From: Nick Burch <nick@apache.org>
Date: Tue, 15 Jul 2008 23:40:09 +0000 (+0000)
Subject: Merged revisions 638786-638802,638805-638811,638813-638814,638816-639230,639233-63924... 
X-Git-Tag: REL_3_5_BETA2~55
X-Git-Url: https://source.dussan.org/?a=commitdiff_plain;h=75b23f2893e7c5d4ccab12236fbf8f1968281bbf;p=poi.git

Merged revisions 638786-638802,638805-638811,638813-638814,638816-639230,639233-639241,639243-639253,639255-639486,639488-639601,639603-639835,639837-639917,639919-640056,640058-640710,640712-641156,641158-641184,641186-641795,641797-641798,641800-641933,641935-641963,641965-641966,641968-641995,641997-642230,642232-642562,642564-642565,642568-642570,642572-642573,642576-642736,642739-642877,642879,642881-642890,642892-642903,642905-642945,642947-643624,643626-643653,643655-643669,643671,643673-643830,643832-643833,643835-644342,644344-644472,644474-644508,644510-645347,645349-645351,645353-645559,645561-645565,645568-645951,645953-646193,646195-646311,646313-646404,646406-646665,646667-646853,646855-646869,646871-647151,647153-647185,647187-647277,647279-647566,647568-647573,647575,647578-647711,647714-647737,647739-647823,647825-648155,648157-648202,648204-648273,648275,648277-648302,648304-648333,648335-648588,648590-648622,648625-648673,648675-649141,649144,649146-649556,649558-649795,649799,649801-649910,649912-649913,649915-650128,650131-650132,650134-650137,650140-650914,650916-651991,651993-652284,652286-652287,652289,652291,652293-652297,652299-652328,652330-652425,652427-652445,652447-652560,652562-652933,652935,652937-652993,652995-653116,653118-653124,653126-653483,653487-653519,653522-653550,653552-653607,653609-653667,653669-653674,653676-653814,653817-653830,653832-653891,653893-653944,653946-654055,654057-654355,654357-654365,654367-654648,654651-655215,655217-655277,655279-655281,655283-655911,655913-656212,656214,656216-656251,656253-656698,656700-656756,656758-656892,656894-657135,657137-657165,657168-657179,657181-657354,657356-657357,657359-657701,657703-657874,657876-658032,658034-658284,658286,658288-658301,658303-658307,658309-658321,658323-658335,658337-658348,658351,658353-658832,658834-658983,658985,658987-659066,659068-659402,659404-659428,659430-659451,659453-659454,659456-659461,659463-659477,659479-659524,659526-659571,659574,659576-660255,660257-660262,660264-660279,660281-660343,660345-660473,660475-660827,660829-660833,660835-660888,660890-663321,663323-663435,663437-663764,663766-663854,663856-664219,664221-664489,664494-664514,664516-668013,668015-668142,668144-668152,668154,668156-668256,668258,668260-669139,669141-669455,669457-669657,669659-669808,669810-670189,670191-671321,671323-672229,672231-672549,672551-672552,672554-672561,672563-672566,672568,672571-673049,673051-673852,673854-673862,673864-673986,673988-673996,673998-674347,674349-674890,674892-674910,674912-674936,674938-674952,674954-675078,675080-675085,675087-675217,675219-675660,675662-675670,675672-675716,675718-675726,675728-675733,675735-675775,675777-675782,675784,675786-675791,675794-675852,675854-676200,676202,676204,676206-676220,676222-676309,676311-676456,676458-676994,676996-677064 via svnmerge from
https://svn.apache.org:443/repos/asf/poi/trunk

........
  r677028 | nick | 2008-07-15 21:19:06 +0100 (Tue, 15 Jul 2008) | 1 line

  Method to check if two fonts have the same contents
........
  r677029 | nick | 2008-07-15 21:24:53 +0100 (Tue, 15 Jul 2008) | 1 line

  Support for removing low level font records
........
  r677041 | nick | 2008-07-15 22:15:16 +0100 (Tue, 15 Jul 2008) | 1 line

  Start on HSSFOptimiser, which removes un-needed cell styles and fonts, fixing up references as it does so
........
  r677057 | nick | 2008-07-15 22:38:38 +0100 (Tue, 15 Jul 2008) | 1 line

  Cell Style optimisations too
........


git-svn-id: https://svn.apache.org/repos/asf/poi/branches/ooxml@677103 13f79535-47bb-0310-9956-ffa450edef68
---

diff --git a/src/documentation/content/xdocs/changes.xml b/src/documentation/content/xdocs/changes.xml
index 1fee98e86e..99106b3a0a 100644
--- a/src/documentation/content/xdocs/changes.xml
+++ b/src/documentation/content/xdocs/changes.xml
@@ -50,6 +50,7 @@
            <action dev="POI-DEVELOPERS" type="add">Created a common interface for handling Excel files, irrespective of if they are .xls or .xlsx</action>
         </release>
         <release version="3.1.1-alpha1" date="2008-??-??">
+           <action dev="POI-DEVELOPERS" type="add">New helper, HSSFOptimiser, which handles removing duplicated font and style records, to avoid going over the limits in Excel</action>
            <action dev="POI-DEVELOPERS" type="fix">45322 - Fixed NPE in HSSFSheet.autoSizeColumn() when cell number format was not found</action>
            <action dev="POI-DEVELOPERS" type="add">45380 - Missing return keyword in ArrayPtg.toFormulaString()</action>
            <action dev="POI-DEVELOPERS" type="add">44958 - Record level support for Data Tables. (No formula parser support though)</action>
diff --git a/src/documentation/content/xdocs/status.xml b/src/documentation/content/xdocs/status.xml
index e6472d6e9a..96ff195f29 100644
--- a/src/documentation/content/xdocs/status.xml
+++ b/src/documentation/content/xdocs/status.xml
@@ -47,6 +47,7 @@
            <action dev="POI-DEVELOPERS" type="add">Created a common interface for handling Excel files, irrespective of if they are .xls or .xlsx</action>
         </release>
         <release version="3.1.1-alpha1" date="2008-??-??">
+           <action dev="POI-DEVELOPERS" type="add">New helper, HSSFOptimiser, which handles removing duplicated font and style records, to avoid going over the limits in Excel</action>
            <action dev="POI-DEVELOPERS" type="fix">45322 - Fixed NPE in HSSFSheet.autoSizeColumn() when cell number format was not found</action>
            <action dev="POI-DEVELOPERS" type="add">45380 - Missing return keyword in ArrayPtg.toFormulaString()</action>
            <action dev="POI-DEVELOPERS" type="add">44958 - Record level support for Data Tables. (No formula parser support though)</action>
diff --git a/src/java/org/apache/poi/hssf/model/Workbook.java b/src/java/org/apache/poi/hssf/model/Workbook.java
index d56051445b..36510eb835 100644
--- a/src/java/org/apache/poi/hssf/model/Workbook.java
+++ b/src/java/org/apache/poi/hssf/model/Workbook.java
@@ -443,6 +443,17 @@ public class Workbook implements Model
         numfonts++;
         return rec;
     }
+    
+    /**
+     * Removes the given font record from the
+     *  file's list. This will make all 
+     *  subsequent font indicies drop by one,
+     *  so you'll need to update those yourself!
+     */
+    public void removeFontRecord(FontRecord rec) {
+    	records.remove(rec); // this updates FontPos for us
+        numfonts--;
+    }
 
     /**
      * gets the number of font records
@@ -702,6 +713,18 @@ public class Workbook implements Model
 
         return retval;
     }
+    
+    /**
+     * Removes the given ExtendedFormatRecord record from the
+     *  file's list. This will make all 
+     *  subsequent font indicies drop by one,
+     *  so you'll need to update those yourself!
+     */
+    public void removeExFormatRecord(ExtendedFormatRecord rec) {
+    	records.remove(rec); // this updates XfPos for us
+        numxfs--;
+    }
+
 
     /**
      * creates a new Cell-type Extneded Format Record and adds it to the end of
diff --git a/src/java/org/apache/poi/hssf/record/FontRecord.java b/src/java/org/apache/poi/hssf/record/FontRecord.java
index d6a5ce859f..11ba3aaa89 100644
--- a/src/java/org/apache/poi/hssf/record/FontRecord.java
+++ b/src/java/org/apache/poi/hssf/record/FontRecord.java
@@ -579,6 +579,31 @@ public class FontRecord
 		result = prime * result + field_10_font_name_len;
 		return result;
 	}
+	
+	/**
+	 * Does this FontRecord have all the same font
+	 *  properties as the supplied FontRecord?
+	 * Note that {@link #equals(Object)} will check
+	 *  for exact objects, while this will check
+	 *  for exact contents, because normally the
+	 *  font record's position makes a big
+	 *  difference too.  
+	 */
+	public boolean sameProperties(FontRecord other) {
+		return 
+		field_1_font_height         == other.field_1_font_height &&
+		field_2_attributes          == other.field_2_attributes &&
+		field_3_color_palette_index == other.field_3_color_palette_index &&
+		field_4_bold_weight         == other.field_4_bold_weight &&
+		field_5_super_sub_script    == other.field_5_super_sub_script &&
+		field_6_underline           == other.field_6_underline &&
+		field_7_family              == other.field_7_family &&
+		field_8_charset             == other.field_8_charset &&
+		field_9_zero                == other.field_9_zero &&
+		field_10_font_name_len      == other.field_10_font_name_len &&
+		field_11_font_name.equals(other.field_11_font_name)
+		;
+	}
 
