fix oracle and bigquery checksums

Author: vvkh

data_diff/databases/bigquery.py

@@ -37,7 +37,7 @@
     apply_query,
     QueryResult,
     CHECKSUM_OFFSET,
-    CHECKSUM_HEXDIGITS
+    CHECKSUM_HEXDIGITS, MD5_HEXDIGITS
 )
 from data_diff.databases.base import TIMESTAMP_PRECISION_POS, ThreadLocalInterpreter, Mixin_RandomSample
 
@@ -64,7 +64,7 @@ def import_bigquery_service_account_impersonation():
 @attrs.define(frozen=False)
 class Mixin_MD5(AbstractMixin_MD5):
     def md5_as_int(self, s: str) -> str:
-        return f"cast(cast( ('0x' || substr(TO_HEX(md5({s})), {CHECKSUM_HEXDIGITS})) as int64) as numeric) - {CHECKSUM_OFFSET}"
+        return f"cast(cast( ('0x' || substr(TO_HEX(md5({s})), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS})) as int64) as numeric) - {CHECKSUM_OFFSET}"
 
 
 @attrs.define(frozen=False)

data_diff/databases/oracle.py

@@ -28,7 +28,7 @@
     QueryError,
     Mixin_RandomSample,
     CHECKSUM_OFFSET,
-    CHECKSUM_HEXDIGITS
+    CHECKSUM_HEXDIGITS, MD5_HEXDIGITS
 )
 from data_diff.databases.base import TIMESTAMP_PRECISION_POS
 
@@ -47,7 +47,7 @@ class Mixin_MD5(AbstractMixin_MD5):
     def md5_as_int(self, s: str) -> str:
         # standard_hash is faster than DBMS_CRYPTO.Hash
         # TODO: Find a way to use UTL_RAW.CAST_TO_BINARY_INTEGER ?
-        return f"to_number(substr(standard_hash({s}, 'MD5'), {CHECKSUM_HEXDIGITS}), 'xxxxxxxxxxxxxxx') - {CHECKSUM_OFFSET}"
+        return f"to_number(substr(standard_hash({s}, 'MD5'), {1+MD5_HEXDIGITS-CHECKSUM_HEXDIGITS}), 'xxxxxxxxxxxxxxx') - {CHECKSUM_OFFSET}"
 
 
 @attrs.define(frozen=False)