高考历年分数最低查询

729 · 发表于 2022-6-27 23:19

本帖最后由 729 于 2022-6-27 23:43 编辑

我肝不动了，整理了一下下的代码，肯定有点问题的代码，但是勉强能跑，我主要用的是get_test函数，哎呀，差不多的啦，仅供学习

[Python] 纯文本查看 复制代码

#encoding :utf-8

import requests
import json
import csv
import os
import pandas as pd
import openpyxl

pro_id = {
'11':'北京市',
'12':'天津市',
'13':'河北省',
'14':'山西省',
'15':'内蒙古自治区',
'21':'辽宁省',
'22':'吉林省',
'23':'黑龙江省',
'31':'上海市',
'32':'江苏省',
'33':'浙江省',
'34':'安徽省',
'35':'福建省',
'36':'江西省',
'37':'山东省',
'41':'河南省',
'42':'湖北省',
'43':'湖南省',
'44':'广东省',
'45':'广西壮族自治区',
'46':'海南省',
'50':'重庆市',
'51':'四川省',
'52':'贵州省',
'53':'云南省',
'54':'西藏自治区',
'61':'陕西省',
'62':'甘肃省',
'63':'青海省',
'64':'宁夏回族自治区',
'65':'新疆维吾尔自治区',
'71':'台湾省',
'81':'香港特别行政区',
'82':'澳门特别行政区',
}

typ_dict = {'2073':'物理类',
            '2074':'历史类',
            '4':'艺术类',
            '1':'理科',
            '2': '文科',
            '3':'综合',
            }
batch_dict = {'7':'一本(广西、陕西)',
              '8':'二本(广西、陕西)',
              '14':'本科(广东、湖南)',
              '1570':'普通类一段(山东)',
              '1571':'普通类二段(山东)',

}
headers = {
            'User-Agent':'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/536.5 (KHTML, like Gecko) Chrome/19.0.1084.9 Safari/536.5'
        }
# 获取单个学校id
def get_school_id(schuool_name):
    school_id = -1
    reque = 'https://api.eol.cn/gkcx/api/?keyword=%s&uri=apigkcx/api/school/hotlists' % (schuool_name)
    re_json = requests.get(reque, headers = headers).text
    re_dict = json.loads(re_json)
    if 'code' in re_dict.keys():
        if re_dict['code'] == '0000':
            for name in re_dict['data']['item']:
                if name['name'] == schuool_name:
                    school_id = name['school_id']
            # print(school_id,type(school_id))
                    return school_id

# 通过年份获取历年成绩
def get_year_grade(year:int, school_id:int, province_id:int, typ:int ,batch:int):
    # 年份、学校ID、招生城市、文科/理科、批次
    ## typ 理科/文科 1/2
    ## province 录取的省份
    # item_dist = {'local_batch_name':0,
    #             'min':0,
    #             'min_section':0,
    #             'spname':0,
    # }
    item_list = []
    # 针对广西
    # if batch == 1:    # 一本
    #     batch = 7
    # elif batch == 2:
    #     batch = 8     # 二本
    # else:
    #     batch = 10    # 转科
    for num in range(1,999):
        ''' https://static-data.gaokao.cn/www/2.0/schoolspecialindex/2021/3593/45/2/10/1.json'''
        url = 'https://static-data.gaokao.cn/www/2.0/schoolspecialindex/%d/%d/%d/%d/%d/%d.json' % (year, school_id, province_id, typ, batch, num)
        re_json = requests.get(url, headers = headers).text
        if re_json == '''""''':
            break
        re_dict = json.loads(re_json)
        if 'code' in re_dict.keys():
            for item in re_dict['data']['item']:
                item_list.append(item)
    return item_list

