Add Replaceable.copy to retain out-of-band info during reordering.
X-SVN-Rev: 1244
This commit is contained in:
Alan Liu
parent
0b9db3eadf
commit
943b8564c9
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/text/Replaceable.java,v $
|
||||
* $Date: 2000/03/10 04:07:22 $
|
||||
* $Revision: 1.2 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.3 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -23,7 +23,7 @@ package com.ibm.text;
|
||||
* <p>Copyright © IBM Corporation 1999. All rights reserved.
|
||||
*
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: Replaceable.java,v $ $Revision: 1.2 $ $Date: 2000/03/10 04:07:22 $
|
||||
* @version $RCSfile: Replaceable.java,v $ $Revision: 1.3 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
public interface Replaceable {
|
||||
/**
|
||||
@ -86,4 +86,26 @@ public interface Replaceable {
|
||||
int charsStart, int charsLen);
|
||||
// Note: We use length rather than limit to conform to StringBuffer
|
||||
// and System.arraycopy.
|
||||
|
||||
/**
|
||||
* Copy a substring of this object, retaining attribute (out-of-band)
|
||||
* information. This method is used to duplicate or reorder substrings.
|
||||
* The destination index must not overlap the source range.
|
||||
* Implementations that do not care about maintaining out-of-band
|
||||
* information during copying may use the naive implementation:
|
||||
*
|
||||
* <pre> char[] text = new char[limit - start];
|
||||
* getChars(start, limit, text, 0);
|
||||
* replace(dest, dest, text, 0, limit - start);</pre>
|
||||
*
|
||||
* @param start the beginning index, inclusive; <code>0 <= start <=
|
||||
* limit</code>.
|
||||
* @param limit the ending index, exclusive; <code>start <= limit <=
|
||||
* length()</code>.
|
||||
* @param dest the destination index. The characters from
|
||||
* <code>start..limit-1</code> will be copied to <code>dest</code>.
|
||||
* Implementations of this method may assume that <code>dest <= start ||
|
||||
* dest >= limit</code>.
|
||||
*/
|
||||
void copy(int start, int limit, int dest);
|
||||
}
|
||||
|
||||
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/text/ReplaceableString.java,v $
|
||||
* $Date: 2000/03/10 04:07:22 $
|
||||
* $Revision: 1.2 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.3 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -24,7 +24,7 @@ package com.ibm.text;
|
||||
*
|
||||
* @see Replaceable
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: ReplaceableString.java,v $ $Revision: 1.2 $ $Date: 2000/03/10 04:07:22 $
|
||||
* @version $RCSfile: ReplaceableString.java,v $ $Revision: 1.3 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
public class ReplaceableString implements Replaceable {
|
||||
private StringBuffer buf;
|
||||
@ -67,16 +67,6 @@ public class ReplaceableString implements Replaceable {
|
||||
return buf.toString();
|
||||
}
|
||||
|
||||
/**
|
||||
* Return the internal storage of this object. <em>Note! Any
|
||||
* changes made to the returned object affect this object's
|
||||
* contents, and vice versa.</em>
|
||||
* @return internal buffer used by this object
|
||||
*/
|
||||
public StringBuffer getStringBuffer() {
|
||||
return buf;
|
||||
}
|
||||
|
||||
/**
|
||||
* Return the number of characters contained in this object.
|
||||
* <code>Replaceable</code> API.
|
||||
@ -168,4 +158,24 @@ public class ReplaceableString implements Replaceable {
|
||||
buf.append(tail);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Copy a substring of this object, retaining attribute (out-of-band)
|
||||
* information. This method is used to duplicate or reorder substrings.
|
||||
* The destination index must not overlap the source range.
|
||||
*
|
||||
* @param start the beginning index, inclusive; <code>0 <= start <=
|
||||
* limit</code>.
|
||||
* @param limit the ending index, exclusive; <code>start <= limit <=
|
||||
* length()</code>.
|
||||
* @param dest the destination index. The characters from
|
||||
* <code>start..limit-1</code> will be copied to <code>dest</code>.
|
||||
* Implementations of this method may assume that <code>dest <= start ||
|
||||
* dest >= limit</code>.
