7151118: Regressions on 7u4 b11 comp. 7u4 b06 on specjvm2008.xml.transform subbenchmark
Summary: roll back XalanJ-2271 that caused the regression
Reviewed-by: lancea
--- a/jaxp/src/com/sun/org/apache/xml/internal/serializer/CharInfo.java Wed Jul 05 18:12:32 2017 +0200
+++ b/jaxp/src/com/sun/org/apache/xml/internal/serializer/CharInfo.java Thu Jun 07 13:47:53 2012 -0700
@@ -55,7 +55,7 @@
final class CharInfo
{
/** Given a character, lookup a String to output (e.g. a decorated entity reference). */
- private HashMap m_charToString;
+ private HashMap m_charToString = new HashMap();
/**
* The name of the HTML entities file.
@@ -72,50 +72,42 @@
"com.sun.org.apache.xml.internal.serializer.XMLEntities";
/** The horizontal tab character, which the parser should always normalize. */
- static final char S_HORIZONAL_TAB = 0x09;
+ public static final char S_HORIZONAL_TAB = 0x09;
/** The linefeed character, which the parser should always normalize. */
- static final char S_LINEFEED = 0x0A;
+ public static final char S_LINEFEED = 0x0A;
/** The carriage return character, which the parser should always normalize. */
- static final char S_CARRIAGERETURN = 0x0D;
- static final char S_SPACE = 0x20;
- static final char S_QUOTE = 0x22;
- static final char S_LT = 0x3C;
- static final char S_GT = 0x3E;
- static final char S_NEL = 0x85;
- static final char S_LINE_SEPARATOR = 0x2028;
+ public static final char S_CARRIAGERETURN = 0x0D;
/** 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.
* @xsl.usage internal
*/
- boolean onlyQuotAmpLtGt;
+ final boolean onlyQuotAmpLtGt;
/** Copy the first 0,1 ... ASCII_MAX values into an array */
- static final int ASCII_MAX = 128;
+ private static final int ASCII_MAX = 128;
/** Array of values is faster access than a set of bits
- * to quickly check ASCII characters in attribute values,
- * the value is true if the character in an attribute value
- * should be mapped to a String.
+ * to quickly check ASCII characters in attribute values.
*/
- private final boolean[] shouldMapAttrChar_ASCII;
+ private boolean[] isSpecialAttrASCII = new boolean[ASCII_MAX];
/** Array of values is faster access than a set of bits
- * to quickly check ASCII characters in text nodes,
- * the value is true if the character in a text node
- * should be mapped to a String.
+ * to quickly check ASCII characters in text nodes.
*/
- private final boolean[] shouldMapTextChar_ASCII;
+ private boolean[] isSpecialTextASCII = new boolean[ASCII_MAX];
+
+ private boolean[] isCleanTextASCII = new boolean[ASCII_MAX];
/** 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 final int array_of_bits[];
+ private int array_of_bits[] = createEmptySetOfIntegers(65535);
// 5 for 32 bit words, 6 for 64 bit words ...
@@ -146,38 +138,33 @@
/**
- * A base constructor just to explicitly create the fields,
- * with the exception of m_charToString which is handled
- * by the constructor that delegates base construction to this one.
- * <p>
- * m_charToString is not created here only for performance reasons,
- * to avoid creating a Hashtable that will be replaced when
- * making a mutable copy, {@link #mutableCopyOf(CharInfo)}.
+ * 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:
*
+ * <pre>
+ * # First char # is a comment
+ * Entity numericValue
+ * quot 34
+ * amp 38
+ * </pre>
+ *
+ * @param entitiesResource Name of properties or resource file that should
+ * be loaded, which describes that mapping of characters to entity
+ * references.