 	/**
 	 * Only returns two for the same exact object -
diff --git a/src/java/org/apache/poi/hssf/record/UnicodeString.java b/src/java/org/apache/poi/hssf/record/UnicodeString.java
index b53fcd485c..9919d52c3d 100644
--- a/src/java/org/apache/poi/hssf/record/UnicodeString.java
+++ b/src/java/org/apache/poi/hssf/record/UnicodeString.java
@@ -439,6 +439,23 @@ public class UnicodeString
       this.field_5_ext_rst = ext_rst;
     }
 
+
+    /**
+     * Swaps all use in the string of one font index 
+     *  for use of a different font index.
+     * Normally only called when fonts have been
+     *  removed / re-ordered
+     */
+    public void swapFontUse(short oldFontIndex, short newFontIndex) {
+    	Iterator i = field_4_format_runs.iterator();
+    	while(i.hasNext()) {
+    		FormatRun run = (FormatRun)i.next();
+    		if(run.fontIndex == oldFontIndex) {
+    			run.fontIndex = newFontIndex;
+    		}
+    	}
+    }
+    
     /**
      * unlike the real records we return the same as "getString()" rather than debug info
      * @see #getDebugInfo()
diff --git a/src/java/org/apache/poi/hssf/usermodel/HSSFCell.java b/src/java/org/apache/poi/hssf/usermodel/HSSFCell.java
index c81929afec..e4a7e37d70 100644
--- a/src/java/org/apache/poi/hssf/usermodel/HSSFCell.java
+++ b/src/java/org/apache/poi/hssf/usermodel/HSSFCell.java
@@ -938,14 +938,13 @@ public class HSSFCell implements Cell
      * object.
      * @see org.apache.poi.hssf.usermodel.HSSFWorkbook#getCellStyleAt(short)
      */
-
     public HSSFCellStyle getCellStyle()
     {
       short styleIndex=record.getXFIndex();
       ExtendedFormatRecord xf = book.getWorkbook().getExFormatAt(styleIndex);
       return new HSSFCellStyle(styleIndex, xf, book);
     }
-
+    
     /**
      * used for internationalization, currently -1 for unchanged, 0 for compressed unicode or 1 for 16-bit
      *
diff --git a/src/java/org/apache/poi/hssf/usermodel/HSSFOptimiser.java b/src/java/org/apache/poi/hssf/usermodel/HSSFOptimiser.java
new file mode 100644
index 0000000000..f2c7acc414
--- /dev/null
+++ b/src/java/org/apache/poi/hssf/usermodel/HSSFOptimiser.java
@@ -0,0 +1,261 @@
+/* ====================================================================
+   Copyright 2002-2004   Apache Software Foundation
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
+==================================================================== */
+package org.apache.poi.hssf.usermodel;
+
+import java.util.HashSet;
+import java.util.Iterator;
+
+import org.apache.poi.hssf.record.ExtendedFormatRecord;
+import org.apache.poi.hssf.record.FontRecord;
+import org.apache.poi.hssf.record.UnicodeString;
+
+/**
+ * Excel can get cranky if you give it files containing too
+ *  many (especially duplicate) objects, and this class can
+ *  help to avoid those.
+ * In general, it's much better to make sure you don't 
+ *  duplicate the objects in your code, as this is likely
+ *  to be much faster than creating lots and lots of
+ *  excel objects+records, only to optimise them down to
+ *  many fewer at a later stage.
+ * However, sometimes this is too hard / tricky to do, which
+ *  is where the use of this class comes in.
+ */
+public class HSSFOptimiser {
+	/**
+	 * Goes through the Workbook, optimising the fonts by
+	 *  removing duplicate ones.
+	 * For now, only works on fonts used in {@link HSSFCellStyle}
+	 *  and {@link HSSFRichTextString}. Any other font uses
+	 *  (eg charts, pictures) may well end up broken!
+	 * This can be a slow operation, especially if you have
+	 *  lots of cells, cell styles or rich text strings
+	 * @param workbook The workbook in which to optimise the fonts
+	 */
+	public static void optimiseFonts(HSSFWorkbook workbook) {
+		// Where each font has ended up, and if we need to
+		//  delete the record for it. Start off with no change
+		short[] newPos = 
+			new short[workbook.getWorkbook().getNumberOfFontRecords()+1];
+		boolean[] zapRecords = new boolean[newPos.length];
+		for(int i=0; i<newPos.length; i++) {
+			newPos[i] = (short)i;
+			zapRecords[i] = false;
+		}
+		
+		// Get each font record, so we can do deletes
+		//  without getting confused
+		FontRecord[] frecs = new FontRecord[newPos.length]; 
+		for(int i=0; i<newPos.length; i++) {
+			// There is no 4!
+			if(i == 4) continue;
+			
+			frecs[i] = workbook.getWorkbook().getFontRecordAt(i);
+		}
+		
+		// Loop over each font, seeing if it is the same
+		//  as an earlier one. If it is, point users of the
+		//  later duplicate copy to the earlier one, and 
+		//  mark the later one as needing deleting
+		// Note - don't change built in fonts (those before 5)
+		for(int i=5; i<newPos.length; i++) {
+			// Check this one for being a duplicate
+			//  of an earlier one
+			int earlierDuplicate = -1;
+			for(int j=0; j<i && earlierDuplicate == -1; j++) {
+				if(j == 4) continue;
+				
+				FontRecord frCheck = workbook.getWorkbook().getFontRecordAt(j);
+				if(frCheck.sameProperties(frecs[i])) {
+					earlierDuplicate = j;
+				}
+			}
+			
+			// If we got a duplicate, mark it as such
+			if(earlierDuplicate != -1) {
+				newPos[i] = (short)earlierDuplicate;
+				zapRecords[i] = true;
+			}
+		}
+		
+		// Update the new positions based on
+		//  deletes that have occurred between
+		//  the start and them
+		// Only need to worry about user fonts
+		for(int i=5; i<newPos.length; i++) {
+			// Find the number deleted to that
