fix: Lambda GPU image OSS compliance and builder consistency (#5739)

Author: junpuf

docker/lambda/Dockerfile

@@ -68,7 +68,12 @@ COPY --from=lambda-python /var/lang/lib /var/lang/lib
 RUN rm -rf /var/lang/lib/python3.13/site-packages
 ENV PATH="/var/lang/bin:$PATH"
 COPY ./docker/lambda/requirements-base.txt /tmp/requirements.txt
-RUN --mount=type=cache,target=/root/.cache/uv uv pip install --python /var/lang/bin/python3.13 -r /tmp/requirements.txt
+RUN --mount=type=cache,target=/root/.cache/uv \
+  uv pip install --python /var/lang/bin/python3.13 -r /tmp/requirements.txt \
+  && uv pip check --python /var/lang/bin/python3.13 \
+  && (find /var/lang -type d -name __pycache__ -exec rm -rf {} + 2>/dev/null; true) \
+  && find /var/lang -type f -name "*.pyc" -delete \
+  && find /var/lang -type f -name "*.pyo" -delete
 
 # ============================================================
 # CuPy builder (Python 3.13 — awslambdaric, CuPy, NumPy, SciPy, etc.)
@@ -236,6 +241,10 @@ RUN chmod +x /usr/local/bin/deep_learning_container.py \
   && echo 'source /usr/local/bin/bash_telemetry.sh' >>/etc/bashrc \
   && echo 'source /usr/local/bin/bash_telemetry.sh' >>/root/.bashrc \
   && bash /tmp/setup_oss_compliance.sh python3 \
+  && printf '\n** FFmpeg; version 6.1 -- https://ffmpeg.org/releases/ffmpeg-6.1.tar.xz\n' >> /root/BUILD_FROM_SOURCE_PACKAGES_LICENCES \
+  && printf '\nFFmpeg 6.1 https://ffmpeg.org/releases/ffmpeg-6.1.tar.xz\n' >> /root/THIRD_PARTY_SOURCE_CODE_URLS \
+  && printf '\n** FFprobe; version 6.1 -- https://ffmpeg.org/releases/ffmpeg-6.1.tar.xz\n' >> /root/BUILD_FROM_SOURCE_PACKAGES_LICENCES \
+  && printf '\nFFprobe 6.1 https://ffmpeg.org/releases/ffmpeg-6.1.tar.xz\n' >> /root/THIRD_PARTY_SOURCE_CODE_URLS \
   && rm /tmp/setup_oss_compliance.sh \
   && rm -rf /var/lang/lib/python3.13/ensurepip/_bundled
 WORKDIR /var/task

test/sanity/scripts/upload_oss_source.py

@@ -48,23 +48,33 @@ def upload_to_s3(s3_client, local_path, s3_key):
 
 
 def clone_and_tar(name, version, url, work_dir):
-    """Clone repo and create tarball. Returns tarball path or None."""
+    """Clone repo or download tarball and create tarball. Returns tarball path or None."""
     dir_name = f"{name}_v{version}_source_code"
     local_dir = os.path.join(work_dir, dir_name)
     tarball = f"{local_dir}.tar.gz"
+    url = url.strip()
+    is_tarball = url.endswith((".tar.gz", ".tar.xz", ".tar.bz2", ".tgz"))
 
     for attempt in range(MAX_RETRIES):
         try:
             if not os.path.isdir(local_dir):
-                subprocess.run(
-                    ["git", "clone", url.strip(), local_dir],
-                    check=True,
-                    capture_output=True,
-                )
+                if is_tarball:
+                    downloaded = os.path.join(work_dir, os.path.basename(url))
+                    subprocess.run(
+                        ["curl", "-fsSL", "-o", downloaded, url], check=True, capture_output=True
+                    )
+                    os.makedirs(local_dir)
+                    subprocess.run(
+                        ["tar", "-xf", downloaded, "-C", local_dir, "--strip-components=1"],
+                        check=True,
+                        capture_output=True,
+                    )
+                else:
+                    subprocess.run(
+                        ["git", "clone", url, local_dir], check=True, capture_output=True
+                    )
             subprocess.run(
-                ["tar", "-czf", tarball, "-C", work_dir, dir_name],
-                check=True,
-                capture_output=True,
+                ["tar", "-czf", tarball, "-C", work_dir, dir_name], check=True, capture_output=True
             )
             return tarball
         except Exception as e: