新浪微博数据抓取(java实现)
2024-05-08 09:51:36
1 多了不说，直接贴出相关部分的实现代码
2
3
4
5 加密部分实现：
6
7
8
9 package token.exe;
10
11 import java.math.BigInteger;
12 import java.util.Random;
13
14 import org.apache.commons.codec.binary.Base64;
15
16 public class WeiboEncoder {
17
18 private static BigInteger n = null;
19 private static BigInteger e = null;
20
21 /**
22 * 使用Base64加密用户名(su的获取)
23 * @param account
24 * @return
25 */
26 @SuppressWarnings("deprecation")
27 public static String encodeAccount(String account){
28 return new String(Base64.encodeBase64(account.getBytes()));
29 }
30
31 /**
32 * 使用RSAEncrypt对用户密码进行加密(sp的获取)
33 * @param pwd
34 * @param nStr
35 * @param eStr
36 * @return
37 */
38 public static String RSAEncrypt(String pwd, String nStr, String eStr){
39 n = new BigInteger(nStr,16);
40 e = new BigInteger(eStr,16);
41
42 BigInteger r = RSADoPublic(pkcs1pad2(pwd,(n.bitLength()+7)>>3));
43 String sp = r.toString(16);
44 if((sp.length()&1) != 0 )
45 sp = "0" + sp;
46 return sp;
47 }
48
49 private static BigInteger RSADoPublic(BigInteger x){
50 return x.modPow(e, n);
51 }
52
53 private static BigInteger pkcs1pad2(String s, int n){
54 if(n < s.length() + 11) { // TODO: fix for utf-8
55 System.err.println("Message too long for RSA");
56 return null;
57 }
58 byte[] ba = new byte[n];
59 int i = s.length()-1;
60 while(i >= 0 && n > 0) {
61 int c = s.codePointAt(i--);
62 if(c < 128) { // encode using utf-8
63 ba[--n] = new Byte(String.valueOf(c));
64 }
65 else if((c > 127) && (c < 2048)) {
66 ba[--n] = new Byte(String.valueOf((c & 63) | 128));
67 ba[--n] = new Byte(String.valueOf((c >> 6) | 192));
68 }
69 else {
70 ba[--n] = new Byte(String.valueOf((c & 63) | 128));
71 ba[--n] = new Byte(String.valueOf(((c >> 6) & 63) | 128));
72 ba[--n] = new Byte(String.valueOf((c >> 12) | 224));
73 }
74 }
75 ba[--n] = new Byte("0");
76
77 byte[] temp = new byte[1];
78 Random rdm = new Random(47L);
79
80 while(n > 2) { // random non-zero pad
81 temp[0] = new Byte("0");
82 while(temp[0] == 0)
83 rdm.nextBytes(temp);
84 ba[--n] = temp[0];
85 }
86 ba[--n] = 2;
87 ba[--n] = 0;
88
89 return new BigInteger(ba);
90 }
91
92
93
94 }
95
96 参数实体：
97
98
99
100 package token.def;
101
102 import java.io.Serializable;
103
104 public class LoginParams implements Serializable {
105
106 private static final long serialVersionUID = -5775728968372860382L;
107 private String pcid;
108 private String servertime;
109 private String nonce;
110 private String rsakv;
111 private String imgUrl;
112 private String sp;
113 private String code;
114 private boolean isLogin = true;
115
116 public String getPcid() {
117 return pcid;
118 }
119
120 public void setPcid(String pcid) {
121 this.pcid = pcid;
122 }
123
124 public String getServertime() {
125 return servertime;
126 }
127
128 public void setServertime(String servertime) {
129 this.servertime = servertime;
130 }
131
132 public String getNonce() {
133 return nonce;
134 }
135 public void setNonce(String nonce) {
136 this.nonce = nonce;
137 }
138
139 public String getRsakv() {
140 return rsakv;
141 }
142
143 public void setRsakv(String rsakv) {
144 this.rsakv = rsakv;
145 }
146
147 public String getImgUrl() {
148 return imgUrl;
149 }
150
