Updates struct2tensor's bundled arrow version to 6.0.1.

Additionally:
- uses `parquet_types` pre-generated by arrow instead of bundling it.
- uses `arrow_format` headers pre-generated by arrow instead of generating it with flatbuffers.
PiperOrigin-RevId: 473794356

Author: iindyk

.bazelrc

@@ -46,4 +46,4 @@ build:macos --host_cxxopt=-std=c++17
 # Suppress all warning messages.
 build:short_logs --output_filter=DONT_MATCH_ANYTHING
 
-build:macos --macos_minimum_os=10.9
+build:macos --macos_minimum_os=10.12

RELEASE.md

@@ -6,6 +6,7 @@
 
 ## Bug Fixes and Other Changes
 
+*   Updates bundled `arrow` version to 6.0.1.
 *   Depends on `tensorflow>=2.10.0,<2.11`.
 
 ## Deprecations

struct2tensor/workspace.bzl

@@ -40,17 +40,18 @@ def struct2tensor_workspace():
     )
     # LINT.ThenChange(third_party/thrift.BUILD:thrift_gen_version)
 
+    # Use the last commit on the relevant release branch to update.
     # LINT.IfChange(arrow_archive_version)
+    ARROW_COMMIT = "347a88ff9d20e2a4061eec0b455b8ea1aa8335dc"  # 6.0.1
+    # LINT.ThenChange(third_party/arrow.BUILD:arrow_gen_version)
+
     http_archive(
         name = "arrow",
         build_file = "//third_party:arrow.BUILD",
-        sha256 = "d7b3838758a365c8c47d55ab0df1006a70db951c6964440ba354f81f518b8d8d",
-        strip_prefix = "arrow-apache-arrow-0.16.0",
-        urls = [
-            "https://github.com/apache/arrow/archive/apache-arrow-0.16.0.tar.gz",
-        ],
+        sha256 = "55fc466d0043c4cce0756bc18e1e62b3233be74c9afe8dc0d18420b9a5fd9714",
+        strip_prefix = "arrow-%s" % ARROW_COMMIT,
+        urls = ["https://github.com/apache/arrow/archive/%s.zip" % ARROW_COMMIT],
     )
-    # LINT.ThenChange(third_party/arrow.BUILD:parquet_gen_version)
 
     _TFMD_COMMIT_HASH = "6703ba1095f8a2fe3567f1a2209c819b530c64de"  # 1.10.0
     http_archive(

third_party/BUILD

@@ -1,4 +1,4 @@
-# Copyright 2020 Google LLC
+# Copyright 2022 Google LLC
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -12,18 +12,4 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-# Description:
-#   librarys for exposing vendored files in third_party.
-
-licenses(["notice"])  # Apache 2.0
-
-exports_files([
-    "parquet/parquet_types.cpp",
-])
-
-cc_library(
-    name = "parquet_types_h",
-    hdrs = ["parquet/parquet_types.h"],
-    includes = ["."],
-    visibility = ["//visibility:public"],
-)
+licenses(["notice"])

third_party/arrow.BUILD

@@ -15,63 +15,43 @@
 # Description:
 #   Apache arrow library
 
-load("@flatbuffers//:build_defs.bzl", "flatbuffer_cc_library")
-
-flatbuffer_cc_library(
-    name = "arrow_format",
-    srcs = [
-        "cpp/src/arrow/ipc/feather.fbs",
-        "format/File.fbs",
-        "format/Message.fbs",
-        "format/Schema.fbs",
-        "format/SparseTensor.fbs",
-        "format/Tensor.fbs",
-    ],
-    flatc_args = [
-        "--scoped-enums",
-        "--gen-object-api",
-    ],
-    out_prefix = "cpp/src/generated/",
-)
-
 package(default_visibility = ["//visibility:public"])
 
+# LINT.IfChange(arrow_gen_version)
+arrow_version_replace_cmd = ("sed " +
+                             "-e 's/@ARROW_VERSION_MAJOR@/6/g' " +
+                             "-e 's/@ARROW_VERSION_MINOR@/0/g' " +
+                             "-e 's/@ARROW_VERSION_PATCH@/1/g' ")
+# LINT.ThenChange(../workspace.bzl:arrow_archive_version)
+
 genrule(
     name = "arrow_util_config",
     srcs = ["cpp/src/arrow/util/config.h.cmake"],
     outs = ["cpp/src/arrow/util/config.h"],
-    cmd = ("sed " +
-           "-e 's/@ARROW_VERSION_MAJOR@/0/g' " +
-           "-e 's/@ARROW_VERSION_MINOR@/16/g' " +
-           "-e 's/@ARROW_VERSION_PATCH@/0/g' " +
+    cmd = (arrow_version_replace_cmd +
            "-e 's/cmakedefine/define/g' " +
            "$< >$@"),
 )
 
-# LINT.IfChange(parquet_gen_version)
 genrule(
     name = "parquet_version_h",
     srcs = ["cpp/src/parquet/parquet_version.h.in"],
     outs = ["cpp/src/parquet/parquet_version.h"],
-    cmd = ("sed " +
-           "-e 's/@PARQUET_VERSION_MAJOR@/1/g' " +
-           "-e 's/@PARQUET_VERSION_MINOR@/5/g' " +
-           "-e 's/@PARQUET_VERSION_PATCH@/1/g' " +
+    # Parquet packaged with arrow uses arrow versioning.
+    cmd = (arrow_version_replace_cmd +
            "$< >$@"),
 )
-# LINT.ThenChange(../workspace.bzl:arrow_archive_version)
 
