src/org/sonews/storage/DrupalMessage.java
author František Kučera <franta-hg@frantovo.cz>
Sun Oct 16 23:41:04 2011 +0200 (2011-10-16)
changeset 83 668014315a54
parent 75 41d6c0cac8b3
child 84 7a3514a9e4fe
permissions -rw-r--r--
Drupal: odstavce – volitelná css třída a zbavení se zytečného xmlns
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
			multipart.addBodyPart(htmlPart);
franta-hg@82
    97
			String xhtmlText = readXhtmlText(rs);
franta-hg@82
    98
			htmlPart.setContent(xhtmlText, XHTML_CONTENT_TYPE);
franta-hg@82
    99
			
franta-hg@74
   100
			/** Plain text part */
franta-hg@72
   101
			MimeBodyPart textPart = new MimeBodyPart();
franta-hg@72
   102
			multipart.addBodyPart(textPart);
franta-hg@82
   103
			textPart.setText(readPlainText(rs, xhtmlText));
franta-hg@72
   104
		} else {
franta-hg@82
   105
			/** empty body, just headers */
franta-hg@72
   106
			setText("");
franta-hg@72
   107
		}
franta-hg@72
   108
	}
franta-hg@72
   109
franta-hg@82
   110
	private String readPlainText(ResultSet rs, String xhtmlText) {
franta-hg@72
   111
		/**
franta-hg@72
   112
		 * TODO: převést na prostý text
franta-hg@72
   113
		 */
franta-hg@82
   114
		return "TODO: obyčejný text\n\n\n" + xhtmlText;
franta-hg@72
   115
	}
franta-hg@72
   116
franta-hg@72
   117
	private String readXhtmlText(ResultSet rs) {
franta-hg@72
   118
		/**
franta-hg@82
   119
		 * TODO: 
franta-hg@82
   120
		 *		- znovupoužívat XSL transformér
franta-hg@82
   121
		 *		- používat cache, ukládat si vygenerované články
franta-hg@72
   122
		 */
franta-hg@74
   123
		try {
franta-hg@82
   124
			String inputText = "<html><body>" + rs.getString("text") + "</body></html>";
franta-hg@75
   125
franta-hg@82
   126
			TransformerFactory tf = TransformerFactory.newInstance();
franta-hg@82
   127
			Transformer paragraphTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart-make-paragraphs.xsl")));
franta-hg@75
   128
franta-hg@82
   129
			String paragraphedText;
franta-hg@82
   130
			boolean tidyWasUsed = false;
franta-hg@82
   131
			try {
franta-hg@82
   132
				StringReader input = new StringReader(inputText);
franta-hg@82
   133
				StringWriter output = new StringWriter(2 * inputText.length());
franta-hg@82
   134
				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@82
   135
				paragraphedText = output.toString();
franta-hg@82
   136
			} catch (Exception e) {
franta-hg@82
   137
				log.log(Level.FINER, "HTML input was shitty – Tidy had to be called.", e);
franta-hg@82
   138
				StringReader input = new StringReader(tidyXhtml(inputText));
franta-hg@82
   139
				StringWriter output = new StringWriter(2 * inputText.length());
franta-hg@82
   140
				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@82
   141
				paragraphedText = output.toString();
franta-hg@82
   142
				tidyWasUsed = true;
franta-hg@82
   143
			}
franta-hg@75
   144
franta-hg@82
   145
			Transformer xhtmlTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart.xsl")));
franta-hg@82
   146
			xhtmlTransformer.setParameter("isRoot", (rs.getInt("parent_id") == 0));
franta-hg@82
   147
			xhtmlTransformer.setParameter("title", rs.getString("subject"));
franta-hg@82
   148
			xhtmlTransformer.setParameter("urlBase", rs.getString("urlBase"));
franta-hg@82
   149
			xhtmlTransformer.setParameter("wwwRead", rs.getString("wwwRead"));
franta-hg@82
   150
			xhtmlTransformer.setParameter("wwwPost", rs.getString("wwwPost"));
franta-hg@82
   151
			xhtmlTransformer.setParameter("headComment", String.format("Drupal-NNTP bridge. Transformed: %1$tc. Tidy had to be used: %2$b", new Date(), tidyWasUsed));
franta-hg@82
   152