151 public void setImgUrl(String imgUrl) {
152 this.imgUrl = imgUrl;
153 }
154
155 public String getSp() {
156 return sp;
157 }
158
159 public void setSp(String sp) {
160 this.sp = sp;
161 }
162
163 public String getCode() {
164 return code;
165 }
166
167 public void setCode(String code) {
168 this.code = code;
169 }
170
171 public boolean isLogin() {
172 return isLogin;
173 }
174
175 public void setLogin(boolean isLogin) {
176 this.isLogin = isLogin;
177 }
178
179 @Override
180 public String toString() {
181 return "LoginParams [pcid=" + pcid + ", servertime=" + servertime
182 + ", nonce=" + nonce + ", rsakv=" + rsakv + ", imgUrl="
183 + imgUrl + ", sp=" + sp + ", code=" + code + ", isLogin="
184 + isLogin + "]";
185 }
186
187 }
188
189
190
191
192
193
194
195 登陆部分实现：
196
197
198
199 package token.exe;
200
201 import java.io.FileOutputStream;
202 import java.io.IOException;
203 import java.net.URLEncoder;
204 import java.security.KeyManagementException;
205 import java.security.NoSuchAlgorithmException;
206 import java.util.ArrayList;
207 import java.util.Date;
208 import java.util.HashMap;
209 import java.util.List;
210 import java.util.Properties;
211 import java.util.Scanner;
212
213 import org.apache.commons.httpclient.Header;
214 import org.apache.commons.httpclient.HttpClient;
215 import org.apache.commons.httpclient.HttpException;
216 import org.apache.commons.httpclient.HttpStatus;
217 import org.apache.commons.httpclient.HttpVersion;
218 import org.apache.commons.httpclient.MultiThreadedHttpConnectionManager;
219 import org.apache.commons.httpclient.NameValuePair;
220 import org.apache.commons.httpclient.cookie.CookiePolicy;
221 import org.apache.commons.httpclient.methods.GetMethod;
222 import org.apache.commons.httpclient.methods.PostMethod;
223 import org.apache.commons.httpclient.params.HttpClientParams;
224 import org.apache.commons.httpclient.params.HttpConnectionManagerParams;
225 import org.apache.commons.httpclient.protocol.Protocol;
226 import org.jsoup.Jsoup;
227 import org.jsoup.nodes.Document;
228 import org.jsoup.nodes.Element;
229
230 import token.SinaWeiboOAuth;
231 import token.def.LoginParams;
232 import weibo4j.model.MySSLSocketFactory;
233
234
235 public class WeiboLoginer {
236
237 private HttpClient httpClient; //httpClient实例初始化
238
239 public WeiboLoginer() {
240
241 //httpclient连接配置
242 MultiThreadedHttpConnectionManager httpManager = new MultiThreadedHttpConnectionManager();
243 HttpConnectionManagerParams connectParams = httpManager.getParams();
244 connectParams.setConnectionTimeout(3000);
245 connectParams.setDefaultMaxConnectionsPerHost(100);
246 connectParams.setSoTimeout(3000);
247 //httpclient参数配置
248 HttpClientParams httpParams = new HttpClientParams();
249 httpParams.setCookiePolicy(CookiePolicy.BROWSER_COMPATIBILITY);
250 httpParams.setVersion(HttpVersion.HTTP_1_1);
251 //设置默认Header
252 List<Header> headers = new ArrayList<Header>();
253 headers.add(new Header("Content-Type", "application/x-www-form-urlencoded"));
254 headers.add(new Header("Host", "login.sina.com.cn"));
255 headers.add(new Header("User-Agent","Mozilla/5.0 (Windows NT 6.1; rv:25.0) Gecko/20100101 Firefox/25.0"));
