From cdcf35642435b08e4df7e0d771dd8906230c5ca3 Mon Sep 17 00:00:00 2001
From: Bert Blommers <bblommers@users.noreply.github.com>
Date: Wed, 19 Jul 2023 09:36:38 +0000
Subject: [PATCH] S3: Optional support for CRC32C (#6534)

---
 moto/s3/models.py              |  8 ++++++++
 moto/s3/utils.py               | 10 +++++++++-
 setup.cfg                      |  4 ++++
 tests/test_s3/test_s3_utils.py |  8 +++++++-
 4 files changed, 28 insertions(+), 2 deletions(-)

diff --git a/moto/s3/models.py b/moto/s3/models.py
index 318978927..061ecdd61 100644
--- a/moto/s3/models.py
+++ b/moto/s3/models.py
@@ -1523,12 +1523,20 @@ class S3Backend(BaseBackend, CloudWatchMetricProvider):
 
     Note that this only works if the environment variable is set **before** the mock is initialized.
 
+    ------------------------------------
+
     When using the MultiPart-API manually, the minimum part size is 5MB, just as with AWS. Use the following environment variable to lower this:
 
     .. sourcecode:: bash
 
         S3_UPLOAD_PART_MIN_SIZE=256
 
+    ------------------------------------
+
+    Install `moto[s3crc32c]` if you use the CRC32C algorithm, and absolutely need the correct value. Alternatively, you can install the `crc32c` dependency manually.
+
+    If this dependency is not installed, Moto will fall-back to the CRC32-computation when computing checksums.
+
     """
 
     def __init__(self, region_name: str, account_id: str):
diff --git a/moto/s3/utils.py b/moto/s3/utils.py
index fafabdcd1..4330111d7 100644
--- a/moto/s3/utils.py
+++ b/moto/s3/utils.py
@@ -199,7 +199,15 @@ class _VersionedKeyStore(dict):  # type: ignore
 def compute_checksum(body: bytes, algorithm: str) -> bytes:
     if algorithm == "SHA1":
         hashed_body = _hash(hashlib.sha1, (body,))
-    elif algorithm == "CRC32" or algorithm == "CRC32C":
+    elif algorithm == "CRC32C":
+        try:
+            import crc32c
+
+            hashed_body = crc32c.crc32c(body).to_bytes(4, "big")
+        except:  # noqa: E722 Do not use bare except
+            # Optional library Can't be found - just revert to CRC32
+            hashed_body = binascii.crc32(body).to_bytes(4, "big")
+    elif algorithm == "CRC32":
         hashed_body = binascii.crc32(body).to_bytes(4, "big")
     else:
         hashed_body = _hash(hashlib.sha256, (body,))
diff --git a/setup.cfg b/setup.cfg
index b90875a0c..344339d02 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -184,6 +184,10 @@ route53resolver = sshpubkeys>=3.1.0
 s3 =
     PyYAML>=5.1
     py-partiql-parser==0.3.3
+s3crc32c =
+    PyYAML>=5.1
+    py-partiql-parser==0.3.3
+    crc32c
 s3control =
 sagemaker =
 sdb =
diff --git a/tests/test_s3/test_s3_utils.py b/tests/test_s3/test_s3_utils.py
index 5edd5a24b..529977cb9 100644
--- a/tests/test_s3/test_s3_utils.py
+++ b/tests/test_s3/test_s3_utils.py
@@ -139,7 +139,13 @@ def test_checksum_crc32():
 
 
 def test_checksum_crc32c():
-    compute_checksum(b"somedata", "CRC32C").should.equal(b"Uwy90A==")
+    try:
+        import crc32c  # noqa # pylint: disable=unused-import
+
+        compute_checksum(b"somedata", "CRC32C").should.equal(b"dB9qBQ==")
+    except:  # noqa: E722 Do not use bare except
+        # Optional library Can't be found - just revert to CRC32
+        compute_checksum(b"somedata", "CRC32C").should.equal(b"Uwy90A==")
 
 
 def test_cors_utils():