show_in_matplot.py 文件源码

def get_data_from_mongo():
    conn = pymongo.Connection('192.168.17.128',27017)
    db = conn.db_lianjia
    results = db.scrapy_zlzp_info.find({},{"zwlb":1,"zwyx":1,"gsdz":1,"gsxz":1,"_id":0})
    #results = db.scrapy_zlzp_info.find({})
    zwlb = ['C??','C++','C#','PYTHON','RUBY','JAVA','IOS','ANDROID','HTML','PHP']
    #zwlb = ['SCALA']
    zwnum_set = {}
    zwyx_set = {}

    je_re = re.compile('([0-9 ]*)-([0-9 ]*)')

    def get_average_salary(slary):
        r_slary = slary.replace(',','')
        m = je_re.match(r_slary)
        try:
            if m:
                low = m.group(1)
                high = m.group(2)
                return (float(low) + float(high)) / 2
        except:
            return 0
        return 0

    #print(get_average_salary('??'))

    for result in results:
        zw = result.get('zwlb')
        yx = result.get('zwyx')
        if isinstance(yx,str) and isinstance(zw,str):
            uzw = zw.upper()
            for zwfl in zwlb:
                if uzw.rfind(zwfl) != -1:
                    zwnum_set[zwfl] = zwnum_set.get(zwfl,0) + 1
                    zwyx_set[zwfl] = zwyx_set.get(zwfl,0) + get_average_salary(yx)
                    #print(zwfl,yx,get_average_salary(yx))

    for key in zwnum_set.keys():
        zwyx_set[key] = zwyx_set[key]/float(zwnum_set[key])


    return zwyx_set,zwnum_set