256 headers.add(new Header("API-RemoteIP", "192.168.0.1"));//伪造新浪验证IP
257 headers.add(new Header("X-Forwarded-For","192.168.0.1"));//伪造真实IP
258 headers.add(new Header("CLIENT-IP", "192.168.0.1"));//伪造客户端IP
259 //初始化httpclient
260 httpClient = new HttpClient(httpParams, httpManager);
261 httpClient.getHostConfiguration().getParams().setParameter("http.default-headers", headers);
262 //设置ssl协议
263 Protocol protocol = new Protocol("https",new MySSLSocketFactory(), 443);
264 Protocol.registerProtocol("https", protocol);
265 //设置代理
266 // httpClient.getHostConfiguration().setProxy("", 0);
267 // httpClient.getParams().setAuthenticationPreemptive(false);
268 }
269
270 /**
271 * 登陆并获取code值,如果出现验证码则返回还有验证码的参数信息
272 * @return
273 */
274 public LoginParams doLogin(String username, String password) {
275
276 Properties properties = initProperties();
277 String base64UserCount = WeiboEncoder.encodeAccount(username);
278 HashMap<String, String> pubkeyMap = null;
279 String sp = null;
280 String imgUrl = null;
281 LoginParams loginParams = new LoginParams();
282 try {
283 pubkeyMap = pubKeyMap(base64UserCount);
284 sp = WeiboEncoder.RSAEncrypt(password, pubkeyMap.get("pubkey"),"10001");
285 imgUrl = getPin(pubkeyMap);
286 if (imgUrl != null) {
287 loginParams.setPcid(pubkeyMap.get("pcid"));
288 loginParams.setNonce(pubkeyMap.get("nonce"));
289 loginParams.setServertime(pubkeyMap.get("servertime"));
290 loginParams.setRsakv(pubkeyMap.get("rsakv"));
291 loginParams.setImgUrl(imgUrl);
292 loginParams.setSp(sp);
293 return loginParams;
294 }
295 } catch (IOException e) {
296 // TODO Auto-generated catch block
297 e.printStackTrace();
298 }
299
300 HashMap<String, String> ticketMap = null;
301 try {
302 ticketMap = getTicket(base64UserCount, sp, pubkeyMap);
303 } catch (Exception e1) {
304 // TODO Auto-generated catch block
305 e1.printStackTrace();
306 }
307
308 //确认在最终登陆后是否再需要验证码(账号为新浪的注册邮箱)
309 String vcUrl = isHasPinAgain(pubkeyMap, ticketMap);
310 if (vcUrl != null) {
311 loginParams.setPcid(pubkeyMap.get("pcid"));
312 loginParams.setNonce(pubkeyMap.get("nonce"));
313 loginParams.setServertime(pubkeyMap.get("servertime"));
314 loginParams.setRsakv(pubkeyMap.get("rsakv"));
315 loginParams.setImgUrl(imgUrl);
316 loginParams.setSp(sp);
317 return loginParams;
318 }
319
320 try {
321 String code = authorize(ticketMap.get("ticket"), properties.getProperty("authorizeURL"),
322 properties.getProperty("redirect_URI"), properties.getProperty("client_ID"),
323 username, ticketMap.get("uid"));
324
325 loginParams.setCode(code);
326 } catch (KeyManagementException e) {
327 // TODO Auto-generated catch block
328 e.printStackTrace();
329 } catch (NoSuchAlgorithmException e) {
330 // TODO Auto-generated catch block
331 e.printStackTrace();
332 } catch (IOException e) {
333 // TODO Auto-generated catch block
334 e.printStackTrace();
335 }
336 return loginParams;
337
338 }
339
340 /**
341 * 有验证码时登陆
342 * @param sp
343 * @param pin
344 * @param pcid
345 * @param servertime
346 * @param nonce
347 * @param rsakv
348 * @return
349 */
350 public LoginParams doLoginByPin(String username, String sp, String pin, String pcid,
351 String servertime,String nonce,String rsakv ) {
352
353 Properties properties = initProperties();
