src/org/sonews/storage/DrupalMessage.java
author František Kučera <franta-hg@frantovo.cz>
Mon, 17 Oct 2011 19:50:23 +0200
changeset 91 70286a6cf338
parent 89 c60625d58158
child 100 08c9fb6fb017
permissions -rw-r--r--
Drupal: text/plain – podpora obrázků a dalších vnořených elementů uvnitř odkazů.
franta-hg@72
     1
/*
franta-hg@72
     2
 *   SONEWS News Server
franta-hg@72
     3
 *   see AUTHORS for the list of contributors
franta-hg@72
     4
 *
franta-hg@72
     5
 *   This program is free software: you can redistribute it and/or modify
franta-hg@72
     6
 *   it under the terms of the GNU General Public License as published by
franta-hg@72
     7
 *   the Free Software Foundation, either version 3 of the License, or
franta-hg@72
     8
 *   (at your option) any later version.
franta-hg@72
     9
 *
franta-hg@72
    10
 *   This program is distributed in the hope that it will be useful,
franta-hg@72
    11
 *   but WITHOUT ANY WARRANTY; without even the implied warranty of
franta-hg@72
    12
 *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
franta-hg@72
    13
 *   GNU General Public License for more details.
franta-hg@72
    14
 *
franta-hg@72
    15
 *   You should have received a copy of the GNU General Public License
franta-hg@72
    16
 *   along with this program.  If not, see <http://www.gnu.org/licenses/>.
franta-hg@72
    17
 */
franta-hg@72
    18
package org.sonews.storage;
franta-hg@72
    19
franta-hg@75
    20
import java.io.BufferedReader;
franta-hg@72
    21
import java.io.ByteArrayOutputStream;
franta-hg@72
    22
import java.io.IOException;
franta-hg@75
    23
import java.io.InputStream;
franta-hg@75
    24
import java.io.InputStreamReader;
franta-hg@75
    25
import java.io.PrintStream;
franta-hg@74
    26
import java.io.StringReader;
franta-hg@74
    27
import java.io.StringWriter;
franta-hg@72
    28
import java.io.UnsupportedEncodingException;
franta-hg@72
    29
import java.sql.ResultSet;
franta-hg@72
    30
import java.sql.SQLException;
franta-hg@72
    31
import java.util.ArrayList;
franta-hg@72
    32
import java.util.Date;
franta-hg@72
    33
import java.util.Enumeration;
franta-hg@74
    34
import java.util.logging.Level;
franta-hg@74
    35
import java.util.logging.Logger;
franta-hg@72
    36
import javax.mail.Header;
franta-hg@72
    37
import javax.mail.MessagingException;
franta-hg@72
    38
import javax.mail.Multipart;
franta-hg@72
    39
import javax.mail.Session;
franta-hg@72
    40
import javax.mail.internet.InternetAddress;
franta-hg@72
    41
import javax.mail.internet.MimeBodyPart;
franta-hg@72
    42
import javax.mail.internet.MimeMessage;
franta-hg@72
    43
import javax.mail.internet.MimeMultipart;
franta-hg@74
    44
import javax.xml.transform.Transformer;
franta-hg@74
    45
import javax.xml.transform.TransformerFactory;
franta-hg@74
    46
import javax.xml.transform.stream.StreamResult;
franta-hg@74
    47
import javax.xml.transform.stream.StreamSource;
franta-hg@74
    48
import org.sonews.util.io.Resource;
franta-hg@72
    49
franta-hg@72
    50
/**
franta-hg@72
    51
 * This is MimeMessage which enables custom Message-ID header
franta-hg@72
    52
 * (this header will not be overwritten by the default one like in MimeMessage).
franta-hg@72
    53
 * 
franta-hg@72
    54
 * Also add header and body separate serialization.