|
||||
*/
|
||||
public void copy(int start, int limit, int dest) {
|
||||
char[] text = new char[limit - start];
|
||||
getChars(start, limit, text, 0);
|
||||
replace(dest, dest, text, 0, limit - start);
|
||||
}
|
||||
}
|
||||
|
||||
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/icu/text/TransliterationRule.java,v $
|
||||
* $Date: 2000/04/25 01:42:58 $
|
||||
* $Revision: 1.19 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.20 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -44,7 +44,7 @@ import com.ibm.util.Utility;
|
||||
* <p>Copyright © IBM Corporation 1999. All rights reserved.
|
||||
*
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: TransliterationRule.java,v $ $Revision: 1.19 $ $Date: 2000/04/25 01:42:58 $
|
||||
* @version $RCSfile: TransliterationRule.java,v $ $Revision: 1.20 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
class TransliterationRule {
|
||||
/**
|
||||
@ -259,28 +259,52 @@ class TransliterationRule {
|
||||
*/
|
||||
public int replace(Replaceable text, int offset,
|
||||
RuleBasedTransliterator.Data data) {
|
||||
String out;
|
||||
if (segments == null) {
|
||||
out = output;
|
||||
text.replace(offset, offset + keyLength, output);
|
||||
return output.length() - keyLength;
|
||||
} else {
|
||||
/* When there are segments to be copied, use the Replaceable.copy()
|
||||
* API in order to retain out-of-band data. Copy everything to the
|
||||
* point after the key, then delete the key. That is, copy things
|
||||
* into offset + keyLength, then replace offset .. offset +
|
||||
* keyLength with the empty string.
|
||||
*
|
||||
* Minimize the number of calls to Replaceable.replace() and
|
||||
* Replaceable.copy().
|
||||
*/
|
||||
int textStart = offset - anteContextLength;
|
||||
int dest = offset + keyLength; // copy new text to here
|
||||
StringBuffer buf = new StringBuffer();
|
||||
for (int i=0; i<output.length(); ++i) {
|
||||
char c = output.charAt(i);
|
||||
int b = data.lookupSegmentReference(c);
|
||||
if (b < 0) {
|
||||
// Accumulate straight (non-segment) text.
|
||||
buf.append(c);
|
||||
} else {
|
||||
for (int j=textStart + segments[2*b];
|
||||
j<textStart + segments[2*b+1]; ++j) {
|
||||
buf.append(text.charAt(j));
|
||||
// Insert any accumulated straight text.
|
||||
if (buf.length() > 0) {
|
||||
text.replace(dest, dest, buf.toString());
|
||||
dest += buf.length();
|
||||
buf.setLength(0);
|
||||
}
|
||||
// Copy segment with out-of-band data
|
||||
b *= 2;
|
||||
text.copy(textStart + segments[b],
|
||||
textStart + segments[b+1], dest);
|
||||
dest += segments[b+1] - segments[b];
|
||||
}
|
||||
|
||||
}
|
||||
out = buf.toString();
|
||||
// Insert any accumulated straight text.
|
||||
if (buf.length() > 0) {
|
||||
text.replace(dest, dest, buf.toString());
|
||||
dest += buf.length();
|
||||
}
|
||||
// Delete the key
|
||||
text.replace(offset, offset + keyLength, "");
|
||||
return dest - (offset + keyLength) - keyLength;
|
||||
}
|
||||
text.replace(offset, offset + keyLength, out);
|
||||
return out.length() - keyLength;
|
||||
}
|
||||
|
||||
/**
|
||||
@ -493,6 +517,9 @@ class TransliterationRule {
|
||||
|
||||
/**
|
||||
* $Log: TransliterationRule.java,v $
|
||||
* Revision 1.20 2000/04/25 17:17:37 alan
|
||||
* Add Replaceable.copy to retain out-of-band info during reordering.
|
||||
*
|
||||
* Revision 1.19 2000/04/25 01:42:58 alan
|
||||
* Allow arbitrary length variable values. Clean up Data API. Update javadocs.
|
||||
*
|
||||
|
||||
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/text/Attic/Replaceable.java,v $
|
||||
* $Date: 2000/03/10 04:07:22 $
|
||||
* $Revision: 1.2 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.3 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -23,7 +23,7 @@ package com.ibm.text;
|
||||
* <p>Copyright © IBM Corporation 1999. All rights reserved.