354 String base64UserCount = WeiboEncoder.encodeAccount(username);
355 HashMap<String, String> ticketMap = null;
356 LoginParams params = new LoginParams();
357 try {
358 ticketMap = getTicket(base64UserCount, sp, pin, pcid,
359 servertime, nonce, rsakv);
360 if (ticketMap.containsKey("reason")) {
361 //意为"输入的验证码不正确"
362 String reply = "\\u8f93\\u5165\\u7684\\u9a8c\\u8bc1\\u7801\\u4e0d\\u6b63\\u786e";
363 String reasonStr = ticketMap.get("reason");
364 if (reasonStr.equals(reply)) {
365 params.setLogin(false);
366 return params;
367 }
368 }
369 String code = authorize(ticketMap.get("ticket"), properties.getProperty("authorizeURL"),
370 properties.getProperty("redirect_URI"), properties.getProperty("client_ID"),
371 username, ticketMap.get("uid"));
372 params.setCode(code);
373 } catch (Exception e) {
374 // TODO Auto-generated catch block
375 e.printStackTrace();
376 }
377
378 return params;
379 }
380
381 /**
382 * 模拟新浪授权
383 * @param ticket ticket参数
384 * @param redirectURI 回调地址
385 * @param clientId appKey
386 * @param username 用户名
387 * @return token
388 * @throws IOException
389 * @throws KeyManagementException
390 * @throws NoSuchAlgorithmException
391 */
392 private String authorize(String ticket, String authorizeURL, String redirectURI,
393 String clientId, String username, String uid) throws IOException,
394 KeyManagementException, NoSuchAlgorithmException {
395
396 String code = null;
397 String url = authorizeURL + "?client_id=" + clientId + "&redirect_uri="
398 + redirectURI + "&response_type=code&forcelogin=true";
399 String regCallback = authorizeURL + "?client_id=" + clientId + "&redirect_uri="
400 + redirectURI + "&response_type=code&display=default&from=&with_cookie=";
401 PostMethod post = new PostMethod(authorizeURL);
402 //模拟申请token的链接，如果不添加，那么回调地址返回则为空
403 post.setRequestHeader("Referer",url);
404 // 模拟登录时所要提交的参数信息
405 NameValuePair[] formpPairs=new NameValuePair[]{
406 new NameValuePair("action", "login"),
407 new NameValuePair("userId",username),
408 new NameValuePair("ticket", ticket),
409 new NameValuePair("response_type", "code"),
410 new NameValuePair("redirect_uri", redirectURI),
411 new NameValuePair("client_id", clientId),
412 new NameValuePair("regCallback", URLEncoder.encode(regCallback, "UTF-8"))
413 };
414 post.setRequestBody(formpPairs);
415 int status = httpClient.executeMethod(post);
416 if (status == HttpStatus.SC_OK) {
417 byte[] htmlDatas = post.getResponseBody();
418 code = authorizeAgain(htmlDatas, ticket, authorizeURL,
419 redirectURI, clientId, username, uid);
420 }else if (status == 302) {
421 Header locationHeader = post.getResponseHeader("location");
422 String location = locationHeader.getValue();
423 code = location.substring(location.indexOf("=")+1);
424 }
425
426 return code;
427 }
428
429 /**
430 * 二次提交授权申请
431 * @param htmlDatas 第一次授权申请返回的页面数据
432 * @return
433 * @throws IOException
434 * @throws HttpException
435 */
436 private String authorizeAgain(byte[] htmlDatas, String ticket, String authorizeURL,
437 String redirectURI,String clientId, String username,
438 String uid) throws HttpException, IOException {
439
440 String verifyToken = null;
441 String html = new String(htmlDatas, "utf-8");