# 获取省份的全部学校信息(包含id)
def get_province_all(province_id:int):
    '''{"address":"广西壮族自治区北海市银海区金海岸大道666号","admissions":2,"answerurl":"","belong":"广西壮族自治区教育厅","central":2,"city_id":4505,"city_name":"北海市","code_enroll":1476500,"colleges_level":"","county_id":450503,"county_name":"银海区","department":2,"doublehigh":0,"dual_class":38003,"dual_class_name":"","f211":2,"f985":2,"id":"gkschool3593","is_logo":1,"is_recruitment":1,"is_top":2,"level":2002,"level_name":"专科（高职）","name":"北海康养职业学院","nature":36001,"nature_name":"民办","province_id":45,"province_name":"广西","rank":2807,"rank_type":10,"school_id":3593,"school_type":6001,"single_province":"","special":[],"type":0,"type_name":"","view_month":"242","view_month_number":242,"view_total":"3692","view_total_number":3692,"view_week":"0","view_week_number":0,"view_year":3692}'''
    item_list = []
    for page in range(1,99):
        url = 'https://api.eol.cn/web/api/?&page=%d&province_id=%d&uri=apidata/api/gk/school/lists' % (page, province_id)
        re_json = requests.get(url, headers = headers).text
        re_dict = json.loads(re_json)
        # print(re_dict)
        if 'code' in re_dict.keys():
            if len(re_dict['data']['item']) != 0:
                for item in re_dict['data']['item']:
                    item_list.append(item)
            else:
                break
    return item_list
    # https://api.eol.cn/web/api/?&page=8&province_id=45&uri=apidata/api/gk/school/lists

# 处理数据，获取学习名字和id
def return_all_school_id(school_info:list):
    # dict_tem = {'name':'','school_id':''}
    item_list = []
    for school in school_info:
        item_list.append({'name':school['name'], 'school_id':school['school_id']})
        # dict_tem['name'] = school['name']
        # dict_tem['school_id'] = school['school_id']
        # item_list.append(dict_tem)
        # print(item_list)
    return item_list
# 提取成绩信息
def return_all_grade(grade_info:list):
    # dict_tem = {}
    item_list = []
    # print(grade_info)
    for school in grade_info:
        item_list.append(
            {
            'local_batch_name': school['local_batch_name'],
            'spname': school['spname'],
            'min': school['min'],
            'min_section': school['min_section'],
            'type':school['type'],
            'province':school['province']   ## 生源地
            }
        )
        # print(type(school['type']))
        # dict_tem['local_batch_name'] = grade_info['local_batch_name']   ## 批次
        # dict_tem['spname'] = grade_info['spname']               ## 专业
        # dict_tem['min'] = grade_info['min']                     ## 最低分数
        # dict_tem['min_section'] = grade_info['min_section']     ## 排名
        # item_list.append(dict_tem)
    return item_list
# 写csv文件
def write_csv(path:str,date:list):
    if not len(date):
        print('抱歉，啥都没有')
        return
    if os.path.exists(path):
        file_ex = True
    else:
        file_ex = False
    with open(path, 'a+', newline='', encoding='utf-8') as csvfile:
        hea = list(date[0])
        writer = csv.DictWriter(csvfile, fieldnames=hea)
        if not file_ex:
            writer.writeheader()
        writer.writerows(date)
    csv_excel = pd.read_csv(path, encoding='utf-8')
    csv_excel.to_excel(path[:-4] + '.xlsx', sheet_name='data')
    os.remove(path)
# 获取某个省份的全部学校录取某个省份的最低分
def get_test(year:int, province_src:int, province_trg:int, typ:int, batch:int, file_path:str):
    # style_hea = {'学校':'','年份':'','批次':'','专业':'','最低分':'','最低排名':''}
    school_info = return_all_school_id(get_province_all(province_src))        ## 获取了所有学校的名字和id
    # print(school_info)
    test_list = []
    num = 0
    for i in school_info:
        num +=1
        print(num)
        gra = get_year_grade(year = year, school_id = i['school_id'], province_id = province_trg, typ = typ, batch = batch)
        # print(gra)
        gra = return_all_grade(gra)
        # print(gra)
        for g in gra:
            test_list.append(
                {
                    '学校' : i['name'],
                    '年份' : 2021,
                    '批次' : g['local_batch_name'],
                    '生源地': pro_id[str(g['province'])],
                    '类别': typ_dict[str(g['type'])],
                    '专业' : g['spname'],
                    '最低分': g['min'],
                    '最低排名' :g['min_section']
                }
            )
            # print(g['type'])
            # style_hea['学校'] = i['name']
            # style_hea['年份'] = 2021
            # style_hea['批次'] = g['local_batch_name']
            # style_hea['专业'] = g['spname']
            # style_hea['最低分'] = g['min']
            # style_hea['最低排名'] = g['min_section']
            # test_list.append(style_hea)
            # print(test_list)
    if file_path == '':
            return test_list
    write_csv(file_path,test_list)