franta-hg@72
    55
 * 
franta-hg@72
    56
 * And can be deserialized from SQL ResultSet
franta-hg@72
    57
 * 
franta-hg@72
    58
 * @author František Kučera (frantovo.cz)
franta-hg@72
    59
 */
franta-hg@72
    60
public class DrupalMessage extends MimeMessage {
franta-hg@72
    61
franta-hg@74
    62
	private static final Logger log = Logger.getLogger(DrupalMessage.class.getName());
franta-hg@72
    63
	private static final String MESSAGE_ID_HEADER = "Message-ID";
franta-hg@72
    64
	private static final String CRLF = "\r\n";
franta-hg@72
    65
	public static final String CHARSET = "UTF-8";
franta-hg@72
    66
	private static final String XHTML_CONTENT_TYPE = "text/html; charset=" + CHARSET;
franta-hg@72
    67
	private String messageID;
franta-hg@72
    68
franta-hg@72
    69
	/**
franta-hg@72
    70
	 * Constructs MIME message from SQL result.
franta-hg@72
    71
	 * @param rs ResultSet containing message data. No {@link ResultSet#next()} will be called, just values from current row will be read.
franta-hg@72
    72
	 * @param constructBody true if whole message should be constructed | false if we need only message headers (body will be dummy).
franta-hg@72
    73
	 */
franta-hg@72
    74
	public DrupalMessage(ResultSet rs, String myDomain, boolean constructBody) throws SQLException, UnsupportedEncodingException, MessagingException {
franta-hg@72
    75
		super(Session.getDefaultInstance(System.getProperties()));
franta-hg@72
    76
franta-hg@72
    77
		addHeader("Message-id", constructMessageId(rs.getInt("id"), rs.getInt("group_id"), rs.getString("group_name"), myDomain));
franta-hg@72
    78
		addHeader("Newsgroups", rs.getString("group_name"));
franta-hg@74
    79
		setFrom(new InternetAddress(rs.getString("sender_email"), rs.getString("sender_name")));
franta-hg@72
    80
		setSubject(rs.getString("subject"));
franta-hg@72
    81
		setSentDate(new Date(rs.getLong("created")));
franta-hg@74
    82
franta-hg@74
    83
		int parentID = rs.getInt("parent_id");
franta-hg@74
    84
		if (parentID > 0) {
franta-hg@72
    85
			String parentMessageID = constructMessageId(parentID, rs.getInt("group_id"), rs.getString("group_name"), myDomain);
franta-hg@72
    86
			addHeader("In-Reply-To", parentMessageID);
franta-hg@72
    87
			addHeader("References", parentMessageID);
franta-hg@72
    88
		}
franta-hg@72
    89
franta-hg@72
    90
		if (constructBody) {
franta-hg@72
    91
			Multipart multipart = new MimeMultipart("alternative");
franta-hg@72
    92
			setContent(multipart);
franta-hg@72
    93
franta-hg@82
    94
			/** XHTML part */
franta-hg@82
    95
			MimeBodyPart htmlPart = new MimeBodyPart();
franta-hg@82
    96
			String xhtmlText = readXhtmlText(rs);
franta-hg@82
    97
			htmlPart.setContent(xhtmlText, XHTML_CONTENT_TYPE);
franta-hg@84
    98
franta-hg@74
    99
			/** Plain text part */
franta-hg@72
   100
			MimeBodyPart textPart = new MimeBodyPart();
franta-hg@89
   101
			String plainText = readPlainText(rs, xhtmlText);
franta-hg@89
   102
			textPart.setText(plainText);
franta-hg@89
   103
			//addHeader("Lines", String.valueOf(plainText.split("\n").length));
franta-hg@87
   104
franta-hg@87
   105
			/**
franta-hg@87
   106
			 * Thunderbirdu záleží, v jakém pořadí části jsou 
franta-hg@87
   107
			 * (když je prostý text druhý, html se nezobrazí),
franta-hg@87
   108
			 * KNode zobrazuje HTML správně, i když je na prvním místě.
