v, _parser())
+                        r[k] = Parse(v, _parser())
         return r
 
     def get_add_handler(self):