src/org/sonews/storage/DrupalMessage.java
author František Kučera <franta-hg@frantovo.cz>
Thu, 13 Oct 2011 03:09:22 +0200
changeset 75 41d6c0cac8b3
parent 74 e1244384cc6f
child 82 21f413541357
permissions -rw-r--r--
Drupal: číštění HTML – Tidy.
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@74
    94
			/** Plain text part */
franta-hg@72
    95
			MimeBodyPart textPart = new MimeBodyPart();
franta-hg@72
    96
			multipart.addBodyPart(textPart);
franta-hg@72
    97
			textPart.setText(readPlainText(rs));
franta-hg@72
    98
franta-hg@74
    99
			/** XHTML part */
franta-hg@72
   100
			MimeBodyPart htmlPart = new MimeBodyPart();
franta-hg@72
   101
			multipart.addBodyPart(htmlPart);
franta-hg@72
   102
			htmlPart.setContent(readXhtmlText(rs), XHTML_CONTENT_TYPE);
franta-hg@72
   103
		} else {
franta-hg@72
   104
			setText("");
franta-hg@72
   105
		}
franta-hg@72
   106
	}
franta-hg@72
   107
franta-hg@72
   108
	private String readPlainText(ResultSet rs) {
franta-hg@72
   109
		/**
franta-hg@72
   110
		 * TODO: převést na prostý text
franta-hg@72
   111
		 */
franta-hg@72
   112
		return "TODO: obyčejný text";
franta-hg@72
   113
	}
franta-hg@72
   114
franta-hg@72
   115
	private String readXhtmlText(ResultSet rs) {
franta-hg@72
   116
		/**
franta-hg@74
   117
		 * TODO: znovupoužívat XSL transformér
franta-hg@72
   118
		 */
franta-hg@74
   119
		try {
franta-hg@74
   120
			String originalText = rs.getString("text");
franta-hg@75
   121
franta-hg@75
   122
			/**
franta-hg@75
   123
			 * TODO: používat cache, ukládat si vygenerované články
franta-hg@75
   124
			 * 
franta-hg@75
   125
			 * 
franta-hg@75
   126
			 * Místo markdownu jen ošetřit:
franta-hg@75
   127
			 *		- odstavce
franta-hg@75
   128
			 *		- nesmyslné entity v odkazech
franta-hg@75
   129
			 *		- neuzavřené značky: br, hr, img
franta-hg@75
   130
			 */
franta-hg@75
   131
			String tidyTexy = tidyXhtml("<html><body>" + originalText + "</body></html>");
franta-hg@75
   132
franta-hg@75
   133
franta-hg@75
   134
franta-hg@75
   135
			StringReader input = new StringReader(tidyTexy);
franta-hg@75
   136
			StringWriter output = new StringWriter(2 * tidyTexy.length());
franta-hg@74
   137
			TransformerFactory tf = TransformerFactory.newInstance();
franta-hg@74
   138
			Transformer t = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart.xsl")));
franta-hg@74
   139
			t.setParameter("isRoot", (rs.getInt("parent_id") == 0));
franta-hg@74
   140
			t.setParameter("title", rs.getString("subject"));
franta-hg@74
   141
			t.setParameter("urlBase", rs.getString("urlBase"));
franta-hg@74
   142
			t.setParameter("wwwRead", rs.getString("wwwRead"));
franta-hg@74
   143
			t.setParameter("wwwPost", rs.getString("wwwPost"));
franta-hg@74
   144
			t.transform(new StreamSource(input), new StreamResult(output));
franta-hg@75
   145
franta-hg@74
   146
			return output.toString();
franta-hg@74
   147
		} catch (Exception e) {
franta-hg@74
   148
			/**
franta-hg@74
   149
			 * TODO: lepší ošetření chyby
franta-hg@74
   150
			 */
franta-hg@74
   151
			log.log(Level.WARNING, "Error while transforming article to XHTML", e);
franta-hg@74
   152
			return "<html><body><p>Při transformaci příspěvku bohužel došlo k chybě.</p></body></html>";
franta-hg@74
   153
		}
franta-hg@72
   154
	}
franta-hg@72
   155
franta-hg@75
   156
	/**
franta-hg@75
   157
	 * TODO: refaktorovat, přesunout
franta-hg@75
   158
	 */
franta-hg@75
   159
	private static String tidyXhtml(String inputText) throws IOException {
franta-hg@75
   160
		Runtime r = Runtime.getRuntime();
franta-hg@75
   161
		Process p = r.exec(new String[]{"tidy",
franta-hg@75
   162
					"-asxml",
franta-hg@75
   163
					"-numeric",
franta-hg@75
   164
					"-utf8",
franta-hg@75
   165
					"-quiet",
franta-hg@75
   166
					"--doctype", "omit",
franta-hg@75
   167
					"--logical-emphasis", "true",
franta-hg@75
   168
					"--show-errors", "0"});
franta-hg@75
   169
franta-hg@75
   170
		PrintStream vstupProcesu = new PrintStream(p.getOutputStream());
franta-hg@75
   171
		vstupProcesu.print(inputText);
franta-hg@75
   172
		vstupProcesu.close();
franta-hg@75
   173
franta-hg@75
   174
		String outputText = streamToString(p.getInputStream());
franta-hg@75
   175
franta-hg@75
   176
		return outputText;
franta-hg@75
   177
	}
franta-hg@75
   178
franta-hg@75
   179
	/**
franta-hg@75
   180
	 * TODO: refaktorovat, přesunout
franta-hg@75
   181
	 */
franta-hg@75
   182
	private static String streamToString(InputStream proud) throws IOException {
