[7576] | 1 | package pkg.crawler.data.conn; |
---|
| 2 | |
---|
| 3 | import java.io.IOException; |
---|
| 4 | import java.util.ArrayList; |
---|
| 5 | import java.util.Date; |
---|
| 6 | import java.util.List; |
---|
| 7 | import java.util.Properties; |
---|
| 8 | |
---|
| 9 | import javax.mail.Address; |
---|
| 10 | import javax.mail.BodyPart; |
---|
| 11 | import javax.mail.Folder; |
---|
| 12 | import javax.mail.Message; |
---|
| 13 | import javax.mail.MessagingException; |
---|
| 14 | import javax.mail.Multipart; |
---|
| 15 | import javax.mail.Part; |
---|
| 16 | import javax.mail.Session; |
---|
| 17 | import javax.swing.JOptionPane; |
---|
| 18 | |
---|
| 19 | import org.apache.solr.client.solrj.SolrServer; |
---|
| 20 | import org.apache.solr.client.solrj.impl.HttpSolrServer; |
---|
| 21 | import org.apache.solr.common.SolrInputDocument; |
---|
| 22 | |
---|
| 23 | import pkg.crawler.entity.Message2SolrEntity; |
---|
| 24 | |
---|
| 25 | import com.sun.mail.imap.ACL; |
---|
| 26 | import com.sun.mail.imap.IMAPFolder; |
---|
| 27 | import com.sun.mail.imap.IMAPMessage; |
---|
| 28 | import com.sun.mail.imap.IMAPStore; |
---|
| 29 | import com.sun.mail.imap.Rights; |
---|
| 30 | import com.sun.mail.imap.Rights.Right; |
---|
| 31 | |
---|
| 32 | public class Crawler implements Runnable{ |
---|
| 33 | |
---|
| 34 | private String host; |
---|
| 35 | private String user; |
---|
| 36 | private String password; |
---|
| 37 | private String mbox; |
---|
| 38 | private String hostSolr; |
---|
| 39 | private String loginStartsWith; |
---|
| 40 | // private ArrayList<Message2SolrEntity> listIMAPMsgs; |
---|
| 41 | private ACL acl; |
---|
| 42 | |
---|
| 43 | public Crawler(String host, String password, String hostSolr, String loginStartsWith) |
---|
| 44 | { |
---|
| 45 | this.host = host; |
---|
| 46 | this.user = "expresso-admin"; |
---|
| 47 | this.password = password; |
---|
| 48 | this.mbox = "user"; |
---|
| 49 | this.loginStartsWith = loginStartsWith; |
---|
| 50 | this.hostSolr = hostSolr; |
---|
| 51 | |
---|
| 52 | //Cria a ACL a ser configurada |
---|
| 53 | acl = new ACL("expresso-admin"); |
---|
| 54 | |
---|
| 55 | //Define direito de leitura |
---|
| 56 | Rights rights = new Rights(); |
---|
| 57 | rights.add(Right.READ); |
---|
| 58 | |
---|
| 59 | //Configura direito de leitura para a ACL |
---|
| 60 | acl.setRights(rights); |
---|
| 61 | } |
---|
| 62 | |
---|
| 63 | |
---|
| 64 | @Override |
---|
| 65 | public void run() |
---|
| 66 | { |
---|
| 67 | while (true) |
---|
| 68 | { |
---|
| 69 | |
---|
| 70 | |
---|
| 71 | //Inicializando as propriedades default da máquina para conexão |
---|
| 72 | //com o servidor IMAP |
---|
| 73 | Properties props = System.getProperties(); |
---|
| 74 | Session session = Session.getInstance(props, null); |
---|
| 75 | |
---|
| 76 | IMAPStore store = new IMAPStore(session, null); |
---|
| 77 | |
---|
| 78 | try { |
---|
| 79 | |
---|
| 80 | //Conectando com o servidor IMAP |
---|
| 81 | store.connect(host, user, password); |
---|
| 82 | //Definindo a pasta base a ser usada |
---|
