Multi cherry picks from master branch (#1339)

- Fix possible KeyError if description is not set (#1328)
- Fix error message when insert mismatch datatypes (#1333)

See also: #1324

Signed-off-by: yangxuan <[email protected]>
Co-authored-by: wayblink <[email protected]>

Author: XuanYang-cn

pymilvus/orm/schema.py

@@ -166,10 +166,10 @@ def to_dict(self):
 
 
 class FieldSchema:
-    def __init__(self, name, dtype, description="", **kwargs):
+    def __init__(self, name: str, dtype: DataType, description="", **kwargs):
         self.name = name
         try:
-            DataType(dtype)
+            dtype = DataType(dtype)
         except ValueError:
             raise DataTypeNotSupportException(message=ExceptionsMessage.FieldDtype) from None
         if dtype == DataType.UNKNOWN:
@@ -224,7 +224,7 @@ def construct_from_dict(cls, raw):
         kwargs['is_primary'] = raw.get("is_primary", False)
         if raw.get("auto_id", None) is not None:
             kwargs['auto_id'] = raw.get("auto_id", None)
-        return FieldSchema(raw['name'], raw['type'], raw['description'], **kwargs)
+        return FieldSchema(raw['name'], raw['type'], raw.get("description", ""), **kwargs)
 
     def to_dict(self):
         _dict = {
@@ -285,7 +285,7 @@ def params(self):
         return self._type_params
 
     @property
-    def dtype(self):
+    def dtype(self) -> DataType:
         return self._dtype
 
 

tests/test_schema.py

@@ -1,8 +1,9 @@
 import numpy
 import pytest
 
-from pymilvus import CollectionSchema, FieldSchema, DataType
+from pymilvus import CollectionSchema, FieldSchema, DataType, MilvusException
 from utils import *
+from pymilvus.orm import schema as s
 
 
 class TestCollectionSchema:
@@ -143,3 +144,23 @@ def test_to_dict(self, raw_dict_norm, raw_dict_float_vector, raw_dict_binary_vec
     #      for f in fields:
     #          if f.dtype == DataType.FLOAT_VECTOR:
     #              assert f.dim == len(dataframe1['float_vec'].values[0])
+
+class TestCheckInsertDataSchema:
+    def test_check_insert_data_schema_issue1324(self):
+        schema = CollectionSchema([
+            FieldSchema(name="id", dtype=DataType.INT64, descrition="int64", is_primary=True, auto_id=True),
+            FieldSchema(name="embedding", dtype=DataType.FLOAT_VECTOR, descrition="float vector", dim=2),
+            FieldSchema(name="work_id2", dtype=5, descrition="work id"),
+            FieldSchema(name='path', dtype=DataType.VARCHAR, description='path to image', max_length=200),
+            FieldSchema(name="uid", dtype=DataType.INT64, descrition="user id"),
+        ])
+
+        data = [
+            [[0.003984056, 0.05035976]],
+            ['15755403'],
+            ['https://xxx.com/app/works/105149/2023-01-11/w_63be653c4643b/963be653c8aa8c.jpg'],
+            ['105149'],
+        ]
+
+        with pytest.raises(MilvusException):
+            s.check_insert_data_schema(schema, data)