538 lines
22 KiB
C#
538 lines
22 KiB
C#
using Aliyun.OSS;
|
||
using HtmlAgilityPack;
|
||
using NPOI.Util.ArrayExtensions;
|
||
using System;
|
||
using System.Collections.Generic;
|
||
using System.Linq;
|
||
using System.Net.Http;
|
||
using System.Text;
|
||
using System.Text.Json;
|
||
using System.Threading.Tasks;
|
||
|
||
namespace New_Spider
|
||
{
|
||
public class XueHtmlAgHelper
|
||
{
|
||
private string bucketName = "static-data-ycymedu"; //
|
||
private string filePrefix = "college-data/page-";
|
||
// 初始化 OSS 客户端
|
||
private OssClient ossClient = new OssClient("https://oss-cn-shanghai.aliyuncs.com", "LTAI5tKs3TXSbt7E4WMDcxwR", "EvC8MjRaQC1kHubgU4MtecZnofOb0v");
|
||
|
||
/// <summary>
|
||
/// 阳光高考完数据解析
|
||
/// </summary>
|
||
public async Task HtmlCreatePageData()
|
||
{
|
||
// API 配置
|
||
string baseUrl = "http://api.xuexingaokao.com/api/college/base_college/index";
|
||
int totalPages = 287;
|
||
string filePrefix = "college-data/page-";
|
||
using var httpClient = new HttpClient();
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{
|
||
try
|
||
{
|
||
// 动态构建 API URL
|
||
var url = $"{baseUrl}?area=&type=&nature=&grade=&is_bz=&cname=&page={page}&perpage=10×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
Console.WriteLine($"Fetching page {page}...");
|
||
|
||
// 获取 JSON 数据(同步请求)
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix}{page}.json";
|
||
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
}
|
||
catch (Exception ex)
|
||
{
|
||
Console.WriteLine($"Error fetching or uploading page {page}: {ex.Message}");
|
||
}
|
||
}
|
||
}
|
||
|
||
|
||
public async Task GetDataDetail()
|
||
{
|
||
int totalPages = 287;
|
||
string baseUrl = "https://static-data-ycymedu.oss-cn-shanghai.aliyuncs.com/";
|
||
string detailUrl = "http://api.xuexingaokao.com/api/college/base_college/collegeDetail?college_id=";
|
||
string facturl = "http://api.xuexingaokao.com/api/college/base_college/faculty?college_id=685×tamp=1733820969";
|
||
string subjectIntroduceurl = "http://api.xuexingaokao.com/api/college/base_college/subjectIntroduce?college_id=685×tamp=1733821065";
|
||
string recruitListUrl = "http://api.xuexingaokao.com/api/college/base_college/recruitList?college_id=685×tamp=1733821136";
|
||
using var httpClient = new HttpClient();
|
||
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{ // 动态构建 API URL
|
||
var url = $"{baseUrl}{filePrefix}{page}.json";
|
||
Console.WriteLine($"Fetching page {page}...");
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = JsonSerializer.Deserialize<universitylistobject>(response.Content.ReadAsStringAsync().Result);
|
||
if (jsonData?.code == 1)
|
||
{
|
||
string filePrefix2 = "college-detail/";
|
||
jsonData.data.data.ToList().ForEach(a =>
|
||
{
|
||
// 动态构建 API URL
|
||
var detailgourl = $"{detailUrl}{a.id}×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
Console.WriteLine($"detailgourl {a.id}...");
|
||
// 获取 JSON 数据(同步请求)
|
||
var response = httpClient.GetAsync(detailgourl).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix2}{a.id}.json";
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
});
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
|
||
/// <summary>
|
||
/// 缺失详情页
|
||
/// </summary>
|
||
public void GetDataRecruitList()
|
||
{
|
||
int totalPages = 287;
|
||
string baseUrl = "https://static-data-ycymedu.oss-cn-shanghai.aliyuncs.com/";
|
||
string recruitListUrl = "http://api.xuexingaokao.com/api/college/base_college/recruitList?college_id=";
|
||
using var httpClient = new HttpClient();
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{ // 动态构建 API URL
|
||
var url = $"{baseUrl}{filePrefix}{page}.json";
|
||
Console.WriteLine($"recruitListUrl page {page}...");
