ContentAnalyzer


1   /**
2    * Title:        Oyster Project
3    * Description:  Creating S/MIME email transport capabilities.
4    * Copyright:    Copyright (c) 2001
5    * @Author       Vladimir Radisic
6    * @Version      2.1.5
7    */
8   package org.enhydra.oyster.mail;
9   
10  import java.io.InputStream  ;
11  import java.io.FileInputStream  ;
12  import java.io.BufferedInputStream  ;
13  import java.io.IOException  ;
14  import java.io.File  ;
15  import org.enhydra.oyster.exception.SMIMEException;
16  
17  /**
18   * Analyzes content which will be placed in MimeBodyPart and suggests which
19   * Content-Transfer-Encoding header parameter should be used (7bit,
20   * quoted-printable or base64). ContentAnalyzer makes the following checks, and
21   * according to them decisions about which Content-Transfer-Encoding to suggest:<BR>
22   * <BR>
23   * 7bit: if all checked characters (or bytes) are in group of ASCII characters
24   * which are less than 127 (7F hexadecimal) and higher than 31 (1F hexadecimal).
25   * Characters: HT (Horizontal Tab) 9 (09 hexadecimal), and combinaton of CR
26   * (Carriage Return) 13 (0D hexadecimal) and LF (Line Feed) 11 (0A hexadecimal)
27   * are also allowed. Single usage of CR or LF characters or it's combination
28   * LFCR are not permitted for this type Content-Transfer-Encoding. Also, acording
29   * to RFC822, no line should be longer than 1000 characters with included trailing
30   * CRLF charachters.<BR>
31   * <BR>
32   * quoted-printable: if at least one character does not satisfy previous condition,
33   * and if number of those characters do not exceed 30 percent of all characters.<BR>
34   * <BR>
35   * base64: if number of those characters which do not satisfy 7bit condition
36   * exceed 30 percent of all characters.<BR>
37   */
38  public class ContentAnalyzer
39  {
40  
41    /**
42     * Number of bytes for processed content data.
43     */
44    private int numberOfBytes = 0;
45  
46    /**
47     * Number of bytes which could not be represented as 7bit for processed
48     * content data.
49     */
50    private int numberOfNon7bitBytes = 0;
51  
52    /**
53     * This variable is used for counting line lenght. It shouldn't be higher
54     * than 1000.
55     */
56    private int lineLenght = 0;
57  
58    /**
59     * Indicates that line lenght at least once exceeded 1000 characters.
60     */
61    private boolean errorInLineLenght = false;
62  
63    /**
64     * Indicates that the previous character was CR (Carriage Return).
65     */
66    private boolean CRCharIndcator = false;
67  
68    /**
69     * Simple constructor.
70     */
71    public ContentAnalyzer() {
72  
73    }
74  
75    /**
76     * Method checks one charachter passed as integer, if it passes condition for
77     * implementation of 7bit Content-Transfer-Encoding. For checking content
78     * data which is constructed from more than one byte, this method must be called
79     * in loop for every byte that belongs to examined content data.
80     * @param dataToCheck integer represetation of passed byte for check
81     */
82    public void isCharFor7BitEncoding(int dataToCheck) {
83      lineLenght++;
84      numberOfBytes++;
85      if(lineLenght > 1000)
86        errorInLineLenght = true;
87      if(dataToCheck < 0x7F && dataToCheck > 0x1F) {
88        return;
89      }
90      if(dataToCheck == 0x09)
91        return;
92      if(dataToCheck == 0x0D) {
93        if(CRCharIndcator) {
94          numberOfNon7bitBytes++;
95          return;
96        }
97        else {
98          CRCharIndcator = true;
99          return;
100       }
101     }
102     if(dataToCheck == 0x0A) {
103       if(CRCharIndcator) {
104         CRCharIndcator = false;
105         lineLenght = 0; // CRLF resets line lenght to 0
106         return;
107       }
108       else {
109         numberOfNon7bitBytes++;
110         return;
111       }
112     }
113     numberOfNon7bitBytes++;
114   }
115 
116   /**
117    * Method checks data passed as InputStream which will be used later in the
118    * process of creation of content in MimeBodyPart, if it passes condition for
119    * implementation of 7bit Content-Transfer-Encoding.
120    * @param dataToCheck content data which has to be checked represented as
121    * InputStream
122    * @exception SMIMEException caused by non SMIMEException which is IOException
123    */
124   public void isFor7BitEncoding(InputStream   dataToCheck) throws SMIMEException {
125     BufferedInputStream   buf = new BufferedInputStream  (dataToCheck, 1000);
126     try {
127       int available = buf.available();
128       int markPoint = 0;
129       while(markPoint < available) {
130         buf.mark(markPoint);
131         buf.reset();
132         int byteVal = buf.read();
133         while(byteVal != -1) {
134           this.isCharFor7BitEncoding(byteVal);
135           byteVal = buf.read();
136         }
137         markPoint+=1000;
138       }
139       buf.close();
140     }
141     catch(IOException   e) {
142             throw SMIMEException.getInstance(this, e, "isFor7BitEncoding");
143     }
144   }
145 
146   /**
147    * Method checks data passed from File which will be used later in the process
148    * of creation of content in MimeBodyPart, if it passes condition for
149    * implementation of 7bit Content-Transfer-Encoding.
150    * @param fileToCheck pointer to file which data has to be checked.
151    * @exception SMIMEException caused by non SMIMEException which is IOException
152    */
153   public void isFor7BitEncoding(File   fileToCheck) throws SMIMEException {
154     try {
155       FileInputStream   fi = new FileInputStream  (fileToCheck);
156       this.isFor7BitEncoding(fi);
157     }
158     catch(IOException   e) {
159     throw SMIMEException.getInstance(this, e, "isFor7BitEncoding");    }
160   }
161 
162   /**
163    * Performs analyze of content data and accoding to that decides which
164    * Content-Transfer-Encoding to suggesst.
165    */
166   public void analyse() {
167 
168   }
169 
170   /**
171    * Returns suggested encoding, after finished analyze.
172    * @return the name of Content-Transfer-Encoding header parameter which can
173    * take values: 7bit, quoted-printable, base64 or null in case of error.
174    */
175   public String   getPreferedEncoding() {
176     if(numberOfBytes == 0)
177       return "7bit";
178     if( numberOfNon7bitBytes == 0 && !errorInLineLenght)
179       return "7bit";
180     if( numberOfNon7bitBytes == 0 && errorInLineLenght)
181       return "quoted-printable";
182     if( numberOfNon7bitBytes != 0) {
183       int percentage = 100 * numberOfNon7bitBytes / numberOfBytes;
184       if(percentage > 30)
185         return "base64";
186       else
187         return "quoted-printable";
188     }
189     return null;
190   }
191 
192   /**
193    * Resets all storaged values in ContentAnalyzer object to default values and
194    * prepares object for new use.
195    */
196   public void reset() {
197     numberOfBytes = 0;
198     numberOfNon7bitBytes = 0;
199     lineLenght = 0;
200     errorInLineLenght = false;
201     CRCharIndcator = false;
202   }
203 }
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags