Use CharUtilities instead of CharClass

git-svn-id: https://svn.apache.org/repos/asf/xmlgraphics/fop/trunk@194673 13f79535-47bb-0310-9956-ffa450edef68

4 files changed, 55 insertions, 25 deletions

diff --git a/src/org/apache/fop/fo/FONode.java b/src/org/apache/fop/fo/FONode.java
index 9f2d5aae5..55fde4fee 100644
--- a/src/org/apache/fop/fo/FONode.java
+++ b/src/org/apache/fop/fo/FONode.java
@@ -11,6 +11,7 @@ package org.apache.fop.fo;
 import org.apache.fop.apps.FOPException;
 import org.apache.fop.apps.StreamRenderer;
 import org.apache.fop.layout.Area;
+import org.apache.fop.util.CharUtilities;
 
 import org.apache.log.Logger;
 
@@ -121,7 +122,7 @@ abstract public class FONode {
     }
 
     public CharIterator charIterator() {
-	return new OneCharIterator(CharClass.CODE_EOT);
+	return new OneCharIterator(CharUtilities.CODE_EOT);
     }
 
 }
diff --git a/src/org/apache/fop/fo/InlineCharIterator.java b/src/org/apache/fop/fo/InlineCharIterator.java
index f8d048814..fe2a8eeda 100644
--- a/src/org/apache/fop/fo/InlineCharIterator.java
+++ b/src/org/apache/fop/fo/InlineCharIterator.java
@@ -1,6 +1,7 @@
 package org.apache.fop.fo;
 
 import org.apache.fop.layout.BorderAndPadding;
+import org.apache.fop.util.CharUtilities;
 import java.util.Iterator;
 import java.util.ListIterator;
 import java.util.NoSuchElementException;
@@ -36,7 +37,7 @@ public class InlineCharIterator extends RecursiveCharIterator {
     public char nextChar() throws NoSuchElementException {
 	if (bStartBoundary) {
 	    bStartBoundary=false;
-	    return CharClass.CODE_EOT;
+	    return CharUtilities.CODE_EOT;
 	}
 	try {
 	    return super.nextChar();
@@ -46,7 +47,7 @@ public class InlineCharIterator extends RecursiveCharIterator {
 	    // Check end boundary char
 	    if (bEndBoundary) {
 		bEndBoundary=false;
-		return CharClass.CODE_EOT;
+		return CharUtilities.CODE_EOT;
 	    }
 	    else throw e;
 	}
diff --git a/src/org/apache/fop/fo/flow/Block.java b/src/org/apache/fop/fo/flow/Block.java
index ba1b22ec4..3dc075b97 100644
--- a/src/org/apache/fop/fo/flow/Block.java
+++ b/src/org/apache/fop/fo/flow/Block.java
@@ -15,6 +15,7 @@ import org.apache.fop.datatypes.*;
 import org.apache.fop.apps.FOPException;
 import org.apache.fop.layoutmgr.LayoutManager;
 import org.apache.fop.layoutmgr.BlockLayoutManager;
+import org.apache.fop.util.CharUtilities;
 
 import org.xml.sax.Attributes;
 
@@ -406,8 +407,8 @@ public class Block extends FObjMixed {
 	    LFchecker lfCheck = new LFchecker(charIter);
 
 	    while (charIter.hasNext()) {
-		switch (CharClass.classOf(charIter.nextChar())) {
-		case CharClass.XMLWHITESPACE:
+		switch (CharUtilities.classOf(charIter.nextChar())) {
+		case CharUtilities.XMLWHITESPACE:
 		    /* Some kind of whitespace character, except linefeed. */
 		    boolean bIgnore=false;
 		    
@@ -440,7 +441,7 @@ public class Block extends FObjMixed {
 		    }
 		    break;
 
-		case CharClass.LINEFEED:
+		case CharUtilities.LINEFEED:
 		    /* A linefeed */
 		    lfCheck.reset();
 		    bPrevWasLF=true; // for following whitespace
@@ -468,14 +469,14 @@ public class Block extends FObjMixed {
 		    }
 		    break;
 