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = JsonSerializer.Deserialize<universitylistobject>(response.Content.ReadAsStringAsync().Result);
|
||
if (jsonData?.code == 1)
|
||
{
|
||
|
||
string filePrefix2 = "college-recruitlist/";
|
||
jsonData.data.data.ToList().ForEach(a =>
|
||
{
|
||
Thread.Sleep(200);
|
||
// 动态构建 API URL
|
||
var gourl = $"{recruitListUrl}{a.id}×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
Console.WriteLine($"recruitListUrl {a.id}...");
|
||
// 获取 JSON 数据(同步请求)
|
||
var response = httpClient.GetAsync(gourl).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix2}{a.id}.json";
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
});
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
|
||
|
||
|
||
public void GetDataSubjectIntroduce()
|
||
{
|
||
int totalPages = 287;
|
||
string baseUrl = "https://static-data-ycymedu.oss-cn-shanghai.aliyuncs.com/";
|
||
string subjectIntroduceurl = "http://api.xuexingaokao.com/api/college/base_college/subjectIntroduce?college_id=";
|
||
using var httpClient = new HttpClient();
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{ // 动态构建 API URL
|
||
var url = $"{baseUrl}{filePrefix}{page}.json";
|
||
Console.WriteLine($"subjectIntroduceurl page {page}...");
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = JsonSerializer.Deserialize<universitylistobject>(response.Content.ReadAsStringAsync().Result);
|
||
if (jsonData?.code == 1)
|
||
{
|
||
|
||
string filePrefix2 = "college-subjectIntroduce/";
|
||
jsonData.data.data.ToList().ForEach(a =>
|
||
{
|
||
Thread.Sleep(200);
|
||
// 动态构建 API URL
|
||
var gourl = $"{subjectIntroduceurl}{a.id}×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
Console.WriteLine($"subjectIntroduceurl {a.id}...");
|
||
// 获取 JSON 数据(同步请求)
|
||
var response = httpClient.GetAsync(gourl).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix2}{a.id}.json";
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
});
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
|
||
/// <summary>
|
||
///
|
||
/// </summary>
|
||
/// <returns></returns>
|
||
public void GetDataFaculty()
|
||
{
|
||
int totalPages = 287;
|
||
string baseUrl = "https://static-data-ycymedu.oss-cn-shanghai.aliyuncs.com/";
|
||
string facturl = "http://api.xuexingaokao.com/api/college/base_college/faculty?college_id=";
|
||
using var httpClient = new HttpClient();
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{ // 动态构建 API URL
|
||
var url = $"{baseUrl}{filePrefix}{page}.json";
|
||
Console.WriteLine($"Fetching page {page}...");
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = JsonSerializer.Deserialize<universitylistobject>(response.Content.ReadAsStringAsync().Result);
|
||
if (jsonData?.code == 1)
|
||
{
|
||
|
||
string filePrefix2 = "college-faculty/";
|
||
jsonData.data.data.ToList().ForEach(a =>
|
||
{
|
||
Thread.Sleep(300);
|
||
// 动态构建 API URL
|
||
var gourl = $"{facturl}{a.id}×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
Console.WriteLine($"facultyurl {a.id}...");
|
||
// 获取 JSON 数据(同步请求)
|
||
var response = httpClient.GetAsync(gourl).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix2}{a.id}.json";
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
});
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
|
||
|
||
|
||
/// <summary>
|
||
/// 招生简介详情
|
||
/// </summary>
|
||
public void GetDataRecruitDetail()
|
||
{
|
||
|
||
int totalPages = 287;
|
||
string baseUrl = "https://static-data-ycymedu.oss-cn-shanghai.aliyuncs.com/";
|
||
string recruitUrl = "http://api.xuexingaokao.com/api/college/base_college/recruitDetail?recruit_id=";
|
||
using var httpClient = new HttpClient();
|
||
string filePrefix2 = "college-recruitlist/";
|
||
string filePrefix3 = "college-recruitdetail/";
