third / fw-freya

Commit 31609a68e28c9fccd6dbefdaaf390ee8e9862c47

Authored by 王明元
1 parent 573ef45b

2023年9月12日14:49:11 修复特殊类型快手账号无法登录问题

Inline Side-by-side
Showing 2 changed files with 61 additions and 12 deletions
src/main/java/cn/fw/freya/service/crawl/impl/KuaiShouCrawl.java
  View file @31609a6
... ... @@ -5,6 +5,7 @@ import cn.fw.freya.dao.LivePoolDao;
5 5 import cn.fw.freya.dao.VideoPoolDao;
6 6 import cn.fw.freya.domain.data.Account;
7 7 import cn.fw.freya.domain.data.FwCookie;
  8 +import cn.fw.freya.domain.data.ResponseReceived;
8 9 import cn.fw.freya.domain.data.pool.LivePool;
9 10 import cn.fw.freya.domain.data.pool.VideoPool;
10 11 import cn.fw.freya.domain.dto.rpc.ReportAccountDto;
... ... @@ -28,8 +29,10 @@ import com.alibaba.fastjson.TypeReference;
28 29 import com.alibaba.fastjson.parser.Feature;
29 30 import lombok.RequiredArgsConstructor;
30 31 import lombok.extern.slf4j.Slf4j;
  32 +import org.apache.http.HttpResponse;
31 33 import org.apache.http.client.CookieStore;
32 34 import org.apache.http.impl.client.BasicCookieStore;
  35 +import org.apache.http.util.EntityUtils;
33 36 import org.apache.logging.log4j.util.PropertiesUtil;
34 37 import org.openqa.selenium.By;
35 38 import org.openqa.selenium.Cookie;
... ... @@ -174,23 +177,35 @@ public class KuaiShouCrawl implements CrawlStrategy, SmartLifecycle {
174 177 if (Objects.isNull(driver)) {
175 178 throw new BusinessException("登陆校验失败，请重新尝试");
176 179 }
177   - WebElement element;
  180 + WebElement element = null;
  181 + JSONObject ksAccountMsg = new JSONObject();
  182 + boolean specialAccount = false;// 某些账号是认证过的账号, 首页右上角无账号信息元素
178 183 try {
179 184 element = new WebDriverWait(driver, 10, 300).until(driver1 ->
180 185 driver1.findElement(By.xpath("//div[contains(text(),'快手号：') or contains(text(),'用户 ID：')]")));
181 186 } catch (Exception e) {
182   - this.exitBrowser(accountNo, null);
183   - throw new BusinessException("网络异常，或该人员暂未扫码登录");
  187 + ksAccountMsg = this.getKSAccountMsg(driver);// 尝试读取http日志获取账号信息
  188 + if (!StringUtils.hasText(ksAccountMsg.getString("userId")) &&
  189 + !StringUtils.hasText(ksAccountMsg.getString("accountNo"))) {
  190 + this.exitBrowser(accountNo, null);
  191 + throw new BusinessException("网络异常，或该人员暂未扫码登录");
  192 + } else {
  193 + specialAccount = true;
  194 + }
184 195 }
185   - if (Objects.nonNull(element)) {
  196 + if (Objects.nonNull(element) || specialAccount) {
186 197 int subLength = 0;
187   - String accountText = element.getText();
188   - if (accountText.startsWith("用户")) {
189   - subLength = 6;
190   - } else if (accountText.startsWith("快手号")) {
191   - subLength = 4;
  198 + String accountText = "";
  199 + if (!specialAccount) {
  200 + accountText = element.getText();
  201 + if (accountText.startsWith("用户"))
  202 + subLength = 6;
  203 + else if (accountText.startsWith("快手号"))
  204 + subLength = 4;
192 205 }
193   - if (accountNo.equals(accountText.substring(subLength))) {
  206 + if (Objects.equals(accountNo, accountText.substring(subLength)) ||
  207 + Objects.equals(accountNo, ksAccountMsg.getString("userId")) ||
  208 + Objects.equals(accountNo, ksAccountMsg.getString("accountNo"))) {
194 209 Integer type = this.getType().getValue();
195 210 driver.get("https://cp.kuaishou.com/statistics/works");// 跳转到'视频数据'页面
196 211 WebElement lockFlag = null;// 未解锁标志
... ... @@ -201,7 +216,7 @@ public class KuaiShouCrawl implements CrawlStrategy, SmartLifecycle {
201 216 try {
202 217 lockFlag = new WebDriverWait(driver, 3, 300).until(driver1 ->
203 218 driver1.findElement(By.xpath("//span[text()='立即开通']")));// '立即开通'元素
204   - } catch (Exception ex){
  219 + } catch (Exception ex) {
205 220 this.exitBrowser(accountNo, null);
206 221 throw new BusinessException("如果已解锁数据授权, 请明日再来尝试登录");
207 222 }
... ... @@ -230,6 +245,40 @@ public class KuaiShouCrawl implements CrawlStrategy, SmartLifecycle {
230 245 }
231 246  
232 247 /**
  248 + * 读取http日志, 获取账号信息
  249 + *
  250 + * @param driver 驱动
  251 + * @return 账号信息
  252 + */
  253 + private JSONObject getKSAccountMsg(WebDriver driver) {
  254 + JSONObject obj = new JSONObject();
  255 + String dataUrl = "https://cp.kuaishou.com/rest/v2/creator/pc/authority/account/current";
  256 + List<ResponseReceived> responseReceivedEvents = common.processHttpTransferData(driver);
  257 + responseReceivedEvents.forEach(item -> {
  258 + if (Objects.isNull(common.getDataUrl(item, dataUrl)))
  259 + return;
  260 + JSONObject data;
  261 + try {
  262 + HttpResponse response = common.getHttpResponse(driver, item, dataUrl);
  263 + if (Objects.nonNull(response)) {
  264 + data = JSON.parseObject(EntityUtils.toString(response.getEntity()))
  265 + .getJSONObject("value")
  266 + .getJSONObject("body")
  267 + .getJSONObject("data")
  268 + ;
  269 + if (StringUtils.hasText(data.getString("userId")))
  270 + obj.put("userId", data.getString("userId"));
  271 + if (StringUtils.hasText(data.getString("userKwaiId")))
  272 + obj.put("accountNo", data.getString("userKwaiId"));
  273 + }
  274 + } catch (Exception e) {
  275 + log.error("获取快手账号信息发生异常, 异常信息: {}", e.getMessage());
  276 + }
  277 + });
  278 + return obj;
  279 + }
  280 +
  281 + /**
233 282 * 分页获取快手视频数据
234 283 *
235 284 * @param accountNo 账号
... ...
src/main/java/cn/fw/freya/utils/CronCheckUtil.java
  View file @31609a6
... ... @@ -40,7 +40,7 @@ public class CronCheckUtil {
40 40 * 测试cron表达式以后的执行时间
41 41 */
42 42 public static void main(String[] args) {
43   - String cronExpression = "0 0 0/1 * * ?";// 待校验的cron
  43 + String cronExpression = "0 0/5 * * * ?";// 待校验的cron
44 44 DateTimeFormatter dateTimeFormatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss");
45 45 final List<LocalDateTime> runtimeList = CronCheckUtil.getRuntimeList(cronExpression, LocalDateTime.now(), 100);
46 46 runtimeList.forEach(item -> System.out.println(dateTimeFormatter.format(item)));
... ...