*/
- private CharInfo()
+ private CharInfo(String entitiesResource, String method)
{
- this.array_of_bits = createEmptySetOfIntegers(65535);
- this.firstWordNotUsed = 0;
- this.shouldMapAttrChar_ASCII = new boolean[ASCII_MAX];
- this.shouldMapTextChar_ASCII = new boolean[ASCII_MAX];
- this.m_charKey = new CharKey();
-
- // Not set here, but in a constructor that uses this one
- // this.m_charToString = new Hashtable();
-
- this.onlyQuotAmpLtGt = true;
-
-
- return;
+ this(entitiesResource, method, false);
}
private CharInfo(String entitiesResource, String method, boolean internal)
{
- // call the default constructor to create the fields
- this();
- m_charToString = new HashMap();
-
ResourceBundle entities = null;
boolean noExtraEntities = true;
@@ -203,10 +190,12 @@
String name = (String) keys.nextElement();
String value = entities.getString(name);
int code = Integer.parseInt(value);
- boolean extra = defineEntity(name, (char) code);
- if (extra)
+ defineEntity(name, (char) code);
+ if (extraEntity(code))
noExtraEntities = false;
}
+ set(S_LINEFEED);
+ set(S_CARRIAGERETURN);
} else {
InputStream is = null;
@@ -290,8 +279,8 @@
int code = Integer.parseInt(value);
- boolean extra = defineEntity(name, (char) code);
- if (extra)
+ defineEntity(name, (char) code);
+ if (extraEntity(code))
noExtraEntities = false;
}
}
@@ -300,6 +289,8 @@
}
is.close();
+ set(S_LINEFEED);
+ set(S_CARRIAGERETURN);
} catch (Exception e) {
throw new RuntimeException(
Utils.messages.createMessage(
@@ -317,8 +308,31 @@
}
}
+ /* 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;
+ }
+
+
+
onlyQuotAmpLtGt = noExtraEntities;
+ // initialize the array with a cache of the BitSet values
+ for (int i=0; i<ASCII_MAX; i++)
+ isSpecialAttrASCII[i] = get(i);
+
/* 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!).
@@ -330,19 +344,8 @@
*/
if (Method.XML.equals(method))
{
- // We choose not to escape the quotation mark as " in text nodes
- shouldMapTextChar_ASCII[S_QUOTE] = false;
+ isSpecialAttrASCII[S_HORIZONAL_TAB] = true;
}
-
- if (Method.HTML.equals(method)) {
- // The XSLT 1.0 recommendation says
- // "The html output method should not escape < characters occurring in attribute values."
- // So we don't escape '<' in an attribute for HTML
- shouldMapAttrChar_ASCII['<'] = false;
-
- // We choose not to escape the quotation mark as " in text nodes.
- shouldMapTextChar_ASCII[S_QUOTE] = false;
- }
}
/**
@@ -350,39 +353,23 @@
* supplied. Nothing happens if the character reference is already defined.
* <p>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. '&lt;' is an
+ * parsing and printing, primarily for HTML documents. '<amp;' is an
* example of a character reference.</p>
*
* @param name The entity's name
* @param value The entity's value
- * @return true if the mapping is not one of:
- * <ul>
- * <li> '<' to "<"
- * <li> '>' to ">"
- * <li> '&' to "&"
- * <li> '"' to """
- * </ul>
*/
- private boolean defineEntity(String name, char value)
+ private void defineEntity(String name, char value)
{
StringBuilder sb = new StringBuilder("&");
sb.append(name);
sb.append(';');
String entityString = sb.toString();
- boolean extra = defineChar2StringMapping(entityString, value);
- return extra;
+ defineChar2StringMapping(entityString, value);
}
/**
- * A utility object, just used to map characters to output Strings,
- * needed because a HashMap needs to map an object as a key, not a
- * Java primitive type, like a char, so this object gets around that
- * and it is reusable.
- */
- private final CharKey m_charKey;
-
- /**
* Map a character to a String. For example given
* the character '>' this method would return the fully decorated
* entity name "<".
@@ -413,21 +400,21 @@
/**
* Tell if the character argument that is from
- * an attribute value has a mapping to a String.
+ * 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,
- * such as when writing out entity references.
+ * or entity references.
