1 | package pkg.crawler.data.conn; |
---|
2 | |
---|
3 | import java.io.IOException; |
---|
4 | import java.io.ObjectOutputStream; |
---|
5 | import java.net.Socket; |
---|
6 | import java.net.UnknownHostException; |
---|
7 | import java.util.ArrayList; |
---|
8 | import java.util.Date; |
---|
9 | import java.util.List; |
---|
10 | import java.util.Properties; |
---|
11 | |
---|
12 | import javax.mail.Address; |
---|
13 | import javax.mail.BodyPart; |
---|
14 | import javax.mail.Folder; |
---|
15 | import javax.mail.Message; |
---|
16 | import javax.mail.MessagingException; |
---|
17 | import javax.mail.Multipart; |
---|
18 | import javax.mail.Part; |
---|
19 | import javax.mail.Session; |
---|
20 | |
---|
21 | import org.apache.solr.client.solrj.SolrServer; |
---|
22 | import org.apache.solr.client.solrj.impl.HttpSolrServer; |
---|
23 | import org.apache.solr.common.SolrInputDocument; |
---|
24 | |
---|
25 | import pkg.crawler.entity.Message2SolrEntity; |
---|
26 | |
---|
27 | import com.sun.mail.imap.ACL; |
---|
28 | import com.sun.mail.imap.IMAPFolder; |
---|
29 | import com.sun.mail.imap.IMAPMessage; |
---|
30 | import com.sun.mail.imap.IMAPStore; |
---|
31 | import com.sun.mail.imap.Rights; |
---|
32 | import com.sun.mail.imap.Rights.Right; |
---|
33 | |
---|
34 | public class Crawler { |
---|
35 | |
---|
36 | private String hostSolr; |
---|
37 | private ACL acl; |
---|
38 | private IMAPFolder imapFINBOXAux; |
---|
39 | private IMAPStore store; |
---|
40 | private String strIMAPFolderAux; |
---|
41 | |
---|
42 | public Crawler(String host, String user, String password, String imapFINBOXAux, String hostSolr) throws MessagingException |
---|
43 | { |
---|
44 | Properties props = System.getProperties(); |
---|
45 | Session session = Session.getInstance(props, null); |
---|
46 | strIMAPFolderAux = imapFINBOXAux; |
---|
47 | store = new IMAPStore(session, null); |
---|
48 | |
---|
49 | store.connect(host, user, password); |
---|
50 | |
---|
51 | this.hostSolr = hostSolr; |
---|
52 | this.imapFINBOXAux = (IMAPFolder)store.getFolder(imapFINBOXAux); |
---|
53 | |
---|
54 | //Cria a ACL a ser configurada |
---|
55 | acl = new ACL("expresso-admin"); |
---|
56 | |
---|
57 | //Define direito de leitura |
---|
58 | Rights rights = new Rights(); |
---|
59 | rights.add(Right.READ); |
---|
60 | |
---|
61 | //Configura direito de leitura para a ACL |
---|
62 | acl.setRights(rights); |
---|
63 | } |
---|
64 | |
---|
65 | public void run() { |
---|
66 | //Cria uma instância de conexão com o servidor Solr |
---|
67 | SolrServer solrServer = new HttpSolrServer(hostSolr); |
---|
68 | |
---|
69 | //Cria |
---|
70 | List<Message2SolrEntity> listIMAPMsgs = new ArrayList<Message2SolrEntity>(); |
---|
71 | //INBOX |
---|
72 | |
---|
73 | //Adiciona ACL de permissão de leitura, para o usuário expresso-admin conseguir fazer a leitura |
---|
74 | try |
---|
75 | { |
---|
76 | imapFINBOXAux.addACL(acl); |
---|
77 | |
---|
78 | |
---|
79 | //Abre a caixa de e-mail |
---|
