From caa07aac4e4580b17095c00d75dcb4516b877ca1 Mon Sep 17 00:00:00 2001
From: Zac <zac.codiamat@mongodb.com>
Date: Thu, 31 Jul 2025 11:17:09 -0700
Subject: [PATCH] SERVER-108478 Move lint_large_files_check to bazel run lint
 (#39413)

GitOrigin-RevId: e6ff6a58c1b1fe83644e859bdfb4e240c6528112
---
 bazel/wrapper_hook/lint.py                    | 17 ++++++++++-
 buildscripts/BUILD.bazel                      | 17 +++++++++++
 buildscripts/large_file_check.py              |  6 +++-
 .../tasks/misc_tasks.yml                      | 29 +------------------
 4 files changed, 39 insertions(+), 30 deletions(-)

diff --git a/bazel/wrapper_hook/lint.py b/bazel/wrapper_hook/lint.py
index dbb667f2799..03aa150909c 100644
--- a/bazel/wrapper_hook/lint.py
+++ b/bazel/wrapper_hook/lint.py
@@ -10,6 +10,7 @@ from typing import List
 REPO_ROOT = pathlib.Path(__file__).parent.parent.parent
 sys.path.append(str(REPO_ROOT))
 
+LARGE_FILE_THRESHOLD = 10 * 1024 * 1024 #10MiB
 
 def create_build_files_in_new_js_dirs() -> None:
     base_dirs = ["src/mongo/db/modules/enterprise/jstests", "jstests"]
@@ -207,7 +208,7 @@ def get_parsed_args(args):
     )
     parser.add_argument(
         "--fail-on-validation",
-        type=bool,
+        action="store_true",
         default=False,
     )
     parser.add_argument(
@@ -216,6 +217,11 @@ def get_parsed_args(args):
         default="origin/master",
         help="Base branch to compare changes against",
     )
+    parser.add_argument(
+        "--large-files",
+        action="store_true",
+        default=False
+    )
     return parser.parse_known_args(args)
 
 def run_rules_lint(bazel_bin: str, args: List[str]) -> bool:
@@ -282,6 +288,15 @@ def run_rules_lint(bazel_bin: str, args: List[str]) -> bool:
     if lint_all or any(file.endswith(".yml") for file in files_to_lint):
         subprocess.run([bazel_bin, "run", "//buildscripts:validate_evg_project_config", "--", f"--evg-project-name={parsed_args.lint_yaml_project}", "--evg-auth-config=.evergreen.yml"], check=True)
 
+    if lint_all or parsed_args.large_files:
+        subprocess.run([bazel_bin, "run", "//buildscripts:large_file_check", "--", "--exclude", "src/third_party/*"], check=True)
+    else:
+        # simple check
+        for file in files_to_lint:
+            if os.path.getsize(file) > LARGE_FILE_THRESHOLD:
+                print(f"File {file} exceeds large file threshold of {LARGE_FILE_THRESHOLD}")
+                return False
+
     # Default to linting everything in rules_lint if no path was passed in.
     if len([arg for arg in args if not arg.startswith("--")]) == 0:
         args = ["//..."] + args
diff --git a/buildscripts/BUILD.bazel b/buildscripts/BUILD.bazel
index b79bcacbc12..8d549633f27 100644
--- a/buildscripts/BUILD.bazel
+++ b/buildscripts/BUILD.bazel
@@ -306,6 +306,23 @@ py_binary(
     ],
 )
 
+py_binary(
+    name = "large_file_check",
+    srcs = ["large_file_check.py"],
+    deps = [
+        "//buildscripts/linter",
+        "//buildscripts/patch_builds",
+        dependency(
+            "structlog",
+            group = "evergreen",
+        ),
+        dependency(
+            "gitpython",
+            group = "evergreen",
+        ),
+    ],
+)
+
 # TODO(SERVER-105817): The following library is autogenerated, please split these out into individual python targets
 py_library(
     name = "all_python_files",
diff --git a/buildscripts/large_file_check.py b/buildscripts/large_file_check.py
index befe8e47ee1..d22c4b1814e 100755
--- a/buildscripts/large_file_check.py
+++ b/buildscripts/large_file_check.py
@@ -57,7 +57,11 @@ MONGO_REVISION_ENV_VAR = "REVISION"
 def _get_repos_and_revisions() -> Tuple[List[Repo], RevisionMap]:
     """Get the repo object and a map of revisions to compare against."""
 
-    repos = [Repo(git.get_base_dir())]
+    repo_dir = os.environ.get("BUILD_WORKSPACE_DIRECTORY", None)
+    if repo_dir:
+        repos = [Repo(repo_dir)]
+    else:
+        repos = [Repo(git.get_base_dir())]
 
     revision_map = generate_revision_map(repos, {"mongo": os.environ.get(MONGO_REVISION_ENV_VAR)})
     return repos, revision_map
diff --git a/etc/evergreen_yml_components/tasks/misc_tasks.yml b/etc/evergreen_yml_components/tasks/misc_tasks.yml
index a92c668cfc8..593cffc1a00 100644
--- a/etc/evergreen_yml_components/tasks/misc_tasks.yml
+++ b/etc/evergreen_yml_components/tasks/misc_tasks.yml
@@ -734,7 +734,7 @@ tasks:
       - func: "bazel run"
         vars:
           target: >-
-            lint //... --lint-yaml=${project}
+            lint //... --lint-yaml=${project} --large-files
 
   # Check that the mutational fuzzer can parse all JS filess.
   - name: lint_fuzzer_sanity_all
@@ -767,33 +767,6 @@ tasks:
       - func: "setup jstestfuzz"
       - func: "lint fuzzer sanity patch"
 
-  - name: lint_large_files_check
-    tags:
-      [
-        "assigned_to_jira_team_devprod_build",
-        "development_critical_single_variant",
-      ]
-    exec_timeout_secs: 600 # 10 minute timeout
-    commands:
-      - command: manifest.load
-      - func: "git get project and add git tag"
-      - func: "f_expansions_write"
-      - func: "kill processes"
-      - func: "cleanup environment"
-      - func: "set up venv"
-      - func: "upload pip requirements"
-      - func: "configure evergreen api credentials"
-      - command: subprocess.exec
-        type: test
-        params:
-          binary: bash
-          args:
-            - "./src/evergreen/run_python_script_with_report.sh"
-            - "lint-large-files-check"
-            - "buildscripts/large_file_check.py"
-            - "--exclude"
-            - "src/third_party/*"
-
   - name: lint_pyright
     tags:
       [