+			//  point, and adjust
+			short preDeletePos = newPos[i];
+			short newPosition = preDeletePos;
+			for(int j=0; j<preDeletePos; j++) {
+				if(zapRecords[j]) newPosition--;
+			}
+			
+			// Update the new position
+			newPos[i] = newPosition;
+		}
+		
+		// Zap the un-needed user font records
+		for(int i=5; i<newPos.length; i++) {
+			if(zapRecords[i]) {
+				workbook.getWorkbook().removeFontRecord(
+						frecs[i]
+				);
+			}
+		}
+		
+		// Tell HSSFWorkbook that it needs to
+		//  re-start its HSSFFontCache
+		workbook.resetFontCache();
+		
+		// Update the cell styles to point at the 
+		//  new locations of the fonts
+		for(int i=0; i<workbook.getWorkbook().getNumExFormats(); i++) {
+			ExtendedFormatRecord xfr = workbook.getWorkbook().getExFormatAt(i);
+			xfr.setFontIndex(
+					newPos[ xfr.getFontIndex() ]
+			);
+		}
+		
+		// Update the rich text strings to point at
+		//  the new locations of the fonts
+		// Remember that one underlying unicode string
+		//  may be shared by multiple RichTextStrings!
+		HashSet doneUnicodeStrings = new HashSet();
+		for(int sheetNum=0; sheetNum<workbook.getNumberOfSheets(); sheetNum++) {
+			HSSFSheet s = workbook.getSheetAt(sheetNum);
+			Iterator rIt = s.rowIterator();
+			while(rIt.hasNext()) {
+				HSSFRow row = (HSSFRow)rIt.next();
+				Iterator cIt = row.cellIterator();
+				while(cIt.hasNext()) {
+					HSSFCell cell = (HSSFCell)cIt.next();
+					if(cell.getCellType() == HSSFCell.CELL_TYPE_STRING) {
+						HSSFRichTextString rtr = cell.getRichStringCellValue();
+						UnicodeString u = rtr.getRawUnicodeString();
+						
+						// Have we done this string already?
+						if(! doneUnicodeStrings.contains(u)) {
+							// Update for each new position
+							for(short i=5; i<newPos.length; i++) {
+								if(i != newPos[i]) {
+									u.swapFontUse(i, newPos[i]);
+								}
+							}
+							
+							// Mark as done
+							doneUnicodeStrings.add(u);
+						}
+					}
+				}
+			}
+		}
+	}
+	
+	/**
+	 * Goes through the Wokrbook, optimising the cell styles
+	 *  by removing duplicate ones.
+	 * For best results, optimise the fonts via a call to
+	 *  {@link #optimiseFonts(HSSFWorkbook)} first.
+	 * @param workbook The workbook in which to optimise the cell styles
+	 */
+	public static void optimiseCellStyles(HSSFWorkbook workbook) {
+		// Where each style has ended up, and if we need to
+		//  delete the record for it. Start off with no change
+		short[] newPos = 
+			new short[workbook.getWorkbook().getNumExFormats()];
+		boolean[] zapRecords = new boolean[newPos.length];
+		for(int i=0; i<newPos.length; i++) {
+			newPos[i] = (short)i;
+			zapRecords[i] = false;
+		}
+		
+		// Get each style record, so we can do deletes
+		//  without getting confused
+		ExtendedFormatRecord[] xfrs = new ExtendedFormatRecord[newPos.length]; 
+		for(int i=0; i<newPos.length; i++) {
+			xfrs[i] = workbook.getWorkbook().getExFormatAt(i);
+		}
+		
+		// Loop over each style, seeing if it is the same
+		//  as an earlier one. If it is, point users of the
+		//  later duplicate copy to the earlier one, and 
+		//  mark the later one as needing deleting
+		// Only work on user added ones, which come after 20
+		for(int i=21; i<newPos.length; i++) {
+			// Check this one for being a duplicate
+			//  of an earlier one
+			int earlierDuplicate = -1;
+			for(int j=0; j<i && earlierDuplicate == -1; j++) {
+				ExtendedFormatRecord xfCheck = workbook.getWorkbook().getExFormatAt(j);
+				if(xfCheck.equals(xfrs[i])) {
+					earlierDuplicate = j;
+				}
+			}
+			
+			// If we got a duplicate, mark it as such
+			if(earlierDuplicate != -1) {
+				newPos[i] = (short)earlierDuplicate;
+				zapRecords[i] = true;
+			}
+		}
+		
+		// Update the new positions based on
+		//  deletes that have occurred between
+		//  the start and them
+		// Only work on user added ones, which come after 20
+		for(int i=21; i<newPos.length; i++) {
+			// Find the number deleted to that
+			//  point, and adjust
+			short preDeletePos = newPos[i];
+			short newPosition = preDeletePos;
+			for(int j=0; j<preDeletePos; j++) {
+				if(zapRecords[j]) newPosition--;
+			}
+			
+			// Update the new position
+			newPos[i] = newPosition;
+		}
+		
+		// Zap the un-needed user style records
+		for(int i=21; i<newPos.length; i++) {
+			if(zapRecords[i]) {
+				workbook.getWorkbook().removeExFormatRecord(
+						xfrs[i]
+				);
+			}
+		}
+		
+		// Finally, update the cells to point at
+		//  their new extended format records
+		for(int sheetNum=0; sheetNum<workbook.getNumberOfSheets(); sheetNum++) {
+			HSSFSheet s = workbook.getSheetAt(sheetNum);
+			Iterator rIt = s.rowIterator();
+			while(rIt.hasNext()) {
+				HSSFRow row = (HSSFRow)rIt.next();
+				Iterator cIt = row.cellIterator();
+				while(cIt.hasNext()) {
+					HSSFCell cell = (HSSFCell)cIt.next();
+					short oldXf = cell.getCellValueRecord().getXFIndex();
+					HSSFCellStyle newStyle = workbook.getCellStyleAt(
+							newPos[oldXf]
+					);
+					cell.setCellStyle(newStyle);
+				}
+			}
+		}
+	}
+}
diff --git a/src/java/org/apache/poi/hssf/usermodel/HSSFRichTextString.java b/src/java/org/apache/poi/hssf/usermodel/HSSFRichTextString.java
index 526925a024..ce5a40acbf 100644
--- a/src/java/org/apache/poi/hssf/usermodel/HSSFRichTextString.java
+++ b/src/java/org/apache/poi/hssf/usermodel/HSSFRichTextString.java
@@ -69,7 +69,7 @@ public class HSSFRichTextString
     
