上次写了个 GitLab 代码统计的脚本,但是手动统计太麻烦了于是改成了自动统计(建议丢到 Linux 上跑),并且支持钉钉群推送(需要在钉钉群加个机器人,并获取对应的 token),真不知道统计这玩意儿到底有啥用 =_=

项目地址

脚本已上传GitHub,下方为旧版脚本,新版请前往 GitHub 获取哦~

安装依赖模块

pip3 install python-dateutil

使用方法

将脚本中的第 14-17 行修改为你自己的内容,上传到服务器即可,执行方式如下

# 统计当天
python 脚本文件.py day
# 统计本周
python 脚本文件.py week
# 统计本月
python 脚本文件.py month

可以加到 crontab 任务中,实现定时自动统计推送,例:

# 代码统计
10 19 * * 1-6  bash /root/GitLab/run.sh day
40 20 * * 1-6  bash /root/GitLab/run.sh day
00 21 * * 6    bash /root/GitLab/run.sh week
40 23 1 1-12 * bash /root/GitLab/run.sh month

注意事项

Git用户名不是指GitLab用户名,而是Git提交工具配置的用户名

脚本源码

#!/usr/bin/python3
# coding=utf8
# @Author: Mario
# @Date  : 2020/12/04
# @Desc  : GitLab 按时间查看各用户代码提交量官方API版(钉钉推送版)

import sys
import json
import time
import requests
from dateutil.parser import parse

# 需修改如下四条信息
gitlab_url = "http://localhost/"  # GitLab 地址
private_token = "xxxxxxxx"  # GitLab Access Tokens(管理员权限)
report_path = "/root"  # 生成的报告文件存放目录
dingding_token = "xxxxxxxx"  # 钉钉机器人token

info_master = []
info_other = []
statistics = {}
date_key = sys.argv[1]  # 获取传入的第一个参数:week,month,day

headers = {
    'Connection': 'close',
}


# UTC时间转时间戳
def utc_time(time):
    dt = parse(time)
    return int(dt.timestamp())


# 获取 GitLab 上的所有项目
def gitlab_projects():
    project_ids = []
    page = 1
    while True:
        url = gitlab_url + "api/v4/projects/?private_token=" + private_token + "&page=" + str(page) + "&per_page=20"
        while True:
            try:
                res = requests.get(url, headers=headers, timeout=10)
                break
            except Exception as e:
                print(e)
                continue
        projects = json.loads(res.text)
        if len(projects) == 0:
            break
        else:
            for project in projects:
                print(project["namespace"]["name"] + " ID:" + str(project["id"]) + " 描述:" + project["description"])
                project_ids.append(project["id"])
            page += 1
    print("共获取到 " + str(len(project_ids)) + " 个有效项目")
    return project_ids


# 获取 GitLab 上的项目 id 中的分支
def project_branches(project_id):
    branch_names = []
    page = 1
    while True:
        url = gitlab_url + "api/v4/projects/" + str(
            project_id) + "/repository/branches?private_token=" + private_token + "&page=" + str(page) + "&per_page=20"
        while True:
            try:
                res = requests.get(url, headers=headers, timeout=10)
                break
            except Exception as e:
                print(e)
                continue
        branches = json.loads(res.text)
        '''Debug
        print(url)
        print('--' * 10)
        print(branches)
        print('*' * 10)
        '''
        if len(branches) == 0:
            break
        else:
            for branch in branches:
                branch_names.append(branch["name"])
            page += 1
    return branch_names


# 获取 GitLab 上的项目分支中的 commits,当 title 或 message 首单词为 Merge 时,表示合并操作,剔除此代码量
def project_commits(project_id, branch, start_time, end_time):
    commit_ids = []
    page = 1
    while True:
        url = gitlab_url + "api/v4/projects/" + str(
            project_id) + "/repository/commits?ref_name=" + branch + "&private_token=" + private_token + "&page=" + str(
            page) + "&per_page=20"
        while True:
            try:
                res = requests.get(url, headers=headers, timeout=10)
                break
            except Exception as e:
                print(e)
                continue
        commits = json.loads(res.text)
        if len(commits) == 0:
            break
        else:
            for commit in commits:
                if "Merge" in commit["title"] or "Merge" in commit["message"] or "合并" in commit["title"] or "合并" in \
                        commit["message"]:  # 不统计合并操作
                    continue
                elif utc_time(commit["authored_date"]) < start_time or utc_time(
                        commit["authored_date"]) > end_time:  # 不满足时间区间
                    continue
                else:
                    commit_ids.append(commit["id"])
            page += 1
    return commit_ids


# 根据 commits 的 id 获取代码量,type: 1 为主分支,2为其他分支
def commit_code(project_id, commit_id, branch_type):
    global info_master, info_other
    url = gitlab_url + "api/v4/projects/" + str(
        project_id) + "/repository/commits/" + commit_id + "?private_token=" + private_token
    while True:
        try:
            res = requests.get(url, headers=headers, timeout=10)
            break
        except Exception as e:
            print(e)
            continue
    data = json.loads(res.text)
    if branch_type == 1:
        obj = {"name": data["author_name"], "additions": data["stats"]["additions"],
               "deletions": data["stats"]["deletions"], "total": data["stats"]["total"]}  # Git工具用户名,新增代码数,删除代码数,总计代码数
        info_master.append(obj)
    elif branch_type == 2:
        obj = {"name": data["author_name"], "additions": data["stats"]["additions"],
               "deletions": data["stats"]["deletions"], "total": data["stats"]["total"]}  # Git工具用户名,新增代码数,删除代码数,总计代码数
        info_other.append(obj)
    # else:
    # do some things