80 | imapFINBOXAux.open(Folder.READ_ONLY); |
---|
81 | |
---|
82 | //Carrega as mensagens da caixa para um array |
---|
83 | Message[] msgsINBOX = imapFINBOXAux.getMessages(); |
---|
84 | |
---|
85 | |
---|
86 | int iCount = 0; |
---|
87 | //Faz a iteração entre as mensagens |
---|
88 | for (Message msgAuxINBOX : msgsINBOX) |
---|
89 | { |
---|
90 | |
---|
91 | IMAPMessage m = (IMAPMessage)msgAuxINBOX; |
---|
92 | |
---|
93 | //adiciona as mensagens na varÃavel listIMAPMsgs |
---|
94 | try |
---|
95 | { |
---|
96 | dumpPart(m, listIMAPMsgs); |
---|
97 | iCount++; |
---|
98 | }catch(Exception e) |
---|
99 | { |
---|
100 | continue; |
---|
101 | } |
---|
102 | msgAuxINBOX = null; |
---|
103 | m = null; |
---|
104 | |
---|
105 | |
---|
106 | SolrInputDocument doc = new SolrInputDocument(); |
---|
107 | doc.addField("id", listIMAPMsgs.get(listIMAPMsgs.size()-1).getId().toString()); |
---|
108 | doc.addField("user", listIMAPMsgs.get(listIMAPMsgs.size()-1).getUser().toString()); |
---|
109 | doc.addField("folder", listIMAPMsgs.get(listIMAPMsgs.size()-1).getFolder().toString()); |
---|
110 | doc.addField("msg_no", listIMAPMsgs.get(listIMAPMsgs.size()-1).getMsgNo()); |
---|
111 | doc.addField("from", listIMAPMsgs.get(listIMAPMsgs.size()-1).getFrom().toString()); |
---|
112 | doc.addField("to", listIMAPMsgs.get(listIMAPMsgs.size()-1).getTo().toString()); |
---|
113 | doc.addField("subject", listIMAPMsgs.get(listIMAPMsgs.size()-1).getSubject().toString()); |
---|
114 | doc.addField("content", listIMAPMsgs.get(listIMAPMsgs.size()-1).getContent().toString()); |
---|
115 | doc.addField("copyto", listIMAPMsgs.get(listIMAPMsgs.size()-1).getCopyto().toString()); |
---|
116 | doc.addField("sent_date", listIMAPMsgs.get(listIMAPMsgs.size()-1).getSent_date()); |
---|
117 | doc.addField("hiddencopyto", listIMAPMsgs.get(listIMAPMsgs.size()-1).getHiddencopyto().toString()); |
---|
118 | |
---|
119 | try { |
---|
120 | solrServer.add(doc); |
---|
121 | } catch (Exception e) { |
---|
122 | System.err.println("solr -> " + e.getMessage()); |
---|
123 | } |
---|
124 | |
---|
125 | |
---|
126 | if(iCount%200 == 0) |
---|
127 | { |
---|
128 | solrServer.commit(); |
---|
129 | |
---|
130 | listIMAPMsgs.clear(); |
---|
131 | |
---|
132 | solrServer = new HttpSolrServer(hostSolr); |
---|
133 | } |
---|
134 | |
---|
135 | } |
---|
136 | imapFINBOXAux.removeACL("expresso-admin"); |
---|
137 | for(int i = 0; i < listIMAPMsgs.size(); i++ ) |
---|
138 | { |
---|
139 | SolrInputDocument doc = new SolrInputDocument(); |
---|
140 | doc.addField("id", listIMAPMsgs.get(i).getId().toString()); |
---|
141 | doc.addField("user", listIMAPMsgs.get(i).getUser().toString()); |
---|
142 | doc.addField("folder", listIMAPMsgs.get(i).getFolder().toString()); |
---|
143 | doc.addField("msg_no", listIMAPMsgs.get(i).getMsgNo()); |
---|
144 | doc.addField("from", listIMAPMsgs.get(i).getFrom().toString()); |
---|
145 | doc.addField("to", listIMAPMsgs.get(i).getTo().toString()); |
---|
