[doc][build/2] upload doc build artifacts (ray-project#46749)

Upload the doc build generated artifact to s3.

Test:
- CI
- aws s3 cp
s3://ray-ci-results/doc_build/6b33c4752fa83d36b1fb7248177d16bdc3fe22ae.tgz
.

Signed-off-by: can <[email protected]>

ci/ray_ci/doc/cmd_build.py

@@ -1,8 +1,14 @@
 import subprocess
+import os
 
+import boto3
 import click
 
-from ci.ray_ci.utils import logger
+from ci.ray_ci.utils import logger, ci_init
+from ray_release.util import get_write_state_machine_aws_bucket
+
+
+AWS_CACHE_KEY = "doc_build"
 
 
 @click.command()
@@ -14,11 +20,19 @@ def main(ray_checkout_dir: str) -> None:
     """
     This script builds ray doc and upload build artifacts to S3.
     """
+    ci_init()
+
     logger.info("Building ray doc.")
     _build(ray_checkout_dir)
 
+    if os.environ.get("BUILDKITE_BRANCH") != "master":
+        logger.info(
+            "Not uploading build artifacts to S3 because this is not the master branch."
+        )
+        return
+
     logger.info("Uploading build artifacts to S3.")
-    _upload_build_artifacts()
+    _upload_build_artifacts(ray_checkout_dir)
 
     return
 
@@ -36,13 +50,36 @@ def _build(ray_checkout_dir):
     )
 
 
-def _upload_build_artifacts():
+def _upload_build_artifacts(ray_checkout_dir):
     """
     Upload the build artifacts to S3.
-
-    TODO(can): to be implemented
     """
-    pass
+    # Get the list of the doc-generated files
+    subprocess.run(
+        ["git", "config", "--global", "--add", "safe.directory", ray_checkout_dir],
+        check=True,
+    )
+    doc_generated_files = subprocess.check_output(
+        ["git", "ls-files", "doc", "--others", "-z"],
+        cwd=ray_checkout_dir,
+    )
+
+    # Create a tarball of the doc-generated files
+    doc_tarball = f'{os.environ["BUILDKITE_COMMIT"]}.tgz'
+    with subprocess.Popen(
+        ["tar", "-cvzf", doc_tarball, "--null", "-T", "-"],
+        stdin=subprocess.PIPE,
+        cwd=ray_checkout_dir,
+    ) as proc:
+        proc.communicate(input=doc_generated_files)
+
+    # Upload the tarball to S3
+    boto3.client("s3").upload_file(
+        os.path.join(ray_checkout_dir, doc_tarball),
+        get_write_state_machine_aws_bucket(),
+        f"{AWS_CACHE_KEY}/{doc_tarball}",
+    )
+    logger.info(f"Successfully uploaded {doc_tarball} to S3.")
 
 
 if __name__ == "__main__":