import java.util.ArrayList;
import java.util.List;

import org.codelibs.core.io.ResourceUtil;

public class ProfileUtil {
    private static final String SPRING_PROFILES_ACTIVE = "spring.profiles.active";

    public static void setup() {
        final List<String> list = new ArrayList<String>();

        final String values = System.getProperty(SPRING_PROFILES_ACTIVE);
        if (values != null) {

    text(
        "profiler",
        defaultProfiler,
        display = ParameterDisplay.PROMPT,
        allowEmpty = false,
        description =
            "Command line option for the performance test task to enable profiling. " +
                "For example `async-profiler`, `async-profiler-heap`, `async-profiler-all` or `jfr`. Use `none` for benchmarking only.",
    )
}

                        listOf("async-profiler", "async-profiler-heap")
                    }.map { FlameGraphGenerationBuildSpec(scenario, os, arch, it) }
                }
            }

    inner class FlameGraphGenerationBuildSpec(
        val performanceScenario: PerformanceScenario,
        override val os: Os,
        override val arch: Arch,
        val profiler: String,
    ) : PerformanceTestBuildSpec {

	if usePkgCache {
		if _, ok := pkgTags[dir]; !ok {
			pkgTags[dir] = info.AllTags
			key = tagKey(dir, context, info.AllTags)
		}
	}

	filenames := append(append([]string{}, info.GoFiles...), info.CgoFiles...)

	// Parse package files.
	var files []*ast.File
	for _, file := range filenames {
		f, err := w.parseFile(dir, file)
		if err != nil {
			log.Fatalf("error parsing package %s: %s", name, err)

if [[ `uname -s | grep -P '^MSYS_NT'` ]]; then
  PROFILE_JSON_PATH=$(replace_drive_letter_with_prefix "$TFCI_OUTPUT_WIN_DOCKER_DIR")
  PROFILE_JSON_PATH="$PROFILE_JSON_PATH/profile.json.gz"
else
  PROFILE_JSON_PATH="$TFCI_OUTPUT_DIR/profile.json.gz"
fi

# TODO(b/361369076) Remove the following block after TF NumPy 1 is dropped
# Move hermetic requirement lock files for NumPy 1 to the root

                    "Flame graphs with $profiler for ${performanceScenario.scenario.scenario} | ${performanceScenario.testProject} " +
                        "on ${os.asName()} (bucket $bucketIndex)",
                performanceSubProject = "performance",
                bucketIndex = bucketIndex,
                extraParameters =

const val INDIVIDUAL_PERFORAMCE_TEST_ARTIFACT_RULES = """
testing/*/build/test-results-*.zip => results
testing/*/build/tmp/**/log.txt => failure-logs
testing/*/build/tmp/**/profile.log => failure-logs
testing/*/build/tmp/**/daemon-*.out.log => failure-logs
"""

// to avoid pathname too long error
fun BuildSteps.substDirOnWindows(os: Os) {
    if (os == Os.WINDOWS) {
        script {

        "//tensorflow/core/platform:stringpiece",
        "//tensorflow/core/platform:types",
        "//tensorflow/core/profiler/internal:print_model_analysis",
        "//tensorflow/core/profiler/lib:traceme",
        "//tensorflow/core/profiler/rpc/client:profiler_client_impl",
        "//tensorflow/core/profiler/rpc:profiler_server_impl",
        "//tensorflow/core/util:managed_stack_trace",
        "//tensorflow/core:all_kernels",

[GC guide](/doc/gc-guide#Linux_transparent_huge_pages), or switch to a Linux
distribution that disables transparent huge pages altogether.

Go 1.22 added contention on runtime-internal locks to the [`mutex`
profile](/pkg/runtime/pprof#Profile). Contention on these locks is always
reported at `runtime._LostContendedRuntimeLock`. Complete stack traces of
runtime locks can be enabled with the [`runtimecontentionstacks`

    analyzing input pipeline performance with TF Profiler.

*   [`tf.distribute.TPUStrategy`](https://www.tensorflow.org/api_docs/python/tf/distribute/TPUStrategy)
    is now a stable API and no longer considered experimental for TensorFlow.
    (earlier `tf.distribute.experimental.TPUStrategy`).

*   [TF Profiler](https://www.tensorflow.org/guide/profiler) introduces two new
    tools: a memory profiler to visualize your model’s memory usage over time