From d71daa7bd8606b14eb0702b67f2b891c4f3201e5 Mon Sep 17 00:00:00 2001
From: Stephanie Kim <kimstephanie@google.com>
Date: Mon, 18 Mar 2024 17:28:47 +0000
Subject: [PATCH] Require size_bytes for first class GCS dep

GCS deps will require a size_bytes field of type `int`. If gclient
sees that the actual byte size is different than the noted size, it
will raise an exception.

Bug: b/328065441
Change-Id: If61496eae39cd372966d7e64ba4a84759708c60b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/tools/depot_tools/+/5367540
Reviewed-by: Robbie Iannucci <iannucci@chromium.org>
Reviewed-by: Joanna Wang <jojwang@chromium.org>
Commit-Queue: Stephanie Kim <kimstephanie@google.com>
---
 gclient.py                    | 15 ++++++++++++++-
 gclient_eval.py               |  1 +
 testing_support/fake_repos.py |  4 ++++
 3 files changed, 19 insertions(+), 1 deletion(-)

diff --git a/gclient.py b/gclient.py
index f245e3e41..10bf68c59 100755
--- a/gclient.py
+++ b/gclient.py
@@ -761,6 +761,7 @@ class Dependency(gclient_utils.WorkItem, DependencySettings):
                                   object_name=dep_value['object_name'],
                                   sha256sum=dep_value['sha256sum'],
                                   output_file=dep_value.get('output_file'),
+                                  size_bytes=dep_value['size_bytes'],
                                   custom_vars=self.custom_vars,
                                   should_process=should_process,
                                   relative=use_relative_paths,
@@ -2504,11 +2505,13 @@ class GcsDependency(Dependency):
     """A Dependency object that represents a single GCS bucket and object"""
 
     def __init__(self, parent, name, bucket, object_name, sha256sum,
-                 output_file, custom_vars, should_process, relative, condition):
+                 output_file, size_bytes, custom_vars, should_process, relative,
+                 condition):
         self.bucket = bucket
         self.object_name = object_name
         self.sha256sum = sha256sum
         self.output_file = output_file
+        self.size_bytes = size_bytes
         url = 'gs://{bucket}/{object_name}'.format(
             bucket=self.bucket,
             object_name=self.object_name,
@@ -2625,10 +2628,13 @@ class GcsDependency(Dependency):
             gsutil.check_call('cp', gcs_url, output_file)
 
         calculated_sha256sum = ''
+        calculated_size_bytes = None
         if os.getenv('GCLIENT_TEST') == '1':
             calculated_sha256sum = 'abcd123'
+            calculated_size_bytes = 10000
         else:
             calculated_sha256sum = self.GetSha256Sum(output_file)
+            calculated_size_bytes = os.path.getsize(output_file)
 
         if calculated_sha256sum != self.sha256sum:
             raise Exception('sha256sum does not match calculated hash. '
@@ -2637,6 +2643,13 @@ class GcsDependency(Dependency):
                                 calculated=calculated_sha256sum,
                             ))
 
+        if calculated_size_bytes != self.size_bytes:
+            raise Exception('size_bytes does not match calculated size bytes. '
+                            '{original} vs {calculated}'.format(
+                                original=self.size_bytes,
+                                calculated=calculated_size_bytes,
+                            ))
+
         if tarfile.is_tarfile(output_file):
             with tarfile.open(output_file, 'r:*') as tar:
                 tar.extractall(path=output_dir)
diff --git a/gclient_eval.py b/gclient_eval.py
index 23b1b1826..08b12fdac 100644
--- a/gclient_eval.py
+++ b/gclient_eval.py
@@ -136,6 +136,7 @@ _GCLIENT_DEPS_SCHEMA = _NodeDictSchema({
             'bucket': str,
             'object_name': str,
             'sha256sum': str,
+            'size_bytes': int,
             schema.Optional('output_file'): str,
             schema.Optional('condition'): str,
             schema.Optional('dep_type', default='gcs'): str,
diff --git a/testing_support/fake_repos.py b/testing_support/fake_repos.py
index 411970a1b..8038c9dc8 100755
--- a/testing_support/fake_repos.py
+++ b/testing_support/fake_repos.py
@@ -896,12 +896,14 @@ deps = {
             'object_name': 'deadbeef',
             'dep_type': 'gcs',
             'sha256sum': 'abcd123',
+            'size_bytes': 10000,
           },
           'src/another_gcs_dep': {
             'bucket': '456bucket',
             'object_name': 'Linux/llvmfile.tar.gz',
             'dep_type': 'gcs',
             'sha256sum': 'abcd123',
+            'size_bytes': 10000,
           },
           'src/gcs_dep_with_output_file': {
             'bucket': '789bucket',
@@ -909,6 +911,7 @@ deps = {
             'dep_type': 'gcs',
             'sha256sum': 'abcd123',
             'output_file': 'clang-format-no-extract',
+            'size_bytes': 10000,
           },
         }"""),
                 'origin':
@@ -943,6 +946,7 @@ deps = {
     'object_name': 'path_to_file.tar.gz',
     'dep_type': 'gcs',
     'sha256sum': 'abcd123',
+    'size_bytes': 10000,
   },
 }
 """,