| 83 | IMAPFolder imapFolder = (IMAPFolder)store.getFolder(mbox); |
---|
| 84 | |
---|
| 85 | //Faz a iteração entre os usuários/pastas do usuário |
---|
| 86 | for (Folder fINBOXAux : imapFolder.list()) |
---|
| 87 | { |
---|
| 88 | //Verifica se esse usuário começa com a letra definida |
---|
| 89 | if(fINBOXAux.getName().charAt(0) != loginStartsWith.charAt(0)) |
---|
| 90 | { |
---|
| 91 | //Caso não seja, pula para o próximo |
---|
| 92 | continue; |
---|
| 93 | } |
---|
| 94 | |
---|
| 95 | //Cria uma instância de conexão com o servidor Solr |
---|
| 96 | SolrServer solrServer = new HttpSolrServer(hostSolr); |
---|
| 97 | |
---|
| 98 | //Cria |
---|
| 99 | List<Message2SolrEntity> listIMAPMsgs = new ArrayList<Message2SolrEntity>(); |
---|
| 100 | //INBOX |
---|
| 101 | IMAPFolder imapFINBOXAux = (IMAPFolder)fINBOXAux; |
---|
| 102 | |
---|
| 103 | imapFINBOXAux.addACL(acl); |
---|
| 104 | |
---|
| 105 | imapFINBOXAux.open(Folder.READ_ONLY); |
---|
| 106 | |
---|
| 107 | Message[] msgsINBOX = imapFINBOXAux.getMessages(); |
---|
| 108 | |
---|
| 109 | for (Message msgAuxINBOX : msgsINBOX) |
---|
| 110 | { |
---|
| 111 | IMAPMessage m = (IMAPMessage)msgAuxINBOX; |
---|
| 112 | dumpPart(m, listIMAPMsgs); |
---|
| 113 | } |
---|
| 114 | imapFINBOXAux.removeACL("expresso-admin"); |
---|
| 115 | for(int i = 0; i < listIMAPMsgs.size(); i++ ) |
---|
| 116 | { |
---|
| 117 | SolrInputDocument doc = new SolrInputDocument(); |
---|
| 118 | doc.addField("id", listIMAPMsgs.get(i).getId().toString()); |
---|
| 119 | doc.addField("user", listIMAPMsgs.get(i).getUser().toString()); |
---|
| 120 | doc.addField("folder", listIMAPMsgs.get(i).getFolder().toString()); |
---|
| 121 | doc.addField("msg_no", listIMAPMsgs.get(i).getMsgNo()); |
---|
| 122 | doc.addField("from", listIMAPMsgs.get(i).getFrom().toString()); |
---|
| 123 | doc.addField("to", listIMAPMsgs.get(i).getTo().toString()); |
---|
| 124 | doc.addField("subject", listIMAPMsgs.get(i).getSubject().toString()); |
---|
| 125 | doc.addField("content", listIMAPMsgs.get(i).getContent().toString()); |
---|
| 126 | doc.addField("copyto", listIMAPMsgs.get(i).getCopyto().toString()); |
---|
| 127 | doc.addField("sent_date", listIMAPMsgs.get(i).getSent_date()); |
---|
| 128 | doc.addField("hiddencopyto", listIMAPMsgs.get(i).getHiddencopyto().toString()); |
---|
| 129 | |
---|
| 130 | try { |
---|
| 131 | solrServer.add(doc); |
---|
| 132 | } catch (Exception e) { |
---|
| 133 | System.err.println("solr -> " + e.getMessage()); |
---|
| 134 | } |
---|
| 135 | |
---|
| 136 | } |
---|
| 137 | |
---|
| 138 | listIMAPMsgs = null; |
---|
| 139 | |
---|
| 140 | solrServer.commit(); |
---|
| 141 | |
---|
| 142 | solrServer = null; |
---|
| 143 | |
---|
| 144 | for (Folder fAux : imapFINBOXAux.list()) |
---|
| 145 | { |
---|
| 146 | //Verifica se não é uma pasta compartilhada |
---|
| 147 | if(fAux.getFullName().split("/").length <= 3 && !fAux.getFullName().split("/")[2].equals("user")) |
---|
| 148 | { |
---|
| 149 | //INBOX |
---|
| 150 | IMAPFolder imapFAux = (IMAPFolder)fAux; |
---|
| 151 | |
---|
| 152 | crawIntoUserFolders(imapFAux); |
---|
| 153 | |
---|
| 154 | } |
---|
| 155 | } |
---|
| 156 | |
---|
| 157 | System.out.println(imapFINBOXAux.getFullName()); |
---|
| 158 | } |
---|
| 159 | store.close(); |
---|
| 160 | |
---|
| 161 | |
---|
| 162 | } catch (MessagingException e) { |
---|
| 163 | System.out.println(e.getMessage()); |
---|
| 164 | e.printStackTrace(); |
---|
| 165 | } |
---|
| 166 | catch (Exception e) |
---|
| 167 | { |
---|
| 168 | System.out.println(e.getMessage()); |
---|
| 169 | e.printStackTrace(); |
---|
| 170 | } |
---|
| 171 | } |
---|
| 172 | } |
---|
| 173 | |
---|
| 174 | private void crawIntoUserFolders(IMAPFolder imapFAux) throws Exception |
---|
| 175 | { |
---|
| 176 | SolrServer solrServer = new HttpSolrServer(hostSolr); |
---|
| 177 | |
---|
| 178 | List<Message2SolrEntity> listIMAPMsgs = new ArrayList<Message2SolrEntity>(); |
---|
| 179 | imapFAux.addACL(acl); |
---|
| 180 | |
---|
| 181 | imapFAux.open(Folder.READ_ONLY); |
---|
| 182 | |
---|
| 183 | Message[] msgs= imapFAux.getMessages(); |
---|
| 184 | |
---|
| 185 | for (Message msgAux: msgs) |
---|
| 186 | { |
---|
| 187 | IMAPMessage m = (IMAPMessage)msgAux; |
---|
| 188 | dumpPart(m, listIMAPMsgs); |
---|
| 189 | } |
---|
| 190 | imapFAux.removeACL("expresso-admin"); |
---|
| 191 | System.out.println(imapFAux.getFullName()); |
---|
| 192 | for(int i = 0; i < listIMAPMsgs.size(); i++ ) |
---|
| 193 | { |
---|
| 194 | SolrInputDocument doc = new SolrInputDocument(); |
---|
| 195 | doc.addField("id", listIMAPMsgs.get(i).getId()); |
---|
| 196 | doc.addField("user", listIMAPMsgs.get(i).getUser()); |
---|
| 197 | doc.addField("folder", listIMAPMsgs.get(i).getFolder()); |
---|
| 198 | doc.addField("msg_no", listIMAPMsgs.get(i).getMsgNo()); |
---|
| 199 | doc.addField("from", listIMAPMsgs.get(i).getFrom()); |
---|
| 200 | doc.addField("to", listIMAPMsgs.get(i).getTo()); |
---|
| 201 | doc.addField("subject", listIMAPMsgs.get(i).getSubject()); |
---|
| 202 | doc.addField("content", listIMAPMsgs.get(i).getContent()); |
---|
| 203 | doc.addField("copyto", listIMAPMsgs.get(i).getCopyto()); |
---|
| 204 | doc.addField("sent_date", listIMAPMsgs.get(i).getSent_date()); |
---|
| 205 | doc.addField("hiddencopyto", listIMAPMsgs.get(i).getHiddencopyto()); |
---|
| 206 | |
---|
| 207 | if(doc.getField("id").toString().trim().equals("")) |
---|
| 208 | { |
---|
| 209 | JOptionPane.showMessageDialog(null, " ¬¬' "); |
---|
| 210 | } |
---|
| 211 | |
---|
| 212 | try { |
---|
| 213 | solrServer.add(doc); |
---|
| 214 | } catch (Exception e) { |
---|
| 215 | System.err.println("solr -> " + e.getMessage()); |
---|
| 216 | } |
---|
| 217 | } |
---|
| 218 | listIMAPMsgs = null; |
---|
| 219 | solrServer.commit(); |
---|
| 220 | solrServer = null; |
---|
| 221 | } |
---|
| 222 | |
---|
| 223 | public void dumpPart(IMAPMessage m, List<Message2SolrEntity> listIMAPMsgs) throws Exception { |
---|
| 224 | |
---|
| 225 | if(m.getMessageID() == null || m.getMessageID().equals("")){ |
---|
| 226 | System.out.println("Damn it!"); |
---|
| 227 | return; |
---|
| 228 | } |
---|
| 229 | |
---|
| 230 | Message2SolrEntity msgEntity = null; |