146 | doc.addField("subject", listIMAPMsgs.get(i).getSubject().toString()); |
---|
147 | doc.addField("content", listIMAPMsgs.get(i).getContent().toString()); |
---|
148 | doc.addField("copyto", listIMAPMsgs.get(i).getCopyto().toString()); |
---|
149 | doc.addField("sent_date", listIMAPMsgs.get(i).getSent_date()); |
---|
150 | doc.addField("hiddencopyto", listIMAPMsgs.get(i).getHiddencopyto().toString()); |
---|
151 | |
---|
152 | try { |
---|
153 | solrServer.add(doc); |
---|
154 | } catch (Exception e) { |
---|
155 | System.err.println("solr -> " + e.getMessage()); |
---|
156 | } |
---|
157 | |
---|
158 | } |
---|
159 | |
---|
160 | listIMAPMsgs = null; |
---|
161 | |
---|
162 | solrServer.commit(); |
---|
163 | |
---|
164 | solrServer = null; |
---|
165 | |
---|
166 | for (Folder fAux : imapFINBOXAux.list()) |
---|
167 | { |
---|
168 | //Verifica se não é uma pasta compartilhada |
---|
169 | if(fAux.getFullName().split("/").length <= 3 && !fAux.getFullName().split("/")[2].equals("user")) |
---|
170 | { |
---|
171 | //INBOX |
---|
172 | IMAPFolder imapFAux = (IMAPFolder)fAux; |
---|
173 | |
---|
174 | crawIntoUserFolders(imapFAux); |
---|
175 | |
---|
176 | } |
---|
177 | } |
---|
178 | imapFINBOXAux.close(true); |
---|
179 | imapFINBOXAux = null; |
---|
180 | |
---|
181 | |
---|
182 | Socket client = null; |
---|
183 | ObjectOutputStream oos = null; |
---|
184 | try { |
---|
185 | client = new Socket("127.1.1.1", 8090); |
---|
186 | |
---|
187 | oos = new ObjectOutputStream(client.getOutputStream()); |
---|
188 | |
---|
189 | oos.writeObject(strIMAPFolderAux); |
---|
190 | |
---|
191 | oos.close(); |
---|
192 | |
---|
193 | client.close(); |
---|
194 | } catch (UnknownHostException e) { |
---|
195 | e.printStackTrace(); |
---|
196 | } catch (IOException e) { |
---|
197 | e.printStackTrace(); |
---|
198 | } |
---|
199 | } |
---|
200 | catch (Exception e) |
---|
201 | { |
---|
202 | Socket client = null; |
---|
203 | ObjectOutputStream oos = null; |
---|
204 | try { |
---|
205 | client = new Socket("127.1.1.1", 8090); |
---|
206 | |
---|
207 | oos = new ObjectOutputStream(client.getOutputStream()); |
---|
208 | |
---|
209 | oos.writeObject(strIMAPFolderAux); |
---|
210 | |
---|
211 | oos.close(); |
---|
212 | |
---|
213 | client.close(); |
---|
214 | } catch (UnknownHostException ee) { |
---|
215 | e.printStackTrace(); |
---|
216 | } catch (IOException ee) { |
---|
217 | e.printStackTrace(); |
---|
218 | } |
---|
219 | return; |
---|
220 | } |
---|
221 | |
---|
222 | |
---|
223 | |
---|
224 | } |
---|
225 | |
---|
226 | private void crawIntoUserFolders(IMAPFolder imapFAux) throws Exception |
---|
227 | { |
---|
228 | SolrServer solrServer = new HttpSolrServer(hostSolr); |
---|
229 | |
---|
230 | List<Message2SolrEntity> listIMAPMsgs = new ArrayList<Message2SolrEntity>(); |
---|
231 | |
---|
232 | try |
---|
233 | { |
---|
234 | imapFAux.addACL(acl); |
---|
235 | }catch(Exception e) |
---|
236 | { |
---|
237 | return; |
---|
238 | } |
---|
239 | |
---|