     /** Called whenever the unicode string is modified. When it is modified
      *  we need to create a new SST index, so that other LabelSSTRecords will not
-     *  be affected by changes tat we make to this string.
+     *  be affected by changes that we make to this string.
      */
     private UnicodeString cloneStringIfRequired() {
       if (book == null)
@@ -169,10 +169,25 @@ public class HSSFRichTextString
         return string.getString();
     }
 
-    /** Used internally by the HSSFCell to get the internal string value*/
+    /** 
+     * Used internally by the HSSFCell to get the internal 
+     * string value.
+     * Will ensure the string is not shared
+     */
     UnicodeString getUnicodeString() {
       return cloneStringIfRequired();
     }
+    
+    /**
+     * Returns the raw, probably shared Unicode String. 
+     * Used when tweaking the styles, eg updating font 
+     *  positions.
+     * Changes to this string may well effect
+     *  other RichTextStrings too! 
+     */
+    UnicodeString getRawUnicodeString() {
+    	return string;
+    }
 
     /** Used internally by the HSSFCell to set the internal string value*/
     void setUnicodeString(UnicodeString str) {
diff --git a/src/java/org/apache/poi/hssf/usermodel/HSSFWorkbook.java b/src/java/org/apache/poi/hssf/usermodel/HSSFWorkbook.java
index 09d66e1f60..d5b801c5c1 100644
--- a/src/java/org/apache/poi/hssf/usermodel/HSSFWorkbook.java
+++ b/src/java/org/apache/poi/hssf/usermodel/HSSFWorkbook.java
@@ -1094,6 +1094,16 @@ public class HSSFWorkbook extends POIDocument implements org.apache.poi.ss.userm
 
         return retval;
     }
+    
+    /**
+     * Reset the fonts cache, causing all new calls
+     *  to getFontAt() to create new objects.
+     * Should only be called after deleting fonts,
+     *  and that's not something you should normally do
+     */
+    protected void resetFontCache() {
+    	fonts = new Hashtable();
+    }
 
