|
| 1 | +#!/usr/bin/env python |
| 2 | +# -*- coding:utf-8 -*- |
| 3 | +# |
| 4 | +# Author : XueWeiHan |
| 5 | + |
| 6 | +# Date : 2019-09-18 22:45 |
| 7 | +# Desc : 生成星图脚本 |
| 8 | +import time |
| 9 | +import datetime |
| 10 | +from collections import OrderedDict |
| 11 | + |
| 12 | +import pygal |
| 13 | +from pygal.style import Style |
| 14 | +import requests |
| 15 | + |
| 16 | + |
| 17 | +class GitHubRepositories(object): |
| 18 | + def __init__(self, name, token, per_page=100, max_star_limit=10000, |
| 19 | + min_star_limit=10, create_days_limit=10, mode='count', |
| 20 | + time_node_count=1): |
| 21 | + self.name = name |
| 22 | + self.token = token |
| 23 | + self.limit_remaining = 0 |
| 24 | + self.reset_limit_time = None |
| 25 | + self.per_page = per_page |
| 26 | + self.info = None |
| 27 | + self.stargazers = set() |
| 28 | + self.max_star_limit = max_star_limit |
| 29 | + self.min_star_limit = min_star_limit |
| 30 | + self.create_days_limit = create_days_limit |
| 31 | + self.star_map = OrderedDict() |
| 32 | + self.days_count = None |
| 33 | + self.mode = mode # 展示star总和('count'), 还是star的增量('change') |
| 34 | + self.time_node_count = time_node_count |
| 35 | + self.x_node_list = None # x轴为时间 |
| 36 | + self.y_node_list = None # y轴为数量 |
| 37 | + # 默认为当前时间 |
| 38 | + self.max_date_str = datetime.datetime.now().strftime('%Y-%m-%d') |
| 39 | + # 项目创建时间 |
| 40 | + self.min_date_str = None |
| 41 | + # 停止请求数据 |
| 42 | + self.end_request = False |
| 43 | + self.interval_days = None # 显示x轴的时间两个点应相隔几天 |
| 44 | + |
| 45 | + @property |
| 46 | + def headers(self): |
| 47 | + # token 授权 |
| 48 | + return { |
| 49 | + 'Authorization': 'token {}'.format(self.token) |
| 50 | + } |
| 51 | + |
| 52 | + @staticmethod |
| 53 | + def fix_time(time_str, rtype='time'): |
| 54 | + # github 返回的时间是 ISO 8601 规则的时间(0时区),下面转化成北京时间(+8时区) |
| 55 | + epoch_time = datetime.datetime.strptime(time_str, |
| 56 | + '%Y-%m-%dT%H:%M:%SZ') |
| 57 | + epoch_time = epoch_time + datetime.timedelta(hours=8) |
| 58 | + if rtype == 'time': |
| 59 | + epoch_time_str = epoch_time.date().strftime('%Y-%m-%d %H:%M:%S') |
| 60 | + else: |
| 61 | + epoch_time_str = epoch_time.date().strftime('%Y-%m-%d') |
| 62 | + return epoch_time_str |
| 63 | + |
| 64 | + def get_repo_info(self): |
| 65 | + url = 'https://api.github.com/repos/{}'.format(self.name) |
| 66 | + response = requests.get(url, headers=self.headers) |
| 67 | + if response.status_code == 200: |
| 68 | + self.limit_remaining = int(response.headers['X-RateLimit-Remaining']) |
| 69 | + self.reset_limit_time = datetime.datetime.fromtimestamp( |
| 70 | + int(response.headers['X-RateLimit-Reset'])).strftime( |
| 71 | + "%Y-%m-%d %H:%M:%S") |
| 72 | + self.info = response.json() |
| 73 | + # 项目名字标准化 |
| 74 | + self.name = self.info['full_name'] |
| 75 | + # 把info中的时间转化成北京时间 |
| 76 | + self.min_date_str = self.fix_time(self.info['created_at'], |
| 77 | + rtype='date') |
| 78 | + self.info['created_at'] = self.fix_time(self.info['created_at']) |
| 79 | + self.info['updated_at'] = self.fix_time(self.info['updated_at']) |
| 80 | + self.info['pushed_at'] = self.fix_time(self.info['pushed_at']) |
| 81 | + |
| 82 | + def generate_all_stargazers_urls(self): |
| 83 | + """ |
| 84 | + 请求所有的 stargazers api url |
| 85 | + """ |
| 86 | + all_stargazers_urls = [] |
| 87 | + stargazers_count = self.info['stargazers_count'] |
| 88 | + stargazers_url_format = 'https://api.github.com/repos/{}/stargazers?' \ |
| 89 | + 'page={}&per_page={}' |
| 90 | + page_count = int(stargazers_count / self.per_page) + 1 |
| 91 | + for i in range(1, page_count + 1): |
| 92 | + all_stargazers_urls.append(stargazers_url_format.format( |
| 93 | + self.name, i, self.per_page)) |
| 94 | + return all_stargazers_urls |
| 95 | + |