240 | imapFAux.open(Folder.READ_ONLY); |
---|
241 | |
---|
242 | Message[] msgs= imapFAux.getMessages(); |
---|
243 | |
---|
244 | int iCount = 0; |
---|
245 | for (Message msgAux: msgs) |
---|
246 | { |
---|
247 | IMAPMessage m = (IMAPMessage)msgAux; |
---|
248 | try |
---|
249 | { |
---|
250 | dumpPart(m, listIMAPMsgs); |
---|
251 | iCount++; |
---|
252 | }catch(Exception e) |
---|
253 | { |
---|
254 | continue; |
---|
255 | } |
---|
256 | msgAux = null; |
---|
257 | m = null; |
---|
258 | |
---|
259 | SolrInputDocument doc = new SolrInputDocument(); |
---|
260 | doc.addField("id", listIMAPMsgs.get(listIMAPMsgs.size()-1).getId().toString()); |
---|
261 | doc.addField("user", listIMAPMsgs.get(listIMAPMsgs.size()-1).getUser().toString()); |
---|
262 | doc.addField("folder", listIMAPMsgs.get(listIMAPMsgs.size()-1).getFolder().toString()); |
---|
263 | doc.addField("msg_no", listIMAPMsgs.get(listIMAPMsgs.size()-1).getMsgNo()); |
---|
264 | doc.addField("from", listIMAPMsgs.get(listIMAPMsgs.size()-1).getFrom().toString()); |
---|
265 | doc.addField("to", listIMAPMsgs.get(listIMAPMsgs.size()-1).getTo().toString()); |
---|
266 | doc.addField("subject", listIMAPMsgs.get(listIMAPMsgs.size()-1).getSubject().toString()); |
---|
267 | doc.addField("content", listIMAPMsgs.get(listIMAPMsgs.size()-1).getContent().toString()); |
---|
268 | doc.addField("copyto", listIMAPMsgs.get(listIMAPMsgs.size()-1).getCopyto().toString()); |
---|
269 | doc.addField("sent_date", listIMAPMsgs.get(listIMAPMsgs.size()-1).getSent_date()); |
---|
270 | doc.addField("hiddencopyto", listIMAPMsgs.get(listIMAPMsgs.size()-1).getHiddencopyto().toString()); |
---|
271 | |
---|
272 | try { |
---|
273 | solrServer.add(doc); |
---|
274 | } catch (Exception e) { |
---|
275 | System.err.println("solr -> " + e.getMessage()); |
---|
276 | } |
---|
277 | |
---|
278 | |
---|
279 | if(iCount%200 == 0) |
---|
280 | { |
---|
281 | solrServer.commit(); |
---|
282 | |
---|
283 | listIMAPMsgs.clear(); |
---|
284 | |
---|
285 | solrServer = new HttpSolrServer(hostSolr); |
---|
286 | } |
---|
287 | |
---|
288 | } |
---|
289 | imapFAux.removeACL("expresso-admin"); |
---|
290 | System.out.println(imapFAux.getFullName()); |
---|
291 | for(int i = 0; i < listIMAPMsgs.size(); i++ ) |
---|
292 | { |
---|
293 | SolrInputDocument doc = new SolrInputDocument(); |
---|
294 | doc.addField("id", listIMAPMsgs.get(i).getId()); |
---|
295 | doc.addField("user", listIMAPMsgs.get(i).getUser()); |
---|
296 | doc.addField("folder", listIMAPMsgs.get(i).getFolder()); |
---|
297 | doc.addField("msg_no", listIMAPMsgs.get(i).getMsgNo()); |
---|
298 | doc.addField("from", listIMAPMsgs.get(i).getFrom()); |
---|
299 | doc.addField("to", listIMAPMsgs.get(i).getTo()); |
---|
300 | doc.addField("subject", listIMAPMsgs.get(i).getSubject()); |
---|
301 | doc.addField("content", listIMAPMsgs.get(i).getContent()); |
---|
302 | doc.addField("copyto", listIMAPMsgs.get(i).getCopyto()); |
---|
303 | doc.addField("sent_date", listIMAPMsgs.get(i).getSent_date()); |
