FileDocCategorySizeDatePackage
CharInfo.javaAPI DocJava SE 5 API22114Fri Aug 26 14:56:02 BST 2005com.sun.org.apache.xml.internal.serializer

CharInfo

public class CharInfo extends Object
This class provides services that tell if a character should have special treatement, such as entity reference substitution or normalization of a newline character. It also provides character to entity reference lookup. DEVELOPERS: See Known Issue in the constructor.
xsl.usage
internal

Fields Summary
private Hashtable
m_charToEntityRef
Lookup table for characters to entity references.
public static String
HTML_ENTITIES_RESOURCE
The name of the HTML entities file. If specified, the file will be resource loaded with the default class loader.
public static String
XML_ENTITIES_RESOURCE
The name of the XML entities file. If specified, the file will be resource loaded with the default class loader.
public static final char
S_HORIZONAL_TAB
The horizontal tab character, which the parser should always normalize.
public static final char
S_LINEFEED
The linefeed character, which the parser should always normalize.
public static char
S_CARRIAGERETURN
The carriage return character, which the parser should always normalize.
final boolean
onlyQuotAmpLtGt
This flag is an optimization for HTML entities. It false if entities other than quot (34), amp (38), lt (60) and gt (62) are defined in the range 0 to 127.
private static final int
ASCII_MAX
Copy the first 0,1 ... ASCII_MAX values into an array
private boolean[]
isSpecialAttrASCII
Array of values is faster access than a set of bits to quickly check ASCII characters in attribute values.
private boolean[]
isSpecialTextASCII
Array of values is faster access than a set of bits to quickly check ASCII characters in text nodes.
private boolean[]
isCleanTextASCII
private int[]
array_of_bits
An array of bits to record if the character is in the set. Although information in this array is complete, the isSpecialAttrASCII array is used first because access to its values is common and faster.
private static final int
SHIFT_PER_WORD
private static final int
LOW_ORDER_BITMASK
private int
firstWordNotUsed
private CharKey
m_charKey
private static Hashtable
m_getCharInfoCache
Table of user-specified char infos.
Constructors Summary
private CharInfo(String entitiesResource, String method)
Constructor that reads in a resource file that describes the mapping of characters to entity references. This constructor is private, just to force the use of the getCharInfo(entitiesResource) factory Resource files must be encoded in UTF-8 and can either be properties files with a .properties extension assumed. Alternatively, they can have the following form, with no particular extension assumed:
# First char # is a comment
Entity numericValue
quot 34
amp 38

param
entitiesResource Name of properties or resource file that should be loaded, which describes that mapping of characters to entity references.



                                                                                                             
        
    
        this(entitiesResource, method, false);
    
private CharInfo(String entitiesResource, String method, boolean internal)

        ResourceBundle entities = null;
        boolean noExtraEntities = true;

        // Make various attempts to interpret the parameter as a properties
        // file or resource file, as follows:
        //
        //   1) attempt to load .properties file using ResourceBundle
        //   2) try using the class loader to find the specified file a resource
        //      file
        //   3) try treating the resource a URI

        if (internal) { 
            try {
                // Load entity property files by using PropertyResourceBundle,
                // cause of security issure for applets
                entities = PropertyResourceBundle.getBundle(entitiesResource);
            } catch (Exception e) {}
        }

        if (entities != null) {
            Enumeration keys = entities.getKeys();
            while (keys.hasMoreElements()){
                String name = (String) keys.nextElement();
                String value = entities.getString(name);
                int code = Integer.parseInt(value);
                defineEntity(name, (char) code);
                if (extraEntity(code))
                    noExtraEntities = false;
            }
            set(S_LINEFEED);
            set(S_CARRIAGERETURN);
        } else {
            InputStream is = null;

            // Load user specified resource file by using URL loading, it
            // requires a valid URI as parameter
            try {
                if (internal) {
                    is = CharInfo.class.getResourceAsStream(entitiesResource);
                } else {
                    ClassLoader cl = ObjectFactory.findClassLoader();
                    if (cl == null) {
                        is = ClassLoader.getSystemResourceAsStream(entitiesResource);
                    } else {
                        is = cl.getResourceAsStream(entitiesResource);
                    }

                    if (is == null) {
                        try {
                            URL url = new URL(entitiesResource);
                            is = url.openStream();
                        } catch (Exception e) {}
                    }
                }

                if (is == null) {
                    throw new RuntimeException(
                        XMLMessages.createXMLMessage(
                            XMLErrorResources.ER_RESOURCE_COULD_NOT_FIND,
                            new Object[] {entitiesResource, entitiesResource}));
                }

                // Fix Bugzilla#4000: force reading in UTF-8
                //  This creates the de facto standard that Xalan's resource 
                //  files must be encoded in UTF-8. This should work in all
                // JVMs.
                //
                // %REVIEW% KNOWN ISSUE: IT FAILS IN MICROSOFT VJ++, which
                // didn't implement the UTF-8 encoding. Theoretically, we should
                // simply let it fail in that case, since the JVM is obviously
                // broken if it doesn't support such a basic standard.  But
                // since there are still some users attempting to use VJ++ for
                // development, we have dropped in a fallback which makes a
                // second attempt using the platform's default encoding. In VJ++
                // this is apparently ASCII, which is subset of UTF-8... and
                // since the strings we'll be reading here are also primarily
                // limited to the 7-bit ASCII range (at least, in English
                // versions of Xalan), this should work well enough to keep us
                // on the air until we're ready to officially decommit from
                // VJ++.

