use numpy 2

Author: rerpha

pyproject.toml

@@ -41,7 +41,7 @@ classifiers = [
 
 dependencies = [
   "flatbuffers",
-  "numpy<2",  # Pinned to <2 due to f142 np.unicode
+  "numpy>2"
 ]
 
 [project.optional-dependencies]

src/streaming_data_types/logdata_f142.py

@@ -490,8 +490,8 @@ def _serialise_value(
 ):
     # We can use a dictionary to map most numpy types to one of the types defined in the flatbuffer schema
     # but we have to handle strings separately as there are many subtypes
-    if np.issubdtype(value.dtype, np.unicode_) or np.issubdtype(
-        value.dtype, np.string_
+    if np.issubdtype(value.dtype, np.str_) or np.issubdtype(
+        value.dtype, np.bytes_
     ):
         string_serialiser(builder, value, source)
     else:
@@ -501,7 +501,7 @@ def _serialise_value(
             # There are a few numpy types we don't try to handle, for example complex numbers
             raise NotImplementedError(
                 f"Cannot serialise data of type {value.dtype}, must use one of "
-                f"{list(_map_scalar_type_to_serialiser.keys()) + [np.unicode_]}"
+                f"{list(_map_scalar_type_to_serialiser.keys()) + [np.str_]}"
             )
 
 
@@ -539,8 +539,8 @@ def _serialise_value(
 
 def _decode_if_scalar_string(value: np.ndarray) -> Union[str, np.ndarray]:
     if value.ndim == 0 and (
-        np.issubdtype(value.dtype, np.unicode_)
-        or np.issubdtype(value.dtype, np.string_)
+        np.issubdtype(value.dtype, np.str_)
+        or np.issubdtype(value.dtype, np.bytes_)
     ):
         return value.item().decode()
     return value