|
||
for (int page = 1; page <= totalPages; page++)
|
||
{ // 动态构建 API URL
|
||
var url = $"{baseUrl}{filePrefix}{page}.json";
|
||
Console.WriteLine($"recruitListUrl page {page}...");
|
||
var response = httpClient.GetAsync(url).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = JsonSerializer.Deserialize<universitylistobject>(response.Content.ReadAsStringAsync().Result);
|
||
if (jsonData?.code == 1)
|
||
{
|
||
jsonData.data.data.ToList().ForEach(a =>
|
||
{
|
||
string objectUrl = $"{baseUrl}{filePrefix2}{a.id}.json";
|
||
Console.WriteLine($"down page {objectUrl}");
|
||
var responseitem = httpClient.GetAsync(objectUrl).Result;
|
||
if (responseitem.IsSuccessStatusCode)
|
||
{
|
||
try
|
||
{
|
||
var items = JsonSerializer.Deserialize<RecruitRootobject>(responseitem.Content.ReadAsStringAsync().Result);
|
||
if (items != null)
|
||
{
|
||
if (items.code == 1)
|
||
{
|
||
items.data.recruit.ToList().ForEach(e =>
|
||
{
|
||
Thread.Sleep(150);
|
||
|
||
try
|
||
{
|
||
var itemGoUrl = $"{recruitUrl}{e.id}×tamp={DateTimeOffset.Now.ToUnixTimeSeconds()}";
|
||
var response = httpClient.GetAsync(itemGoUrl).Result;
|
||
if (response.IsSuccessStatusCode)
|
||
{
|
||
var jsonData = response.Content.ReadAsStringAsync().Result;
|
||
// 构建文件名和路径
|
||
string objectName = $"{filePrefix3}{e.id}.json";
|
||
// 上传到 OSS(同步上传)
|
||
using var stream = new MemoryStream(Encoding.UTF8.GetBytes(jsonData));
|
||
ossClient.PutObject(bucketName, objectName, stream);
|
||
Console.WriteLine($"Uploaded page {page} to OSS as {objectName}");
|
||
}
|
||
else
|
||
{
|
||
Console.WriteLine($"Failed to fetch page {page}, Status Code: {response.StatusCode}");
|
||
}
|
||
}
|
||
catch (Exception ex)
|
||
{
|
||
Console.WriteLine(ex.Message);
|
||
}
|
||
});
|
||
|
||
|
||
}
|
||
}
|
||
}
|
||
catch (Exception exx)
|
||
{
|
||
Console.WriteLine(exx.Message);
|
||
Console.WriteLine($"{objectUrl}解析失败");
|
||
}
|
||
|
||
}
|
||
});
|
||
}
|
||
}
|
||
}
|
||
|
||
}
|
||
|
||
|
||
|
||
}
|
||
|
||
}
|
||
|
||
|
||
|
||
public class universitylistobject
|
||
{
|
||
public int code { get; set; }
|
||
public string msg { get; set; }
|
||
public Data data { get; set; }
|
||
}
|
||
|
||
public class Data
|
||
{
|
||
public int total { get; set; }
|
||
public string has_next { get; set; }
|
||
public string page { get; set; }
|
||
public string perpage { get; set; }
|
||
public Params _params { get; set; }
|
||
public Paramslist paramsList { get; set; }
|
||
public Datum[] data { get; set; }
|
||
}
|
||
|
||
public class Params
|
||
{
|
||
public string area { get; set; }
|
||
public string type { get; set; }
|
||
public string nature { get; set; }
|
||
public string grade { get; set; }
|
||
public string is_bz { get; set; }
|
||
public string cname { get; set; }
|
||
public string page { get; set; }
|
||
public string perpage { get; set; }
|
||
public string timestamp { get; set; }
|
||
}
|
||
|
||
public class Paramslist
|
||
{
|
||
public Arealist[] arealist { get; set; }
|
||
public Typelist[] typelist { get; set; }
|
||
public Naturelist[] naturelist { get; set; }
|
||
public Xingzhilist[] xingzhilist { get; set; }
|
||
public Classlist[] classlist { get; set; }
|
||
}
|
||
|
||
public class Arealist
|
||
{
|
||
public int id { get; set; }
|
||
public string area { get; set; }
|
||
public string province { get; set; }
|
||
public string city { get; set; }
|
||
public int create_time { get; set; }
|
||
public string text { get; set; }
|
||
public Child[] children { get; set; }
|
||
}
|
||
|
||
public class Child
|
||
{
|
||
public int id { get; set; }
|
||
public string area { get; set; }
|
||
public string province { get; set; }
|
||
public string city { get; set; }
|
||
public int create_time { get; set; }
|
||
