0.0.195

Author: joocer

orso/cityhash/__init__.py

@@ -1,12 +1,14 @@
-
 def CityHash32(data: bytes) -> int:
-    from opteryx.third_party.cyan4973.xxhash import hash_bytes
+    from xxhash import xxh32
+
     if not isinstance(data, bytes):
         data = str(data).encode()
-    return hash_bytes(data) & 0xFFFFFFFF
+    return xxh32(data)
+
 
 def CityHash64(data: bytes) -> int:
-    from opteryx.third_party.cyan4973.xxhash import hash_bytes
+    from xxhash import xxh64
+
     if not isinstance(data, bytes):
         data = str(data).encode()
-    return hash_bytes(data) & 0xFFFFFFFFFFFFFFFF
+    return xxh64(data)

orso/dataframe.py

@@ -421,11 +421,11 @@ def schema(self) -> RelationSchema:
         return self._schema
 
     def __hash__(self):
-        from opteryx.third_party.cyan4973.xxhash import hash_bytes
+        from xxhash import xxh64
 
         _hash = 0
         for i, row in enumerate(self._rows):
-            row_hash = hash_bytes(str(row).encode())
+            row_hash = xxh64(str(row).encode())
             _hash = i ^ _hash ^ row_hash
         return _hash
 

requirements.txt

@@ -1,7 +1,7 @@
 data-expectations>=1.6.0
 orjson
 ormsgpack
-opteryx>=0.20.0
 pandas
 pyarrow
-numpy
+numpy
+xxhash