442 Document doc = Jsoup.parse(html);
443 Element verifyTokeneElement = doc.select("input[name=verifyToken]").first();
444 verifyToken = verifyTokeneElement.attr("value");
445
446
447 String code = null;
448 String url = authorizeURL + "?client_id=" + clientId + "&redirect_uri="
449 + redirectURI + "&response_type=code&forcelogin=true";
450 String regCallback = authorizeURL + "?client_id=" + clientId + "&redirect_uri="
451 + redirectURI + "&response_type=code&display=default&from=&with_cookie=";
452 PostMethod post = new PostMethod(authorizeURL);
453 //模拟申请token的链接，如果不添加，那么回调地址返回则为空
454 post.setRequestHeader("Referer",authorizeURL);
455 // 模拟登录时所要提交的参数信息
456 NameValuePair[] formpPairs=new NameValuePair[]{
457 new NameValuePair("action", "authorize"),
458 new NameValuePair("uid",uid),
459 new NameValuePair("url", url),
460 new NameValuePair("response_type", "code"),
461 new NameValuePair("redirect_uri", redirectURI),
462 new NameValuePair("client_id", clientId),
463 new NameValuePair("verifyToken", verifyToken),
464 new NameValuePair("regCallback", URLEncoder.encode(regCallback, "UTF-8"))
465 };
466 post.setRequestBody(formpPairs);
467 int status = httpClient.executeMethod(post);
468 if (status == 302) {
469 Header locationHeader = post.getResponseHeader("location");
470 String location = locationHeader.getValue();
471 if (location == null) {
472 throw new NullPointerException("redirect_uri is null");
473 }
474 code = location.substring(location.indexOf("=")+1);
475 }
476 return code;
477 }
478
479 /**
480 * 模拟用户预登录
481 * @param unameBase64
482 * @return
483 * @throws IOException
484 */
485 private HashMap<String, String> pubKeyMap(String unameBase64)
486 throws IOException {
487
488 String url = "https://login.sina.com.cn/sso/prelogin.php?"
489 + "entry=openapi&"
490 + "callback=sinaSSOController.preloginCallBack&" + "su="
491 + unameBase64 + "&" + "rsakt=mod&" + "checkpin=1&"
492 + "client=ssologin.js(v1.4.5)" + "&_=" + new Date().getTime();
493 return getParaFromResult(get(url));
494 }
495
496 /**
497 * 预登陆是否需要验证码
498 * @param pubkeyMap
499 * @return
500 */
501 private String getPin(HashMap<String, String> pubkeyMap) {
502
503 String imgUrl = null;
504 int isShowpin = 0;
505 if (pubkeyMap != null) {
506 String showpin = pubkeyMap.get("showpin");
507 if (showpin != null) {
508 isShowpin = Integer.parseInt(showpin);
509 if (isShowpin == 1) {
510 String url = "https://login.sina.com.cn/cgi/pin.php?"
511 + "r=" + Math.floor(Math.random() * 100000000)
512 + "&s=0"
513 + "&p=" + pubkeyMap.get("pcid");
514
515 imgUrl = url;
516 }
517 }
518 }
519 return imgUrl;
520 }
521
522
523 /**
524 * 确认登陆后是否需要再验证
525 * @return
526 */
527 private String isHasPinAgain(HashMap<String, String> pubkeyMap,
528 HashMap<String, String> ticketMap) {
529
530 String imgUrl = null;
531 int isHasPin = 0;
532 if ((pubkeyMap != null) && (ticketMap != null)) {
533 //意为"为了您的帐号安全，请输入验证码"
534 String str = "\\u4e3a\\u4e86\\u60a8\\u7684\\u5e10\\u53f7\\u5b89" +
535 "\\u5168\\uff0c\\u8bf7\\u8f93\\u5165\\u9a8c\\u8bc1\\u7801";
536
537 if (ticketMap.containsKey("reason")) {
538 String reasonStr = ticketMap.get("reason");
539 if (reasonStr.equals(str)) {
540 isHasPin = 1;
