sbdiff.py

#!/usr/bin/env python3

# SBDiff v25.02
# See https://gist.github.com/Tobiaqs/3b37a3ce35d81237b597c6a2f9ce8e3b for the current version

import os
import shlex
import subprocess
import sys

if len(sys.argv) != 4:
    print("Usage: sbdiff.py <folder> <older_tag> <newer_tag>")
    exit(1)

repo_folder = sys.argv[1]

if not os.path.isdir(repo_folder):
    print(f"Folder {repo_folder} does not exist.")
    exit(1)

tag_older = sys.argv[2]
tag_newer = sys.argv[3]

subprocess.run(shlex.split("rm -rf /tmp/_sbdiff_1 /tmp/_sbdiff_2"))

print("Copying repo...")
subprocess.run(["cp", "-r", repo_folder, "/tmp/_sbdiff_1"])

print("Cleaning repo...")
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 clean -fdx"), stdout=subprocess.DEVNULL
)

print("GC...")
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 gc --prune=now --aggressive"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)

subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 submodule deinit --all"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)

subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 reset --hard HEAD"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)

print("Fetching from remote...")
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 fetch"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)

subprocess.run(
    ["cp", "-r", "/tmp/_sbdiff_1", "/tmp/_sbdiff_2"], stdout=subprocess.DEVNULL
)


def count_files(folder):
    return int(
        subprocess.run(
            "find . -type f | grep -v ./.git/ | wc -l",
            shell=True,
            cwd=folder,
            stdout=subprocess.PIPE,
        )
        .stdout.decode()
        .strip()
    )


subprocess.run(
    shlex.split(f"git -C /tmp/_sbdiff_1 checkout -f {tag_older}"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)
if (
    subprocess.run(
        shlex.split("git -C /tmp/_sbdiff_1 status -s"), stdout=subprocess.PIPE
    )
    .stdout.decode()
    .strip()
):
    print("E: Repo is not clean")
    exit(1)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 submodule init"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 submodule update"), stdout=subprocess.DEVNULL
)

subprocess.run(
    shlex.split(f"git -C /tmp/_sbdiff_2 checkout -f {tag_newer}"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)
if (
    subprocess.run(
        shlex.split("git -C /tmp/_sbdiff_2 status -s"), stdout=subprocess.PIPE
    )
    .stdout.decode()
    .strip()
):
    print("E: Repo is not clean")
    exit(1)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_2 submodule init"),
    stdout=subprocess.DEVNULL,
    stderr=subprocess.DEVNULL,
)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_2 submodule update"), stdout=subprocess.DEVNULL
)

file_count_older = count_files("/tmp/_sbdiff_1")
file_count_newer = count_files("/tmp/_sbdiff_2")

subprocess.run(shlex.split("rm -rf /tmp/_sbdiff_1/.git"), stdout=subprocess.DEVNULL)
subprocess.run(shlex.split("git -C /tmp/_sbdiff_1 init"), stdout=subprocess.DEVNULL)
subprocess.run(shlex.split("git -C /tmp/_sbdiff_1 add ."), stdout=subprocess.DEVNULL)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_1 commit -m 1st"), stdout=subprocess.DEVNULL
)
subprocess.run(shlex.split("rm -rf /tmp/_sbdiff_2/.git"), stdout=subprocess.DEVNULL)
subprocess.run(
    shlex.split("mv /tmp/_sbdiff_1/.git /tmp/_sbdiff_2/.git"), stdout=subprocess.DEVNULL
)
subprocess.run(shlex.split("git -C /tmp/_sbdiff_2 add ."), stdout=subprocess.DEVNULL)
subprocess.run(
    shlex.split("git -C /tmp/_sbdiff_2 commit -m 2nd"), stdout=subprocess.DEVNULL
)
diff = (
    subprocess.run(
        shlex.split("git -C /tmp/_sbdiff_2 diff --raw HEAD^ HEAD"),
        stdout=subprocess.PIPE,
    )
    .stdout.decode()
    .strip()
)
subprocess.run(shlex.split("rm -rf /tmp/_sbdiff_1"), stdout=subprocess.DEVNULL)
subprocess.run(shlex.split("rm -rf /tmp/_sbdiff_2"), stdout=subprocess.DEVNULL)

modified_files = 0
added_files = 0
deleted_files = 0

for line in diff.split("\n"):
    if line:
        if line[31] == "M":
            modified_files += 1
        if line[31] == "A":
            added_files += 1
        if line[31] == "D":
            deleted_files += 1

print(f"Modified files: {modified_files}")
print(f"Added files: {added_files}")
print(f"Deleted files: {deleted_files}")
print()
print(f"Files count in {tag_older}: {file_count_older}")
print(f"Files count in {tag_newer}: {file_count_newer}")
print()
print(f"Modified percentage: {round(modified_files / min(file_count_older, file_count_newer) * 1000) / 10}%")
print(f"Added percentage: {round(added_files / min(file_count_older, file_count_newer) * 1000) / 10}%")
print(f"Deleted percentage: {round(deleted_files / min(file_count_older, file_count_newer) * 1000) / 10}%")
print(f"Added + deleted + modified percentage: {round((modified_files + added_files + deleted_files) / min(file_count_older, file_count_newer) * 1000) / 10}%")
print(f"Percentual different in file count: {round((file_count_newer - file_count_older) / file_count_older * 1000) / 10}%")