Test fixing tensorflow-text nightly builds.

PiperOrigin-RevId: 816948811

Author: ybaturina

WORKSPACE

@@ -4,11 +4,11 @@ load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
 
 http_archive(
     name = "icu",
-    strip_prefix = "icu-release-64-2",
-    sha256 = "dfc62618aa4bd3ca14a3df548cd65fe393155edd213e49c39f3a30ccd618fc27",
+    strip_prefix = "icu-release-76-1",
+    sha256 = "bf9bf92b514dcd9de461961fd922c24a97be4d6e29a0b60b1f8d0bb029e1e449",
     urls = [
-        "https://storage.googleapis.com/mirror.tensorflow.org/github.com/unicode-org/icu/archive/release-64-2.zip",
-        "https://github.com/unicode-org/icu/archive/release-64-2.zip",
+        "https://storage.googleapis.com/mirror.tensorflow.org/github.com/unicode-org/icu/archive/release-76-1.zip",
+        "https://github.com/unicode-org/icu/archive/release-76-1.zip",
     ],
     build_file = "//third_party/icu:BUILD.bzl",
     patches = ["//third_party/icu:udata.patch"],
@@ -95,7 +95,7 @@ load("@org_tensorflow//third_party/py:python_init_rules.bzl", "python_init_rules
 
 python_init_rules()
 
-load("//tensorflow_text:tftext.bzl", "py_deps_profile")
+load("//tensorflow_text:py_deps_profile.bzl", "py_deps_profile")
 
 py_deps_profile(
     name = "release_or_nightly",
@@ -158,7 +158,22 @@ load(
 python_wheel_version_suffix_repository(name = "tf_wheel_version_suffix")
 
 load(
-    "@local_xla//third_party/gpus/cuda/hermetic:cuda_json_init_repository.bzl",
+    "@rules_ml_toolchain//cc/deps:cc_toolchain_deps.bzl",
+    "cc_toolchain_deps",
+)
+
+cc_toolchain_deps()
+
+register_toolchains("@rules_ml_toolchain//cc:linux_x86_64_linux_x86_64")
+
+register_toolchains("@rules_ml_toolchain//cc:linux_x86_64_linux_x86_64_cuda")
+
+register_toolchains("@rules_ml_toolchain//cc:linux_aarch64_linux_aarch64")
+
+register_toolchains("@rules_ml_toolchain//cc:linux_aarch64_linux_aarch64_cuda")
+
+load(
+    "@rules_ml_toolchain//third_party/gpus/cuda/hermetic:cuda_json_init_repository.bzl",
     "cuda_json_init_repository",
 )
 
@@ -170,7 +185,7 @@ load(
     "CUDNN_REDISTRIBUTIONS",
 )
 load(
-    "@local_xla//third_party/gpus/cuda/hermetic:cuda_redist_init_repositories.bzl",
+    "@rules_ml_toolchain//third_party/gpus/cuda/hermetic:cuda_redist_init_repositories.bzl",
     "cuda_redist_init_repositories",
     "cudnn_redist_init_repository",
 )
@@ -184,21 +199,21 @@ cudnn_redist_init_repository(
 )
 
 load(
-    "@local_xla//third_party/gpus/cuda/hermetic:cuda_configure.bzl",
+    "@rules_ml_toolchain//third_party/gpus/cuda/hermetic:cuda_configure.bzl",
     "cuda_configure",
 )
 
 cuda_configure(name = "local_config_cuda")
 
 load(
-    "@local_xla//third_party/nccl/hermetic:nccl_redist_init_repository.bzl",
+    "@rules_ml_toolchain//third_party/nccl/hermetic:nccl_redist_init_repository.bzl",
     "nccl_redist_init_repository",
 )
 
 nccl_redist_init_repository()
 
 load(
-    "@local_xla//third_party/nccl/hermetic:nccl_configure.bzl",
+    "@rules_ml_toolchain//third_party/nccl/hermetic:nccl_configure.bzl",
     "nccl_configure",
 )
 

oss_scripts/configure.sh

@@ -54,6 +54,12 @@ curl https://raw.githubusercontent.com/tensorflow/tensorflow/master/.bazelrc -o
 sed -i -e 's/build --noincompatible_remove_legacy_whole_archive//' .bazelrc
 
 write_to_bazelrc "build:manylinux2014 --config=release_cpu_linux"
+# JUST FOR TESTING
+git clone https://github.com/tensorflow/tensorflow.git org_tensorflow
+cd org_tensorflow
+git checkout 735467e89ccfd7ace190363412bb5698164628b5
+cd ..
+write_to_bazelrc "common:linux --override_repository=org_tensorflow=./org_tensorflow"
 
 if (which python3) | grep -q "python3"; then
   installed_python="python3"
@@ -76,8 +82,16 @@ if [[ "$TF_VERSION" == *"rc"* ]]; then
   REQUIREMENTS_EXTRA_FLAGS="$REQUIREMENTS_EXTRA_FLAGS --pre"
 fi
 