			StringReader input = new StringReader(paragraphedText);
franta-hg@82
   153
			StringWriter output = new StringWriter(2 * paragraphedText.length());
franta-hg@82
   154
			xhtmlTransformer.transform(new StreamSource(input), new StreamResult(output));
franta-hg@75
   155
franta-hg@74
   156
			return output.toString();
franta-hg@74
   157
		} catch (Exception e) {
franta-hg@74
   158
			/**
franta-hg@74
   159
			 * TODO: lepší ošetření chyby
franta-hg@74
   160
			 */
franta-hg@74
   161
			log.log(Level.WARNING, "Error while transforming article to XHTML", e);
franta-hg@74
   162
			return "<html><body><p>Při transformaci příspěvku bohužel došlo k chybě.</p></body></html>";
franta-hg@74
   163
		}
franta-hg@72
   164
	}
franta-hg@72
   165
franta-hg@75
   166
	/**
franta-hg@75
   167
	 * TODO: refaktorovat, přesunout
franta-hg@75
   168
	 */
franta-hg@75
   169
	private static String tidyXhtml(String inputText) throws IOException {
franta-hg@82
   170
		// https://sourceforge.net/tracker/index.php?func=detail&aid=3424437&group_id=27659&atid=390966
franta-hg@82
   171
		inputText = inputText.replaceAll("\\n", "◆\n");
franta-hg@82
   172
franta-hg@75
   173
		Runtime r = Runtime.getRuntime();
franta-hg@82
   174
		Process p = r.exec(new String[]{"tidy", // http://tidy.sourceforge.net
franta-hg@82
   175
					"-asxml", // well formed XHTML
franta-hg@82
   176
					"-numeric", // číselné entity
franta-hg@82
   177
					"-utf8", // kódování
franta-hg@82
   178
					"--show-warnings", "false", // žádná varování nás nezajímají
franta-hg@82
   179
					"--show-errors", "0", // ani chyby
franta-hg@82
   180
					"--doctype", "omit", // doctype nepotřebujeme (doplníme si případně vlastní v XSLT)
franta-hg@82
   181
					"--logical-emphasis", "true", // em a strong místo i a b
franta-hg@82
   182
					"--literal-attributes", "true", // zachovat mezery a konce řádků v atributech
franta-hg@82
   183
					"--force-output", "true" // neznámé značky zahodíme, vložíme jen jejich obsah
franta-hg@82
   184
				});
franta-hg@75
   185
franta-hg@75
   186
		PrintStream vstupProcesu = new PrintStream(p.getOutputStream());
franta-hg@75
   187
		vstupProcesu.print(inputText);
franta-hg@75
   188
		vstupProcesu.close();
franta-hg@75
   189
franta-hg@75
   190
		String outputText = streamToString(p.getInputStream());
franta-hg@75
   191
franta-hg@82
   192
		outputText = outputText.replaceAll("◆\\n", "\n");
franta-hg@82
   193
		outputText = outputText.replaceAll("◆", "\n");
franta-hg@82
   194
franta-hg@75
   195
		return outputText;
franta-hg@75
   196
	}
franta-hg@75
   197
franta-hg@75
   198
	/**
franta-hg@75
   199
	 * TODO: refaktorovat, přesunout
franta-hg@75
   200
	 */
franta-hg@75
   201
	private static String streamToString(InputStream proud) throws IOException {
franta-hg@75
   202
		StringBuilder výsledek = new StringBuilder();
franta-hg@75
   203
		BufferedReader buf = new BufferedReader(new InputStreamReader(proud));
franta-hg@75
   204
		while (true) {
franta-hg@75
   205
			String radek = buf.readLine();
franta-hg@75
   206
			if (radek == null) {
franta-hg@75
   207
				break;
franta-hg@75
   208
			} else {
franta-hg@75
   209
				výsledek.append(radek);
franta-hg@75
   210
				výsledek.append("\n");
franta-hg@75
   211
			}
franta-hg@75
   212
		}
franta-hg@75
   213
		return výsledek.toString();
franta-hg@75
   214
	}
franta-hg@75
   215
franta-hg@72
   216
	private static String constructMessageId(int articleID, int groupID, String groupName, String domainName) {
franta-hg@72
   217
		StringBuilder sb = new StringBuilder();
franta-hg@72
   218
		sb.append("<");
franta-hg@72
   219
