Python爬虫 抓肺炎疫情实时数据
Talent、茂ღ茂 人气:1数据下载
网上一搜,首先搜到的是腾讯的疫情实时追踪,那就用这个数据源吧。
有了网址怎么抓数据呢?这里,可以从纷乱中找到最靠谱的下载方式。我习惯用FireFox浏览器,下面的讲解就以FireFox为例(其他浏览器基本类似)。
- 打开菜单,点击“Web开发者”,在递进菜单中选择"网络":
- 刷新页面,我们很快就能发现,应答类型为json格式的这个请求,最有可能包含我们需要的数据了:
- 深入分析,我们就得到了url地址、请求方法、参数、应答格式等信息。查询参数中,callback是回调函数名,我们可以尝试置空,_应该是以毫秒为单位的当前时间戳。有了这些信息,分分钟就可以抓到数据了。我们先在IDLE中以交互方式抓一下看看效果:
>> import time, json, requests >> url = 'https://view.inews.qq.com/g2/getOnsInfo?name=disease_h5&callback=&_=%d'%int(time.time()*1000) >> data = json.loads(requests.get(url=url).json()['data'])
- 只要两行代码,就可以抓到数据了。怎么样,是不是超级简单?我们在来看看数据结构:
>>> data.keys() dict_keys(['chinaTotal', 'chinaAdd', 'lastUpdateTime', 'areaTree', 'chinaDayList', 'chinaDayAddList', 'isShowAdd']) >>> d = data['areaTree'][0]['children'] >>> len(d) 34 >>> [item['name'] for item in d] ['湖北', '浙江', '广东', '河南', '湖南', '江西', '安徽', '重庆', '山东', '江苏', '四川', '上海', '北京', '福建', '黑龙江', '广西', '陕西', '河北', '云南', '海南', '山西', '辽宁', '天津', '贵州', '甘肃', '吉林', '内蒙古', '宁夏', '新疆', '香港', '青海', '台湾', '澳门', '西藏'] >>> d[0]['children'] [{'name': '武汉', 'total': {'confirm': 10117, 'suspect': 0, 'dead': 414, 'heal': 431}, 'today': {'confirm': 1766, 'suspect': 0, 'dead': 52, 'heal': 58}}, {'name': '孝感', 'total': {'confirm': 1886, 'suspect': 0, 'dead': 25, 'heal': 9}, 'today': {'confirm': 424, 'suspect': 0, 'dead': 7, 'heal': 3}}, {'name': '黄冈', 'total': {'confirm': 1807, 'suspect': 0, 'dead': 29, 'heal': 60}, 'today': {'confirm': 162, 'suspect': 0, 'dead': 4, 'heal': 8}}, {'name': '随州', 'total': {'confirm': 834, 'suspect': 0, 'dead': 9, 'heal': 9}, 'today': {'confirm': 128, 'suspect': 0, 'dead': 1, 'heal': 0}}, {'name': '荆州', 'total': {'confirm': 801, 'suspect': 0, 'dead': 10, 'heal': 18}, 'today': {'confirm': 88, 'suspect': 0, 'dead': 1, 'heal': 6}}, {'name': '襄阳', 'total': {'confirm': 787, 'suspect': 0, 'dead': 2, 'heal': 10}, 'today': {'confirm': 52, 'suspect': 0, 'dead': 0, 'heal': 3}}, {'name': '黄石', 'total': {'confirm': 566, 'suspect': 0, 'dead': 2, 'heal': 25}, 'today': {'confirm': 57, 'suspect': 0, 'dead': 0, 'heal': 7}}, {'name': '宜昌', 'total': {'confirm': 563, 'suspect': 0, 'dead': 6, 'heal': 9}, 'today': {'confirm': 67, 'suspect': 0, 'dead': 2, 'heal': 0}}, {'name': '荆门', 'total': {'confirm': 508, 'suspect': 0, 'dead': 17, 'heal': 21}, 'today': {'confirm': 86, 'suspect': 0, 'dead': 1, 'heal': 5}}, {'name': '鄂州', 'total': {'confirm': 423, 'suspect': 0, 'dead': 18, 'heal': 8}, 'today': {'confirm': 41, 'suspect': 0, 'dead': 0, 'heal': 2}}, {'name': '咸宁', 'total': {'confirm': 399, 'suspect': 0, 'dead': 1, 'heal': 3}, 'today': {'confirm': 15, 'suspect': 0, 'dead': 1, 'heal': 1}}, {'name': '十堰', 