src/org/sonews/storage/DrupalMessage.java
author František Kučera <franta-hg@frantovo.cz>
Sun, 16 Oct 2011 22:36:46 +0200
changeset 82 21f413541357
parent 75 41d6c0cac8b3
child 84 7a3514a9e4fe
permissions -rw-r--r--
Drupal: tidy, odstavce, XSLT
     1 /*
     2  *   SONEWS News Server
     3  *   see AUTHORS for the list of contributors
     4  *
     5  *   This program is free software: you can redistribute it and/or modify
     6  *   it under the terms of the GNU General Public License as published by
     7  *   the Free Software Foundation, either version 3 of the License, or
     8  *   (at your option) any later version.
     9  *
    10  *   This program is distributed in the hope that it will be useful,
    11  *   but WITHOUT ANY WARRANTY; without even the implied warranty of
    12  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    13  *   GNU General Public License for more details.
    14  *
    15  *   You should have received a copy of the GNU General Public License
    16  *   along with this program.  If not, see <http://www.gnu.org/licenses/>.
    17  */
    18 package org.sonews.storage;
    19 
    20 import java.io.BufferedReader;
    21 import java.io.ByteArrayOutputStream;
    22 import java.io.IOException;
    23 import java.io.InputStream;
    24 import java.io.InputStreamReader;
    25 import java.io.PrintStream;
    26 import java.io.StringReader;
    27 import java.io.StringWriter;
    28 import java.io.UnsupportedEncodingException;
    29 import java.sql.ResultSet;
    30 import java.sql.SQLException;
    31 import java.util.ArrayList;
    32 import java.util.Date;
    33 import java.util.Enumeration;
    34 import java.util.logging.Level;
    35 import java.util.logging.Logger;
    36 import javax.mail.Header;
    37 import javax.mail.MessagingException;
    38 import javax.mail.Multipart;
    39 import javax.mail.Session;
    40 import javax.mail.internet.InternetAddress;
    41 import javax.mail.internet.MimeBodyPart;
    42 import javax.mail.internet.MimeMessage;
    43 import javax.mail.internet.MimeMultipart;
    44 import javax.xml.transform.Transformer;
    45 import javax.xml.transform.TransformerFactory;
    46 import javax.xml.transform.stream.StreamResult;
    47 import javax.xml.transform.stream.StreamSource;
    48 import org.sonews.util.io.Resource;
    49 
    50 /**
    51  * This is MimeMessage which enables custom Message-ID header
    52  * (this header will not be overwritten by the default one like in MimeMessage).
    53  * 
    54  * Also add header and body separate serialization.
    55  * 
    56  * And can be deserialized from SQL ResultSet
    57  * 
    58  * @author František Kučera (frantovo.cz)
    59  */
    60 public class DrupalMessage extends MimeMessage {
    61 
    62 	private static final Logger log = Logger.getLogger(DrupalMessage.class.getName());
    63 	private static final String MESSAGE_ID_HEADER = "Message-ID";
    64 	private static final String CRLF = "\r\n";
    65 	public static final String CHARSET = "UTF-8";
    66 	private static final String XHTML_CONTENT_TYPE = "text/html; charset=" + CHARSET;
    67 	private String messageID;
    68 
    69 	/**
    70 	 * Constructs MIME message from SQL result.
    71 	 * @param rs ResultSet containing message data. No {@link ResultSet#next()} will be called, just values from current row will be read.
    72 	 * @param constructBody true if whole message should be constructed | false if we need only message headers (body will be dummy).