# GitLab 数据查询
def gitlab_info(start_time, end_time):
    for project_id in gitlab_projects():  # 遍历所有项目ID
        for branch_name in project_branches(project_id):  # 遍历每个项目中的分支
            if branch_name == "master":  # 主分支
                for commit_id in project_commits(project_id, branch_name, start_time, end_time):  # 遍历每个分支中的 commit id
                    commit_code(project_id, commit_id, 1)  # 获取代码提交量
            else:  # 其他分支
                for commit_id in project_commits(project_id, branch_name, start_time, end_time):
                    commit_code(project_id, commit_id, 2)  # 获取代码提交量


#  统计数据处理
def gitlab_statistics_data(branch_type):
    global statistics
    statistics.clear()

    name = []  # Git工具用户名
    additions = []  # 新增代码数
    deletions = []  # 删除代码数
    total = []  # 总计代码数
    times = []  # 提交次数

    if branch_type == 1:
        info = info_master
    elif branch_type == 2:
        info = info_other
    # else
    # do some things

    for i in info:
        for key, value in i.items():
            if key == "name":
                name.append(value)
            if key == "additions":
                additions.append(value)
            if key == "deletions":
                deletions.append(value)
            if key == "total":
                total.append(value)
            times.append(1)  # 提交次数
    array = list(zip(name, additions, deletions, total, times))
    # print(array)

    # 去重累加
    for j in array:
        name = j[0]
        additions = j[1]
        deletions = j[2]
        total = j[3]
        times = j[4]
        if name in statistics.keys():
            statistics[name][0] += additions
            statistics[name][1] += deletions
            statistics[name][2] += total
            statistics[name][3] += times
        else:
            statistics.update({name: [additions, deletions, total, times]})
        # else:
        # do some things


# 代码统计内容列表处理
def gitlab_statistics_content():
    for i in statistics.keys():
        content_save("<tr><td>" + str(i) + "</td><td>" + str(statistics[i][0]) + "</td><td>" + str(
            statistics[i][1]) + "</td><td>" + str(statistics[i][2]) + "</td><td>" + str(
            statistics[i][3]) + "</td></tr>", "a")


# 保存数据到文件
def content_save(data, mode):
    file = open(report_path + "/" + file_name, mode, encoding='utf-8')
    file.write(data)
    file.close()


# 钉钉通知
def dingding(mes):
    content = {
        "msgtype": "actionCard",
        "actionCard": {
            "title": "GitLab代码提交量统计",
            "text": "![](" + imgurl + ")GitLab代码提交量统计",
            "hideAvatar": "0",
            "btnOrientation": "0",
            "singleTitle": "点击查看详情",
            "singleURL": mes
        }
    }
    json_headers = {'Content-Type': 'application/json;charset=utf-8'}
    result = requests.post("https://oapi.dingtalk.com/robot/send?access_token=" + dingding_token, json.dumps(content),
                           headers=json_headers).text
    print(result)


if __name__ == "__main__":
    print("正在统计数据,请耐心等待,这将花费不少时间~")
    print(date_key)
    now_time = time.time()  # 获取当前所在时间戳
    if date_key == "week":  # 周
        file_name = "week.html"
        imgurl = "https://ae01.alicdn.com/kf/U2a5dbe53da1a442ea8fdc54cb511dfefM.jpg"
        gitlab_info(now_time - 511200, now_time)  # 起-止时间(时间戳)
    elif date_key == "month":  # 月
        file_name = "month.html"
        imgurl = "https://ae01.alicdn.com/kf/U7f01e81b945b4069864be36843c6a0f8I.jpg"
        gitlab_info(int(utc_time(time.strftime("%Y-%m-01 00:00:00"))), now_time)
    elif date_key == "day":  # 天
        file_name = "day.html"
        imgurl = "https://ae01.alicdn.com/kf/U56b2fb24ff7b4e5fa05c5acf7d3d0318r.jpg"
        gitlab_info(int(utc_time(time.strftime("%Y-%m-%d 00:00:00"))), now_time)
    else:
        print("您的参数错误!")
        sys.exit(1)

    #  保存结果
    #  header
    content_save(
        "<!DOCTYPE html><html><head><title>GitLab代码提交量统计</title><meta charset=\"utf-8\"/><link href=\"https://cdn.jsdelivr.net/gh/Fog-Forest/cdn@2.1.6.2/markdown/markdown.css\" rel=\"stylesheet\" type=\"text/css\" /></head><body><h3>GitLab代码提交量统计(单位:行)   TAG:" + date_key + "</h3><h5>上次更新时间:" + time.strftime(
            "%Y-%m-%d %H:%M:%S",
            time.localtime()) + "</h5><br><blockquote><p>master</p></blockquote><table><thead><tr><th>Git用户名</th><th>新增代码数</th><th>删除代码数</th><th>总计代码数</th><th>提交次数</th></tr></thead><tbody>",
        "w")

    #  content
    gitlab_statistics_data(1)  # 主分支
    gitlab_statistics_content()
    content_save(
        "</tbody></table><blockquote><p>dev</p></blockquote><table><thead><tr><th>Git用户名</th><th>新增代码数</th><th>删除代码数</th><th>总计代码数</th><th>提交次数</th></tr></thead><tbody>",
        "a")

    gitlab_statistics_data(2)  # 其他分支
    gitlab_statistics_content()

    # floor
    content_save("</tbody></table></body></html>", "a")
    # dingding("http://www.baidu.com/" + date_key + ".html")  # 填写生成的 html 文件链接,不需要推送可以注释掉
    print(time.strftime("%Y-%m-%d %H:%M:%S", time.localtime()) + "   OK~")

执行结果


Never give up your dreams.