def get_max_mean_min(grade_info:list):
    item_list = []
    # print(grade_info)
    for school in grade_info:
        item_list.append(
            {
            '批次': school['local_batch_name'],
            '专业': school['spname'],
            '最低分': school['min'],
            '最高分': school['max'],
            '平均分':school['average'],
            '最低排名': school['min_section'],
            # '理科':school['type'],
            # 'province':school['province']   ## 生源地
            }
        )
    for i in item_list:
        print(i,'\n')
# 获取全国对某个省份的专业录取最低分
def get_test_1(year:int, province_trg:int, typ:int, batch:int, file_path):
    # 年、目标省代码、类别(理科/文科/物理类/历史类...)、批次(本科一批次、二、三之类的)、保存路径
    for id in pro_id.keys():
        re = get_test(year, int(id), province_trg, typ,batch, '')
        write_csv(file_path,re)

def get_test_2(year:int, school_id:int, typ:int, batch:int, file_path):
    # test_list = []
    # gra = get_year_grade(year, school_id, typ, batch)
    # gra = return_all_grade(gra)
    # for g in gra:
    #     test_list.append(
    #     {
    #         '学校' : i['name'],
    #         '年份' : 2021,
    #         '批次' : g['local_batch_name'],
    #         '生源地': pro_id[str(g['province'])],
    #         '类别': typ_dict[str(g['type'])],
    #         '专业' : g['spname'],
    #         '最低分': g['min'],
    #         '最低排名' :g['min_section']
    #     }
    # )
    # write_csv(file_path,test_list)
    pass
if __name__ =='__main__':
#     typ_dict = {'2073':'物理类',
#             '2074':'历史类',
#             '4':'艺术类',
#             '1':'理科',
#             '2': '文科',
#             '3':'综合',
#             }
# batch_dict = {'7':'一本(广西、陕西)',
#               '8':'二本(广西、陕西)',
#               '10':'专科',
#               '14':'本科(广东、湖南)',
#               '1570':'普通类一段(山东)',
#               '1571':'普通类二段(山东)',
#               '44': '本科二批A段',
#               '45': '本科二批B段',
#               '51': '本科一批A段',
 
# }
    # id = get_school_id('武汉理工大学')
    # if id != -1:
    #     print(get_year_grade(45,2021,id,1,1))
    # ll = get_province_all(45)
    # for l in ll:
    #     print(l['name'])
    # print(os.path.exists(r'./123.csv'))
    # date = [{'test_1':1,'test_2':2},{'test_1':2,'test_2':1}]
    # write_csv(r'./123.csv',date)
    # school_info = return_all_school_id(get_province_all())
    # get_test()

    get_test(2021, 45, 45, 2, 7,r'./20220627/2021年广西本科一批次专业最低分数线.csv')
    get_test(2021, 45, 45, 2, 8,r'./20220627/2021年广西本科二批次专业最低分数线.csv')

    get_test(2021, 44, 43, 2073, 14,r'./20220627/2021年广东(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 45, 43, 2073, 14,r'./20220627/2021年广西(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 36, 43, 2073, 14,r'./20220627/2021年江西(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 44, 43, 2074, 14,r'./20220627/2021年广东(生源地：湖南)历史类本科批次专业最低分数线.csv')
    get_test(2021, 45, 43, 2074, 14,r'./20220627/2021年广西(生源地：湖南)历史类本科批次专业最低分数线.csv')
    get_test(2021, 36, 43, 2074, 14,r'./20220627/2021年江西(生源地：湖南)历史类本科批次专业最低分数线.csv')

