中国88个超500万人口的大中城市都在哪里？Python动态图告诉你！

import numpy as np 
import pandas as pd 
import geopandas as gpd 
import shapely 
from shapely import geometry as geo 
from shapely import wkt 
import geopandas as gpd 
import matplotlib.pyplot as plt 
import matplotlib.animation as  animation 
import contextily as ctx

import imageio
import os 
from PIL import Image

plt.rcParams['font.family'] = 'sans-serif'
plt.rcParams['font.sans-serif'] = ['SimHei']
plt.rcParams['axes.unicode_minus'] = False
plt.rcParams['animation.writer'] = 'html'
plt.rcParams['animation.embed_limit'] = 100

def rgba_to_rgb(img_rgba):
    img_rgb = Image.new("RGB", img_rgba.size, (255, 255, 255))
    img_rgb.paste(img_rgba, mask=img_rgba.split()[3]) 
    return img_rgb 

def html_to_gif(html_file, gif_file, duration=0.5):
    path = html_file.replace(".html","_frames")
    images = [os.path.join(path,x) for x in sorted(os.listdir(path))]
    frames = [imageio.imread(x) for x in images]
    if frames[0].shape[-1]==4:
        frames = [np.array(rgba_to_rgb(Image.fromarray(x))) for x in frames]
    imageio.mimsave(gif_file, frames, 'gif', duration=duration)
    return gif_file

cmap = [
'#2E91E5',
'#1CA71C',
'#DA16FF',
'#B68100',
'#EB663B',
'#00A08B',
'#FC0080',
'#6C7C32',
'#862A16',
'#620042',
'#DA60CA',
'#0D2A63']*100

def getCoords(geom):
    if isinstance(geom,geo.MultiPolygon):
        return [np.array(g.exterior) for g in geom.geoms]
    elif isinstance(geom,geo.Polygon):
        return [np.array(geom.exterior)]
    elif isinstance(geom,geo.LineString):
        return [np.array(geom)]
    elif isinstance(geom,geo.MultiLineString):
        return [np.array(x) for x in list(geom.geoms)]
    else:
        raise Exception("geom must be one of [polygon,MultiPolygon,LineString,MultiLineString]!")

#底图数据
dfprovince = gpd.read_file("./data/dfprovince.geojson").set_crs("epsg:4326").to_crs("epsg:2343")
dfnanhai = gpd.read_file("./data/dfnanhai.geojson").set_crs("epsg:4326").to_crs("epsg:2343")
dfline9 =  dfnanhai[(dfnanhai["LENGTH"]>1.0)&(dfnanhai["LENGTH"]<2.0)]

#散点数据
dfpoints = gpd.read_file("./data/china_big_cities.geojson").set_crs("epsg:4326").to_crs("epsg:2343")
dfpoints["point"] = dfpoints.representative_point()
dfpoints = dfpoints.query("population>=5000000") 

df = pd.DataFrame({"x":[pt.x for pt in dfpoints["point"]],
                  "y": [pt.y for pt in dfpoints["point"]],
                  "z":[x for x in dfpoints["population"]]})
df.index = [x for x in dfpoints["city"]] 

def bubble_map_dance(df,title = "中国超500万人口城市",
                     filename = None,
                     figsize = (8,6),dpi = 144,
                     duration = 0.5,
                     anotate_points = ["北京市","上海市","重庆市","赣州市","沈阳市"]):

    fig, ax_base =plt.subplots(figsize=figsize,dpi=dpi)
    ax_child=fig.add_axes([0.800,0.125,0.10,0.20])
    
    def plot_frame(i):

            ax_base.clear()
            ax_child.clear()

            #============================================================
            #绘制底图
            #============================================================

            #绘制省边界
            polygons = [getCoords(x) for x in dfprovince["geometry"]]
            for j,coords in enumerate(polygons):
                for x in coords:
                    poly = plt.Polygon(x, fill=True, ec = "gray", fc = "white",alpha=0.5,linewidth=.8)
                    poly_child = plt.Polygon(x, fill=True, ec = "gray", fc = "white",alpha=0.5,linewidth=.8)
                    ax_base.add_patch(poly)
                    ax_child.add_patch(poly_child )

            #绘制九段线
            coords = [getCoords(x) for x in dfline9["geometry"]]
            lines = [y for x in coords for y in x ]
            for ln in lines:
                x, y = np.transpose(ln)
                line = plt.Line2D(x,y,color="gray",linestyle="-.",linewidth=1.5)
                line_child = plt.Line2D(x,y,color="gray",linestyle="-.",linewidth=1.5)
                ax_base.add_artist(line)
                ax_child.add_artist(line_child)



            #设置spine格式
            for spine in['top','left',"bottom","right"]:
                ax_base.spines[spine].set_color("none")
                ax_child.spines[spine].set_alpha(0.5)
            ax_base.axis("off")
            

