feat(client): auto detect input_type

jina-ai · May 27, 2020 · fda5790 · fda5790
1 parent 6702c61
commit fda5790
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 12 deletions.
diff --git a/jina/clients/python/request.py b/jina/clients/python/request.py
@@ -2,18 +2,19 @@
 __license__ = "Apache-2.0"
 
 import ctypes
+import os
 import random
+import urllib.parse
 from typing import Iterator, Union
 
-from ...enums import ClientInputType, ClientMode
+from ...enums import ClientMode
 from ...helper import batch_iterator
 from ...proto import jina_pb2
 
 
 def _generate(data: Union[Iterator[bytes], Iterator['jina_pb2.Document'], Iterator[str]], batch_size: int = 0,
               first_doc_id: int = 0, first_request_id: int = 0,
               random_doc_id: bool = False, mode: ClientMode = ClientMode.INDEX, top_k: int = 50,
-              input_type: ClientInputType = ClientInputType.BUFFER,
               mime_type: str = None,
               *args, **kwargs) -> Iterator['jina_pb2.Message']:
     if isinstance(mode, str):
@@ -31,18 +32,18 @@ def _generate(data: Union[Iterator[bytes], Iterator['jina_pb2.Document'], Iterat
 
         for _raw in pi:
             d = getattr(req, str(mode).lower()).docs.add()
-            if input_type == ClientInputType.PROTOBUF:
+            if isinstance(_raw, jina_pb2.Document):
                 d.CopyFrom(_raw)
-            elif input_type == ClientInputType.DATA_URI:
-                d.data_uri = _raw
-            elif input_type == ClientInputType.FILE_PATH:
-                d.file_path = _raw
-            elif input_type == ClientInputType.BUFFER:
-                if isinstance(_raw, str):
-                    _raw = _raw.encode()  # auto-fix for str
+            elif isinstance(_raw, bytes):
                 d.buffer = _raw
                 if mime_type:
                     d.mime_type = mime_type
+            elif isinstance(_raw, str):
+                scheme = urllib.parse.urlparse(_raw).scheme
+                if scheme in {'http', 'https'} or os.path.exists(_raw) or os.access(os.path.dirname(_raw), os.W_OK):
+                    d.file_path = _raw
+                elif scheme == 'data':
+                    d.data_uri = _raw
             d.doc_id = first_doc_id if not random_doc_id else random.randint(0, ctypes.c_uint(-1).value)
             d.weight = 1.0
             first_doc_id += 1

diff --git a/jina/executors/crafters/convert.py b/jina/executors/crafters/convert.py
@@ -38,9 +38,9 @@ def craft(self, file_path: str, data_uri: str, buffer: bytes, *args, **kwargs):
         if buffer:
             pass
         elif file_path:
-            return super(FilePath2Buffer, self).craft(file_path)
+            return FilePath2Buffer.craft(self, file_path)
         elif data_uri:
-            return super(DataURI2Buffer, self).craft(data_uri)
+            return DataURI2Buffer.craft(self, data_uri)
         else:
             raise ValueError('this document has no "file_path", no "data_uri" and no "buffer" set')