franta-hg@87
   109
			 */
franta-hg@72
   110
			multipart.addBodyPart(textPart);
franta-hg@87
   111
			multipart.addBodyPart(htmlPart);
franta-hg@72
   112
		} else {
franta-hg@82
   113
			/** empty body, just headers */
franta-hg@72
   114
			setText("");
franta-hg@72
   115
		}
franta-hg@72
   116
	}
franta-hg@72
   117
franta-hg@82
   118
	private String readPlainText(ResultSet rs, String xhtmlText) {
franta-hg@89
   119
		try {
franta-hg@89
   120
			TransformerFactory tf = TransformerFactory.newInstance();
franta-hg@89
   121
			Transformer textTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeTextPart.xsl")));
franta-hg@89
   122
franta-hg@89
   123
			StringReader input = new StringReader(xhtmlText);
franta-hg@89
   124
			StringWriter output = new StringWriter(xhtmlText.length());
franta-hg@89
   125
			textTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@89
   126
franta-hg@89
   127
			return output.toString();
franta-hg@89
   128
		} catch (Exception e) {
franta-hg@89
   129
			/**
franta-hg@89
   130
			 * TODO: lepší ošetření chyby
franta-hg@89
   131
			 */
franta-hg@89
   132
			log.log(Level.WARNING, "Error while transforming article to plain text", e);
franta-hg@89
   133
			return makeSimpleXHTML("Při transformaci příspěvku bohužel došlo k chybě.");
franta-hg@89
   134
		}
franta-hg@72
   135
	}
franta-hg@72
   136
franta-hg@72
   137
	private String readXhtmlText(ResultSet rs) {
franta-hg@72
   138
		/**
franta-hg@82
   139
		 * TODO: 
franta-hg@82
   140
		 *		- znovupoužívat XSL transformér
franta-hg@82
   141
		 *		- používat cache, ukládat si vygenerované články
franta-hg@72
   142
		 */
franta-hg@74
   143
		try {
franta-hg@84
   144
			String inputText = makeSimpleXHTML(rs.getString("text"));
franta-hg@75
   145
franta-hg@82
   146
			TransformerFactory tf = TransformerFactory.newInstance();
franta-hg@82
   147
			Transformer paragraphTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart-make-paragraphs.xsl")));
franta-hg@75
   148
franta-hg@82
   149
			String paragraphedText;
franta-hg@82
   150
			boolean tidyWasUsed = false;
franta-hg@82
   151
			try {
franta-hg@82
   152
				StringReader input = new StringReader(inputText);
franta-hg@82
   153
				StringWriter output = new StringWriter(2 * inputText.length());
franta-hg@82
   154
				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@82
   155
				paragraphedText = output.toString();
franta-hg@82
   156
			} catch (Exception e) {
franta-hg@82
   157
				log.log(Level.FINER, "HTML input was shitty – Tidy had to be called.", e);
franta-hg@82
   158
				StringReader input = new StringReader(tidyXhtml(inputText));
franta-hg@82
   159
				StringWriter output = new StringWriter(2 * inputText.length());
franta-hg@82
   160
				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@82
   161
				paragraphedText = output.toString();
franta-hg@82
   162
				tidyWasUsed = true;
franta-hg@82
   163
			}
franta-hg@75
   164
franta-hg@82
   165
			Transformer xhtmlTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart.xsl")));
franta-hg@82
   166
			xhtmlTransformer.setParameter("isRoot", (rs.getInt("parent_id") == 0));
franta-hg@82
   167
			xhtmlTransformer.setParameter("title", rs.getString("subject"));
franta-hg@82
   168
			xhtmlTransformer.setParameter("urlBase", rs.getString("urlBase"));
franta-hg@82
   169
			xhtmlTransformer.setParameter("wwwRead", rs.getString("wwwRead"));
franta-hg@82
   170
			xhtmlTransformer.setParameter("wwwPost", rs.getString("wwwPost"));
franta-hg@82
   171
			xhtmlTransformer.setParameter("headComment", String.format("Drupal-NNTP bridge. Transformed: %1$tc. Tidy had to be used: %2$b", new Date(), tidyWasUsed));
franta-hg@82
   172
			StringReader input = new StringReader(paragraphedText);
franta-hg@82
   173