* @xsl.usage internal
*/
- final boolean shouldMapAttrChar(int value)
+ final boolean isSpecialAttrChar(int value)
{
// 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 shouldMapAttrChar_ASCII[value];
+ return isSpecialAttrASCII[value];
// rather than java.util.BitSet, our private
// implementation is faster (and less general).
@@ -436,27 +423,48 @@
/**
* Tell if the character argument that is from a
- * text node has a mapping to a String, for example
- * to map '<' to "<".
+ * text node should have special treatment.
*
* @param value the value of a character that is in a text node
- * @return true if the character has a mapping to a String,
- * such as when writing out entity references.
+ * @return true if the character should have any special treatment,
+ * such as when writing out attribute values,
+ * or entity references.
* @xsl.usage internal
*/
- final boolean shouldMapTextChar(int value)
+ final boolean isSpecialTextChar(int value)
{
// 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 shouldMapTextChar_ASCII[value];
+ return isSpecialTextASCII[value];
// rather than java.util.BitSet, our private
// implementation is faster (and less general).
return get(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
+ */
+ final boolean isTextASCIIClean(int value)
+ {
+ return isCleanTextASCII[value];
+ }
+
+// In the future one might want to use the array directly and avoid
+// the method call, but I think the JIT alreay inlines this well enough
+// so don't do it (for now) - bjm
+// public final boolean[] getASCIIClean()
+// {
+// return isCleanTextASCII;
+// }
+
+
private static CharInfo getCharInfoBasedOnPrivilege(
final String entitiesFileName, final String method,
final boolean internal){
@@ -491,17 +499,15 @@
{
CharInfo charInfo = (CharInfo) m_getCharInfoCache.get(entitiesFileName);
if (charInfo != null) {
- return mutableCopyOf(charInfo);
+ return charInfo;
}
// try to load it internally - cache
try {
charInfo = getCharInfoBasedOnPrivilege(entitiesFileName,
method, true);
- // Put the common copy of charInfo in the cache, but return
- // a copy of it.
m_getCharInfoCache.put(entitiesFileName, charInfo);
- return mutableCopyOf(charInfo);
+ return charInfo;
} catch (Exception e) {}
// try to load it externally - do not cache
@@ -528,41 +534,7 @@
method, false);
}
- /**
- * Create a mutable copy of the cached one.
- * @param charInfo The cached one.
- * @return
- */
- private static CharInfo mutableCopyOf(CharInfo charInfo) {
- CharInfo copy = new CharInfo();
-
- int max = charInfo.array_of_bits.length;
- System.arraycopy(charInfo.array_of_bits,0,copy.array_of_bits,0,max);
-
- copy.firstWordNotUsed = charInfo.firstWordNotUsed;
-
- max = charInfo.shouldMapAttrChar_ASCII.length;
- System.arraycopy(charInfo.shouldMapAttrChar_ASCII,0,copy.shouldMapAttrChar_ASCII,0,max);
-
- max = charInfo.shouldMapTextChar_ASCII.length;
- System.arraycopy(charInfo.shouldMapTextChar_ASCII,0,copy.shouldMapTextChar_ASCII,0,max);
-
- // utility field copy.m_charKey is already created in the default constructor
-
- copy.m_charToString = (HashMap) charInfo.m_charToString.clone();
-
- copy.onlyQuotAmpLtGt = charInfo.onlyQuotAmpLtGt;
-
- return copy;
- }
-
- /**
- * Table of user-specified char infos.
- * The table maps entify file names (the name of the
- * property file without the .properties extension)
- * to CharInfo objects populated with entities defined in
- * corresponding property file.
- */
+ /** Table of user-specified char infos. */
private static HashMap m_getCharInfoCache = new HashMap();
/**
@@ -604,8 +576,7 @@
* the creation of the set.
*/
private final void set(int i) {
- setASCIItextDirty(i);
- setASCIIattrDirty(i);
+ setASCIIdirty(i);
int j = (i >> SHIFT_PER_WORD); // this word is used
int k = j + 1;
@@ -640,43 +611,24 @@
return in_the_set;
}
+ // record if there are any entities other than
+ // quot, amp, lt, gt (probably user defined)
/**
- * This method returns true if there are some non-standard mappings to
- * entities other than quot, amp, lt, gt, and its only purpose is for
- * performance.