                BufferedReader reader;
                try {
                    reader = new BufferedReader(new InputStreamReader(is, "UTF-8"));
                } catch (UnsupportedEncodingException e) {
                    reader = new BufferedReader(new InputStreamReader(is));
                }

                String line = reader.readLine();

                while (line != null) {
                    if (line.length() == 0 || line.charAt(0) == '#") {
                        line = reader.readLine();

                        continue;
                    }

                    int index = line.indexOf(' ");

                    if (index > 1) {
                        String name = line.substring(0, index);

                        ++index;

                        if (index < line.length()) {
                            String value = line.substring(index);
                            index = value.indexOf(' ");

                            if (index > 0) {
                                value = value.substring(0, index);
                            }

                            int code = Integer.parseInt(value);

                            defineEntity(name, (char) code);
                            if (extraEntity(code))
                                noExtraEntities = false;
                        }
                    }

                    line = reader.readLine();
                }

                is.close();
                set(S_LINEFEED);
                set(S_CARRIAGERETURN);
            } catch (Exception e) {
                throw new RuntimeException(
                    XMLMessages.createXMLMessage(
                        XMLErrorResources.ER_RESOURCE_COULD_NOT_LOAD,
                        new Object[] { entitiesResource,
                                       e.toString(),
                                       entitiesResource,
                                       e.toString()}));
            } finally {
                if (is != null) {
                    try {
                        is.close();
                    } catch (Exception except) {}
                }
            }
        }
          
        /* initialize the array isCleanTextASCII[] with a cache of values
         * for use by ToStream.character(char[], int , int)
         * and the array isSpecialTextASCII[] with the opposite values
         * (all in the name of performance!)
         */
        for (int ch = 0; ch <ASCII_MAX; ch++)
        if((((0x20 <= ch || (0x0A == ch || 0x0D == ch || 0x09 == ch)))
             && (!get(ch))) || ('"" == ch))
        {
            isCleanTextASCII[ch] = true;
            isSpecialTextASCII[ch] = false;
        }
        else {
            isCleanTextASCII[ch] = false;
            isSpecialTextASCII[ch] = true;     
        }       
        
        /* Now that we've used get(ch) just above to initialize the
         * two arrays we will change by adding a tab to the set of 
         * special chars for XML (but not HTML!).
         * We do this because a tab is always a
         * special character in an XML attribute, 
         * but only a special character in XML text 
         * if it has an entity defined for it.
         * This is the reason for this delay.
         */
        if (Method.XML.equals(method)) 
        {
            set(S_HORIZONAL_TAB);
        }
        

        onlyQuotAmpLtGt = noExtraEntities;

        // initialize the array with a cache of the BitSet values
        for (int i=0; i<ASCII_MAX; i++)
            isSpecialAttrASCII[i] = get(i);    

    
Methods Summary
private static intarrayIndex(int i)
Returns the array element holding the bit value for the given integer

param
i the integer that might be in the set of integers


                                  
         
        return (i >> SHIFT_PER_WORD);
    
private static intbit(int i)
For a given integer in the set it returns the single bit value used within a given word that represents whether the integer is in the set or not.

        int ret = (1 << (i & LOW_ORDER_BITMASK));
        return ret;
    
private int[]createEmptySetOfIntegers(int max)
Creates a new empty set of integers (characters)

param
max the maximum integer to be in the set.

        firstWordNotUsed = 0; // an optimization 

        int[] arr = new int[arrayIndex(max - 1) + 1];
            return arr;
 
    
private voiddefineEntity(java.lang.String name, char value)
Defines a new character reference. The reference's name and value are supplied. Nothing happens if the character reference is already defined.