    get_test(2021, 14, 14, 1, 8,r'./20220627/2021年山西理科本科二批次专业最低分数线.csv')

    get_test(2021, 45, 44, 2074, 14,r'./20220627/2021年广西(生源地：广东)历史类本科批次专业最低分数线.csv')
    get_test(2021, 36, 44, 2074, 14,r'./20220627/2021年江西(生源地：广东)历史类本科批次专业最低分数线.csv')
    get_test(2021, 43, 44, 2074, 14,r'./20220627/2021年湖南(生源地：广东)历史类本科批次专业最低分数线.csv')

    get_test(2021, 65, 43, 2073, 14,r'./20220627/2021年新疆(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 46, 43, 2073, 14,r'./20220627/2021年海南(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 35, 43, 2073, 14,r'./20220627/2021年福建(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 34, 43, 2073, 14,r'./20220627/2021年安徽(生源地：湖南)物理类本科批次专业最低分数线.csv')
    get_test(2021, 61, 43, 2073, 14,r'./20220627/2021年陕西(生源地：湖南)物理类本科批次专业最低分数线.csv')
    
    
    get_test(2021, 65, 65, 1, 7,r'./20220627/2021年新疆理科本科一批次专业最低分数线.csv')
    get_test(2021, 65, 65, 1, 8,r'./20220627/2021年新疆文科本科二批次专业最低分数线.csv')
    get_test(2021, 65, 65, 2, 7,r'./20220627/2021年新疆文科本科一批次专业最低分数线.csv')
    get_test(2021, 65, 65, 2, 8,r'./20220627/2021年新疆文科本科二批次专业最低分数线.csv')

    get_test(2021, 53, 53, 1, 7,r'./20220627/2021年云南理科本科一批次专业最低分数线.csv')
    get_test(2021, 53, 53, 1, 8,r'./20220627/2021年云南理科本科二批次专业最低分数线.csv')

    get_test(2021, 45, 45, 1, 10,r'./20220627/2021年广西理科专科批次专业最低分数线.csv')

    get_test(2021, 22, 22, 1, 44,r'./20220627/2021年吉林理科本科二A段批次专业最低分数线.csv')
    get_test(2021, 22, 22, 2, 44,r'./20220627/2021年吉林理文科本科二A段批次专业最低分数线.csv')

    get_test(2021, 15, 15, 2, 8,r'./20220627/2021年内蒙古文科本科二批次专业最低分数线.csv')

kexue8 · 发表于 2022-6-28 06:28

感谢大神

温馨提示：最低分数可能不等于录取分数。并非达到了最低分数就一定能被录取的哟。

LeMonThELord · 发表于 2022-6-28 12:03

shine_tsai 发表于 2022-6-28 09:41
楼主，程序出错，返回如下：
Traceback (most recent call last):
File "D:/PycharmProjects/studyProj ...

你这是少文件夹，加个 20220627 的文件夹就好了

yjh_b · 发表于 2022-6-28 01:11

第一个打卡，有现成的吗

innovationing · 发表于 2022-6-28 05:38

第二个打卡，楼主大神啊！

82686839 · 发表于 2022-6-28 05:55

楼主大神啊

Deuxlinne · 发表于 2022-6-28 06:07

真心祝願大家，不要選擇生化環材專業。

逍遥贝勒爷 · 发表于 2022-6-28 06:08

大神大神！！！！

lgh978 · 发表于 2022-6-28 06:20

楼主有成品吗

xuwupiaomiao · 发表于 2022-6-28 06:44

虽然不会用还是👍 👍

mestyle · 发表于 2022-6-28 07:46

厉害👍

帐号		自动登录	找回密码
密码			注册[Register]

[Python 转载] 高考历年分数最低查询

免费评分

本帖被以下淘专辑推荐:

个人中心