| 96 | + def parse_stargazers_data(self, stargazers_data): |
| 97 | + for fi_data in stargazers_data: |
| 98 | + # 根据 star 时间数据聚合到天 |
| 99 | + epoch_date_str = self.fix_time(fi_data['starred_at'], 'date') |
| 100 | + epoch_date = datetime.datetime.strptime(epoch_date_str, '%Y-%m-%d') |
| 101 | + max_date = datetime.datetime.strptime(self.max_date_str, '%Y-%m-%d') |
| 102 | + if epoch_date > max_date: |
| 103 | + self.end_request = True |
| 104 | + break |
| 105 | + self.stargazers.add(fi_data['user']['id']) |
| 106 | + if epoch_date_str in self.star_map: |
| 107 | + self.star_map[epoch_date_str] += 1 |
| 108 | + else: |
| 109 | + self.star_map[epoch_date_str] = 1 |
| 110 | + |
| 111 | + def get_stargazer(self, req_s, url, headers): |
| 112 | + rs = req_s.get(url, headers=headers) |
| 113 | + self.parse_stargazers_data(rs.json()) |
| 114 | + |
| 115 | + def get_all_stargazers(self): |
| 116 | + """ |
| 117 | + 请求所有的 stargazers api url |
| 118 | + """ |
| 119 | + all_stargazers_urls = self.generate_all_stargazers_urls() |
| 120 | + if self.limit_remaining < len(all_stargazers_urls): |
| 121 | + msg = '当前时次请求数量超出' |
| 122 | + print(msg) |
| 123 | + raise Exception(msg) |
| 124 | + else: |
| 125 | + self.limit_remaining -= len(all_stargazers_urls) |
| 126 | + headers = self.headers |
| 127 | + headers['Accept'] = 'application/vnd.github.v3.star+json' |
| 128 | + s = requests.session() |
| 129 | + for stargazers_url in all_stargazers_urls: |
| 130 | + if not self.end_request: |
| 131 | + print(stargazers_url, self.end_request) |
| 132 | + self.get_stargazer(s, stargazers_url, headers) |
| 133 | + # lock = threading.Lock() |
| 134 | + # threads = [threading.Thread(target=self.get_stargazer, |
| 135 | + # args=(s, stargazers_url, headers, lock)) |
| 136 | + # for stargazers_url in all_stargazers_urls] |
| 137 | + # [thread.start() for thread in threads] |
| 138 | + # [thread.join() for thread in threads] |
| 139 | + # del s |
| 140 | + print('request limit remaining: {}'.format(self.limit_remaining)) |
| 141 | + |
| 142 | + def make_time_node(self): |
| 143 | + max_date = datetime.datetime.strptime(self.max_date_str, '%Y-%m-%d') |
| 144 | + min_date = datetime.datetime.strptime(self.min_date_str, '%Y-%m-%d') |
| 145 | + date_list = [self.min_date_str] |
| 146 | + self.days_count = (max_date - min_date).days # 该项目共创建了多少天 |
| 147 | + self.interval_days = self.days_count / self.time_node_count |
| 148 | + flag_date = max_date - datetime.timedelta(days=self.interval_days) |
| 149 | + tmp_date = min_date |
| 150 | + # 生成所有的 x_node(时间节点) |
| 151 | + while tmp_date < flag_date: |
| 152 | + date_node = tmp_date + datetime.timedelta(days=self.interval_days) |
| 153 | + tmp_date = date_node |
| 154 | + date_list.append(date_node.strftime('%Y-%m-%d')) |
| 155 | + date_list.append(self.max_date_str) |
| 156 | + self.x_node_list = date_list |
| 157 | + |
| 158 | + def make_value_node(self): |
| 159 | + min_date = datetime.datetime.strptime(self.min_date_str, '%Y-%m-%d') |
| 160 | + # 初始化结果dict,防止某段时间的数为空造成异常 |
| 161 | + result_data = {date_node: 0 for date_node in self.x_node_list} |
| 162 | + for k, v in self.star_map.items(): |
| 163 | + starred_datetime = datetime.datetime.strptime(k, '%Y-%m-%d') |
| 164 | + # 把原始数据的日期对应到node轴,并把value累加(原始数据聚合到x轴的量级) |
| 165 | + date_index = int((starred_datetime - min_date).days / self.interval_days) |
| 166 | + if (starred_datetime - min_date).days % self.interval_days: |
| 167 | + date_index += 1 |
| 168 | + result_k = self.x_node_list[date_index] |
| 169 | + result_data[result_k] += v |
| 170 | + |
| 171 | + if self.mode == 'count': |
| 172 | + # 某x轴上日期时的star总数 |
| 173 | + for index, x_node in enumerate(self.x_node_list): |
| 174 | + if index != 0: |
| 175 | + result_data[x_node] += result_data[ |