     /**
      * create a new Cell style and add it to the workbook's style table
diff --git a/src/testcases/org/apache/poi/hssf/model/TestWorkbook.java b/src/testcases/org/apache/poi/hssf/model/TestWorkbook.java
index 99663d2038..f89297390c 100644
--- a/src/testcases/org/apache/poi/hssf/model/TestWorkbook.java
+++ b/src/testcases/org/apache/poi/hssf/model/TestWorkbook.java
@@ -32,6 +32,7 @@ public final class TestWorkbook extends TestCase {
 		Workbook wb = (new HW()).getWorkbook();
 		
 		assertEquals(4, wb.getNumberOfFontRecords());
+		assertEquals(68, wb.getRecords().size());
 		
 		FontRecord f1 = wb.getFontRecordAt(0);
 		FontRecord f4 = wb.getFontRecordAt(3);
@@ -45,9 +46,41 @@ public final class TestWorkbook extends TestCase {
 		// There is no 4! new ones go in at 5
 		
 		FontRecord n = wb.createNewFont();
+		assertEquals(69, wb.getRecords().size());
 		assertEquals(5, wb.getNumberOfFontRecords());
 		assertEquals(5, wb.getFontIndex(n));
 		assertEquals(n, wb.getFontRecordAt(5));
+		
+		// And another
+		FontRecord n6 = wb.createNewFont();
+		assertEquals(70, wb.getRecords().size());
+		assertEquals(6, wb.getNumberOfFontRecords());
+		assertEquals(6, wb.getFontIndex(n6));
+		assertEquals(n6, wb.getFontRecordAt(6));
+		
+		
+		// Now remove the one formerly at 5
+		assertEquals(70, wb.getRecords().size());
+		wb.removeFontRecord(n);
+
+		// Check that 6 has gone to 5
+		assertEquals(69, wb.getRecords().size());
+		assertEquals(5, wb.getNumberOfFontRecords());
+		assertEquals(5, wb.getFontIndex(n6));
+		assertEquals(n6, wb.getFontRecordAt(5));
+		
+		// Check that the earlier ones are unchanged
+		assertEquals(0, wb.getFontIndex(f1));
+		assertEquals(3, wb.getFontIndex(f4));
+		assertEquals(f1, wb.getFontRecordAt(0));
+		assertEquals(f4, wb.getFontRecordAt(3));
+		
+		// Finally, add another one
+		FontRecord n7 = wb.createNewFont();
+		assertEquals(70, wb.getRecords().size());
+		assertEquals(6, wb.getNumberOfFontRecords());
+		assertEquals(6, wb.getFontIndex(n7));
+		assertEquals(n7, wb.getFontRecordAt(6));
 	}
 	
