【Python统计git提交代码量】

# # 克隆仓库（如果已经克隆，则不需要这一步）
# git clone https://github.com/username/repo-name.git
 
# # 进入仓库目录
# cd repo-name
 
# # 统计代码行数（包括了所有的commit）
# git log --pretty=tformat: --numstat | awk '{add += $1; subs += $2; loc += $1 - $2} END {print "Added lines:", add, "Removed lines:", subs, "Total lines of code:", loc}'
 
# # 如果只想统计最新一次commit的代码行数
# git log -1 --pretty=tformat: --numstat | awk '{add += $1; subs += $2; loc += $1 - $2} END {print "Added lines:", add, "Removed lines:", subs, "Total lines of code:", loc}'


import subprocess
additions = 0
subtractions = 0
total = 0
# 统计所有commit的代码行数
def count_lines_of_code(repo_path, sincetime="2025-03-25", untiltime="2025-05-30"):
    global additions
    global subtractions
    global total
    additions = 0
    subtractions = 0 
    total = 0
    process = subprocess.Popen(['git', 'log', "--since={}".format(sincetime), "--until={}".format(untiltime) ,'--pretty=tformat:', '--numstat'], cwd=repo_path, stdout=subprocess.PIPE)
    output, _ = process.communicate()
    lines = output.decode('utf-8').replace("\t", " ").splitlines()
    for _ in lines:
        if _.split(" ")[0].isdigit():
            print("##A formatted string:   {}".format(_))
            additions += int(_.strip(" ").split(" ")[0]) 
            subtractions += int(_.strip(" ").split(" ")[1]) 
        else:
            print("##Ignore non formatted strings:   {}".format(_))
    total = additions - subtractions
    return additions, subtractions, total
 
# 统计最新一次commit的代码行数
def count_lines_of_code_latest(repo_path):
    global additions
    global subtractions
    global total
    additions = 0
    subtractions = 0 
    total = 0
    process = subprocess.Popen(['git', 'log', '-1', '--pretty=tformat:', '--numstat'], cwd=repo_path, stdout=subprocess.PIPE)
    output, _ = process.communicate()
    lines = output.decode('utf-8').replace("\t", " ").splitlines()
    for _ in lines:
        if _.split(" ")[0].isdigit():
            print("##A formatted string:   {}".format(_))
            additions += int(_.strip(" ").split(" ")[0]) 
            subtractions += int(_.strip(" ").split(" ")[1]) 
        else:
            print("##Ignore non formatted strings:   {}".format(_))
    total = additions - subtractions
    return additions, subtractions, total
 
# 使用方法
repo_path = r'D:\workspace\product_code'  # 替换为你的仓库路径
additions, subtractions, total = count_lines_of_code(repo_path, sincetime="2025-05-01", untiltime="2025-05-30")
print(f"Added lines: {additions}, Removed lines: {subtractions}, Total lines of code: {total}") 
 
# additions_latest, subtractions_latest, total_latest = count_lines_of_code_latest(repo_path)
# print(f"Added lines (latest commit): {additions_latest}, Removed lines (latest commit): {subtractions_latest}, Total lines of code (latest commit): {total_latest}")