| 176 | + self.x_node_list[index - 1]] |
| 177 | + |
| 178 | + self.y_node_list = [result_data[fi_x_node] for fi_x_node in |
| 179 | + self.x_node_list] |
| 180 | + |
| 181 | + def init(self): |
| 182 | + start_time = time.time() |
| 183 | + self.get_repo_info() |
| 184 | + error_msg = '项目名称:{},生成失败原因:'.format(self.name) |
| 185 | + if self.info: |
| 186 | + created_datetime = datetime.datetime.strptime( |
| 187 | + self.info['created_at'], '%Y-%m-%d %H:%M:%S') |
| 188 | + created_days = (datetime.datetime.now() - created_datetime).days |
| 189 | + if created_days < self.create_days_limit: |
| 190 | + error_msg += '不支持创建天数小于 {} 天的项目'.format(self.create_days_limit) |
| 191 | + raise Exception(error_msg) |
| 192 | + elif self.info['stargazers_count'] > self.max_star_limit: |
| 193 | + error_msg += '暂不支持 star 数量大于 {} 的项目'.format(self.max_star_limit) |
| 194 | + raise Exception(error_msg) |
| 195 | + elif self.info['stargazers_count'] < self.min_star_limit: |
| 196 | + error_msg += '暂不支持 star 数量小于 {} 的项目'.format(self.min_star_limit) |
| 197 | + raise Exception(error_msg) |
| 198 | + elif not self.limit_remaining: |
| 199 | + error_msg += '(测试阶段)请求数达到限制次数' |
| 200 | + raise Exception(error_msg) |
| 201 | + else: |
| 202 | + try: |
| 203 | + self.get_all_stargazers() |
| 204 | + self.make_time_node() |
| 205 | + self.make_value_node() |
| 206 | + except Exception as e: |
| 207 | + raise e |
| 208 | + else: |
| 209 | + error_msg += '未找到项目' |
| 210 | + raise Exception(error_msg) |
| 211 | + print('Get {} data speed time: {}'.format( |
| 212 | + self.name, time.time() - start_time)) |
| 213 | + |
| 214 | + |
| 215 | +class GenerateStarSVG(object): |
| 216 | + @staticmethod |
| 217 | + def make_svg(svg_type, x_nodes, y_nodes): |
| 218 | + # 参数分别修改:x轴描述旋转角度、圆角、颜色(Python 蓝) |
| 219 | + line_chart = pygal.Line(x_label_rotation=20, tooltip_border_radius=10, |
| 220 | + background='transparent', |
| 221 | + plot_background='transparent', |
| 222 | + style=Style(colors=('#376fa0',))) |
| 223 | + line_chart.x_labels = x_nodes |
| 224 | + line_chart.add("Star", y_nodes) |
| 225 | + if svg_type == 'uri': |
| 226 | + return line_chart.render_data_uri() |
| 227 | + elif svg_type == 'svg': |
| 228 | + line_chart.render_to_file('local.svg') |
| 229 | + |
| 230 | + |
| 231 | +def main(config, repo_name='521xueweihan/HelloGitHub'): |
| 232 | + _g = GitHubRepositories(repo_name, config['token'], |
| 233 | + per_page=config['per_page'], |
| 234 | + max_star_limit=config['max_star_limit'], |
| 235 | + min_star_limit=config['min_star_limit'], |
| 236 | + create_days_limit=config['create_days_limit']) |
| 237 | + _g.init() |
| 238 | + starchart = { |
| 239 | + 'repo_name': _g.info['full_name'], |
| 240 | + 'repo_url': _g.info['html_url'], |
| 241 | + 'repo_created_at': _g.info['created_at'], |
| 242 | + 'days_count': _g.days_count, |
| 243 | + 'star_count': len(_g.stargazers), |
| 244 | + 'y_nodes': _g.y_node_list, |
| 245 | + 'x_nodes': _g.x_node_list, |
| 246 | + 'create_time': datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S') |
| 247 | + } |
| 248 | + return starchart |
| 249 | + |
| 250 | + |
| 251 | +if __name__ == '__main__': |
| 252 | + config = { |
| 253 | + 'token': 'xxxx', |
| 254 | + 'per_page': 100, |
| 255 | + 'max_star_limit': 200000, |
| 256 | + 'min_star_limit': 10, |
| 257 | + 'create_days_limit': 10, |
| 258 | + } |
| 259 | + g = GitHubRepositories('username/project_name', config['token'], |
| 260 | + per_page=config['per_page'], |
| 261 | + max_star_limit=config['max_star_limit'], |
| 262 | + min_star_limit=config['min_star_limit'], |
| 263 | + create_days_limit=config['create_days_limit']) |
| 264 | + g.init() |
| 265 | + svg = GenerateStarSVG() |
| 266 | + svg.make_svg('svg', g.x_node_list, g.y_node_list) |
| 267 | + |
0 commit comments