Add CLI flag to calculate hashes of collection

acquire/acquire.py

@@ -44,7 +44,9 @@
     filter_out_by_path_match,
     filter_out_by_value_match,
     filter_out_huge_files,
+    md5sum,
     serialize_into_csv,
+    sha256sum,
 )
 from acquire.log import get_file_handler, reconfigure_log_file, setup_logging
 from acquire.outputs import OUTPUTS
@@ -2127,13 +2129,21 @@ def acquire_target(target: Target, args: argparse.Namespace, output_ts: str | No
         else:
             report_file_name = f"{output_path.name}.report.json"
 
+        if args.hash_collection:
+            hashes = {"md5": md5sum(output.path), "sha256": sha256sum(output.path)}
+            execution_report["hashes"] = hashes
+
         report_file_path = output_path.parent / report_file_name
         persist_execution_report(report_file_path, execution_report)
 
         files.append(report_file_path)
         log.info("Acquisition report for %s is written to %s", target, report_file_path)
 
     log.info("Output: %s", output.path)
+
+    if args.hash_collection:
+        log.info("Output MD5: %s", hashes["md5"])
+        log.info("Output SHA-256: %s", hashes["sha256"])
     return files
 
 

acquire/hashes.py

@@ -244,3 +244,17 @@ def serialize_into_csv(rows: Iterator[list], compress: bool = True) -> tuple[int
                 )
 
     return (counter, raw_buffer.getvalue())
+
+
+def md5sum(path: Path) -> str:
+    with path.open(mode="rb") as f:
+        digest = hashlib.file_digest(f, "md5")
+
+    return digest.hexdigest()
+
+
+def sha256sum(path: Path) -> str:
+    with path.open(mode="rb") as f:
+        digest = hashlib.file_digest(f, "sha256")
+
+    return digest.hexdigest()

acquire/utils.py

@@ -180,6 +180,11 @@ def create_argument_parser(profiles: dict, volatile: dict, modules: dict) -> arg
         nargs="+",
         help="upload specified files (all other acquire actions are ignored)",
     )
+    parser.add_argument(
+        "--hash-collection",
+        action=argparse.BooleanOptionalAction,
+        help="Calculate hashes of the collection",
+    )
     parser.add_argument("--no-proxy", action="store_true", help="don't autodetect proxies")
 
     parser.add_argument("-K", "--keychain-file", type=Path, help="keychain file in CSV format")