+# JUST FOR TESTING
+wget https://storage.googleapis.com/ml-sysroot-testing/tensorflow_metadata-1.18.0.dev0-py3-none-any.whl
+mv tensorflow_metadata-1.18.0.dev0-py3-none-any.whl /github/tensorflow_metadata-1.18.0.dev0-py3-none-any.whl
+echo "tensorflow-metadata @ file:///github/tensorflow_metadata-1.18.0.dev0-py3-none-any.whl" >> oss_scripts/pip_package/requirements.in
+
 bazel run //oss_scripts/pip_package:requirements.update -- $REQUIREMENTS_EXTRA_FLAGS
 
+# JUST FOR TESTING
+sed -i 's#file://tensorflow_metadata-1.18.0.dev0-py3-none-any.whl#file:///github/tensorflow_metadata-1.18.0.dev0-py3-none-any.whl#g' oss_scripts/pip_package/requirements_lock_3_9.txt
+
 TF_ABIFLAG=$(bazel run //oss_scripts/pip_package:tensorflow_build_info -- abi)
 SHARED_LIBRARY_NAME="libtensorflow_framework.so.2"
 if is_macos; then

oss_scripts/pip_package/BUILD

@@ -1,6 +1,7 @@
 # Tools for building the TF.Text pip package.
-load("@python//:defs.bzl", "compile_pip_requirements")
+
 load("@python_version_repo//:py_version.bzl", "REQUIREMENTS")
+load("@rules_python//python:pip.bzl", "compile_pip_requirements")
 
 package(default_visibility = ["//visibility:private"])
 

oss_scripts/pip_package/requirements.in

@@ -1,8 +1,7 @@
 setuptools==70.0.0
 dm-tree==0.1.8  # Limit for macos support.
 numpy
-protobuf==4.25.3  # b/397977335 - Fix crash on python 3.9, 3.10.
 tensorflow
 tf-keras
 tensorflow-datasets
-tensorflow-metadata
+#tensorflow-metadata

tensorflow_text/BUILD

@@ -1,7 +1,8 @@
-# Placeholder: load py_library
-# Placeholder: load py_test
 load("@org_tensorflow//tensorflow/lite:build_def.bzl", "tflite_cc_shared_object")
 load("@rules_cc//cc:cc_library.bzl", "cc_library")
+load("@rules_python//python:py_library.bzl", "py_library")
+load("@rules_python//python:py_test.bzl", "py_test")
+load("//third_party/tensorflow:tensorflow.bzl", "if_oss")
 load("//tensorflow_text:tftext.bzl", "extra_py_deps", "py_tf_text_library")
 # [internal] load build_test.bzl
 
@@ -123,9 +124,11 @@ py_library(
         ":wordpiece_tokenizer",
         ":wordshape_ops",
         # python/util:all_util tensorflow dep,
-        "//tensorflow_text/core/pybinds:tflite_registrar",
         "//tensorflow_text/tools/wordpiece_vocab",
-    ] + extra_py_deps(),
+    ] + if_oss(
+        ["//tensorflow_text/core/pybinds:tflite_registrar_library"],
+        ["//tensorflow_text/core/pybinds:tflite_registrar"],
+    ) + extra_py_deps(),
 )
 
 # build_test

tensorflow_text/core/kernels/BUILD

@@ -337,6 +337,7 @@ cc_library(
     deps = [
         "@com_google_absl//absl/container:flat_hash_set",
         "@com_google_absl//absl/status:statusor",
+        "@com_google_absl//absl/strings",
         "@darts_clone",
     ],
 )

tensorflow_text/core/kernels/darts_clone_trie_builder.cc

@@ -19,6 +19,7 @@
 #include <numeric>
 
 #include "absl/container/flat_hash_set.h"
+#include "absl/strings/str_cat.h"
 #include "include/darts.h"
 
 namespace tensorflow {

tensorflow_text/core/pybinds/BUILD

@@ -1,7 +1,10 @@
 # Code that exposes C++ libraries to Python via pybind11.
 
-# Placeholder: load py_test
-load("@org_tensorflow//tensorflow:tensorflow.bzl", "pybind_extension")
+load("@rules_python//python:py_test.bzl", "py_test")
+
+# This comment separates the load statements, so copybara reverse transformation works.
+
+load("@org_tensorflow//tensorflow:tensorflow.default.bzl", "pybind_extension", "pywrap_library")
 
 licenses(["notice"])
 
@@ -11,6 +14,11 @@ package(default_visibility = [
     "//tensorflow_text:__subpackages__",
 ])
 