		sb.append(articleID);
franta-hg@72
   220
		sb.append("-");
franta-hg@72
   221
		sb.append(groupID);
franta-hg@72
   222
		sb.append("-");
franta-hg@72
   223
		sb.append(groupName);
franta-hg@72
   224
		sb.append("@");
franta-hg@72
   225
		sb.append(domainName);
franta-hg@72
   226
		sb.append(">");
franta-hg@72
   227
		return sb.toString();
franta-hg@72
   228
	}
franta-hg@72
   229
franta-hg@72
   230
	@Override
franta-hg@72
   231
	public void setHeader(String name, String value) throws MessagingException {
franta-hg@72
   232
		super.setHeader(name, value);
franta-hg@72
   233
franta-hg@72
   234
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   235
			messageID = value;
franta-hg@72
   236
		}
franta-hg@72
   237
	}
franta-hg@72
   238
franta-hg@72
   239
	@Override
franta-hg@72
   240
	public final void addHeader(String name, String value) throws MessagingException {
franta-hg@72
   241
		super.addHeader(name, value);
franta-hg@72
   242
franta-hg@72
   243
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   244
			messageID = value;
franta-hg@72
   245
		}
franta-hg@72
   246
	}
franta-hg@72
   247
franta-hg@72
   248
	@Override
franta-hg@72
   249
	public void removeHeader(String name) throws MessagingException {
franta-hg@72
   250
		super.removeHeader(name);
franta-hg@72
   251
franta-hg@72
   252
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   253
			messageID = null;
franta-hg@72
   254
		}
franta-hg@72
   255
	}
franta-hg@72
   256
franta-hg@72
   257
	public void setMessageID(String messageID) {
franta-hg@72
   258
		this.messageID = messageID;
franta-hg@72
   259
	}
franta-hg@72
   260
franta-hg@72
   261
	@Override
franta-hg@72
   262
	protected void updateMessageID() throws MessagingException {
franta-hg@72
   263
		if (messageID == null) {
franta-hg@72
   264
			super.updateMessageID();
franta-hg@72
   265
		} else {
franta-hg@72
   266
			setHeader(MESSAGE_ID_HEADER, messageID);
franta-hg@72
   267
		}
franta-hg@72
   268
	}
franta-hg@72
   269
franta-hg@72
   270
	/**
franta-hg@72
   271
	 * Call {@link #saveChanges()} before this method, if you want all headers including such ones like:
franta-hg@72
   272
	 * 
franta-hg@72
   273
	 * <pre>MIME-Version: 1.0
franta-hg@72
   274
	 *Content-Type: multipart/alternative;</pre>
franta-hg@72
   275
	 * 
franta-hg@72
   276
	 * @return serialized headers
franta-hg@72
   277
	 * @throws MessagingException if getAllHeaders() fails
franta-hg@72
   278
	 */
franta-hg@72
   279
	public String getHeaders() throws MessagingException {
franta-hg@72
   280
		StringBuilder sb = new StringBuilder();
franta-hg@72
   281
		for (Enumeration eh = getAllHeaderLines(); eh.hasMoreElements();) {
franta-hg@72
   282
			sb.append(eh.nextElement());
franta-hg@72
   283
			sb.append(CRLF);
franta-hg@72
   284
		}
franta-hg@72
   285
		return sb.toString();
franta-hg@72
   286
	}
franta-hg@72
   287
franta-hg@72
   288
	public byte[] getBody() throws IOException, MessagingException {
franta-hg@72
   289
		saveChanges();
franta-hg@72
   290
franta-hg@72
   291
		ArrayList<String> skipHeaders = new ArrayList<String>();
franta-hg@72
   292
		for (Enumeration eh = getAllHeaders(); eh.hasMoreElements();) {
franta-hg@72
   293
			Header h = (Header) eh.nextElement();
franta-hg@72
   294
			skipHeaders.add(h.getName());
franta-hg@72
   295
		}
franta-hg@72
   296
franta-hg@72
   297
		ByteArrayOutputStream baos = new ByteArrayOutputStream(1024);
franta-hg@72
   298
		writeTo(baos, skipHeaders.toArray(new String[skipHeaders.size()]));
franta-hg@72
   299
		return baos.toByteArray();
franta-hg@72
   300
	}
franta-hg@72
   301
}