 	private class HW extends HSSFWorkbook {
diff --git a/src/testcases/org/apache/poi/hssf/record/TestFontRecord.java b/src/testcases/org/apache/poi/hssf/record/TestFontRecord.java
index 6485f586f9..d5051e9915 100644
--- a/src/testcases/org/apache/poi/hssf/record/TestFontRecord.java
+++ b/src/testcases/org/apache/poi/hssf/record/TestFontRecord.java
@@ -121,4 +121,21 @@ public class TestFontRecord
         for (int i = 0; i < data.length; i++)
             assertEquals("At offset " + i, data[i], recordBytes[i+4]);
     }
+    
+    public void testSameProperties() throws Exception {
+        FontRecord f1 = new FontRecord(new TestcaseRecordInputStream((short)0x31, (short)data.length, data));
+        FontRecord f2 = new FontRecord(new TestcaseRecordInputStream((short)0x31, (short)data.length, data));
+    	
+        assertTrue(f1.sameProperties(f2));
+        
+        f2.setFontName("Arial2");
+        assertFalse(f1.sameProperties(f2));
+        f2.setFontName("Arial");
+        assertTrue(f1.sameProperties(f2));
+        
+        f2.setFontHeight((short)11);
+        assertFalse(f1.sameProperties(f2));
+        f2.setFontHeight((short)0xc8);
+        assertTrue(f1.sameProperties(f2));
+    }
 }
diff --git a/src/testcases/org/apache/poi/hssf/usermodel/AllUserModelTests.java b/src/testcases/org/apache/poi/hssf/usermodel/AllUserModelTests.java
index 363e58c142..15a19c0912 100755
--- a/src/testcases/org/apache/poi/hssf/usermodel/AllUserModelTests.java
+++ b/src/testcases/org/apache/poi/hssf/usermodel/AllUserModelTests.java
@@ -47,6 +47,7 @@ public class AllUserModelTests {
 		result.addTestSuite(TestHSSFDateUtil.class);
 		result.addTestSuite(TestHSSFHeaderFooter.class);
 		result.addTestSuite(TestHSSFHyperlink.class);
+		result.addTestSuite(TestHSSFOptimiser.class);
 		result.addTestSuite(TestHSSFPalette.class);
 		result.addTestSuite(TestHSSFPatriarch.class);
 		result.addTestSuite(TestHSSFPicture.class);
diff --git a/src/testcases/org/apache/poi/hssf/usermodel/TestHSSFOptimiser.java b/src/testcases/org/apache/poi/hssf/usermodel/TestHSSFOptimiser.java
new file mode 100644
index 0000000000..b998fdd519
--- /dev/null
+++ b/src/testcases/org/apache/poi/hssf/usermodel/TestHSSFOptimiser.java
@@ -0,0 +1,240 @@
+/* ====================================================================
+   Copyright 2002-2004   Apache Software Foundation
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
+==================================================================== */
+package org.apache.poi.hssf.usermodel;
+
+import org.apache.poi.hssf.model.Workbook;
+
+import junit.framework.TestCase;
+
+public class TestHSSFOptimiser extends TestCase {
+	public void testDoesNoHarmIfNothingToDo() throws Exception {
+		HSSFWorkbook wb = new HSSFWorkbook();
+		
+		HSSFFont f = wb.createFont();
+		f.setFontName("Testing");
+		HSSFCellStyle s = wb.createCellStyle();
+		s.setFont(f);
+		
+		assertEquals(5, wb.getNumberOfFonts());
+		assertEquals(22, wb.getNumCellStyles());
+		
+		// Optimise fonts
+		HSSFOptimiser.optimiseFonts(wb);
+		
+		assertEquals(5, wb.getNumberOfFonts());
+		assertEquals(22, wb.getNumCellStyles());
+		
+		assertEquals(f, s.getFont(wb));
+		
+		// Optimise styles
+		HSSFOptimiser.optimiseCellStyles(wb);
+		
+		assertEquals(5, wb.getNumberOfFonts());
+		assertEquals(22, wb.getNumCellStyles());
+		
+		assertEquals(f, s.getFont(wb));
+	}
+	
+	public void testOptimiseFonts() throws Exception {
+		HSSFWorkbook wb = new HSSFWorkbook();
+		
+		// Add 6 fonts, some duplicates
+		HSSFFont f1 = wb.createFont();
+		f1.setFontHeight((short)11);
+		f1.setFontName("Testing");
+		
+		HSSFFont f2 = wb.createFont();
+		f2.setFontHeight((short)22);
+		f2.setFontName("Also Testing");
+		
+		HSSFFont f3 = wb.createFont();
+		f3.setFontHeight((short)33);
+		f3.setFontName("Unique");
+		
+		HSSFFont f4 = wb.createFont();
+		f4.setFontHeight((short)11);
+		f4.setFontName("Testing");
+		
+		HSSFFont f5 = wb.createFont();
+		f5.setFontHeight((short)22);
+		f5.setFontName("Also Testing");
+		
+		HSSFFont f6 = wb.createFont();
+		f6.setFontHeight((short)66);
+		f6.setFontName("Also Unique");
+		
+		
+		
+		// Use all three of the four in cell styles
+		assertEquals(21, wb.getNumCellStyles());
+		
+		HSSFCellStyle cs1 = wb.createCellStyle();
+		cs1.setFont(f1);
+		assertEquals(5, cs1.getFontIndex());
+		
+		HSSFCellStyle cs2 = wb.createCellStyle();
+		cs2.setFont(f4);
+		assertEquals(8, cs2.getFontIndex());
+		
+		HSSFCellStyle cs3 = wb.createCellStyle();
+		cs3.setFont(f5);
+		assertEquals(9, cs3.getFontIndex());
+		
+		HSSFCellStyle cs4 = wb.createCellStyle();