    73 	 */
    74 	public DrupalMessage(ResultSet rs, String myDomain, boolean constructBody) throws SQLException, UnsupportedEncodingException, MessagingException {
    75 		super(Session.getDefaultInstance(System.getProperties()));
    76 
    77 		addHeader("Message-id", constructMessageId(rs.getInt("id"), rs.getInt("group_id"), rs.getString("group_name"), myDomain));
    78 		addHeader("Newsgroups", rs.getString("group_name"));
    79 		setFrom(new InternetAddress(rs.getString("sender_email"), rs.getString("sender_name")));
    80 		setSubject(rs.getString("subject"));
    81 		setSentDate(new Date(rs.getLong("created")));
    82 
    83 		int parentID = rs.getInt("parent_id");
    84 		if (parentID > 0) {
    85 			String parentMessageID = constructMessageId(parentID, rs.getInt("group_id"), rs.getString("group_name"), myDomain);
    86 			addHeader("In-Reply-To", parentMessageID);
    87 			addHeader("References", parentMessageID);
    88 		}
    89 
    90 		if (constructBody) {
    91 			Multipart multipart = new MimeMultipart("alternative");
    92 			setContent(multipart);
    93 
    94 			/** XHTML part */
    95 			MimeBodyPart htmlPart = new MimeBodyPart();
    96 			multipart.addBodyPart(htmlPart);
    97 			String xhtmlText = readXhtmlText(rs);
    98 			htmlPart.setContent(xhtmlText, XHTML_CONTENT_TYPE);
    99 			
   100 			/** Plain text part */
   101 			MimeBodyPart textPart = new MimeBodyPart();
   102 			multipart.addBodyPart(textPart);
   103 			textPart.setText(readPlainText(rs, xhtmlText));
   104 		} else {
   105 			/** empty body, just headers */
   106 			setText("");
   107 		}
   108 	}
   109 
   110 	private String readPlainText(ResultSet rs, String xhtmlText) {
   111 		/**
   112 		 * TODO: převést na prostý text
   113 		 */
   114 		return "TODO: obyčejný text\n\n\n" + xhtmlText;
   115 	}
   116 
   117 	private String readXhtmlText(ResultSet rs) {
   118 		/**
   119 		 * TODO: 
   120 		 *		- znovupoužívat XSL transformér
   121 		 *		- používat cache, ukládat si vygenerované články
   122 		 */
   123 		try {
   124 			String inputText = "<html><body>" + rs.getString("text") + "</body></html>";
   125 
   126 			TransformerFactory tf = TransformerFactory.newInstance();
   127 			Transformer paragraphTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart-make-paragraphs.xsl")));
   128 
   129 			String paragraphedText;
   130 			boolean tidyWasUsed = false;
   131 			try {
   132 				StringReader input = new StringReader(inputText);
   133 				StringWriter output = new StringWriter(2 * inputText.length());
   134 				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
   135 				paragraphedText = output.toString();
   136 			} catch (Exception e) {
   137 				log.log(Level.FINER, "HTML input was shitty – Tidy had to be called.", e);
   138 				StringReader input = new StringReader(tidyXhtml(inputText));
   139 				StringWriter output = new StringWriter(2 * inputText.length());
   140 				paragraphTransformer.transform(new StreamSource(input), new StreamResult(output));
   141 				paragraphedText = output.toString();
   142 				tidyWasUsed = true;
   143 			}
   144 
   145 			Transformer xhtmlTransformer = tf.newTransformer(new StreamSource(Resource.getAsStream("helpers/mimeXhtmlPart.xsl")));
   146 			xhtmlTransformer.setParameter("isRoot", (rs.getInt("parent_id") == 0));
   147 			xhtmlTransformer.setParameter("title", rs.getString("subject"));
   148 			xhtmlTransformer.setParameter("urlBase", rs.getString("urlBase"));
   149 			xhtmlTransformer.setParameter("wwwRead", rs.getString("wwwRead"));
   150 			xhtmlTransformer.setParameter("wwwPost", rs.getString("wwwPost"));
   151 			xhtmlTransformer.setParameter("headComment", String.format("Drupal-NNTP bridge. Transformed: %1$tc. Tidy had to be used: %2$b", new Date(), tidyWasUsed));
   152 			StringReader input = new StringReader(paragraphedText);
   153 			StringWriter output = new StringWriter(2 * paragraphedText.length());
   154 			xhtmlTransformer.transform(new StreamSource(input), new StreamResult(output));
   155 
   156 			return output.toString();
   157 		} catch (Exception e) {
   158 			/**
   159 			 * TODO: lepší ošetření chyby
   160 			 */
   161 			log.log(Level.WARNING, "Error while transforming article to XHTML", e);
   162 			return "<html><body><p>Při transformaci příspěvku bohužel došlo k chybě.</p></body></html>";
   163 		}
   164 	}
   165 
   166 	/**
   167 	 * TODO: refaktorovat, přesunout
   168 	 */
   169 	private static String tidyXhtml(String inputText) throws IOException {