---|
| 231 | msgEntity = new Message2SolrEntity(); |
---|
| 232 | // System.out.println(m.getSize()); |
---|
| 233 | |
---|
| 234 | msgEntity.setId(new StringBuilder( m.getMessageID())); |
---|
| 235 | |
---|
| 236 | String user = m.getFolder().getFullName().split("/")[1]; |
---|
| 237 | msgEntity.setUser(new StringBuilder(user)); |
---|
| 238 | |
---|
| 239 | String folder = m.getFolder().getFullName().split("/") |
---|
| 240 | [m.getFolder().getFullName().split("/").length-1]; |
---|
| 241 | |
---|
| 242 | if(!user.trim().equals(folder.trim())) |
---|
| 243 | { |
---|
| 244 | msgEntity.setFolder(new StringBuilder(folder)); |
---|
| 245 | } |
---|
| 246 | else |
---|
| 247 | { |
---|
| 248 | msgEntity.setFolder(new StringBuilder("INBOX")); |
---|
| 249 | } |
---|
| 250 | |
---|
| 251 | msgEntity.setMsgNo(new StringBuilder(String.valueOf(m.getMessageNumber()))); |
---|
| 252 | |
---|
| 253 | |
---|
| 254 | if(m.getSubject() != null){ |
---|
| 255 | msgEntity.setSubject(new StringBuilder( m.getSubject() )); |
---|
| 256 | } |
---|
| 257 | Address[] a; |
---|
| 258 | // FROM |
---|
| 259 | if ((a = m.getFrom()) != null) { |
---|
| 260 | for (int j = 0; j < a.length; j++) |
---|
| 261 | { |
---|
| 262 | if(msgEntity.getFrom() == null) |
---|
| 263 | { |
---|
| 264 | msgEntity.setFrom(new StringBuilder(a[j].toString())); |
---|
| 265 | } |
---|
| 266 | else |
---|
| 267 | { |
---|
| 268 | msgEntity.setFrom(msgEntity.getFrom().append(", ").append(a[j].toString())); |
---|
| 269 | } |
---|
| 270 | // System.out.println("FROM: " + a[j].toString()); |
---|
| 271 | } |
---|
| 272 | |
---|
| 273 | } |
---|
| 274 | |
---|
| 275 | // TO |
---|
| 276 | if ((a = m.getRecipients(Message.RecipientType.TO)) != null) { |
---|
| 277 | for (int j = 0; j < a.length; j++) |
---|
| 278 | { |
---|
| 279 | if(msgEntity.getTo() == null) |
---|
| 280 | { |
---|
| 281 | msgEntity.setTo(new StringBuilder(a[j].toString())); |
---|
| 282 | } |
---|
| 283 | else |
---|
| 284 | { |
---|
| 285 | msgEntity.setTo(msgEntity.getTo().append(", ").append(a[j].toString())); |
---|
| 286 | } |
---|
| 287 | } |
---|
| 288 | } |
---|
| 289 | |
---|
| 290 | // CC |
---|
| 291 | if ((a = m.getRecipients(Message.RecipientType.CC)) != null) { |
---|
| 292 | for (int j = 0; j < a.length; j++) |
---|
| 293 | { |
---|
| 294 | if(msgEntity.getCopyto() == null) |
---|
| 295 | { |
---|
| 296 | msgEntity.setCopyto(new StringBuilder(a[j].toString())); |
---|
| 297 | } |
---|
| 298 | else |
---|
| 299 | { |
---|
| 300 | msgEntity.setCopyto(msgEntity.getCopyto().append(", ").append(a[j].toString())); |
---|
| 301 | } |
---|
| 302 | } |
---|
| 303 | } |
---|
| 304 | // CC |
---|
| 305 | if ((a = m.getRecipients(Message.RecipientType.BCC)) != null) { |
---|
| 306 | for (int j = 0; j < a.length; j++) |
---|
| 307 | { |
---|
| 308 | if(msgEntity.getHiddencopyto() == null) |
---|
| 309 | { |
---|
| 310 | msgEntity.setHiddencopyto(new StringBuilder(a[j].toString())); |
---|
| 311 | } |
---|
| 312 | else |
---|
| 313 | { |
---|
| 314 | msgEntity.setHiddencopyto(new StringBuilder(msgEntity.getHiddencopyto() + ","+a[j].toString())); |
---|
| 315 | } |
---|
| 316 | } |
---|
| 317 | } |
---|
| 318 | |
---|
| 319 | // DATE |
---|
| 320 | Date d = m.getSentDate(); |
---|
| 321 | if(msgEntity != null) |
---|
| 322 | { |
---|
| 323 | if(d != null) |
---|
| 324 | { |
---|
| 325 | msgEntity.setSent_date(new StringBuilder(String.valueOf(d.getTime()))); |
---|
| 326 | } |
---|
| 327 | else |
---|
| 328 | { |
---|
| 329 | msgEntity.setSent_date(new StringBuilder(String.valueOf(new Date().getTime()))); |
---|
| 330 | } |
---|
| 331 | } |
---|
| 332 | |
---|
| 333 | try{ |
---|
| 334 | Object o = m.getContent(); |
---|
| 335 | if (o instanceof String) { |
---|
| 336 | if( msgEntity != null ) |
---|
| 337 | { |
---|
| 338 | msgEntity.setContent(new StringBuilder( o.toString())); |
---|
| 339 | } |
---|
| 340 | } else if (o instanceof Multipart) { |
---|
| 341 | Multipart multipart = (Multipart)o; |
---|
| 342 | int count = multipart.getCount(); |
---|
| 343 | |
---|
| 344 | for (int i = 0; i < count; i++) |
---|
| 345 | { |
---|
| 346 | if(msgEntity.getContent() == null || msgEntity.getContent().toString().trim().equals("")) |
---|
| 347 | { |
---|
| 348 | msgEntity.setContent(new StringBuilder("")); |
---|
| 349 | } |
---|
| 350 | if(multipart.getBodyPart(i).getContentType() != Part.ATTACHMENT) |
---|
| 351 | { |
---|
| 352 | msgEntity.setContent(msgEntity.getContent().append(" ").append(getPlainContent(multipart.getBodyPart(i)))); |
---|
| 353 | } |
---|
| 354 | |
---|
| 355 | } |
---|
| 356 | |
---|
| 357 | } |
---|
| 358 | }catch (Exception e){ |
---|
| 359 | System.out.println("Exception ocurred!"); |
---|
| 360 | e.printStackTrace(); |
---|
| 361 | } |
---|
| 362 | |
---|
| 363 | if(msgEntity != null && msgEntity.getId() != null){ |
---|
| 364 | listIMAPMsgs.add(msgEntity); |
---|
| 365 | } |
---|
| 366 | |
---|
| 367 | } |
---|
| 368 | |
---|
| 369 | private StringBuilder getPlainContent(BodyPart bodyPart) throws IOException, MessagingException |
---|
| 370 | { |
---|
| 371 | try |
---|
| 372 | { |
---|
| 373 | |
---|
| 374 | if(bodyPart.getContent() instanceof String) |
---|
| 375 | { |
---|
| 376 | // ByteArrayOutputStream out = new ByteArrayOutputStream(); |
---|
| 377 | // InputStream in = bodyPart.getInputStream(); |
---|
| 378 | // |
---|
| 379 | // byte[] buffer = new byte[1024]; |
---|
| 380 | // byte[] data = null; |
---|
| 381 | // |
---|
| 382 | // while(in.read(buffer) != -1) { |
---|
| 383 | // out.write(buffer); |
---|
| 384 | // } |
---|
| 385 | // |
---|
| 386 | // data = out.toByteArray(); |
---|
| 387 | |
---|
| 388 | StringBuilder text = new StringBuilder(bodyPart.getContent().toString()); |
---|
| 389 | |
---|
| 390 | // out.close(); |
---|
| 391 | // in.close(); |
---|
| 392 | |
---|
| 393 | return text; |
---|
| 394 | } |
---|
| 395 | }catch(Exception e) |
---|
| 396 | { |
---|
| 397 | System.out.println(e.getMessage()); |
---|
| 398 | e.printStackTrace(); |
---|
| 399 | return new StringBuilder(""); |
---|
| 400 | } |
---|
| 401 | |
---|
| 402 | |
---|
| 403 | return new StringBuilder(""); |
---|
| 404 | } |
---|
| 405 | |
---|
| 406 | } |
---|