franta-hg@75
   183
		StringBuilder výsledek = new StringBuilder();
franta-hg@75
   184
		BufferedReader buf = new BufferedReader(new InputStreamReader(proud));
franta-hg@75
   185
		while (true) {
franta-hg@75
   186
			String radek = buf.readLine();
franta-hg@75
   187
			if (radek == null) {
franta-hg@75
   188
				break;
franta-hg@75
   189
			} else {
franta-hg@75
   190
				výsledek.append(radek);
franta-hg@75
   191
				výsledek.append("\n");
franta-hg@75
   192
			}
franta-hg@75
   193
		}
franta-hg@75
   194
		return výsledek.toString();
franta-hg@75
   195
	}
franta-hg@75
   196
franta-hg@72
   197
	private static String constructMessageId(int articleID, int groupID, String groupName, String domainName) {
franta-hg@72
   198
		StringBuilder sb = new StringBuilder();
franta-hg@72
   199
		sb.append("<");
franta-hg@72
   200
		sb.append(articleID);
franta-hg@72
   201
		sb.append("-");
franta-hg@72
   202
		sb.append(groupID);
franta-hg@72
   203
		sb.append("-");
franta-hg@72
   204
		sb.append(groupName);
franta-hg@72
   205
		sb.append("@");
franta-hg@72
   206
		sb.append(domainName);
franta-hg@72
   207
		sb.append(">");
franta-hg@72
   208
		return sb.toString();
franta-hg@72
   209
	}
franta-hg@72
   210
franta-hg@72
   211
	@Override
franta-hg@72
   212
	public void setHeader(String name, String value) throws MessagingException {
franta-hg@72
   213
		super.setHeader(name, value);
franta-hg@72
   214
franta-hg@72
   215
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   216
			messageID = value;
franta-hg@72
   217
		}
franta-hg@72
   218
	}
franta-hg@72
   219
franta-hg@72
   220
	@Override
franta-hg@72
   221
	public final void addHeader(String name, String value) throws MessagingException {
franta-hg@72
   222
		super.addHeader(name, value);
franta-hg@72
   223
franta-hg@72
   224
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   225
			messageID = value;
franta-hg@72
   226
		}
franta-hg@72
   227
	}
franta-hg@72
   228
franta-hg@72
   229
	@Override
franta-hg@72
   230
	public void removeHeader(String name) throws MessagingException {
franta-hg@72
   231
		super.removeHeader(name);
franta-hg@72
   232
franta-hg@72
   233
		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
franta-hg@72
   234
			messageID = null;
franta-hg@72
   235
		}
franta-hg@72
   236
	}
franta-hg@72
   237
franta-hg@72
   238
	public void setMessageID(String messageID) {
franta-hg@72
   239
		this.messageID = messageID;
franta-hg@72
   240
	}
franta-hg@72
   241
franta-hg@72
   242
	@Override
franta-hg@72
   243
	protected void updateMessageID() throws MessagingException {
franta-hg@72
   244
		if (messageID == null) {
franta-hg@72
   245
			super.updateMessageID();
franta-hg@72
   246
		} else {
franta-hg@72
   247
			setHeader(MESSAGE_ID_HEADER, messageID);
franta-hg@72
   248
		}
franta-hg@72
   249
	}
franta-hg@72
   250
franta-hg@72
   251
	/**
franta-hg@72
   252
	 * Call {@link #saveChanges()} before this method, if you want all headers including such ones like:
franta-hg@72
   253
	 * 
franta-hg@72
   254
	 * <pre>MIME-Version: 1.0
franta-hg@72
   255
	 *Content-Type: multipart/alternative;</pre>
franta-hg@72
   256
	 * 
franta-hg@72
   257
	 * @return serialized headers
franta-hg@72
   258
	 * @throws MessagingException if getAllHeaders() fails
franta-hg@72
   259
	 */
franta-hg@72
   260
	public String getHeaders() throws MessagingException {
franta-hg@72
   261
		StringBuilder sb = new StringBuilder();
franta-hg@72
   262
		for (Enumeration eh = getAllHeaderLines(); eh.hasMoreElements();) {
franta-hg@72
   263
			sb.append(eh.nextElement());
franta-hg@72
   264
			sb.append(CRLF);
franta-hg@72
   265
		}
franta-hg@72
   266
		return sb.toString();
franta-hg@72
   267
	}
franta-hg@72
   268
franta-hg@72
   269
	public byte[] getBody() throws IOException, MessagingException {
franta-hg@72
   270
		saveChanges();
franta-hg@72
   271
franta-hg@72
   272
		ArrayList<String> skipHeaders = new ArrayList<String>();
franta-hg@72
   273
		for (Enumeration eh = getAllHeaders(); eh.hasMoreElements();) {
franta-hg@72
   274
			Header h = (Header) eh.nextElement();
franta-hg@72
   275
			skipHeaders.add(h.getName());
franta-hg@72
   276
		}
franta-hg@72
   277
franta-hg@72
   278
		ByteArrayOutputStream baos = new ByteArrayOutputStream(1024);
franta-hg@72
   279
		writeTo(baos, skipHeaders.toArray(new String[skipHeaders.size()]));
franta-hg@72
   280
		return baos.toByteArray();
franta-hg@72
   281
	}
franta-hg@72
   282
}