541 String url = "https://login.sina.com.cn/cgi/pin.php?"
542 + "r=" + Math.floor(Math.random() * 100000000)
543 + "&s=0"
544 + "&p=" + pubkeyMap.get("pcid");
545
546 imgUrl = url;
547 }
548 }
549 }
550 return imgUrl;
551 }
552
553 /**
554 * 获取验证码
555 */
556 public String getVCode(String pcid) {
557
558 String imgUrl = null;
559 if (pcid != null) {
560 String url = "https://login.sina.com.cn/cgi/pin.php?"
561 + "r=" + Math.floor(Math.random() * 100000000)
562 + "&s=0"
563 + "&p=" + pcid;
564
565 imgUrl = url;
566 }
567 return imgUrl;
568 }
569
570 /**
571 * 保存验证码
572 * @param url 验证码链接
573 */
574 public void saveVCodeImg(String url) {
575
576 GetMethod getImages = new GetMethod(url);
577 try {
578 int status = httpClient.executeMethod(getImages);
579 if (status == HttpStatus.SC_OK) {
580 FileOutputStream outputStream = new FileOutputStream("vc.jpg");
581 outputStream.write(getImages.getResponseBody());
582 outputStream.close();
583 }
584 } catch (HttpException e) {
585 // TODO Auto-generated catch block
586 e.printStackTrace();
587 } catch (IOException e) {
588 // TODO Auto-generated catch block
589 e.printStackTrace();
590 }
591
592 }
593
594 /**
595 * 无验证码时模拟用户登录,并获取ticket
596 * @param usernameBase64 使用Base64加密的用户名
597 * @param sp 使用SHA1加密后的用户密码
598 * @return
599 * @throws Exception
600 */
601 private HashMap<String, String> getTicket(String usernameBase64,
602 String sp, HashMap<String, String> pubkeyMap) throws Exception {
603 String url = null;
604 if (pubkeyMap != null) {
605 url = "https://login.sina.com.cn/sso/login.php?"
606 + "entry=openapi&"
607 + "gateway=1&"
608 + "from=&"
609 + "savestate=0&"
610 + "useticket=1&"
611 + "pagerefer=&"
612 + "ct=1800&"
613 + "s=1&"
614 + "vsnf=1&"
615 + "vsnval=&"
616 + "door=&"
617 + "su="+ usernameBase64
618 + "&"
619 + "service=miniblog&"
620 + "servertime="+ pubkeyMap.get("servertime")
621 + "&"
622 + "nonce="+ pubkeyMap.get("nonce")
623 + "&"
624 + "pwencode=rsa&"
625 + "rsakv="+ pubkeyMap.get("rsakv")
626 + "&"
627 + "sp="+ sp
628 + "&"
629 + "encoding=UTF-8&"
630 + "callback=sinaSSOController.loginCallBack&"
631 + "cdult=2&"
632 + "domain=weibo.com&"
633 + "prelt=37&"
634 + "returntype=TEXT&"
635 + "client=ssologin.js(v1.4.5)&" + "_=" + new Date().getTime();
636
637 }
638 return getParaFromResult(get(url));
639 }
640
641
642 /**
643 * 有验证码时模拟用户登录,并获取ticket
644 * @param usernameBase64
645 * @param sp
646 * @param pin
647 * @param pcid
648 * @param servertime
649 * @param nonce
650 * @param rsakv
651 * @return
652 * @throws Exception
653 */
654 public HashMap<String, String> getTicket(String usernameBase64, String sp, String pin,
