博客

技巧篇：常用的python代码汇总

模块1：读写excel文件

from datetime import datetime
import odps
import xlwt
import os
from odps import DataFrame
import pandas as pd
import xlrd
import numpy as np
from collections import defaultdict
from collections import Counter


# 写入工作簿
def write_imf(fl_save_path, data):
    wb = xlwt.Workbook(encoding='utf-8')  # 不写encoding会出现编码错误
    sh = wb.add_sheet(u'data', cell_overwrite_ok=True)

    # 表头部分,单独写
    colnames = data.columns.values
    for i in range(0, data.shape[1]):
        sh.write(0, i, colnames[i])
    # 表内容,循环写入,好像没简便的方法

    for i in range(1, len(data) + 1):
        for j in range(0, data.shape[1]):
            value = data.iloc[i - 1, j]
            # print(value)
            # 这里的坑特别多！！！数据读进来之后就成numpy.xxx64的类型了，在dataframe的时候就需要统一干掉！
            try:
                value.dtype
                if value.dtype == 'int64':
                    value = int(value)
                # print('value is:%d,type is:%s'%(value,type(value)))
                if value.dtype == 'float64':
                    value = float(value)
                    # print('value is:%d,type is:%s' % (value, type(value)))
            except(RuntimeError, TypeError, NameError, ValueError, AttributeError):
                pass
            sh.write(i, j, value)

    wb.save(fl_save_path)
    print('congratulation save successful!')

def save_pd_to_csv(fl_save_path, data):
    try:
        # 直接转csv不加encoding，中文会乱码
        data.to_csv(fl_save_path, encoding="utf_8_sig", header=True, index=False)  # 存储
        return True
    except:
        return False

def get_excel_content(file_path):
    # 获取excel内的SQL语句，需要通过xlrd获取workbook中的SQL内容，或者读txt，后续改为配置文件
    wb = xlrd.open_workbook(file_path, encoding_override='utf-8')
    sht = wb.sheet_by_index(0)  # 默认第一个工作表
    # print(sht.name)
    wb_cont_imf = []
    nrows = sht.nrows  # 行数
    wb_cont_imf = [sht.row_values(i) for i in range(0, nrows)]  # 第一个工作表内容按行循环写入
    df = pd.DataFrame(wb_cont_imf[1:], columns=wb_cont_imf[0])
    return df

模块2：获取各种时间

# 获取年月第一天最后一天
def getMonthFirstDayAndLastDay(year=None, month=None):
    """
    :param year: 年份，默认是本年，可传int或str类型
    :param month: 月份，默认是本月，可传int或str类型
    :return: firstDay: 当月的第一天，datetime.date类型
              lastDay: 当月的最后一天，datetime.date类型
    """
    if year:
        year = int(year)
    else:
        year = datetime.date.today().year
    if month:
        month = int(month)
    else:
        month = datetime.date.today().month

    # 获取当月第一天的星期和当月的总天数
    firstDayWeekDay, monthRange = calendar.monthrange(year, month)

    # 获取当月的第一天
    firstDay = datetime.date(year=year, month=month, day=1)
    lastDay = datetime.date(year=year, month=month, day=monthRange)

    # return firstDay, lastDay
    return lastDay

模块3：pd中的dataframe转png

# dataframe2png
def render_mpl_table(data, col_width=5.0, row_height=0.625, font_size=1,
                     header_color='#40466e', row_colors=['#f1f1f2', 'w'], edge_color='w',
                     bbox=[0, 0, 1, 1], header_columns=0,
                     ax=None,**kwargs):
    if ax is None:
        # size = (np.array(data.shape[::-1]) + np.array([0, 1])) * np.array([col_width, row_height])
        # fig, ax = plt.subplots(figsize=size)
        fig, ax = plt.subplots() # 创建一个空的绘图区
        # 衍生知识点，服务器上安装中文字体
        plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
        # plt.rcParams['font.sans-serif'] = ['WenQuanYi Zen Hei Mono']
        plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
        plt.style.use('ggplot')
        ax.axis('off')

    mpl_table = ax.table(cellText=data.values, bbox=bbox, colLabels=data.columns, **kwargs)

    mpl_table.auto_set_font_size(False)
    mpl_table.set_fontsize(font_size)

    for k, cell in six.iteritems(mpl_table._cells):
        cell.set_edgecolor(edge_color)
        nrow = k[0]
        ncol = k[1]
        # 设置表格底色
        if nrow == 0 or ncol < header_columns:
            cell.set_text_props(weight='bold', color='w')
            cell.set_facecolor(header_color)
        else:
            cell.set_facecolor(row_colors[k[0] % len(row_colors)])