-		case CharClass.EOT:
+		case CharUtilities.EOT:
 		    //   A "boundary" objects such as non-character inline
 		    // or nested block object was encountered.
 		    // If any whitespace run in progress, finish it.
 		    // FALL THROUGH
 
-		case CharClass.UCWHITESPACE: // Non XML-whitespace
-		case CharClass.NONWHITESPACE:
+		case CharUtilities.UCWHITESPACE: // Non XML-whitespace
+		case CharUtilities.NONWHITESPACE:
 		    /* Any other character */
 		    bInWS = bPrevWasLF=false;
 		    lfCheck.reset();
@@ -503,7 +504,8 @@ public class Block extends FObjMixed {
 			bNextIsLF=true;
 			break;
 		    }
-		    else if (CharClass.classOf(c)!=CharClass.XMLWHITESPACE) {
+		    else if (CharUtilities.classOf(c) !=
+			     CharUtilities.XMLWHITESPACE) {
 			break;
 		    }
 		}
diff --git a/src/org/apache/fop/util/CharUtilities.java b/src/org/apache/fop/util/CharUtilities.java
index f8bcc7edc..14187a5ea 100644
--- a/src/org/apache/fop/util/CharUtilities.java
+++ b/src/org/apache/fop/util/CharUtilities.java
@@ -10,9 +10,36 @@ package org.apache.fop.util;
 import org.apache.fop.layout.FontState;
 
 /**
+ * This class provides utilities to distinguish various kinds of Unicode
+ * whitespace and to get character widths in a given FontState.
  */
 public class CharUtilities {
 
+    /** Character code used to signal a character boundary in
+     * inline content, such as an inline with borders and padding
+     * or a nested block object.
+     */
+    public static final char CODE_EOT=0;
+
+    public static final int UCWHITESPACE=0; // unicode white space
+    public static final int LINEFEED=1;
+    public static final int EOT=2; // Boundary beteween text runs
+    public static final int NONWHITESPACE=3;
+    public static final int XMLWHITESPACE=4;
+
+
+    /**
+     * Return the appropriate CharClass constant for the type
+     * of the passed character.
+     */
+    public static int classOf(char c) {
+	if (c == CODE_EOT) return EOT;
+	if (c == '\n') return LINEFEED;
+	if ( c==' '|| c == '\r' || c=='\t' ) return XMLWHITESPACE;
+	if (isAnySpace(c)) return UCWHITESPACE;
+	return NONWHITESPACE;
+    }
+
     /**
      * Helper method for getting the width of a unicode char
      * from the current fontstate.
@@ -76,21 +103,20 @@ public class CharUtilities {
      * it's not non-breaking
      */
     public static boolean isSpace(char c) {
-        if (c == ' ' || c == '\u2000' ||    // en quad
-        c == '\u2001' ||                    // em quad
-        c == '\u2002' ||                    // en space
-        c == '\u2003' ||                    // em space
-        c == '\u2004' ||                    // three-per-em space
-        c == '\u2005' ||                    // four--per-em space
-        c == '\u2006' ||                    // six-per-em space
-        c == '\u2007' ||                    // figure space
-        c == '\u2008' ||                    // punctuation space
-        c == '\u2009' ||                    // thin space
-        c == '\u200A' ||                    // hair space
-        c == '\u200B')                      // zero width space
-            return true;
-        else
-            return false;
+        return (c == ' ' ||
+		(c >= '\u2000' && c <= '\u200B'));
+//         c == '\u2000'                   // en quad
+//         c == '\u2001'                   // em quad
+//         c == '\u2002'                   // en space
+//         c == '\u2003'                   // em space
+//         c == '\u2004'                   // three-per-em space
+//         c == '\u2005'                   // four--per-em space
+//         c == '\u2006'                   // six-per-em space
+//         c == '\u2007'                   // figure space
+//         c == '\u2008'                   // punctuation space
+//         c == '\u2009'                   // thin space
+//         c == '\u200A'                   // hair space
+//         c == '\u200B'                   // zero width space
     }
 
     /**