   170 		// https://sourceforge.net/tracker/index.php?func=detail&aid=3424437&group_id=27659&atid=390966
   171 		inputText = inputText.replaceAll("\\n", "◆\n");
   172 
   173 		Runtime r = Runtime.getRuntime();
   174 		Process p = r.exec(new String[]{"tidy", // http://tidy.sourceforge.net
   175 					"-asxml", // well formed XHTML
   176 					"-numeric", // číselné entity
   177 					"-utf8", // kódování
   178 					"--show-warnings", "false", // žádná varování nás nezajímají
   179 					"--show-errors", "0", // ani chyby
   180 					"--doctype", "omit", // doctype nepotřebujeme (doplníme si případně vlastní v XSLT)
   181 					"--logical-emphasis", "true", // em a strong místo i a b
   182 					"--literal-attributes", "true", // zachovat mezery a konce řádků v atributech
   183 					"--force-output", "true" // neznámé značky zahodíme, vložíme jen jejich obsah
   184 				});
   185 
   186 		PrintStream vstupProcesu = new PrintStream(p.getOutputStream());
   187 		vstupProcesu.print(inputText);
   188 		vstupProcesu.close();
   189 
   190 		String outputText = streamToString(p.getInputStream());
   191 
   192 		outputText = outputText.replaceAll("◆\\n", "\n");
   193 		outputText = outputText.replaceAll("◆", "\n");
   194 
   195 		return outputText;
   196 	}
   197 
   198 	/**
   199 	 * TODO: refaktorovat, přesunout
   200 	 */
   201 	private static String streamToString(InputStream proud) throws IOException {
   202 		StringBuilder výsledek = new StringBuilder();
   203 		BufferedReader buf = new BufferedReader(new InputStreamReader(proud));
   204 		while (true) {
   205 			String radek = buf.readLine();
   206 			if (radek == null) {
   207 				break;
   208 			} else {
   209 				výsledek.append(radek);
   210 				výsledek.append("\n");
   211 			}
   212 		}
   213 		return výsledek.toString();
   214 	}
   215 
   216 	private static String constructMessageId(int articleID, int groupID, String groupName, String domainName) {
   217 		StringBuilder sb = new StringBuilder();
   218 		sb.append("<");
   219 		sb.append(articleID);
   220 		sb.append("-");
   221 		sb.append(groupID);
   222 		sb.append("-");
   223 		sb.append(groupName);
   224 		sb.append("@");
   225 		sb.append(domainName);
   226 		sb.append(">");
   227 		return sb.toString();
   228 	}
   229 
   230 	@Override
   231 	public void setHeader(String name, String value) throws MessagingException {
   232 		super.setHeader(name, value);
   233 
   234 		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
   235 			messageID = value;
   236 		}
   237 	}
   238 
   239 	@Override
   240 	public final void addHeader(String name, String value) throws MessagingException {
   241 		super.addHeader(name, value);
   242 
   243 		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
   244 			messageID = value;
   245 		}
   246 	}
   247 
   248 	@Override
   249 	public void removeHeader(String name) throws MessagingException {
   250 		super.removeHeader(name);
   251 
   252 		if (MESSAGE_ID_HEADER.equalsIgnoreCase(name)) {
   253 			messageID = null;
   254 		}
   255 	}
   256 
   257 	public void setMessageID(String messageID) {
   258 		this.messageID = messageID;
   259 	}
   260 
   261 	@Override
   262 	protected void updateMessageID() throws MessagingException {
   263 		if (messageID == null) {
   264 			super.updateMessageID();
   265 		} else {
   266 			setHeader(MESSAGE_ID_HEADER, messageID);
   267 		}
   268 	}
   269 
   270 	/**
   271 	 * Call {@link #saveChanges()} before this method, if you want all headers including such ones like:
   272 	 * 
   273 	 * <pre>MIME-Version: 1.0
   274 	 *Content-Type: multipart/alternative;</pre>
   275 	 * 
   276 	 * @return serialized headers
   277 	 * @throws MessagingException if getAllHeaders() fails
   278 	 */
   279 	public String getHeaders() throws MessagingException {
   280 		StringBuilder sb = new StringBuilder();
   281 		for (Enumeration eh = getAllHeaderLines(); eh.hasMoreElements();) {
   282 			sb.append(eh.nextElement());
   283 			sb.append(CRLF);
   284 		}
   285 		return sb.toString();
   286 	}
   287 
   288 	public byte[] getBody() throws IOException, MessagingException {
   289 		saveChanges();
   290 
   291 		ArrayList<String> skipHeaders = new ArrayList<String>();
   292 		for (Enumeration eh = getAllHeaders(); eh.hasMoreElements();) {
   293 			Header h = (Header) eh.nextElement();
   294 			skipHeaders.add(h.getName());
   295 		}
   296 
   297 		ByteArrayOutputStream baos = new ByteArrayOutputStream(1024);
   298 		writeTo(baos, skipHeaders.toArray(new String[skipHeaders.size()]));
   299 		return baos.toByteArray();
   300 	}
   301 }