    # # 对当日异常数据为0的部分，着重体现
    # row_num = []
    # for k, cell in mpl_table._cells.items():
    #     nrow = k[0]
    #     ncol = k[1]
    #     val = cell.get_text().get_text()
    #     if nrow > 0 and ncol == 2 and val != '0':
    #         row_num.append(nrow)
    # for k, cell in six.iteritems(mpl_table._cells):
    #     nrow = k[0]
    #     # 设置表格底色
    #     if nrow in row_num:
    #         cell.set_facecolor('gold')

    # 保留原图的设置
    # fig.set_size_inches(width/100.0,height/100.0)#输出width*height像素
    plt.gca().xaxis.set_major_locator(plt.NullLocator())
    plt.gca().yaxis.set_major_locator(plt.NullLocator())
    plt.subplots_adjust(top=1, bottom=0, left=0, right=1, hspace=0, wspace=0)
    plt.margins(0, 0)

    return ax

模块4：绘制词云

#!/user/bin/python
# -*- coding:utf-8 -*-
_author_ = 'xisuo'

import datetime
import calendar
import xlwt
import os
import pandas as pd
import xlrd
import openpyxl
import numpy as np
from collections import defaultdict
import platform
from wordcloud import WordCloud,STOPWORDS
import matplotlib.pyplot as plt
from PIL import Image


def create_wordcloud(docs=None,imgs=None,filename=None):
    '''
    :param docs:读入词汇txt，尽量不重复
    :param imgs: 读入想要生成的图形，网上随便找
    :param filename: 保存图片文件名
    :return:
    '''
    # Read the whole text.
    text = open(os.path.join(current_file, docs)).read()
    alice_mask = np.array(Image.open(os.path.join(current_file, imgs)))
    print(font_path)
    wc = WordCloud(background_color="white",
                   max_words=2000,
                   font_path=font_path,  # 设置字体格式，如不设置显示不了中文
                   mask=alice_mask,
                   stopwords=STOPWORDS.add("said")
                   )
    # generate word cloud
    wc.generate(text)
    # store to file
    if filename is None:filename="词云结果.png"
    wc.to_file(os.path.join(current_file, filename))

def main():
    docs='demo.txt'    #读入的文本
    imgs="eg.jpg"     #需要绘制的图像
    filename='res_eg.png'     #保存图片文件名
    create_wordcloud(docs=docs,imgs=imgs,filename=filename)
    print('create wordcloud successful')

if __name__ == '__main__':
    start_time = datetime.datetime.now()
    print('start running program at:%s' % start_time)

    systemp_type = platform.system()
    if (systemp_type == 'Windows'):
        plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
        plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
        font_path='simfang.ttf'
        try:
            current_path = os.getcwd()
        except:
            current_path = os.path.dirname(__file__)
        current_file = os.path.join(current_path, 'docs')
        current_file = current_path
    elif (systemp_type == 'Linux'):
        font_path = 'Arial Unicode MS.ttf'
        plt.rcParams['font.family'] = ['Arial Unicode MS']  # 用来正常显示中文标签
        plt.rcParams['axes.unicode_minus'] = False  # 用来正常显示负号
        current_file = '/home/xisuo/mhc_work/docs/'  # 服务器上的路径
    else:
        quit()
    if not os.path.exists(current_file):
        os.mkdir(current_file)
        print('目录中部存在docs文件夹，完成新文件夹创建过程。')
    print('当前操作系统：%s，文件存储路径为：%s' % (systemp_type, current_file))

    main()

    end_time = datetime.datetime.now()
    tt = end_time - start_timepython
    print('ending time:%s', end_time)
    print('this analysis total spend time:%s' % tt.seconds)