655 String pcid, String servertime,String nonce,String rsakv) throws Exception {
656
657 String url = "https://login.sina.com.cn/sso/login.php?"
658 + "entry=openapi&"
659 + "gateway=1&"
660 + "from=&"
661 + "savestate=0&"
662 + "useticket=1&"
663 + "pagerefer=&"
664 + "pcid=" + pcid + "&"
665 + "ct=1800&"
666 + "s=1&"
667 + "vsnf=1&"
668 + "vsnval=&"
669 + "door=" + pin + "&"
670 + "su="+ usernameBase64
671 + "&"
672 + "service=miniblog&"
673 + "servertime="+ servertime
674 + "&"
675 + "nonce="+ nonce
676 + "&"
677 + "pwencode=rsa&"
678 + "rsakv="+ rsakv
679 + "&"
680 + "sp="+ sp
681 + "&"
682 + "encoding=UTF-8&"
683 + "callback=sinaSSOController.loginCallBack&"
684 + "cdult=2&"
685 + "domain=weibo.com&"
686 + "prelt=37&"
687 + "returntype=TEXT&"
688 + "client=ssologin.js(v1.4.5)&" + "_=" + new Date().getTime();
689
690 return getParaFromResult(get(url));
691 }
692
693 /**
694 * 分析结果,取出所需参数
695 * @param result 页面内容
696 * @return
697 */
698 private HashMap<String, String> getParaFromResult(String result) {
699
700 HashMap<String, String> hm = new HashMap<String, String>();
701 result = result.substring(result.indexOf("{") + 1, result.indexOf("}"));
702 String[] r = result.split(",");
703 String[] temp;
704 for (int i = 0; i < r.length; i++) {
705 temp = r[i].split(":");
706 for (int j = 0; j < 2; j++) {
707 if (temp[j].contains("\""))
708 temp[j] = temp[j].substring(1, temp[j].length() - 1);
709 }
710 hm.put(temp[0], temp[1]);
711 }
712 return hm;
713 }
714
715 /**
716 * 执行给定的URL,并输出目标URL返回的页面结果
717 * @param url
718 * @return
719 * @throws IOException
720 */
721 private String get(String url) throws IOException {
722
723 String surl = null;
724 GetMethod getMethod = new GetMethod(url);
725 int status = httpClient.executeMethod(getMethod);
726 if (status == HttpStatus.SC_OK) {
727 surl = new String(getMethod.getResponseBody(), "UTF-8");
728 }
729 getMethod.releaseConnection();
730 return surl;
731 }
732
733 /**
734 * 配置信息初始化
735 * @return
736 */
737 private Properties initProperties() {
738
739 Properties prop = new Properties();
740 try {
741 prop.load(Thread.currentThread().getContextClassLoader().
742 getResourceAsStream("config.properties"));
743
744 } catch (IOException e) {
745 // TODO Auto-generated catch block
746 e.printStackTrace();
747 }
748 return prop;
749 }
750
751 /**
752 * @param args
753 */
754 public static void main(String[] args) {
755
756 WeiboLoginer loginer = new WeiboLoginer();
757 LoginParams loginParams = loginer.doLogin("","");
758 //有验证码时
759 if (loginParams.getCode() == null) {
760 String pcid = loginParams.getPcid();
761 String nonce = loginParams.getNonce();
762 String rsakv = loginParams.getRsakv();
763 String servertime = loginParams.getServertime();
764 String sp = loginParams.getSp();
765
766 System.err.println(loginParams.getImgUrl());
767 //再次获取验证码
768 System.err.println(loginer.getVCode(pcid));
769
770 Scanner input = new Scanner(System.in);
771 String pin = input.nextLine();
772
773 LoginParams loginResult = loginer.doLoginByPin("",sp, pin, pcid, servertime, nonce, rsakv);
774 if (!loginResult.isLogin()) {
775 System.err.println("验证码错误！重新录入");
776
777 //获取验证码并保存（测试）
778 String imgUrl = loginer.getVCode(pcid);
779 loginer.saveVCodeImg(imgUrl);
780
781 Scanner input1= new Scanner(System.in);
782 String pin1 = input1.nextLine();
783
784 String code = loginer.doLoginByPin("",sp, pin1, pcid, servertime, nonce, rsakv).getCode();
785 System.out.println(SinaWeiboOAuth.getToken(code));
786 }
787
788 }else {
789 //无验证码时
790 String code = loginParams.getCode();
791 System.out.println(SinaWeiboOAuth.getToken(code));
792 }
793 }
794
795 }
796
797
798
799 上述代码完整模拟了微博登陆的全过程，并获得最终授权
800
801
802
803
新浪微博数据抓取(java实现)相关推荐
- php抓取新浪微博数据抓取,php获取新浪微博数据API实例
php获取新浪微博数据API实例 发布于 2014-11-29 12:36:06 | 118 次阅读 | 评论: 0 | 来源: 网友投递 PHP开源脚本语言PHP(外文名: Hypertext Pr ...