---|
304 | doc.addField("hiddencopyto", listIMAPMsgs.get(i).getHiddencopyto()); |
---|
305 | |
---|
306 | } |
---|
307 | listIMAPMsgs = null; |
---|
308 | imapFAux.close(true); |
---|
309 | imapFAux = null; |
---|
310 | solrServer.commit(); |
---|
311 | solrServer = null; |
---|
312 | } |
---|
313 | |
---|
314 | private void dumpPart(IMAPMessage m, List<Message2SolrEntity> listIMAPMsgs) throws Exception |
---|
315 | { |
---|
316 | //Verifica se possui identificador para poder continuar |
---|
317 | if(m.getMessageID() == null || m.getMessageID().equals("")){ |
---|
318 | return; |
---|
319 | } |
---|
320 | |
---|
321 | Message2SolrEntity msgEntity = null; |
---|
322 | msgEntity = new Message2SolrEntity(); |
---|
323 | |
---|
324 | msgEntity.setId(new StringBuilder( m.getMessageID())); |
---|
325 | |
---|
326 | String user = m.getFolder().getFullName().split("/")[1]; |
---|
327 | msgEntity.setUser(new StringBuilder(user)); |
---|
328 | |
---|
329 | String folder = m.getFolder().getFullName().split("/") |
---|
330 | [m.getFolder().getFullName().split("/").length-1]; |
---|
331 | |
---|
332 | if(!user.trim().equals(folder.trim())) |
---|
333 | { |
---|
334 | msgEntity.setFolder(new StringBuilder(folder)); |
---|
335 | } |
---|
336 | else |
---|
337 | { |
---|
338 | msgEntity.setFolder(new StringBuilder("INBOX")); |
---|
339 | } |
---|
340 | |
---|
341 | msgEntity.setMsgNo(new StringBuilder(String.valueOf(m.getMessageNumber()))); |
---|
342 | |
---|
343 | |
---|
344 | if(m.getSubject() != null){ |
---|
345 | msgEntity.setSubject(new StringBuilder( m.getSubject() )); |
---|
346 | } |
---|
347 | Address[] a; |
---|
348 | // FROM |
---|
349 | if ((a = m.getFrom()) != null) { |
---|
350 | for (int j = 0; j < a.length; j++) |
---|
351 | { |
---|
352 | if(msgEntity.getFrom() == null) |
---|
353 | { |
---|
354 | msgEntity.setFrom(new StringBuilder(a[j].toString())); |
---|
355 | } |
---|
356 | else |
---|
357 | { |
---|
358 | msgEntity.setFrom(msgEntity.getFrom().append(", ").append(a[j].toString())); |
---|
359 | } |
---|
360 | // System.out.println("FROM: " + a[j].toString()); |
---|
361 | } |
---|
362 | |
---|
363 | } |
---|
364 | |
---|
365 | // TO |
---|
366 | if ((a = m.getRecipients(Message.RecipientType.TO)) != null) { |
---|
367 | for (int j = 0; j < a.length; j++) |
---|
368 | { |
---|
369 | if(msgEntity.getTo() == null) |
---|
370 | { |
---|
371 | msgEntity.setTo(new StringBuilder(a[j].toString())); |
---|
372 | } |
---|
373 | else |
---|
374 | { |
---|
375 | msgEntity.setTo(msgEntity.getTo().append(", ").append(a[j].toString())); |
---|
376 | } |
---|
377 | } |
---|
378 | } |
---|
379 | |
---|
380 | // CC |
---|
381 | if ((a = m.getRecipients(Message.RecipientType.CC)) != null) { |
---|
382 | for (int j = 0; j < a.length; j++) |
---|
383 | { |
---|
384 | if(msgEntity.getCopyto() == null) |
---|
385 | { |
---|
386 | msgEntity.setCopyto(new StringBuilder(a[j].toString())); |
---|
387 | } |
---|
388 | else |
---|
389 | { |
---|
