From 3a1bd1185f87cf133321a90b0c0616bf68cc16c6 Mon Sep 17 00:00:00 2001
From: Tomasz Sobczyk <tomasz.sobczyk1997@gmail.com>
Date: Sat, 5 Dec 2020 15:00:02 +0100
Subject: [PATCH] Add binpack coarse shuffle tool.

---
 script/shuffle_binpack.py | 69 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 69 insertions(+)
 create mode 100644 script/shuffle_binpack.py

diff --git a/script/shuffle_binpack.py b/script/shuffle_binpack.py
new file mode 100644
index 00000000..409d4907
--- /dev/null
+++ b/script/shuffle_binpack.py
@@ -0,0 +1,69 @@
+import struct
+import sys
+import os
+import random
+from pathlib import Path
+
+def index_binpack(file):
+    print('Indexing...')
+    index = []
+    offset = 0
+    report_every = 100
+    prev_mib = -report_every
+    while file.peek():
+        chunk_header = file.read(8)
+        assert chunk_header[0:4] == b'BINP'
+        size = struct.unpack('<I', chunk_header[4:])[0]
+        file.seek(size, os.SEEK_CUR)
+        index.append((offset, size + 8))
+        offset += size + 8
+
+        mib = offset // 1024 // 1024
+        if mib // 100 != prev_mib // 100:
+            print('Indexed {} MiB'.format(mib))
+            prev_mib = mib
+
+    return index
+
+def copy_binpack_indexed(in_file, index, out_file):
+    print('Copying...')
+    total_size = 0
+    report_every = 100
+    prev_mib = -report_every
+    for offset, size in index:
+        in_file.seek(offset, os.SEEK_SET)
+        data = in_file.read(size)
+        assert len(data) == size
+        out_file.write(data)
+
+        total_size += size
+        mib = total_size // 1024 // 1024
+        if mib // 100 != prev_mib // 100:
+            print('Copied {} MiB'.format(mib))
+            prev_mib = mib
+
+def main():
+    if len(sys.argv) < 3:
+        print('Usage: python shuffle_binpack.py infile outfile')
+        return
+
+    in_filename = sys.argv[1]
+    out_filename = sys.argv[2]
+
+    if (Path(out_filename).exists()):
+        print('Output path already exists. Please specify a path to a file that does not exist.')
+        return
+
+    in_file = open(in_filename, 'rb')
+    out_file = open(out_filename, 'wb')
+
+    index = index_binpack(in_file)
+    print('Shuffling...')
+    random.shuffle(index)
+
+    copy_binpack_indexed(in_file, index, out_file)
+
+    in_file.close()
+    out_file.close()
+
+main()