Bump to v0.1.3 with leb128

Author: juntyr

pyproject.toml

@@ -4,15 +4,15 @@ build-backend = "hatchling.build"
 
 [project]
 name = "numcodecs-tokenize"
-version = "0.1.2"
+version = "0.1.3"
 description = "Tokenization codec for the `numcodecs` buffer compression API"
 readme = "README.md"
 license = "MPL-2.0"
 requires-python = ">=3.10"
 dependencies = [
+    "leb128~=1.0.8",
     "numcodecs>=0.13.0,<0.17",
     "numpy~=2.0",
-    "varint~=1.0.2",
 ]
 
 [dependency-groups]
@@ -35,5 +35,5 @@ addopts = ["--import-mode=importlib"]
 xfail_strict = true
 
 [[tool.mypy.overrides]]
-module = ["numcodecs.*", "varint.*"]
+module = ["leb128.*", "numcodecs.*"]
 follow_untyped_imports = true

src/numcodecs_tokenize/__init__.py

@@ -6,10 +6,10 @@
 
 from io import BytesIO
 
+import leb128
 import numcodecs.compat
 import numcodecs.registry
 import numpy as np
-import varint
 from numcodecs.abc import Codec
 
 from .typing import S, T, U
@@ -60,14 +60,14 @@ def encode(
         # message: dtype shape [padding] table indices
         message = []
 
-        message.append(varint.encode(len(dtype.str)))
+        message.append(leb128.u.encode(len(dtype.str)))
         message.append(dtype.str.encode("ascii"))
 
-        message.append(varint.encode(len(shape)))
+        message.append(leb128.u.encode(len(shape)))
         for s in shape:
-            message.append(varint.encode(s))
+            message.append(leb128.u.encode(s))
 
-        message.append(varint.encode(unique.size))
+        message.append(leb128.u.encode(unique.size))
 
         # select the smallest output data type that can encode the indices
         utype: np.dtype[np.unsignedinteger]
@@ -132,13 +132,13 @@ def decode(
         b = numcodecs.compat.ensure_bytes(buf)
         b_io = BytesIO(b)
 
-        dtype = np.dtype(b_io.read(varint.decode_stream(b_io)).decode("ascii"))
+        dtype = np.dtype(b_io.read(leb128.u.decode_reader(b_io)).decode("ascii"))
 
         shape = tuple(
-            varint.decode_stream(b_io) for _ in range(varint.decode_stream(b_io))
+            leb128.u.decode_reader(b_io) for _ in range(leb128.u.decode_reader(b_io))
         )
 
-        table_len = varint.decode_stream(b_io)
+        table_len = leb128.u.decode_reader(b_io)
 
         # select the smallest output data type that can encode the indices
         utype: np.dtype[np.unsignedinteger]