+		cs4.setFont(f6);
+		assertEquals(10, cs4.getFontIndex());
+		
+		assertEquals(25, wb.getNumCellStyles());
+		
+		
+		// And three in rich text
+		HSSFSheet s = wb.createSheet();
+		HSSFRow r = s.createRow(0);
+		
+		HSSFRichTextString rtr1 = new HSSFRichTextString("Test");
+		rtr1.applyFont(0, 2, f1);
+		rtr1.applyFont(3, 4, f2);
+		r.createCell((short)0).setCellValue(rtr1);
+		
+		HSSFRichTextString rtr2 = new HSSFRichTextString("AlsoTest");
+		rtr2.applyFont(0, 2, f3);
+		rtr2.applyFont(3, 5, f5);
+		rtr2.applyFont(6, 8, f6);
+		r.createCell((short)1).setCellValue(rtr2);
+		
+		
+		// Check what we have now
+		assertEquals(10, wb.getNumberOfFonts());
+		assertEquals(25, wb.getNumCellStyles());
+		
+		// Optimise
+		HSSFOptimiser.optimiseFonts(wb);
+		
+		// Check font count
+		assertEquals(8, wb.getNumberOfFonts());
+		assertEquals(25, wb.getNumCellStyles());
+		
+		// Check font use in cell styles
+		assertEquals(5, cs1.getFontIndex());
+		assertEquals(5, cs2.getFontIndex()); // duplicate of 1
+		assertEquals(6, cs3.getFontIndex()); // duplicate of 2
+		assertEquals(8, cs4.getFontIndex()); // two have gone
+		
+		
+		// And in rich text
+		
+		// RTR 1 had f1 and f2, unchanged 
+		assertEquals(5, r.getCell(0).getRichStringCellValue().getFontAtIndex(0));
+		assertEquals(5, r.getCell(0).getRichStringCellValue().getFontAtIndex(1));
+		assertEquals(6, r.getCell(0).getRichStringCellValue().getFontAtIndex(3));
+		assertEquals(6, r.getCell(0).getRichStringCellValue().getFontAtIndex(4));
+		
+		// RTR 2 had f3 (unchanged), f5 (=f2) and f6 (moved down)
+		assertEquals(7, r.getCell(1).getRichStringCellValue().getFontAtIndex(0));
+		assertEquals(7, r.getCell(1).getRichStringCellValue().getFontAtIndex(1));
+		assertEquals(6, r.getCell(1).getRichStringCellValue().getFontAtIndex(3));
+		assertEquals(6, r.getCell(1).getRichStringCellValue().getFontAtIndex(4));
+		assertEquals(8, r.getCell(1).getRichStringCellValue().getFontAtIndex(6));
+		assertEquals(8, r.getCell(1).getRichStringCellValue().getFontAtIndex(7));
+	}
+	
+	public void testOptimiseStyles() throws Exception {
+		HSSFWorkbook wb = new HSSFWorkbook();
+
+		// Two fonts
+		assertEquals(4, wb.getNumberOfFonts());
+		
+		HSSFFont f1 = wb.createFont();
+		f1.setFontHeight((short)11);
+		f1.setFontName("Testing");
+		
+		HSSFFont f2 = wb.createFont();
+		f2.setFontHeight((short)22);
+		f2.setFontName("Also Testing");
+
+		assertEquals(6, wb.getNumberOfFonts());
+	
+		
+		// Several styles
+		assertEquals(21, wb.getNumCellStyles());
+		
+		HSSFCellStyle cs1 = wb.createCellStyle();
+		cs1.setFont(f1);
+		
+		HSSFCellStyle cs2 = wb.createCellStyle();
+		cs2.setFont(f2);
+		
+		HSSFCellStyle cs3 = wb.createCellStyle();
+		cs3.setFont(f1);
+		
+		HSSFCellStyle cs4 = wb.createCellStyle();
+		cs4.setFont(f1);
+		cs4.setAlignment((short)22);
+		
+		HSSFCellStyle cs5 = wb.createCellStyle();
+		cs5.setFont(f2);
+		cs5.setAlignment((short)111);
+		
+		HSSFCellStyle cs6 = wb.createCellStyle();
+		cs6.setFont(f2);
+		
+		assertEquals(27, wb.getNumCellStyles());
+		
+		
+		// Use them
+		HSSFSheet s = wb.createSheet();
+		HSSFRow r = s.createRow(0);
+		
+		r.createCell((short)0).setCellStyle(cs1);
+		r.createCell((short)1).setCellStyle(cs2);
+		r.createCell((short)2).setCellStyle(cs3);
+		r.createCell((short)3).setCellStyle(cs4);
+		r.createCell((short)4).setCellStyle(cs5);
+		r.createCell((short)5).setCellStyle(cs6);
+		r.createCell((short)6).setCellStyle(cs1);
+		r.createCell((short)7).setCellStyle(cs2);
+		
+		assertEquals(21, r.getCell(0).getCellValueRecord().getXFIndex());
+		assertEquals(26, r.getCell(5).getCellValueRecord().getXFIndex());
+		assertEquals(21, r.getCell(6).getCellValueRecord().getXFIndex());
+		
+		
+		// Optimise
+		HSSFOptimiser.optimiseCellStyles(wb);
+		
+		
+		// Check
+		assertEquals(6, wb.getNumberOfFonts());
+		assertEquals(25, wb.getNumCellStyles());
+		
+		// cs1 -> 21
+		assertEquals(21, r.getCell(0).getCellValueRecord().getXFIndex());
+		// cs2 -> 22
+		assertEquals(22, r.getCell(1).getCellValueRecord().getXFIndex());
+		// cs3 = cs1 -> 21
+		assertEquals(21, r.getCell(2).getCellValueRecord().getXFIndex());
+		// cs4 --> 24 -> 23
+		assertEquals(23, r.getCell(3).getCellValueRecord().getXFIndex());
+		// cs5 --> 25 -> 24
+		assertEquals(24, r.getCell(4).getCellValueRecord().getXFIndex());
+		// cs6 = cs2 -> 22
+		assertEquals(22, r.getCell(5).getCellValueRecord().getXFIndex());
+		// cs1 -> 21
+		assertEquals(21, r.getCell(6).getCellValueRecord().getXFIndex());
+		// cs2 -> 22
+		assertEquals(22, r.getCell(7).getCellValueRecord().getXFIndex());
+	}
+}