|
||||
*
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: Replaceable.java,v $ $Revision: 1.2 $ $Date: 2000/03/10 04:07:22 $
|
||||
* @version $RCSfile: Replaceable.java,v $ $Revision: 1.3 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
public interface Replaceable {
|
||||
/**
|
||||
@ -86,4 +86,26 @@ public interface Replaceable {
|
||||
int charsStart, int charsLen);
|
||||
// Note: We use length rather than limit to conform to StringBuffer
|
||||
// and System.arraycopy.
|
||||
|
||||
/**
|
||||
* Copy a substring of this object, retaining attribute (out-of-band)
|
||||
* information. This method is used to duplicate or reorder substrings.
|
||||
* The destination index must not overlap the source range.
|
||||
* Implementations that do not care about maintaining out-of-band
|
||||
* information during copying may use the naive implementation:
|
||||
*
|
||||
* <pre> char[] text = new char[limit - start];
|
||||
* getChars(start, limit, text, 0);
|
||||
* replace(dest, dest, text, 0, limit - start);</pre>
|
||||
*
|
||||
* @param start the beginning index, inclusive; <code>0 <= start <=
|
||||
* limit</code>.
|
||||
* @param limit the ending index, exclusive; <code>start <= limit <=
|
||||
* length()</code>.
|
||||
* @param dest the destination index. The characters from
|
||||
* <code>start..limit-1</code> will be copied to <code>dest</code>.
|
||||
* Implementations of this method may assume that <code>dest <= start ||
|
||||
* dest >= limit</code>.
|
||||
*/
|
||||
void copy(int start, int limit, int dest);
|
||||
}
|
||||
|
||||
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/text/Attic/ReplaceableString.java,v $
|
||||
* $Date: 2000/03/10 04:07:22 $
|
||||
* $Revision: 1.2 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.3 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -24,7 +24,7 @@ package com.ibm.text;
|
||||
*
|
||||
* @see Replaceable
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: ReplaceableString.java,v $ $Revision: 1.2 $ $Date: 2000/03/10 04:07:22 $
|
||||
* @version $RCSfile: ReplaceableString.java,v $ $Revision: 1.3 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
public class ReplaceableString implements Replaceable {
|
||||
private StringBuffer buf;
|
||||
@ -67,16 +67,6 @@ public class ReplaceableString implements Replaceable {
|
||||
return buf.toString();
|
||||
}
|
||||
|
||||
/**
|
||||
* Return the internal storage of this object. <em>Note! Any
|
||||
* changes made to the returned object affect this object's
|
||||
* contents, and vice versa.</em>
|
||||
* @return internal buffer used by this object
|
||||
*/
|
||||
public StringBuffer getStringBuffer() {
|
||||
return buf;
|
||||
}
|
||||
|
||||
/**
|
||||
* Return the number of characters contained in this object.
|
||||
* <code>Replaceable</code> API.
|
||||
@ -168,4 +158,24 @@ public class ReplaceableString implements Replaceable {
|
||||
buf.append(tail);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* Copy a substring of this object, retaining attribute (out-of-band)
|
||||
* information. This method is used to duplicate or reorder substrings.
|
||||
* The destination index must not overlap the source range.
|
||||
*
|
||||
* @param start the beginning index, inclusive; <code>0 <= start <=
|
||||
* limit</code>.
|
||||
* @param limit the ending index, exclusive; <code>start <= limit <=
|
||||
* length()</code>.
|
||||
* @param dest the destination index. The characters from
|
||||
* <code>start..limit-1</code> will be copied to <code>dest</code>.
|
||||
* Implementations of this method may assume that <code>dest <= start ||
|
||||
* dest >= limit</code>.