-cc_library(
-    name = "xxhash",
-    srcs = [],
-    hdrs = [
-        "cpp/src/arrow/vendored/xxhash/xxh3.h",
-        "cpp/src/arrow/vendored/xxhash/xxhash.c",
-        "cpp/src/arrow/vendored/xxhash/xxhash.h",
-    ],
-    copts = ["-Wno-implicit-fallthrough"],
-    visibility = ["//visibility:private"],
-)
+# These are originally generated by flatbuffers. We use them instead of
+# re-generating from fbs files under "format/" and "cpp/src/arrow/ipc/".
+fbs_headers = [
+    "cpp/src/generated/feather_generated.h",
+    "cpp/src/generated/File_generated.h",
+    "cpp/src/generated/Message_generated.h",
+    "cpp/src/generated/Schema_generated.h",
+    "cpp/src/generated/SparseTensor_generated.h",
+    "cpp/src/generated/Tensor_generated.h",
+]
 
 cc_library(
     name = "arrow",
@@ -97,8 +77,16 @@ cc_library(
             "cpp/src/arrow/**/*.h",
             "cpp/src/parquet/**/*.h",
             "cpp/src/parquet/**/*.cc",
+            "cpp/src/arrow/tensor/*.cc",
+            "cpp/src/generated/parquet_types.cpp",
         ],
         exclude = [
+            # Excluding files which we don't depend on, but needs
+            # additional dependencies like boost, snappy etc.
+            "cpp/src/arrow/util/bpacking_avx2.*",
+            "cpp/src/arrow/util/bpacking_avx512.*",
+            "cpp/src/arrow/util/bpacking_neon*",
+            "cpp/src/arrow/util/bpacking_simd*",
             "cpp/src/arrow/util/compression_brotli*",
             "cpp/src/arrow/util/compression_bz2*",
             "cpp/src/arrow/util/compression_lz4*",
@@ -115,27 +103,87 @@ cc_library(
             "cpp/src/**/stream_to_file.cc",
             "cpp/src/arrow/ipc/json*.cc",
             "cpp/src/arrow/vendored/xxhash/**",
-            "cpp/src/parquet/encryption_internal.cc",
+            #"cpp/src/arrow/vendored/datetime/**",
+            "cpp/src/parquet/encryption/encryption_internal.cc",
         ],
-    ) + [
-        "@struct2tensor//third_party:parquet/parquet_types.cpp",
-    ],
+    ),
     hdrs = [
-        # declare header from above genrule
+        # Headers from above genrules.
         "cpp/src/arrow/util/config.h",
         "cpp/src/parquet/parquet_version.h",
-    ],
+        # This is originally generated by thrift compiler. We include the
+        # files generated by arrow to avoid requiring the compiler
+        # (thrift library is still needed though).
+        "cpp/src/generated/parquet_types.h",
+    ] + fbs_headers,
     includes = [
         "cpp/src",
         "cpp/src/arrow/vendored/xxhash",
+        "cpp/thirdparty/flatbuffers/include",
     ],
     textual_hdrs = [
         "cpp/src/arrow/vendored/xxhash/xxhash.c",
     ],
     deps = [
-        ":arrow_format",
+        ":datetime",
+        ":flatbuffers",
         ":xxhash",
-        "@struct2tensor//third_party:parquet_types_h",
         "@thrift",
     ],
 )
+
+cc_library(
+    name = "xxhash",
+    srcs = [],
+    hdrs = [
+        "cpp/src/arrow/vendored/xxhash/xxhash.c",
+        "cpp/src/arrow/vendored/xxhash/xxhash.h",
+    ],
+    copts = ["-Wno-implicit-fallthrough"],
+    includes = ["."],
+    visibility = ["//visibility:private"],
+)
+
+config_setting(
+    name = "windows",
+    constraint_values = [
+        "@bazel_tools//platforms:windows",
+    ],
+)
+
+cc_library(
+    name = "datetime",
+    srcs = [
+        "cpp/src/arrow/vendored/datetime/tz.cpp",
+    ],
+    hdrs = [
+        "cpp/src/arrow/vendored/datetime/date.h",
+        "cpp/src/arrow/vendored/datetime/ios.h",
+        "cpp/src/arrow/vendored/datetime/tz.h",
+        "cpp/src/arrow/vendored/datetime/tz_private.h",
+        "cpp/src/arrow/vendored/datetime/visibility.h",
+    ],
+    includes = ["."],
+    linkopts = select({
+        ":windows": [
+            "ole32.lib",  # for CoTaskMemFree
+            "shell32.lib",  # for SHGetKnownFolderPath
+        ],
+        "//conditions:default": [
+        ],
+    }),
+    visibility = ["//visibility:private"],
+)
+
+cc_library(
+    name = "flatbuffers",
+    srcs = [],
+    hdrs = [
+        "cpp/thirdparty/flatbuffers/include/flatbuffers/base.h",
+        "cpp/thirdparty/flatbuffers/include/flatbuffers/flatbuffers.h",
+        "cpp/thirdparty/flatbuffers/include/flatbuffers/stl_emulation.h",
+    ],
+    copts = ["-Wno-implicit-fallthrough"],
+    includes = ["."],
+    visibility = ["//visibility:private"],
+)