public string text { get; set; }
|
||
}
|
||
|
||
public class Typelist
|
||
{
|
||
public int id { get; set; }
|
||
public string type_name { get; set; }
|
||
public int create_time { get; set; }
|
||
public int list_order { get; set; }
|
||
}
|
||
|
||
public class Naturelist
|
||
{
|
||
public int id { get; set; }
|
||
public string nature_name { get; set; }
|
||
public int create_time { get; set; }
|
||
public int list_order { get; set; }
|
||
public int staue { get; set; }
|
||
}
|
||
|
||
public class Xingzhilist
|
||
{
|
||
public int id { get; set; }
|
||
public string nature_name { get; set; }
|
||
public int create_time { get; set; }
|
||
public int list_order { get; set; }
|
||
public int staue { get; set; }
|
||
}
|
||
|
||
public class Classlist
|
||
{
|
||
public int id { get; set; }
|
||
public string nature_name { get; set; }
|
||
public int create_time { get; set; }
|
||
public int list_order { get; set; }
|
||
public int staue { get; set; }
|
||
}
|
||
|
||
public class Datum
|
||
{
|
||
public int id { get; set; }
|
||
public string college_name { get; set; }
|
||
public string college_name_display { get; set; }
|
||
public string rk_ranking_2024 { get; set; }
|
||
public int rk_ranking_list { get; set; }
|
||
public string college_class { get; set; }
|
||
public string college_nature { get; set; }
|
||
public Detail detail { get; set; }
|
||
public string rk_ranking { get; set; }
|
||
}
|
||
|
||
public class Detail
|
||
{
|
||
public int id { get; set; }
|
||
public string college_name { get; set; }
|
||
public string college_logo { get; set; }
|
||
public object ex_rate { get; set; }
|
||
public string college_class { get; set; }
|
||
public string college_create_time { get; set; }
|
||
public string master_num { get; set; }
|
||
public string doctor_num { get; set; }
|
||
public string college_nature { get; set; }
|
||
public string college_type { get; set; }
|
||
public string college_area { get; set; }
|
||
public string college_competent { get; set; }
|
||
public string[] college_grade { get; set; }
|
||
public object wsl_ranking { get; set; }
|
||
public string rk_ranking { get; set; }
|
||
public string enrollment_rate_in { get; set; }
|
||
public string enrollment_rate_out { get; set; }
|
||
public string job_report { get; set; }
|
||
}
|
||
|
||
|
||
|
||
|
||
|
||
|
||
public class RecruitRootobject
|
||
{
|
||
public int code { get; set; }
|
||
public string msg { get; set; }
|
||
public RecruitData data { get; set; }
|
||
}
|
||
|
||
public class RecruitData
|
||
{
|
||
public Tag tag { get; set; }
|
||
public List<Recruit> recruit { get; set; }
|
||
}
|
||
|
||
public class Tag
|
||
{
|
||
public string id { get; set; }
|
||
public string college_name { get; set; }
|
||
public string college_logo { get; set; }
|
||
public string ex_rate { get; set; }
|
||
public string college_class { get; set; }
|
||
public string college_create_time { get; set; }
|
||
public string master_num { get; set; }
|
||
public string doctor_num { get; set; }
|
||
public string college_nature { get; set; }
|
||
public string college_type { get; set; }
|
||
public string college_area { get; set; }
|
||
public string college_competent { get; set; }
|
||
public string[] college_grade { get; set; }
|
||
public int wsl_ranking { get; set; }
|
||
public string rk_ranking { get; set; }
|
||
public string enrollment_rate_in { get; set; }
|
||
public string enrollment_rate_out { get; set; }
|
||
public string job_report { get; set; }
|
||
}
|
||
|
||
public class Recruit
|
||
{
|
||
public int id { get; set; }
|
||
public int base_college_id { get; set; }
|
||
public string year { get; set; }
|
||
public string college_name { get; set; }
|
||
public string title { get; set; }
|
||
}
|