			StringWriter output = new StringWriter(2 * paragraphedText.length());
franta-hg@82
   174
			xhtmlTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@75
   175
franta-hg@74
   176
			return output.toString();
franta-hg@74
   177
		} catch (Exception e) {
franta-hg@74
   178
			/**
franta-hg@74
   179
			 * TODO: lepší ošetření chyby
franta-hg@74
   180
			 */
franta-hg@74
   181
			log.log(Level.WARNING, "Error while transforming article to XHTML", e);
franta-hg@84
   182
			return makeSimpleXHTML("<p>Při transformaci příspěvku bohužel došlo k chybě.</p>");
franta-hg@74
   183
		}
franta-hg@72
   184
	}
franta-hg@72
   185
franta-hg@84
   186
	private static String makeSimpleXHTML(String body) {
franta-hg@84
   187
		return "<html xmlns=\"http://www.w3.org/1999/xhtml\"><body>" + body + "</body></html>";
franta-hg@84
   188
	}
franta-hg@84
   189
franta-hg@75
   190
	/**
franta-hg@75
   191
	 * TODO: refaktorovat, přesunout
franta-hg@75
   192
	 */
franta-hg@75
   193
	private static String tidyXhtml(String inputText) throws IOException {
franta-hg@89
   194
		/*
franta-hg@89
   195
		 * Viz https://sourceforge.net/tracker/index.php?func=detail&aid=3424437&group_id=27659&atid=390966
franta-hg@89
   196
		 *
franta-hg@89
   197
		 * TODO:
franta-hg@89
   198
		 *		- použít delší zástupný řetězec, ne jen jeden znak
franta-hg@89
   199
		 *		- umísťovat ho jen tam, kde už nějaký text je (ne mezi >\s*<)
franta-hg@89
   200
		 */
franta-hg@82
   201
		inputText = inputText.replaceAll("\\n", "◆\n");
franta-hg@82
   202
franta-hg@75
   203
		Runtime r = Runtime.getRuntime();
franta-hg@82
   204
		Process p = r.exec(new String[]{"tidy", // http://tidy.sourceforge.net
franta-hg@82
   205
					"-asxml", // well formed XHTML
franta-hg@82
   206
					"-numeric", // číselné entity
franta-hg@82
   207
					"-utf8", // kódování
franta-hg@82
   208
					"--show-warnings", "false", // žádná varování nás nezajímají
franta-hg@82
   209
					"--show-errors", "0", // ani chyby
franta-hg@82
   210
					"--doctype", "omit", // doctype nepotřebujeme (doplníme si případně vlastní v XSLT)
franta-hg@82
   211
					"--logical-emphasis", "true", // em a strong místo i a b
franta-hg@82
   212
					"--literal-attributes", "true", // zachovat mezery a konce řádků v atributech
franta-hg@82
   213
					"--force-output", "true" // neznámé značky zahodíme, vložíme jen jejich obsah
franta-hg@82
   214
				});
franta-hg@75
   215
franta-hg@75
   216
		PrintStream vstupProcesu = new PrintStream(p.getOutputStream());
franta-hg@75
   217
		vstupProcesu.print(inputText);
franta-hg@75
   218
		vstupProcesu.close();
franta-hg@75
   219
franta-hg@75
   220
		String outputText = streamToString(p.getInputStream());
franta-hg@75
   221
franta-hg@82
   222
		outputText = outputText.replaceAll("◆\\n", "\n");
franta-hg@82
   223
		outputText = outputText.replaceAll("◆", "\n");
franta-hg@82
   224
franta-hg@75
   225
		return outputText;
franta-hg@75
   226
	}
franta-hg@75
   227
franta-hg@75
   228
	/**
franta-hg@75
   229
	 * TODO: refaktorovat, přesunout
franta-hg@75
   230
	 */
franta-hg@75
   231
	private static String streamToString(InputStream proud) throws IOException {
franta-hg@75
   232
		StringBuilder výsledek = new StringBuilder();
franta-hg@75
   233
		BufferedReader buf = new BufferedReader(new InputStreamReader(proud));
franta-hg@75
   234
		while (true) {
franta-hg@75
   235
			String radek = buf.readLine();
franta-hg@75
   236
			if (radek == null) {
franta-hg@75
   237
				break;
franta-hg@75
   238
			} else {
franta-hg@75
   239
				výsledek.append(radek);
franta-hg@75
   240
				výsledek.append("\n");
franta-hg@75
   241
			}
franta-hg@75
   242
		}
franta-hg@75
   243
		return výsledek.toString();
franta-hg@75
   244
	}
franta-hg@75
   245
franta-hg@72
   246