            #设置绘图范围
            bounds = dfprovince.total_bounds
            ax_base.set_xlim(bounds[0]-(bounds[2]-bounds[0])/10, bounds[2]+(bounds[2]-bounds[0])/10)
            ax_base.set_ylim(bounds[1]+(bounds[3]-bounds[1])/3.5, bounds[3]+(bounds[3]-bounds[1])/100)

            ax_child.set_xlim(bounds[2]-(bounds[2]-bounds[0])/2.5, bounds[2]-(bounds[2]-bounds[0])/20)
            ax_child.set_ylim(bounds[1]-(bounds[3]-bounds[1])/20, bounds[1]+(bounds[3]-bounds[1])/2)

            #移除坐标轴刻度
            ax_child.set_xticks([]);
            ax_child.set_yticks([]);

            #============================================================
            #绘制散点
            #============================================================

            k = i//3+1
            m = i%3
            text = "NO."+str(len(df)+1-k) 

            dfdata = df.iloc[:k,:].copy()
            dftmp = df.iloc[:k-1,:].copy()

            # 绘制散点图像
            if len(dftmp)>0:
                ax_base.scatter(dftmp["x"],dftmp["y"],s = 100*dftmp["z"]/df["z"].mean(),
                       c = (cmap*100)[0:len(dftmp)],alpha = 0.3,zorder = 3)
                ax_child.scatter(dftmp["x"],dftmp["y"],s = 100*dftmp["z"]/df["z"].mean(),
                       c = (cmap*100)[0:len(dftmp)],alpha = 0.3,zorder = 3)

                # 添加注释文字
                for i,p in enumerate(dftmp.index):
                    px,py,pz = dftmp.loc[p,["x","y","z"]].tolist() 
                    if p in anotate_points:
                        ax_base.annotate(p,xy = (px,py),  xycoords = "data",xytext = (-15,10),
                        fontsize = 10,fontweight = "bold",color = cmap[i], textcoords = "offset points")
            
            # 添加标题和排名序号
            #ax_base.set_title(title,color = "black",fontsize = 12)
            ax_base.text(0.5, 0.95, title, va="center", ha="center", 
                        size = 12,transform = ax_base.transAxes)
            ax_base.text(0.5, 0.5, text, va="center", ha="center", 
                         alpha=0.3, size = 50,transform = ax_base.transAxes)

            # 添加注意力动画
            if m==0:
                px,py,pz = dfdata["x"][[-1]],dfdata["y"][[-1]],dfdata["z"][-1]
                p = dfdata.index[-1]+":"+str(pz//10000)+"万"
                ax_base.scatter(px,py,s = 800*pz/df["z"].mean(),
                   c = cmap[len(dfdata)-1:len(dfdata)],alpha = 0.5,zorder = 4)
                ax_base.annotate(p,xy = (px,py),  xycoords = "data",
                        xytext = (-15,10),fontsize = 20,fontweight = "bold",
                        color = cmap[k-1], textcoords = "offset points",zorder = 5)
                
            if m==1:
                px,py,pz = dfdata["x"][[-1]],dfdata["y"][[-1]],dfdata["z"][-1]
                p = dfdata.index[-1]+":"+str(pz//10000)+"万"
                ax_base.scatter(px,py,s = 400*pz/df["z"].mean(),
                   c = cmap[len(dfdata)-1:len(dfdata)],alpha = 0.5,zorder = 4)
                ax_base.annotate(p,xy = (px,py),  xycoords = "data",
                        xytext = (-15,10),fontsize = 15,fontweight = "bold",
                        color = cmap[k-1], textcoords = "offset points",zorder = 5)

            if m==2:
                px,py,pz = dfdata["x"][[-1]],dfdata["y"][[-1]],dfdata["z"][-1]
                p = dfdata.index[-1]+":"+str(pz//10000)+"万"
                ax_base.scatter(px,py,s = 100*pz/df["z"].mean(),
                   c = cmap[len(dfdata)-1:len(dfdata)],alpha = 0.5,zorder = 4)
                ax_base.annotate(p,xy = (px,py),  xycoords = "data",
                        xytext = (-15,10),fontsize = 10,fontweight = "bold",
                        color = cmap[k-1], textcoords = "offset points",zorder = 5)
                
    my_animation = animation.FuncAnimation(fig,plot_frame,frames = range(0,3*len(df)),interval = int(duration*1000))
    
    if filename is None:
        try:
            from IPython.display import HTML
            HTML(my_animation.to_jshtml())
            return HTML(my_animation.to_jshtml())
        except ImportError:
            pass
    else:
        my_animation.save(filename)
        return filename

html_file = "中国超500万人口城市.html"
bubble_map_dance(df,filename = html_file)

gif_file = html_file.replace(".html",".gif")
html_to_gif(html_file,gif_file,duration=0.5)