390 | msgEntity.setCopyto(msgEntity.getCopyto().append(", ").append(a[j].toString())); |
---|
391 | } |
---|
392 | } |
---|
393 | } |
---|
394 | // CC |
---|
395 | if ((a = m.getRecipients(Message.RecipientType.BCC)) != null) { |
---|
396 | for (int j = 0; j < a.length; j++) |
---|
397 | { |
---|
398 | if(msgEntity.getHiddencopyto() == null) |
---|
399 | { |
---|
400 | msgEntity.setHiddencopyto(new StringBuilder(a[j].toString())); |
---|
401 | } |
---|
402 | else |
---|
403 | { |
---|
404 | msgEntity.setHiddencopyto(new StringBuilder(msgEntity.getHiddencopyto() + ","+a[j].toString())); |
---|
405 | } |
---|
406 | } |
---|
407 | } |
---|
408 | |
---|
409 | // DATE |
---|
410 | Date d = m.getSentDate(); |
---|
411 | if(msgEntity != null) |
---|
412 | { |
---|
413 | if(d != null) |
---|
414 | { |
---|
415 | msgEntity.setSent_date(new StringBuilder(String.valueOf(d.getTime()))); |
---|
416 | } |
---|
417 | else |
---|
418 | { |
---|
419 | msgEntity.setSent_date(new StringBuilder(String.valueOf(new Date().getTime()))); |
---|
420 | } |
---|
421 | } |
---|
422 | |
---|
423 | try{ |
---|
424 | Object o = m.getContent(); |
---|
425 | if (o instanceof String) { |
---|
426 | if( msgEntity != null ) |
---|
427 | { |
---|
428 | msgEntity.setContent(new StringBuilder( o.toString())); |
---|
429 | } |
---|
430 | } else if (o instanceof Multipart) { |
---|
431 | Multipart multipart = (Multipart)o; |
---|
432 | int count = multipart.getCount(); |
---|
433 | |
---|
434 | for (int i = 0; i < count; i++) |
---|
435 | { |
---|
436 | if(msgEntity.getContent() == null || msgEntity.getContent().toString().trim().equals("")) |
---|
437 | { |
---|
438 | msgEntity.setContent(new StringBuilder("")); |
---|
439 | } |
---|
440 | if(multipart.getBodyPart(i).getContentType() != Part.ATTACHMENT) |
---|
441 | { |
---|
442 | msgEntity.setContent(msgEntity.getContent().append(" ").append(getPlainContent(multipart.getBodyPart(i)))); |
---|
443 | } |
---|
444 | |
---|
445 | } |
---|
446 | |
---|
447 | } |
---|
448 | }catch (Exception e){ |
---|
449 | System.out.println("Exception ocurred!"); |
---|
450 | e.printStackTrace(); |
---|
451 | } |
---|
452 | |
---|
453 | if(msgEntity != null && msgEntity.getId() != null){ |
---|
454 | listIMAPMsgs.add(msgEntity); |
---|
455 | } |
---|
456 | m = null; |
---|
457 | } |
---|
458 | |
---|
459 | private StringBuilder getPlainContent(BodyPart bodyPart) throws IOException, MessagingException |
---|
460 | { |
---|
461 | try |
---|
462 | { |
---|
463 | |
---|
464 | if(bodyPart.getContent() instanceof String) |
---|
465 | { |
---|
466 | StringBuilder text = new StringBuilder(bodyPart.getContent().toString()); |
---|
467 | |
---|
468 | return text; |
---|
469 | } |
---|
470 | }catch(Exception e) |
---|
471 | { |
---|
472 | System.out.println(e.getMessage()); |
---|
473 | e.printStackTrace(); |
---|
474 | return new StringBuilder(""); |
---|
475 | } |
---|
476 | |
---|
477 | |
---|
478 | return new StringBuilder(""); |
---|
479 | } |
---|
480 | |
---|
481 | } |
---|