模块5：下载ppt素材

#!/user/bin/python
#-*- coding:utf-8 -*-
_author_ = 'xisuo'

import urllib.request
import requests
from bs4 import BeautifulSoup
from lxml import etree
import os

url='http://www.pptschool.com/1491.html'

response=requests.get(url).text
# soup=BeautifulSoup(response,'lxml')
# cont=soup.find('article', class_='article-content')

html=etree.HTML(response)

src_list=html.xpath('//div/article/p/img/@src')

current_path=os.path.dirname(__file__)
save_path=os.path.join(current_path,'ppt_img')

if os.path.exists(save_path):
    pass
else:
    os.mkdir(save_path)
    print('img folder create successful')

i=1
for src in src_list:
    save_img_path=os.path.join(save_path,'%d.jpg'%i)
    try:
        with open(save_img_path,'wb') as f:
            f.write(urllib.request.urlopen(src).read())
            f.close()
        i=i+1
        print('save true')
    except Exception as e:
        print('save img fail')

模块6：模型存储和读取

rom sklearn import joblib
from sklearn import svm
from sklearn2pmml import PMMLPipeline, sklearn2pmml
import pickle

def save_model(train_X,train_y):
    ''''
    save model
    :return:
    '''
    clf = svm.SVC()
    clf.fit(X, y)
    joblib.dump(clf, "train_model.m")
    sklearn2pmml(clf, "train_model.pmml")
    with open('train_model.pickle', 'wb') as f:
        pickle.dump(clf, f)
    return True

def load_model():
    '''
    laod model
    :return:
    '''
    clf_joblib=joblib.load('train_model.m')
    clf_pickle== pickle.load(open('linearregression.pickle','rb'))
    return clf_joblib,clf_pickle

模块7：TF-IDF

import time
import pandas as pd
import numpy as np
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.feature_extraction.text import TfidfVectorizer

# 读取数据 - 性能不好待优化
print('开始读取KeyTag标签...')
read_data_path = 'D:/untitled/incomelevel_kwtag_20190801.txt'
load_data = pd.read_csv(read_data_path, sep='\t',encoding='utf-8')
data = pd.DataFrame(load_data,columns = ['income_level','kw_tag'])
print('...读取KeyTag标签完成')


# 将数据分组处理
print('开始分组处理KeyTag标签...')
# 高收入
incomelevel_top = data[data['income_level'] == '高']
incomelevel_top = incomelevel_top.head() #test
kw_tag_top = ' '.join(incomelevel_top['kw_tag'])
print('kw_tag_top : \n',kw_tag_top)
# 中收入
incomelevel_mid = data[data['income_level'] == '中']
incomelevel_mid = incomelevel_mid.head()  #test
kw_tag_mid = ' '.join(incomelevel_mid['kw_tag'])
print('kw_tag_mid : \n',kw_tag_mid)
# 低收入
incomelevel_low = data[data['income_level'] == '低']
incomelevel_low = incomelevel_low.head()  #test
kw_tag_low = ' '.join(incomelevel_low['kw_tag'])
print('kw_tag_low : \n',kw_tag_low)

print('...分组处理KeyTag标签完成')

# 开始加载TF-IDF
vectorizer = CountVectorizer()
result = vectorizer.fit_transform([kw_tag_top, kw_tag_mid, kw_tag_low])
transformer = TfidfVectorizer()
kw_tag_score = transformer.fit_transform([kw_tag_top, kw_tag_mid, kw_tag_low])
print('...KeyTag分词结束')
# 获取全量标签
kw_tag_value = transformer.get_feature_names()
result_target = pd.DataFrame(kw_tag_value,columns = ['kw_tag'])
print('result_target : \n',result_target)
# 分词得分处理
tf_score = kw_tag_score.toarray()
print('tf_score : \n',tf_score)
kw_tag_score_top = pd.DataFrame(tf_score[0],columns = ['kw_tag_score_top']) # 217
kw_tag_score_mid = pd.DataFrame(tf_score[1],columns = ['kw_tag_score_mid'])
kw_tag_score_low = pd.DataFrame(tf_score[2],columns = ['kw_tag_score_low'])
print(len(kw_tag_score_top))

模块8：生成省市地图

import time
import pandas as pd
import xlrd
import re
import matplotlib.pyplot as plt
import six
import numpy as np
# 载入ppt和pyecharts相关的包
from pyecharts.render import make_snapshot
from snapshot_phantomjs import snapshot
from pyecharts import options as opts
from collections import defaultdict
from pyecharts.charts import Bar, Geo, Map, Line,Funnel,Page
import os
from example.commons import Faker

def create_zjs_map():
    folder_path = os.getcwd()
    file_name = "白皮书数据地图.xlsx"
    file_path = os.path.join(folder_path, file_name)
    dat = get_excel_content(file_path, sheet_name="省份地图")

    df = dat[['城市', '渗透率']]
    df.columns = ['city', 'penarate']
    print(df)