'total': {'confirm': 353, 'suspect': 0, 'dead': 0, 'heal': 14}, 'today': {'confirm': 35, 'suspect': 0, 'dead': 0, 'heal': 5}}, {'name': '仙桃', 'total': {'confirm': 265, 'suspect': 0, 'dead': 5, 'heal': 0}, 'today': {'confirm': 40, 'suspect': 0, 'dead': 1, 'heal': 0}}, {'name': '恩施州', 'total': {'confirm': 144, 'suspect': 0, 'dead': 0, 'heal': 10}, 'today': {'confirm': 6, 'suspect': 0, 'dead': 0, 'heal': 4}}, {'name': '天门', 'total': {'confirm': 138, 'suspect': 0, 'dead': 10, 'heal': 1}, 'today': {'confirm': 10, 'suspect': 0, 'dead': 0, 'heal': 1}}, {'name': '潜江', 'total': {'confirm': 64, 'suspect': 0, 'dead': 1, 'heal': 0}, 'today': {'confirm': 10, 'suspect': 0, 'dead': 0, 'heal': 0}}, {'name': '神农架', 'total': {'confirm': 10, 'suspect': 0, 'dead': 0, 'heal': 2}, 'today': {'confirm': 0, 'suspect': 0, 'dead': 0, 'heal': 0}}, {'name': '地区待确认', 'total': {'confirm': 0, 'suspect': 0, 'dead': 0, 'heal': 3}, 'today': {'confirm': 0, 'suspect': 0, 'dead': 0, 'heal': 0}}]
数据处理
- 以省为单位画疫情图,我们只需要统计同属一个省的所有地市的确诊数据即可。最终的数据抓取代码如下:
import time, json, requests def catch_distribution(): """抓取行政区域确诊分布数据""" data = {} url = 'https://view.inews.qq.com/g2/getOnsInfo?name=disease_h5&callback=&_=%d'%int(time.time()*1000) for item in json.loads(requests.get(url=url).json()['data'])['areaTree'][0]['children']: if item['name'] not in data: data.update({item['name']:0}) for city_data in item['children']: data[item['name']] += int(city_data['total']['confirm']) return data
数据可视化
数据可视化,我习惯使用matplotlib模块。matplotlib有很多扩展工具包(toolkits),比如,画3D需要mplot3d工具包,画地图的话,则需要basemap工具包,以及处理地图投影的pyproj模块。另外画海陆分界线、国界线、行政分界线等还需要shape数据。所需模块请自行安装,shape文件可以从这里下载,绘图用到的矢量字库可以从自己的电脑上随便找一个(我用的是simsun.ttf)。我的主程序是2019nCoV.py,shape文件下载下来之后,是这样保存的:
- 以下为全部代码,除了疫情地图,还包括了全国每日武汉肺炎确诊数据的下载和可视化。
# -*- coding: utf-8 -*- import time import json import requests from datetime import datetime import numpy as np import matplotlib import matplotlib.figure from matplotlib.font_manager import FontProperties from matplotlib.backends.backend_agg import FigureCanvasAgg from matplotlib.patches import Polygon from matplotlib.collections import PatchCollection from mpl_toolkits.basemap import Basemap import matplotlib.pyplot as plt import matplotlib.dates as mdates plt.rcParams['font.sans-serif'] = ['FangSong'] # 设置默认字体 plt.rcParams['axes.unicode_minus'] = False # 解决保存图像时'-'显示为方块的问题 def catch_daily(): """抓取每日确诊和死亡数据""" url = 'https://view.inews.qq.com/g2/getOnsInfo?name=wuwei_ww_cn_day_counts&callback=&_=%d'%int(time.time()*1000) data = json.loads(requests.get(url=url).json()['data']) data.sort(key=lambda x:x['date']) date_list = list() # 日期 confirm_list = list() # 