|
||||
*/
|
||||
public void copy(int start, int limit, int dest) {
|
||||
char[] text = new char[limit - start];
|
||||
getChars(start, limit, text, 0);
|
||||
replace(dest, dest, text, 0, limit - start);
|
||||
}
|
||||
}
|
||||
|
||||
@ -5,8 +5,8 @@
|
||||
*******************************************************************************
|
||||
*
|
||||
* $Source: /xsrl/Nsvn/icu/icu4j/src/com/ibm/text/Attic/TransliterationRule.java,v $
|
||||
* $Date: 2000/04/25 01:42:58 $
|
||||
* $Revision: 1.19 $
|
||||
* $Date: 2000/04/25 17:17:37 $
|
||||
* $Revision: 1.20 $
|
||||
*
|
||||
*****************************************************************************************
|
||||
*/
|
||||
@ -44,7 +44,7 @@ import com.ibm.util.Utility;
|
||||
* <p>Copyright © IBM Corporation 1999. All rights reserved.
|
||||
*
|
||||
* @author Alan Liu
|
||||
* @version $RCSfile: TransliterationRule.java,v $ $Revision: 1.19 $ $Date: 2000/04/25 01:42:58 $
|
||||
* @version $RCSfile: TransliterationRule.java,v $ $Revision: 1.20 $ $Date: 2000/04/25 17:17:37 $
|
||||
*/
|
||||
class TransliterationRule {
|
||||
/**
|
||||
@ -259,28 +259,52 @@ class TransliterationRule {
|
||||
*/
|
||||
public int replace(Replaceable text, int offset,
|
||||
RuleBasedTransliterator.Data data) {
|
||||
String out;
|
||||
if (segments == null) {
|
||||
out = output;
|
||||
text.replace(offset, offset + keyLength, output);
|
||||
return output.length() - keyLength;
|
||||
} else {
|
||||
/* When there are segments to be copied, use the Replaceable.copy()
|
||||
* API in order to retain out-of-band data. Copy everything to the
|
||||
* point after the key, then delete the key. That is, copy things
|
||||
* into offset + keyLength, then replace offset .. offset +
|
||||
* keyLength with the empty string.
|
||||
*
|
||||
* Minimize the number of calls to Replaceable.replace() and
|
||||
* Replaceable.copy().
|
||||
*/
|
||||
int textStart = offset - anteContextLength;
|
||||
int dest = offset + keyLength; // copy new text to here
|
||||
StringBuffer buf = new StringBuffer();
|
||||
for (int i=0; i<output.length(); ++i) {
|
||||
char c = output.charAt(i);
|
||||
int b = data.lookupSegmentReference(c);
|
||||
if (b < 0) {
|
||||
// Accumulate straight (non-segment) text.
|
||||
buf.append(c);
|
||||
} else {
|
||||
for (int j=textStart + segments[2*b];
|
||||
j<textStart + segments[2*b+1]; ++j) {
|
||||
buf.append(text.charAt(j));
|
||||
// Insert any accumulated straight text.
|
||||
if (buf.length() > 0) {
|
||||
text.replace(dest, dest, buf.toString());
|
||||
dest += buf.length();
|
||||
buf.setLength(0);
|
||||
}
|
||||
// Copy segment with out-of-band data
|
||||
b *= 2;
|
||||
text.copy(textStart + segments[b],
|
||||
textStart + segments[b+1], dest);
|
||||
dest += segments[b+1] - segments[b];
|
||||
}
|
||||
|
||||
}
|
||||
out = buf.toString();
|
||||
// Insert any accumulated straight text.
|
||||
if (buf.length() > 0) {
|
||||
text.replace(dest, dest, buf.toString());
|
||||
dest += buf.length();
|
||||
}
|
||||
// Delete the key
|
||||
text.replace(offset, offset + keyLength, "");
|
||||
return dest - (offset + keyLength) - keyLength;
|
||||
}
|
||||
text.replace(offset, offset + keyLength, out);
|
||||
return out.length() - keyLength;
|
||||
}
|
||||
|
||||
/**
|
||||
@ -493,6 +517,9 @@ class TransliterationRule {
|
||||
|
||||
/**
|
||||
* $Log: TransliterationRule.java,v $
|
||||
* Revision 1.20 2000/04/25 17:17:37 alan
|
||||
* Add Replaceable.copy to retain out-of-band info during reordering.
|
||||
*
|
||||
* Revision 1.19 2000/04/25 01:42:58 alan
|
||||
* Allow arbitrary length variable values. Clean up Data API. Update javadocs.
|
||||
*
|
||||
|
||||
Loading…
Reference in New Issue
Block a user