- * @param charToMap The value of the character that is mapped to a String
- * @param outputString The String to which the character is mapped, usually
- * an entity reference such as "<".
- * @return true if the mapping is not one of:
- * <ul>
- * <li> '<' to "<"
- * <li> '>' to ">"
- * <li> '&' to "&"
- * <li> '"' to """
- * </ul>
+ * @return true if the entity
+ * @param code The value of the character that has an entity defined
+ * for it.
*/
- private boolean extraEntity(String outputString, int charToMap)
+ private boolean extraEntity(int entityValue)
{
boolean extra = false;
- if (charToMap < ASCII_MAX)
+ if (entityValue < 128)
{
- switch (charToMap)
+ switch (entityValue)
{
- case '"' : // quot
- if (!outputString.equals("""))
- extra = true;
- break;
- case '&' : // amp
- if (!outputString.equals("&"))
- extra = true;
- break;
- case '<' : // lt
- if (!outputString.equals("<"))
- extra = true;
- break;
- case '>' : // gt
- if (!outputString.equals(">"))
- extra = true;
+ case 34 : // quot
+ case 38 : // amp
+ case 60 : // lt
+ case 62 : // gt
break;
default : // other entity in range 0 to 127
extra = true;
@@ -686,61 +638,49 @@
}
/**
- * If the character is in the ASCII range then
- * mark it as needing replacement with
- * a String on output if it occurs in a text node.
+ * If the character is a printable ASCII character then
+ * mark it as not clean and needing replacement with
+ * a String on output.
* @param ch
*/
- private void setASCIItextDirty(int j)
+ private void setASCIIdirty(int j)
{
if (0 <= j && j < ASCII_MAX)
{
- shouldMapTextChar_ASCII[j] = true;
+ isCleanTextASCII[j] = false;
+ isSpecialTextASCII[j] = true;
}
}
/**
- * If the character is in the ASCII range then
- * mark it as needing replacement with
- * a String on output if it occurs in a attribute value.
+ * If the character is a printable ASCII character then
+ * mark it as and not needing replacement with
+ * a String on output.
* @param ch
*/
- private void setASCIIattrDirty(int j)
+ private void setASCIIclean(int j)
{
if (0 <= j && j < ASCII_MAX)
{
- shouldMapAttrChar_ASCII[j] = true;
+ isCleanTextASCII[j] = true;
+ isSpecialTextASCII[j] = false;
}
}
- /**
- * Call this method to register a char to String mapping, for example
- * to map '<' to "<".
- * @param outputString The String to map to.
- * @param inputChar The char to map from.
- * @return true if the mapping is not one of:
- * <ul>
- * <li> '<' to "<"
- * <li> '>' to ">"
- * <li> '&' to "&"
- * <li> '"' to """
- * </ul>
- */
- boolean defineChar2StringMapping(String outputString, char inputChar)
+ private void defineChar2StringMapping(String outputString, char inputChar)
{
CharKey character = new CharKey(inputChar);
m_charToString.put(character, outputString);
- set(inputChar); // mark the character has having a mapping to a String
-
- boolean extraMapping = extraEntity(outputString, inputChar);
- return extraMapping;
-
+ set(inputChar);
}
/**
* Simple class for fast lookup of char values, when used with
* hashtables. You can set the char, then use it as a key.
*
+ * This class is a copy of the one in com.sun.org.apache.xml.internal.utils.
+ * It exists to cut the serializers dependancy on that package.
+ *
* @xsl.usage internal
*/
private static class CharKey extends Object
--- a/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToHTMLStream.java Wed Jul 05 18:12:32 2017 +0200
+++ b/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToHTMLStream.java Thu Jun 07 13:47:53 2012 -0700
@@ -58,7 +58,7 @@
* Map that tells which XML characters should have special treatment, and it
* provides character to entity name lookup.
*/
- private final CharInfo m_htmlcharInfo =
+ private static final CharInfo m_htmlcharInfo =
// new CharInfo(CharInfo.HTML_ENTITIES_RESOURCE);
CharInfo.getCharInfo(CharInfo.HTML_ENTITIES_RESOURCE, Method.HTML);
@@ -1369,7 +1369,7 @@
// System.out.println("ch: "+(int)ch);
// System.out.println("m_maxCharacter: "+(int)m_maxCharacter);
// System.out.println("m_attrCharsMap[ch]: "+(int)m_attrCharsMap[ch]);
- if (escapingNotNeeded(ch) && (!m_charInfo.shouldMapAttrChar(ch)))
+ if (escapingNotNeeded(ch) && (!m_charInfo.isSpecialAttrChar(ch)))
{
cleanLength++;
}
--- a/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToStream.java Wed Jul 05 18:12:32 2017 +0200
+++ b/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToStream.java Thu Jun 07 13:47:53 2012 -0700
@@ -919,8 +919,7 @@
{
// This is the old/fast code here, but is this
// correct for all encodings?
- if (ch >= CharInfo.S_SPACE || (CharInfo.S_LINEFEED == ch ||
- CharInfo.S_CARRIAGERETURN == ch || CharInfo.S_HORIZONAL_TAB == ch))
+ if (ch >= 0x20 || (0x0A == ch || 0x0D == ch || 0x09 == ch))
ret= true;
else
ret = false;
@@ -1029,7 +1028,7 @@
*
* @throws java.io.IOException
*/
- int accumDefaultEntity(
+ protected int accumDefaultEntity(
java.io.Writer writer,
char ch,
int i,
@@ -1048,7 +1047,7 @@
{
// if this is text node character and a special one of those,
// or if this is a character from attribute value and a special one of those
- if ((fromTextNode && m_charInfo.shouldMapTextChar(ch)) || (!fromTextNode && m_charInfo.shouldMapAttrChar(ch)))
+ if ((fromTextNode && m_charInfo.isSpecialTextChar(ch)) || (!fromTextNode && m_charInfo.isSpecialAttrChar(ch)))
{
String outputStringForChar = m_charInfo.getOutputStringForChar(ch);
@@ -1399,6 +1398,7 @@
if (m_cdataTagOpen)
closeCDATA();
+ // the check with _escaping is a bit of a hack for XLSTC
if (m_disableOutputEscapingStates.peekOrFalse() || (!m_escaping))
{
@@ -1421,173 +1421,82 @@
try
{
int i;
+ char ch1;
int startClean;
// skip any leading whitspace
// don't go off the end and use a hand inlined version
// of isWhitespace(ch)
final int end = start + length;
- int lastDirtyCharProcessed = start - 1; // last non-clean character that was processed
- // that was processed
- final Writer writer = m_writer;
- boolean isAllWhitespace = true;
-
- // process any leading whitspace
- i = start;
- while (i < end && isAllWhitespace) {
- char ch1 = chars[i];
-
- if (m_charInfo.shouldMapTextChar(ch1)) {
- // The character is supposed to be replaced by a String
- // so write out the clean whitespace characters accumulated
- // so far
- // then the String.
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- String outputStringForChar = m_charInfo
- .getOutputStringForChar(ch1);
- writer.write(outputStringForChar);
- // We can't say that everything we are writing out is
- // all whitespace, we just wrote out a String.
- isAllWhitespace = false;
- lastDirtyCharProcessed = i; // mark the last non-clean
- // character processed
- i++;
- } else {
- // The character is clean, but is it a whitespace ?
- switch (ch1) {
- // TODO: Any other whitespace to consider?
- case CharInfo.S_SPACE:
- // Just accumulate the clean whitespace
- i++;
- break;
- case CharInfo.S_LINEFEED:
- lastDirtyCharProcessed = processLineFeed(chars, i,
- lastDirtyCharProcessed, writer);
- i++;
- break;
- case CharInfo.S_CARRIAGERETURN:
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write(" ");
- lastDirtyCharProcessed = i;
- i++;
- break;
- case CharInfo.S_HORIZONAL_TAB:
- // Just accumulate the clean whitespace
- i++;
- break;
- default:
- // The character was clean, but not a whitespace
- // so break the loop to continue with this character
- // (we don't increment index i !!)
- isAllWhitespace = false;
- break;
+ int lastDirty = start - 1; // last character that needed processing
+ for (i = start;
+ ((i < end)
+ && ((ch1 = chars[i]) == 0x20
+ || (ch1 == 0xA && m_lineSepUse)
+ || ch1 == 0xD
+ || ch1 == 0x09));
+ i++)
+ {
+ /*
+ * We are processing leading whitespace, but are doing the same
+ * processing for dirty characters here as for non-whitespace.
+ *
+ */
+ if (!m_charInfo.isTextASCIIClean(ch1))
+ {
+ lastDirty = processDirty(chars,end, i,ch1, lastDirty, true);
+ i = lastDirty;
}
}
- }
/* If there is some non-whitespace, mark that we may need
* to preserve this. This is only important if we have indentation on.
*/
- if (i < end || !isAllWhitespace)
+ if (i < end)
m_ispreserve = true;
+
+// int lengthClean; // number of clean characters in a row
+// final boolean[] isAsciiClean = m_charInfo.getASCIIClean();
+
+ final boolean isXML10 = XMLVERSION10.equals(getVersion());
+ // we've skipped the leading whitespace, now deal with the rest
for (; i < end; i++)
{
- char ch = chars[i];
-
- if (m_charInfo.shouldMapTextChar(ch)) {
- // The character is supposed to be replaced by a String
- // e.g. '&' --> "&"
- // e.g. '<' --> "<"
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- String outputStringForChar = m_charInfo.getOutputStringForChar(ch);
- writer.write(outputStringForChar);
- lastDirtyCharProcessed = i;
- }
- else {
- if (ch <= 0x1F) {
- // Range 0x00 through 0x1F inclusive
- //
- // This covers the non-whitespace control characters
- // in the range 0x1 to 0x1F inclusive.
- // It also covers the whitespace control characters in the same way:
- // 0x9 TAB
- // 0xA NEW LINE
- // 0xD CARRIAGE RETURN
- //
- // We also cover 0x0 ... It isn't valid
- // but we will output "�"
-
- // The default will handle this just fine, but this
- // is a little performance boost to handle the more
- // common TAB, NEW-LINE, CARRIAGE-RETURN
- switch (ch) {
-
- case CharInfo.S_HORIZONAL_TAB:
- // Leave whitespace TAB as a real character
+ {
+ // A tight loop to skip over common clean chars
+ // This tight loop makes it easier for the JIT
+ // to optimize.
+ char ch2;
+ while (i<end
+ && ((ch2 = chars[i])<127)
+ && m_charInfo.isTextASCIIClean(ch2))
+ i++;
+ if (i == end)
break;
- case CharInfo.S_LINEFEED:
- lastDirtyCharProcessed = processLineFeed(chars, i, lastDirtyCharProcessed, writer);
- break;
- case CharInfo.S_CARRIAGERETURN:
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write(" ");
- lastDirtyCharProcessed = i;
- // Leave whitespace carriage return as a real character
- break;
- default:
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- lastDirtyCharProcessed = i;
- break;
-
- }
- }
- else if (ch < 0x7F) {
- // Range 0x20 through 0x7E inclusive
- // Normal ASCII chars, do nothing, just add it to
- // the clean characters
-
}
- else if (ch <= 0x9F){
- // Range 0x7F through 0x9F inclusive
- // More control characters, including NEL (0x85)
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- lastDirtyCharProcessed = i;
+
+ final char ch = chars[i];
+ /* The check for isCharacterInC0orC1Ranger and
+ * isNELorLSEPCharacter has been added
+ * to support Control Characters in XML 1.1
+ */
+ if (!isCharacterInC0orC1Range(ch) &&
+ (isXML10 || !isNELorLSEPCharacter(ch)) &&
+ (escapingNotNeeded(ch) && (!m_charInfo.isSpecialTextChar(ch)))
+ || ('"' == ch))
+ {
+ ; // a character needing no special processing
}
- else if (ch == CharInfo.S_LINE_SEPARATOR) {
- // LINE SEPARATOR
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write("
");
- lastDirtyCharProcessed = i;
- }
- else if (m_encodingInfo.isInEncoding(ch)) {
- // If the character is in the encoding, and
- // not in the normal ASCII range, we also
- // just leave it get added on to the clean characters
-
- }
- else {
- // This is a fallback plan, we should never get here
- // but if the character wasn't previously handled
- // (i.e. isn't in the encoding, etc.) then what
- // should we do? We choose to write out an entity
- writeOutCleanChars(chars, i, lastDirtyCharProcessed);
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- lastDirtyCharProcessed = i;
- }
+ else
+ {
+ lastDirty = processDirty(chars,end, i, ch, lastDirty, true);
+ i = lastDirty;
}
}
// we've reached the end. Any clean characters at the
// end of the array than need to be written out?
- startClean = lastDirtyCharProcessed + 1;
+ startClean = lastDirty + 1;
if (i > startClean)
{
int lengthClean = i - startClean;
@@ -1606,32 +1515,6 @@
if (m_tracer != null)
super.fireCharEvent(chars, start, length);
}
-
- private int processLineFeed(final char[] chars, int i, int lastProcessed, final Writer writer) throws IOException {
- if (!m_lineSepUse
- || (m_lineSepLen ==1 && m_lineSep[0] == CharInfo.S_LINEFEED)){
- // We are leaving the new-line alone, and it is just
- // being added to the 'clean' characters,
- // so the last dirty character processed remains unchanged
- }
- else {
- writeOutCleanChars(chars, i, lastProcessed);
- writer.write(m_lineSep, 0, m_lineSepLen);
- lastProcessed = i;
- }
- return lastProcessed;
- }
-
- private void writeOutCleanChars(final char[] chars, int i, int lastProcessed) throws IOException {
- int startClean;
- startClean = lastProcessed + 1;
- if (startClean < i)
- {
- int lengthClean = i - startClean;
- m_writer.write(chars, startClean, lengthClean);
- }
- }
-
/**
* This method checks if a given character is between C0 or C1 range
* of Control characters.
@@ -1751,7 +1634,7 @@
*
* @throws org.xml.sax.SAXException
*/
- private int accumDefaultEscape(
+ protected int accumDefaultEscape(
Writer writer,
char ch,
int i,
@@ -1815,15 +1698,16 @@
* to write it out as Numeric Character Reference(NCR) regardless of XML Version
* being used for output document.
*/
- if (isCharacterInC0orC1Range(ch) || isNELorLSEPCharacter(ch))
+ if (isCharacterInC0orC1Range(ch) ||
+ (XMLVERSION11.equals(getVersion()) && isNELorLSEPCharacter(ch)))
{
writer.write("&#");
writer.write(Integer.toString(ch));
writer.write(';');
}
else if ((!escapingNotNeeded(ch) ||
- ( (fromTextNode && m_charInfo.shouldMapTextChar(ch))
- || (!fromTextNode && m_charInfo.shouldMapAttrChar(ch))))
+ ( (fromTextNode && m_charInfo.isSpecialTextChar(ch))
+ || (!fromTextNode && m_charInfo.isSpecialAttrChar(ch))))
&& m_elemContext.m_currentElemDepth > 0)
{
writer.write("&#");
@@ -2087,86 +1971,28 @@
string.getChars(0,len, m_attrBuff, 0);
final char[] stringChars = m_attrBuff;
- for (int i = 0; i < len;)
+ for (int i = 0; i < len; )
{
char ch = stringChars[i];
-
- if (m_charInfo.shouldMapAttrChar(ch) || !(escapingNotNeeded(ch))) {
- // The character is supposed to be replaced by a String
- // e.g. '&' --> "&"
- // e.g. '<' --> "<"
+ if (escapingNotNeeded(ch) && (!m_charInfo.isSpecialAttrChar(ch)))
+ {
+ writer.write(ch);
+ i++;
+ }
+ else
+ { // I guess the parser doesn't normalize cr/lf in attributes. -sb
+// if ((CharInfo.S_CARRIAGERETURN == ch)
+// && ((i + 1) < len)
+// && (CharInfo.S_LINEFEED == stringChars[i + 1]))
+// {
+// i++;
+// ch = CharInfo.S_LINEFEED;
+// }
+
i = accumDefaultEscape(writer, ch, i, stringChars, len, false, true);
}
- else {
- i++;
- if (0x0 <= ch && ch <= 0x1F) {
- // Range 0x00 through 0x1F inclusive
- // This covers the non-whitespace control characters
- // in the range 0x1 to 0x1F inclusive.
