[Python] Fix wheel building; update version to 1.0.0b1 (#362)

* Fix cibuildwheel network proxy.

Signed-off-by: Xiong HuaiDong <[email protected]>

* Fix dependent urllib3 version.

Signed-off-by: Xiong HuaiDong <[email protected]>

* Fix callback debug prints.

Signed-off-by: Xiong HuaiDong <[email protected]>

* Update LakeSoul wheel version to 1.0.0b1 for ray dataset.

Signed-off-by: Xiong HuaiDong <[email protected]>

---------

Signed-off-by: Xiong HuaiDong <[email protected]>
Co-authored-by: Xiong HuaiDong <[email protected]>

pyproject.toml

@@ -8,7 +8,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "lakesoul"
-version = "1.0.0b0"
+version = "1.0.0b1"
 authors = [
   { name="LakeSoul Team", email="[email protected]" },
 ]
@@ -41,12 +41,13 @@ include = ["lakesoul*"]
 [project.optional-dependencies]
 torch = ["torch>=1.11"]
 pandas = ["pandas>=1.4"]
-datasets = ["datasets>=2.14"]
+datasets = ["datasets>=2.14", "urllib3<2,>=1.21.1"]
 ray = ["ray>=2.7"]
 all = [
     "torch>=1.11",
     "pandas>=1.4",
     "datasets>=2.14",
+    "urllib3<2,>=1.21.1",
     "ray>=2.7",
 ]
 

python/docker_build_all_wheels.py

@@ -35,7 +35,7 @@ def _patch_cibuildwheel(self, dir_path):
         file_path = os.path.join(site_path, 'cibuildwheel', 'oci_container.py')
         with io.open(file_path) as fin:
             text = fin.read()
-        pattern = r'^( +network_args = \["--network=host"])$\n\n( +)(shell_args = \[)'
+        pattern = r'^( +network_args = \["--network=host"])$\n\n( +)(.+)'
         match = re.search(pattern, text, re.M)
         if match is None:
             return
@@ -73,7 +73,7 @@ def context():
                 py_path = os.path.join(dir_path, 'bin', 'python')
                 args = [py_path, '-m', 'pip', 'install', '--upgrade', 'pip']
                 subprocess.check_call(args)
-                args = [py_path, '-m', 'pip', 'install', 'cibuildwheel~=2.15']
+                args = [py_path, '-m', 'pip', 'install', 'cibuildwheel==2.16.2']
                 subprocess.check_call(args)
                 self._patch_cibuildwheel(dir_path)
                 yield dir_path

python/lakesoul/metadata/native_client.py

@@ -44,7 +44,7 @@ def __init__(self):
         self._bool = False
 
         def callback(bool, msg):
-            print("create connection callback: status={} msg={}".format(bool, msg.decode("utf-8")))
+            #print("create connection callback: status={} msg={}".format(bool, msg.decode("utf-8")))
             if not bool:
                 message = "fail to initialize lakesoul.metadata.native_client.NativeMetadataClient"
                 raise RuntimeError(message)
@@ -89,12 +89,11 @@ def execute_query(self, query_type, params):
         joined_params = PARAM_DELIM.join(params).encode("utf-8")
 
         def execute_query_callback(len, msg):
-            print("execute_query query_type={} callback: len={} msg={}".format(query_type, len, msg.decode("utf-8")))
+            #print("execute_query query_type={} callback: len={} msg={}".format(query_type, len, msg.decode("utf-8")))
             self._query_result_len = len
 
         def export_bytes_result_callback(bool, msg):
-            print(
-                "export_bytes_result callback: bool={} msg={}".format(query_type, bool, msg.decode("utf-8")))
+            #print("export_bytes_result callback: bool={} msg={}".format(query_type, bool, msg.decode("utf-8")))
             self._bool = bool
 
         with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor:

python/lakesoul/metadata/utils.py

@@ -225,5 +225,5 @@ def to_arrow_schema(schema_json_str, exclude_columns=None):
         if field['name'] in exclude_columns:
             continue
         arrow_fields.append(to_arrow_field(field))
-    print(arrow_fields)
+    #print(arrow_fields)
     return pyarrow.schema(arrow_fields)