Unlike internal entities, character references are a string to single character mapping. They are used to map non-ASCII characters both on parsing and printing, primarily for HTML documents. '<amp;' is an example of a character reference.

param
name The entity's name
param
value The entity's value

        CharKey character = new CharKey(value);

        m_charToEntityRef.put(character, name);
        set(value);
    
private booleanextraEntity(int entityValue)

return
true if the entity
param
code The value of the character that has an entity defined for it.

        boolean extra = false;
        if (entityValue < 128)
        {
            switch (entityValue)
            {
                case 34 : // quot
                case 38 : // amp
                case 60 : // lt
                case 62 : // gt
                    break;
                default : // other entity in range 0 to 127  
                    extra = true;
            }
        }
        return extra;
    
private final booleanget(int i)
Return true if the integer (character)is in the set of integers. This implementation uses an array of integers with 32 bits per integer. If a bit is set to 1 the corresponding integer is in the set of integers.

param
i an integer that is tested to see if it is the set of integers, or not.


        boolean in_the_set = false;
        int j = (i >> SHIFT_PER_WORD); // wordIndex(i)
        // an optimization here, ... a quick test to see
        // if this integer is beyond any of the words in use
        if(j < firstWordNotUsed)
            in_the_set = (array_of_bits[j] & 
                          (1 << (i & LOW_ORDER_BITMASK))
            ) != 0;  // 0L for 64 bit words
        return in_the_set;
    
public static com.sun.org.apache.xml.internal.serializer.CharInfogetCharInfo(java.lang.String entitiesFileName, java.lang.String method)
Factory that reads in a resource file that describes the mapping of characters to entity references. Resource files must be encoded in UTF-8 and have a format like:
# First char # is a comment
Entity numericValue
quot 34
amp 38
(Note: Why don't we just switch to .properties files? Oct-01 -sc)

param
entitiesResource Name of entities resource file that should be loaded, which describes that mapping of characters to entity references.
param
method the output method type, which should be one of "xml", "html", "text"...
xsl.usage
internal

        CharInfo charInfo = (CharInfo) m_getCharInfoCache.get(entitiesFileName);
        if (charInfo != null) {
            return charInfo;
        }

        // try to load it internally - cache
        try {
            charInfo = new CharInfo(entitiesFileName, method, true);
            m_getCharInfoCache.put(entitiesFileName, charInfo);
            return charInfo;
        } catch (Exception e) {}

        // try to load it externally - do not cache
        try {
            return new CharInfo(entitiesFileName, method);
        } catch (Exception e) {}

        String absoluteEntitiesFileName;

        if (entitiesFileName.indexOf(':") < 0) {
            absoluteEntitiesFileName =
                SystemIDResolver.getAbsoluteURIFromRelative(entitiesFileName);
        } else {
            try {
                absoluteEntitiesFileName =
                    SystemIDResolver.getAbsoluteURI(entitiesFileName, null);
            } catch (TransformerException te) {
                throw new WrappedRuntimeException(te);
            }
        }

        return new CharInfo(absoluteEntitiesFileName, method, false);
    
public synchronized java.lang.StringgetEntityNameForChar(char value)
Resolve a character to an entity reference name. This is reusing a stored key object, in an effort to avoid heap activity. Unfortunately, that introduces a threading risk. Simplest fix for now is to make it a synchronized method, or to give up the reuse; I see very little performance difference between them. Long-term solution would be to replace the hashtable with a sparse array keyed directly from the character's integer value; see DTM's string pool for a related solution.

param
value character value that should be resolved to a name.
return
name of character entity, or null if not found.
xsl.usage
internal


                                                                                                                
        
    
        // CharKey m_charKey = new CharKey(); //Alternative to synchronized
        m_charKey.setChar(value);
        return (String) m_charToEntityRef.get(m_charKey);
    
public final booleanisSpecialAttrChar(int value)
Tell if the character argument that is from an attribute value should have special treatment.

param
value the value of a character that is in an attribute value
return
true if the character should have any special treatment, such as when writing out attribute values, or entity references.
xsl.usage
internal

        // for performance try the values in the boolean array first,
        // this is faster access than the BitSet for common ASCII values

        if (value < ASCII_MAX)
            return isSpecialAttrASCII[value];

        // rather than java.util.BitSet, our private
        // implementation is faster (and less general).
        return get(value);
    
public final booleanisSpecialTextChar(int value)
Tell if the character argument that is from a text node should have special treatment.

param
value the value of a character that is in a text node
return
true if the character should have any special treatment, such as when writing out attribute values, or entity references.
xsl.usage
internal

        // for performance try the values in the boolean array first,
        // this is faster access than the BitSet for common ASCII values

        if (value < ASCII_MAX)
            return isSpecialTextASCII[value];

        // rather than java.util.BitSet, our private
        // implementation is faster (and less general).
        return get(value);
    
public final booleanisTextASCIIClean(int value)
This method is used to determine if an ASCII character in a text node (not an attribute value) is "clean".

param
value the character to check (0 to 127).
return
true if the character can go to the writer as-is
xsl.usage
internal

        return isCleanTextASCII[value];
    
private final voidset(int i)
Adds the integer (character) to the set of integers.

param
i the integer to add to the set, valid values are 0, 1, 2 ... up to the maximum that was specified at the creation of the set.

        
        int j = (i >> SHIFT_PER_WORD); // this word is used
        int k = j + 1;       
        
        if(firstWordNotUsed < k) // for optimization purposes.
            firstWordNotUsed = k;
            
        array_of_bits[j] |= (1 << (i & LOW_ORDER_BITMASK));