	private static String constructMessageId(int articleID, int groupID, String groupName, String domainName) {
franta-hg@72
   247
		StringBuilder sb = new StringBuilder();
franta-hg@72
   248
		sb.append("<");
franta-hg@72
   249
		sb.append(articleID);
franta-hg@72
   250
		sb.append("-");
franta-hg@72
   251
		sb.append(groupID);
franta-hg@72
   252
		sb.append("-");
franta-hg@72
   253
		sb.append(groupName);
franta-hg@72
   254
		sb.append("@");
franta-hg@72
   255
		sb.append(domainName);
franta-hg@72
   256
		sb.append(">");
franta-hg@72
   257
		return sb.toString();
franta-hg@72
   258
	}
franta-hg@72
   259
franta-hg@72
   260
	@Override
franta-hg@72
   261
	public void setHeader(String name, String value) throws MessagingException {
franta-hg@72
   262
		super.setHeader(name, value);
franta-hg@72
   263
franta-hg@72
   264
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   265
			messageID = value;
franta-hg@72
   266
		}
franta-hg@72
   267
	}
franta-hg@72
   268
franta-hg@72
   269
	@Override
franta-hg@72
   270
	public final void addHeader(String name, String value) throws MessagingException {
franta-hg@72
   271
		super.addHeader(name, value);
franta-hg@72
   272
franta-hg@72
   273
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   274
			messageID = value;
franta-hg@72
   275
		}
franta-hg@72
   276
	}
franta-hg@72
   277
franta-hg@72
   278
	@Override
franta-hg@72
   279
	public void removeHeader(String name) throws MessagingException {
franta-hg@72
   280
		super.removeHeader(name);
franta-hg@72
   281
franta-hg@72
   282
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   283
			messageID = null;
franta-hg@72
   284
		}
franta-hg@72
   285
	}
franta-hg@72
   286
franta-hg@72
   287
	public void setMessageID(String messageID) {
franta-hg@72
   288
		this.messageID = messageID;
franta-hg@72
   289
	}
franta-hg@72
   290
franta-hg@72
   291
	@Override
franta-hg@72
   292
	protected void updateMessageID() throws MessagingException {
franta-hg@72
   293
		if (messageID == null) {
franta-hg@72
   294
			super.updateMessageID();
franta-hg@72
   295
		} else {
franta-hg@72
   296
			setHeader(MESSAGE_ID_HEADER, messageID);
franta-hg@72
   297
		}
franta-hg@72
   298
	}
franta-hg@72
   299
franta-hg@72
   300
	/**
franta-hg@72
   301
	 * Call {@link #saveChanges()} before this method, if you want all headers including such ones like:
franta-hg@72
   302
	 * 
franta-hg@72
   303
	 * <pre>MIME-Version: 1.0
franta-hg@72
   304
	 *Content-Type: multipart/alternative;</pre>
franta-hg@72
   305
	 * 
franta-hg@72
   306
	 * @return serialized headers
franta-hg@72
   307
	 * @throws MessagingException if getAllHeaders() fails
franta-hg@72
   308
	 */
franta-hg@72
   309
	public String getHeaders() throws MessagingException {
franta-hg@72
   310
		StringBuilder sb = new StringBuilder();
franta-hg@72
   311
		for (Enumeration eh = getAllHeaderLines(); eh.hasMoreElements();) {
franta-hg@72
   312
			sb.append(eh.nextElement());
franta-hg@72
   313
			sb.append(CRLF);
franta-hg@72
   314
		}
franta-hg@72
   315
		return sb.toString();
franta-hg@72
   316
	}
franta-hg@72
   317
franta-hg@72
   318
	public byte[] getBody() throws IOException, MessagingException {
franta-hg@72
   319
		saveChanges();
franta-hg@72
   320
franta-hg@72
   321
		ArrayList<String> skipHeaders = new ArrayList<String>();
franta-hg@72
   322
		for (Enumeration eh = getAllHeaders(); eh.hasMoreElements();) {
franta-hg@72
   323
			Header h = (Header) eh.nextElement();
franta-hg@72
   324
			skipHeaders.add(h.getName());
franta-hg@72
   325
		}
franta-hg@72
   326
franta-hg@72
   327
		ByteArrayOutputStream baos = new ByteArrayOutputStream(1024);
franta-hg@72
   328
		writeTo(baos, skipHeaders.toArray(new String[skipHeaders.size()]));
franta-hg@72
   329
		return baos.toByteArray();
franta-hg@72
   330
	}
franta-hg@72
   331
}