确诊 suspect_list = list() # 疑似 dead_list = list() # 死亡 heal_list = list() # 治愈 for item in data: month, day = item['date'].split('/') date_list.append(datetime.strptime('2020-%s-%s'%(month, day), '%Y-%m-%d')) confirm_list.append(int(item['confirm'])) suspect_list.append(int(item['suspect'])) dead_list.append(int(item['dead'])) heal_list.append(int(item['heal'])) return date_list, confirm_list, suspect_list, dead_list, heal_list def catch_distribution(): """抓取行政区域确诊分布数据""" data = {} url = 'https://view.inews.qq.com/g2/getOnsInfo?name=disease_h5&callback=&_=%d'%int(time.time()*1000) for item in json.loads(requests.get(url=url).json()['data'])['areaTree'][0]['children']: if item['name'] not in data: data.update({item['name']:0}) for city_data in item['children']: data[item['name']] += int(city_data['total']['confirm']) return data def plot_daily(): """绘制每日确诊和死亡数据""" date_list, confirm_list, suspect_list, dead_list, heal_list = catch_daily() # 获取数据 plt.figure('2019-nCoV疫情统计图表', facecolor='#f4f4f4', figsize=(10, 8)) plt.title('2019-nCoV疫情曲线', fontsize=20) plt.plot(date_list, confirm_list, label='确诊') plt.plot(date_list, suspect_list, label='疑似') plt.plot(date_list, dead_list, label='死亡') plt.plot(date_list, heal_list, label='治愈') plt.gca().xaxis.set_major_formatter(mdates.DateFormatter('%m-%d')) # 格式化时间轴标注 plt.gcf().autofmt_xdate() # 优化标注(自动倾斜) plt.grid(linestyle=':') # 显示网格 plt.legend(loc='best') # 显示图例 plt.savefig('2019-nCoV疫情曲线.png') # 保存为文件 #plt.show() def plot_distribution(): """绘制行政区域确诊分布数据""" data = catch_distribution() font_14 = FontProperties(fname='res/simsun.ttf', size=14) font_11 = FontProperties(fname='res/simsun.ttf', size=11) width = 1600 height = 800 rect = [0.1, 0.12, 0.8, 0.8] lat_min = 0 lat_max = 60 lon_min = 77 lon_max = 140 '''全球等经纬投影模式使用以下设置,否则使用上面的对应设置 width = 3000 height = 1500 rect = [0, 0, 1, 1] lat_min = -90 lat_max = 90 lon_min = 0 lon_max = 360 ''' handles = [ matplotlib.patches.Patch(color='#ffaa85', alpha=1, linewidth=0), matplotlib.patches.Patch(color='#ff7b69', alpha=1, linewidth=0), matplotlib.patches.Patch(color='#bf2121', alpha=1, linewidth=0), matplotlib.patches.Patch(color='#7f1818', alpha=1, linewidth=0), ] labels = [ '1-9人', '10-99人', '100-999人', '>1000人'] provincePos = { "辽宁省":[121.7,40.9], "吉林省":[124.5,43.5], "黑龙江省":[125.6,46.5], "北京市":[116.0,39.9], "天津市":[117.0,38.7], "内蒙古自治区":[110.0,41.5], "宁夏回族自治区":[105.2,37.0], "山西省":[111.0,37.0], "河北省":[114.0,37.8], "山东省":[116.5,36.0], "河南省":[111.8,33.5], "陕西省":[107.5,33.5], "湖北省":[111.0,30.5], "江苏省":[119.2,32.5], "安徽省":[115.5,31.8], "上海市":[121.0,31.0], "湖南省":[110.3,27.0], "江西省":[114.0,27.0], "浙江省":[118.8,28.5], "福建省":[116.2,25.5], "广东省":[113.2,23.1], "台湾省":[120.5,23.5], "海南省":[108.0,19.0], "广西壮族自治区":[107.3,23.0], "重庆市":[106.5,29.5], "云南省":[101.0,24.0], "贵州省":[106.0,26.5], "四川省":[102.0,30.5], "甘肃省":[103.0,35.0], "青海省":[95.0,35.0], "新疆维吾尔自治区":[85.5,42.5], "西藏自治区":[85.0,31.5], "香港特别行政区":[115.1,21.2], "澳门特别行政区":[112.5,21.2] } fig = matplotlib.figure.Figure() fig.set_size_inches(width/100, height/100) # 设置绘图板尺寸 axes = fig.add_axes(rect) # 兰博托投影模式,局部 m = Basemap(projection='lcc', llcrnrlon=77, llcrnrlat=14, urcrnrlon=140, urcrnrlat=51, lat_1=33, lat_2=45, lon_0=100, ax=axes) # 兰博托投影模式,全图 #m = Basemap(projection='lcc', llcrnrlon=80, llcrnrlat=0, urcrnrlon=140, urcrnrlat=51, lat_1=33, lat_2=45, lon_0=100, ax=axes) # 圆柱投影模式,局部 #m = Basemap(llcrnrlon=lon_min, urcrnrlon=lon_max, llcrnrlat=lat_min, urcrnrlat=lat_max, resolution='l', ax=axes) # 正射投影模式 #m = Basemap(projection='ortho', lat_0=36, lon_0=102, resolution='l', ax=axes) # 全球等经纬投影模式, #m = Basemap(llcrnrlon=lon_min, urcrnrlon=lon_max, llcrnrlat=lat_min, urcrnrlat=lat_max, resolution='l', ax=axes) #m.etopo() m.readshapefile('res/china-shapefiles-master/china', 'province', drawbounds=True) m.readshapefile('res/china-shapefiles-master/china_nine_dotted_line', 'section', drawbounds=True) m.drawcoastlines(color='black') # 洲际线 m.drawcountries(color='black') # 国界线 m.drawparallels(np.arange(lat_min,lat_max,10), labels=[1,0,0,0]) #画经度线 m.drawmeridians(np.arange(lon_min,lon_max,10), labels=[0,0,0,1]) #画纬度线 pset = set() for info, shape in zip(m.province_info, m.province): pname = info['OWNER'].strip('\x00') fcname = info['FCNAME'].strip('\x00') if pname != fcname: # 不绘制海岛 continue for key in data.keys(): if key in pname: if data[key] == 0: color = '#f0f0f0' elif data[key] < 10: color = '#ffaa85' elif data[key] <100: color = '#ff7b69' elif data[key] < 1000: color = '#bf2121' else: color = '#7f1818' break poly = Polygon(shape, facecolor=color, edgecolor=color) axes.add_patch(poly) pos = provincePos[pname] text = pname.replace("自治区", "").replace("特别行政区", "").replace("壮族", "").replace("维吾尔", "").replace("回族", "").replace("省", "").replace("市", "") if text not in pset: x, y = m(pos[0], pos[1]) axes.text(x, y, text, fontproperties=font_11, color='#00FFFF') pset.add(text) axes.legend(handles, labels, bbox_to_anchor=(0.5, -0.11), loc='lower center', ncol=4, prop=font_14) axes.set_title("2019-nCoV疫情地图", fontproperties=font_14) FigureCanvasAgg(fig) fig.savefig('2019-nCoV疫情地图.png') if __name__ == '__main__': plot_daily() plot_distribution()
nCoV图表
2019-nCoV疫情曲线:
2019-nCoV疫情地图(兰勃托投影):
2019-nCoV疫情地图(圆柱投影):
2019-nCoV疫情地图(正射投影):
2019-nCoV疫情地图(全球等经纬投影模式):
加载全部内容