From 1b31d8331ea3a1f848becac956ef2b6f8f583458 Mon Sep 17 00:00:00 2001
From: meejah <meejah@meejah.ca>
Date: Mon, 6 Mar 2023 19:50:51 +0000
Subject: [PATCH 1/2] Add script to create test directories & files

Copy-pasted from https://whetstone.private.storage/privatestorage/privatestoragedesktop/-/issues/761#note_27299
---
 integration/make_small_files.py | 126 ++++++++++++++++++++++++++++++++
 1 file changed, 126 insertions(+)
 create mode 100644 integration/make_small_files.py

diff --git a/integration/make_small_files.py b/integration/make_small_files.py
new file mode 100644
index 000000000..9356bdc2d
--- /dev/null
+++ b/integration/make_small_files.py
@@ -0,0 +1,126 @@
+#!/usr/bin/env python3
+import math
+import random
+from pathlib import Path
+from hashlib import sha256
+
+import click
+
+
+# Description: Creates a lot of directories and small files to
+# performance test GridSync/Magic Folder with.
+# Author: Meejah
+# Date: 2023-01-20
+
+
+_hash = sha256("a random seed".encode("utf8"))
+
+
+def random_file_segment():
+    """
+    A random filename segment for a directory or file
+    """
+    _hash.update("a".encode("utf8"))
+    digest = _hash.hexdigest()
+    return digest[:random.randrange(1, len(digest))]
+
+
+def generate_directories(base, count):
+    """
+    A generator that creates a certain number of random
+    directories-names below the base
+    """
+    # XXX would be nice to use hypothesis strategies to generate
+    # stuff, but .. that's hard?
+    for _ in range(count):
+        seg = random_file_segment()
+        yield base / seg
+
+
+def generate_filename_segments(files):
+    """
+    Generate some number of random filename segments
+    """
+    for _ in range(files):
+        yield random_file_segment()
+
+
+def generate_local_paths(output, files, directories):
+    """
+    generator for a sequence of path names
+    """
+    dir_names = generate_directories(output, directories)
+    file_names = generate_filename_segments(files)
+
+    # since we need at least one file in each directory (because we
+    # don't directly store directories) we place one of our files in
+    # each subdir
+
+    reusable_dirs = []
+
+    for d in dir_names:
+        reusable_dirs.append(d)
+        f = next(file_names)
+        d.mkdir()
+        path = d / f
+        yield path
+
+    idx = 0
+    for f in file_names:
+        idx = (idx + 1) % len(reusable_dirs)
+        yield reusable_dirs[idx] / f
+
+
+@click.command()
+@click.option(
+    "--files",
+    default=665,
+    help="Number of files to put data in"
+)
+@click.option(
+    "--directories",
+    default=237,
+    help="Number of folders to split data into"
+)
+@click.option(
+    "--output",
+    type=click.Path(exists=False, file_okay=False, dir_okay=True),
+    default="./small-test-case",
+)
+@click.option(
+    "--size",
+    default=3*1024*1024,
+)
+def small_files(files, directories, output, size):
+    """
+    small-files test-case creator
+    """
+    print(output)
+    if directories > files:
+        raise click.UsageError(
+            "Must have more files than directories"
+        )
+
+    data_per_file = math.ceil(float(size) / files)
+
+    with open("/dev/urandom", "rb") as urandom:
+
+        def generate_data():
+            # could introduce some variance...
+            for _ in range(files):
+                yield urandom.read(data_per_file)
+
+        outp = Path(output)
+        outp.mkdir()
+
+        data = generate_data()
+
+        for path in generate_local_paths(outp, files, directories):
+            print(path)
+            with path.open('wb') as output:
+                output.write(next(data))
+
+
+if __name__ == "__main__":
+    small_files()
+

From 52c19ed7a18b2b2f9b6c34bf35cdb461fc8c36d5 Mon Sep 17 00:00:00 2001
From: "Christopher R. Wood" <crwood@users.noreply.github.com>
Date: Tue, 7 Mar 2023 12:42:05 +0000
Subject: [PATCH 2/2] Add TODO about random source

From @crwood, see
https://github.com/gridsync/gridsync/pull/604#issuecomment-1456718805
---
 integration/make_small_files.py | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/integration/make_small_files.py b/integration/make_small_files.py
index 9356bdc2d..0b8cc841a 100644
--- a/integration/make_small_files.py
+++ b/integration/make_small_files.py
@@ -103,6 +103,8 @@ def small_files(files, directories, output, size):
 
     data_per_file = math.ceil(float(size) / files)
 
+    # TODO: this should use Python's os.urandom() so
+    # that Windows folks can use it too
     with open("/dev/urandom", "rb") as urandom:
 
         def generate_data():