+pywrap_library(
+    name = "tflite_registrar_library",
+    deps = [":tflite_registrar"],
+)
+
 pybind_extension(
     name = "tflite_registrar",
     srcs = [

tensorflow_text/py_deps_profile.bzl

@@ -0,0 +1,49 @@
+"""
+Repository rule for open source dependencies.
+"""
+
+def _py_deps_profile_impl(ctx):
+    is_switch = False
+    for var_name, var_val in ctx.attr.switch.items():
+        is_switch = is_switch or ctx.os.environ.get(var_name, "") == var_val
+
+    prefix = ctx.attr.pip_repo_name
+    reqirements_name = ctx.attr.requirements_in.name
+    requirements_in_substitutions = {}
+    build_content = ['exports_files(["{}"])'.format(reqirements_name)]
+    for k, v in ctx.attr.deps_map.items():
+        repo_name = v[0] if is_switch else k
+        requirements_in_substitutions[k + "\n"] = repo_name + "\n"
+        requirements_in_substitutions[k + "\r\n"] = repo_name + "\r\n"
+        aliased_targets = ["pkg"] + v[1:]
+        norm_repo_name = repo_name.replace("-", "_")
+        norm_alas_name = k.replace("-", "_")
+        for target in aliased_targets:
+            alias_name = "{}_{}".format(norm_alas_name, target)
+            alias_value = "@{}_{}//:{}".format(prefix, norm_repo_name, target)
+            build_content.append("""
+alias(
+    name = "{}",
+    actual = "{}",
+    visibility = ["//visibility:public"]
+)
+""".format(alias_name, alias_value))
+
+    ctx.file("BUILD", "".join(build_content))
+    ctx.template(
+        reqirements_name,
+        ctx.attr.requirements_in,
+        executable = False,
+        substitutions = requirements_in_substitutions,
+    )
+
+py_deps_profile = repository_rule(
+    implementation = _py_deps_profile_impl,
+    attrs = {
+        "requirements_in": attr.label(mandatory = True),
+        "deps_map": attr.string_list_dict(mandatory = True),
+        "pip_repo_name": attr.string(mandatory = True),
+        "switch": attr.string_dict(mandatory = True),
+    },
+    local = True,
+)

tensorflow_text/tftext.bzl

@@ -1,6 +1,8 @@
 """
 Build rule for open source tf.text libraries.
 """
+load("@rules_python//python:py_library.bzl", "py_library")
+
 
 def py_tf_text_library(
         name,
@@ -85,7 +87,7 @@ def py_tf_text_library(
             }),
         )
 
-    native.py_library(
+    py_library(
         name = name,
         srcs = srcs,
         srcs_version = "PY2AND3",
@@ -139,6 +141,12 @@ def tf_cc_library(
     oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/time")
     oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/types:variant")
     oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/functional:any_invocable")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/log:check")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/log:log")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/log:absl_check")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/log:absl_log")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/container:btree")
+    oss_deps = oss_deps + _dedupe(deps, "@com_google_absl//absl/container:flat_hash_set")
     deps += select({
         "@org_tensorflow//tensorflow:mobile": [
             "@org_tensorflow//tensorflow/core:portable_tensorflow_lib_lite",
@@ -227,49 +235,3 @@ def extra_py_deps():
         "@pypi_tensorflow_datasets//:pkg",
         "@pypi_tensorflow_metadata//:pkg",
     ]
-
-def _py_deps_profile_impl(ctx):
-    is_switch = False
-    for var_name, var_val in ctx.attr.switch.items():
-        is_switch = is_switch or ctx.os.environ.get(var_name, "") == var_val
-
-    prefix = ctx.attr.pip_repo_name
-    reqirements_name = ctx.attr.requirements_in.name
-    requirements_in_substitutions = {}
-    build_content = ['exports_files(["{}"])'.format(reqirements_name)]
-    for k, v in ctx.attr.deps_map.items():
-        repo_name = v[0] if is_switch else k
-        requirements_in_substitutions[k + "\n"] = repo_name + "\n"
-        requirements_in_substitutions[k + "\r\n"] = repo_name + "\r\n"
-        aliased_targets = ["pkg"] + v[1:]
-        norm_repo_name = repo_name.replace("-", "_")
-        norm_alas_name = k.replace("-", "_")
-        for target in aliased_targets:
-            alias_name = "{}_{}".format(norm_alas_name, target)
-            alias_value = "@{}_{}//:{}".format(prefix, norm_repo_name, target)
-            build_content.append("""
-alias(
-    name = "{}",
-    actual = "{}",
-    visibility = ["//visibility:public"]
-)
-""".format(alias_name, alias_value))
-
-    ctx.file("BUILD", "".join(build_content))
-    ctx.template(
-        reqirements_name,
-        ctx.attr.requirements_in,
-        executable = False,
-        substitutions = requirements_in_substitutions,
-    )
-
-py_deps_profile = repository_rule(
-    implementation = _py_deps_profile_impl,
-    attrs = {
-        "requirements_in": attr.label(mandatory = True),
-        "deps_map": attr.string_list_dict(mandatory = True),
-        "pip_repo_name": attr.string(mandatory = True),
-        "switch": attr.string_dict(mandatory = True),
-    },
-    local = True,
-)