    # df['city'] = df['city'].apply(lambda x: reg.sub('', x))
    citys = df['city'].values.tolist()
    values = df['penarate'].values.tolist()
    print(citys)
    print('{:.0f}%'.format(max(values)*100),'{:.0f}%'.format(min(values)*100))

    city_name='浙江'
    penetration_map = (
        Map(init_opts=opts.InitOpts(width='1200px', height='1000px', bg_color='white'))
            .add("{}透率分布".format(city_name), [list(z) for z in zip(citys, values)], city_name)
            .set_series_opts(
                label_opts=opts.LabelOpts(
                    is_show=True,
                    font_size=15
            )
        )
            .set_global_opts(
                visualmap_opts=opts.VisualMapOpts(
                    is_show=True,
                    max_=max(values),
                    min_=min(values),
                    is_calculable=False,
                    orient='horizontal',
                    split_number=3,
                    range_color=['#C2D5F8', '#88B0FB', '#4D8AFD'],
                    range_text=['{:.0f}%'.format(max(values)*100),'{:.0f}%'.format(min(values)*100)],
                    pos_left='10%',
                    pos_bottom='15%'
                ),
                legend_opts=opts.LegendOpts(is_show=False)
        )
    )
    # penetration_map.render()
    make_snapshot(snapshot, penetration_map.render(), "zj_map.png")
    print('保存 zj_map.png')
    return penetration_map

def create_county_map(city_name):

    folder_path = os.getcwd()
    file_name = "白皮书数据地图.xlsx"
    file_path = os.path.join(folder_path, file_name)
    dat = get_excel_content(file_path, sheet_name="城市地图")

    df = dat[['city', 'county', 'penarate']][dat.city == city_name]

    citys = df['county'].values.tolist()
    values = df['penarate'].values.tolist()
    max_insurance = max(values)

    print(citys)

    province_penetration_map = (
        Map(init_opts=opts.InitOpts(width='1200px', height='1000px', bg_color='white'))
            .add("{}透率分布".format(city_name), [list(z) for z in zip(citys, values)], reg.sub('',city_name))
            .set_series_opts(
                label_opts=opts.LabelOpts(
                is_show=True,
                font_size=15
            )
        )
            .set_global_opts(
            visualmap_opts=opts.VisualMapOpts(
                is_show=True,
                max_=max(values),
                min_=min(values),
                is_calculable=False,
                orient='horizontal',
                split_number=3,
                range_color=['#C2D5F8', '#88B0FB', '#4D8AFD'],
                range_text=['{:.0f}%'.format(max(values) * 100), '{:.0f}%'.format(min(values) * 100)],
                pos_left='10%',
                pos_bottom='5%'
            ),
            legend_opts=opts.LegendOpts(is_show=False)
        )
    )
    # insurance_map.render()
    make_snapshot(snapshot, province_penetration_map.render(), "city_map_{}.png".format(city_name))
    print('保存 city_map_{}.png'.format(city_name))
    return province_penetration_map

def create_funnel_label():

    folder_path=os.getcwd()
    file_name = "白皮书数据地图.xlsx"
    file_path = os.path.join(folder_path, file_name)
    dat = get_excel_content(file_path, sheet_name="漏斗图")

    df = dat[['category', 'cnt']]
    print(df)

    category = df['category'].values.tolist()
    values = df['cnt'].values.tolist()

    funnel_map = (
        Funnel(init_opts=opts.InitOpts(width='1200px', height='1000px', bg_color='white'))
            .add("漏斗图", [list(z) for z in zip(category, values)])
            .set_series_opts(
                label_opts=opts.LabelOpts(
                    position='inside',
                    font_size=16,
                )
            )
            .set_global_opts(
                legend_opts=opts.LegendOpts(is_show=False)
            )
    )
    # insurance_map.render()
    make_snapshot(snapshot, funnel_map.render(), "funnel.png")
    print('保存 funnel.png')
    return funnel_map

city_list=['温州市','杭州市','绍兴市','嘉兴市','湖州市','宁波市','金华市','台州市','衢州市','丽水市','舟山市']

for city_name in city_list:
    create_county_map(city_name)