- // It also covers the whitespace control characters in the same way:
- // 0x9 TAB
- // 0xA NEW LINE
- // 0xD CARRIAGE RETURN
- //
- // We also cover 0x0 ... It isn't valid
- // but we will output "�"
-
- // The default will handle this just fine, but this
- // is a little performance boost to handle the more
- // common TAB, NEW-LINE, CARRIAGE-RETURN
- switch (ch) {
-
- case CharInfo.S_HORIZONAL_TAB:
- writer.write("	");
- break;
- case CharInfo.S_LINEFEED:
- writer.write(" ");
- break;
- case CharInfo.S_CARRIAGERETURN:
- writer.write(" ");
- break;
- default:
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- break;
-
}
- }
- else if (ch < 0x7F) {
- // Range 0x20 through 0x7E inclusive
- // Normal ASCII chars
- writer.write(ch);
- }
- else if (ch <= 0x9F){
- // Range 0x7F through 0x9F inclusive
- // More control characters
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- }
- else if (ch == CharInfo.S_LINE_SEPARATOR) {
- // LINE SEPARATOR
- writer.write("
");
- }
- else if (m_encodingInfo.isInEncoding(ch)) {
- // If the character is in the encoding, and
- // not in the normal ASCII range, we also
- // just write it out
- writer.write(ch);
- }
- else {
- // This is a fallback plan, we should never get here
- // but if the character wasn't previously handled
- // (i.e. isn't in the encoding, etc.) then what
- // should we do? We choose to write out a character ref
- writer.write("&#");
- writer.write(Integer.toString(ch));
- writer.write(';');
- }
-
- }
- }
+
}
/**
@@ -2936,14 +2762,6 @@
closeCDATA();
m_cdataTagOpen = false;
}
- if (m_writer != null) {
- try {
- m_writer.flush();
- }
- catch(IOException e) {
- // what? me worry?
- }
- }
}
public void setContentHandler(ContentHandler ch)
--- a/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToXMLStream.java Wed Jul 05 18:12:32 2017 +0200
+++ b/jaxp/src/com/sun/org/apache/xml/internal/serializer/ToXMLStream.java Thu Jun 07 13:47:53 2012 -0700
@@ -56,7 +56,7 @@
* Map that tells which XML characters should have special treatment, and it
* provides character to entity name lookup.
*/
- private CharInfo m_xmlcharInfo =
+ private static CharInfo m_xmlcharInfo =
// new CharInfo(CharInfo.XML_ENTITIES_RESOURCE);
CharInfo.getCharInfo(CharInfo.XML_ENTITIES_RESOURCE, Method.XML);
@@ -329,11 +329,12 @@
/**
* Before Xalan 1497, a newline char was printed out if not inside of an
- * element. The whitespace is not significant if the output is standalone
+ * element. The whitespace is not significant is the output is standalone
*/
if (m_elemContext.m_currentElemDepth <= 0 && m_isStandalone)
writer.write(m_lineSep, 0, m_lineSepLen);
+
/*
* Don't write out any indentation whitespace now,
* because there may be non-whitespace text after this.