open source parquet_dataset.

PiperOrigin-RevId: 303827624

Author: andylou2

BUILD

@@ -23,6 +23,11 @@ sh_binary(
     testonly = True,  # Some files are testonly
     srcs = ["build_pip_package.sh"],
     data = [
+        # :struct2tensor_kernels_and_ops is not required to build the wheel,
+        # but is included here as a sanity check to make sure our kernels
+        # and ops are not broken.
+        "//struct2tensor:struct2tensor_kernels_and_ops",
+        "//struct2tensor:parquet",
         "//struct2tensor/ops:struct2tensor_ops",
         "//struct2tensor/proto:query_metadata_py_pb2",
         "//struct2tensor/test:dependent_test_py_pb2",

setup.py

@@ -61,6 +61,7 @@ def has_ext_modules(self):
         'protobuf>=3.8.0',
         'tensorflow>=1.15,<3',
         'tensorflow_metadata>=0.13.0',
+        'pyarrow>=0.15,<1',
     ],
     # Add in any packaged data.
     include_package_data=True,

struct2tensor/BUILD

@@ -143,6 +143,21 @@ s2t_pytype_library(
     ],
 )
 
+s2t_pytype_library(
+    name = "parquet",
+    srcs = [
+        "expression_impl/parquet.py",
+    ],
+    deps = [
+        ":expression",
+        ":map_prensor_to_prensor",
+        ":path",
+        ":placeholder",
+        ":prensor",
+        "//struct2tensor/ops:gen_parquet_dataset_py",
+    ],
+)
+
 s2t_pytype_library(
     name = "proto_test_util",
     testonly = 1,