- php抓取新浪微博数据抓取,php利用curl抓取新浪微博内容示例
很多人都喜欢在网站上DIY自己的微博,所以我也写了一个. 这里直接抓取了新浪微博工具中的微博秀地址. 代码如下: set_time_limit(0); $url="http://widget ...
- php抓取新浪微博数据抓取,php利用curl抓取新浪微博内容实例
很多人都喜欢在网站上DIY自己的微博,所以我也写了一个. 这里直接抓取了新浪微博工具中的微博秀地址. 复制代码 代码如下: set_time_limit(0); $url="http://w ...
- java模拟新浪微博_Java模拟新浪微博登陆抓取数据
前言: 兄弟们来了来了,最近有人在问如何模拟新浪微博登陆抓取数据,我听后默默地抽了一口老烟,暗暗的对自己说,老汉是时候该你出场了,所以今天有时间就整理整理,浅谈一二. 首先: 要想登陆新浪微博需要预登 ...
- 分享：Python使用cookielib和urllib2模拟登陆新浪微博并抓取数据
Python使用cookielib和urllib2模拟登陆新浪微博并抓取数据 http://my.oschina.net/leopardsaga/blog/94774
- 基于Java的阿里妈妈数据抓取技术
基于Java的阿里妈妈数据抓取技术 前言: 对于需要登录的网站爬虫最大的困难就是需要登录,然后才能获取到数据,如微博,阿里妈妈,webqq等.之前也有看过使用浏览器登录到网站后直接从浏览器中获取coo ...
- java爬虫实现数据抓取
这篇博客我们将用java来实现基本的爬虫数据抓取操作,基本思路就是通过java访问url,然后拿到返回的html文档,并用jsoup解析. 首先我们来看下需要抓取的页面,以我自己的csdn博客列表页面 ...
- java分页抓取数据_网页分页数据抓取的几种方式
相信所有个人网站的站长都有抓取别人数据的经历吧,目前抓取别人网站数据的方式无非两种方式: 一.使用第三方工具,其中最著名的是火车头采集器,在此不做介绍. 二.自己写程序抓取,这种方式要求站长自己写程序 ...
- 查询数据 抓取 网站数据_有了数据，我就学会了如何在几个小时内抓取网站，您也可以...
查询数据 抓取 网站数据 I had a shameful secret. It is one that affects a surprising number of people in the da ...
最新文章
- Java泛型：泛型类、泛型接口和泛型方法
- Ubuntu 18.04 LTS安装配置驱动、CUDA9.1和Amber18 AmberTools18
- 江苏“超牛”女博导：16岁考入北大，跨专业读博，成为全球第4个获此奖项的学者！...
- IP Header包
- JDBC连接方式有哪几种
- mxnet java 例子_mxnet(gluon) 实现DQN简单小例子
- linux 5005端口,Linux配置防火墙端口 8080端口
- 信息学奥赛一本通（1226：装箱问题）
- SpringBoot学习（五）
- 拿走不谢：一份历经线上考验的大规模系统的消息队列技术方案！
- ALSA驱动、设备函数调用关系
- 计算机应用技术专业的未来,计算机应用技术专业召开“2020级人才培养方案修订”主题网络研讨会...
- protel dxp教程 protel dxp经典视频教程
- 电感值、电容值的理解（上）
- L0到L4超全介绍！30+自动驾驶方案汇总
- python_sklearn机器学习算法系列之LogisticRegression(逻辑回归)----识别垃圾邮件（短信）
- POJ - 3311 Hie with the Pie
- 适用于Win和Mac的专业电脑数据恢复软件EasyRecovery易恢复14
- Android Studio 安装应用失败总结
- VM虚拟机扩容centos8的硬盘