From 77b5e6762fbc13f31940680ea987019356a077ff Mon Sep 17 00:00:00 2001
From: Seth Grover <seth.d.grover@gmail.com>
Date: Thu, 31 Oct 2024 14:24:23 -0600
Subject: [PATCH] work in progress

---
 src/maltest/maltest.py | 18 ++++++++++++++----
 src/maltest/utils.py   | 10 ++++++++++
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/src/maltest/maltest.py b/src/maltest/maltest.py
index efedcab..270bc96 100755
--- a/src/maltest/maltest.py
+++ b/src/maltest/maltest.py
@@ -12,7 +12,13 @@
 import signal
 import sys
 
-from maltest.utils import MalcolmVM, MalcolmTestCollection, ShuttingDown, set_malcolm_vm_info
+from maltest.utils import (
+    MalcolmVM,
+    MalcolmTestCollection,
+    ShuttingDown,
+    set_malcolm_vm_info,
+    shakey_file_hash,
+)
 
 ###################################################################################################
 script_name = os.path.basename(__file__)
@@ -321,9 +327,13 @@ def main():
                     for pcapFile in pcaps:
                         # TODO: would it be better to use SFTP for this? or even the upload interface?
                         # TODO: Assuming the Malcolm directory like this might not be very robust
-                        malcolmVm.CopyFile(
-                            pcapFile, f'/home/{args.vmImageUsername}/Malcolm/pcap/upload/', tolerateFailure=True
-                        )
+                        pcapFileParts = os.path.splitext(pcapFile)
+                        if pcapHash := shakey_file_hash(pcapFile):
+                            malcolmVm.CopyFile(
+                                pcapFile,
+                                f'/home/{args.vmImageUsername}/Malcolm/pcap/upload/{pcapHash}{pcapFileParts[1]}',
+                                tolerateFailure=True,
+                            )
 
                 # TODO: wait until all data has been processed (no new documents are being indexed for X amount of time)
 
diff --git a/src/maltest/utils.py b/src/maltest/utils.py
index 0a69433..3a0cb96 100644
--- a/src/maltest/utils.py
+++ b/src/maltest/utils.py
@@ -2,6 +2,7 @@
 
 import ast
 import glob
+import hashlib
 import json
 import mmguero
 import os
@@ -31,6 +32,15 @@
 )
 
 
+###################################################################################################
+def shakey_file_hash(filename, digest_len=8):
+    try:
+        with open(filename, 'rb', buffering=0) as f:
+            return hashlib.file_digest(f, 'shake_256').hexdigest(digest_len)
+    except:
+        return None
+
+
 ###################################################################################################
 def set_malcolm_vm_info(info):
     global MalcolmVmInfo