Optimize formula evaluation of row-references

We currently walk 1 million rows for every formula which
uses a column-reference like "$A"

Execution time of test-case TestVlookup.testFullColumnAreaRef61841
went from more than 16 seconds to around 2 seconds

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1894675 13f79535-47bb-0310-9956-ffa450edef68

diff --git a/poi/src/main/java/org/apache/poi/ss/formula/LazyAreaEval.java b/poi/src/main/java/org/apache/poi/ss/formula/LazyAreaEval.java
index bb33f870f8ddbb39585755e35ab0e5a158455b3d..583ffacccc22fe84127a2e45b55dd849acf10443 100644 (file)
--- a/poi/src/main/java/org/apache/poi/ss/formula/LazyAreaEval.java
+++ b/poi/src/main/java/org/apache/poi/ss/formula/LazyAreaEval.java
@@ -37,7 +37,7 @@ final class LazyAreaEval extends AreaEvalBase {
 
     public LazyAreaEval(int firstRowIndex, int firstColumnIndex, int lastRowIndex,
             int lastColumnIndex, SheetRangeEvaluator evaluator) {
-        super(evaluator, firstRowIndex, firstColumnIndex, lastRowIndex, lastColumnIndex);
+        super(evaluator, firstRowIndex, firstColumnIndex, evaluator.adjustRowNumber(lastRowIndex), lastColumnIndex);
         _evaluator = evaluator;
     }
 

diff --git a/poi/src/main/java/org/apache/poi/ss/formula/SheetRangeEvaluator.java b/poi/src/main/java/org/apache/poi/ss/formula/SheetRangeEvaluator.java
index 806dc08f3571311ba58516c910cabebe6a668afa..3afe4a4851506fc45240953f69a418cb1d24e5ad 100644 (file)
--- a/poi/src/main/java/org/apache/poi/ss/formula/SheetRangeEvaluator.java
+++ b/poi/src/main/java/org/apache/poi/ss/formula/SheetRangeEvaluator.java
@@ -25,7 +25,7 @@ import org.apache.poi.ss.formula.eval.ValueEval;
 final class SheetRangeEvaluator implements SheetRange {
     private final int _firstSheetIndex;
     private final int _lastSheetIndex;
-    private SheetRefEvaluator[] _sheetEvaluators;
+    private final SheetRefEvaluator[] _sheetEvaluators;
 
     public SheetRangeEvaluator(int firstSheetIndex, int lastSheetIndex, SheetRefEvaluator[] sheetEvaluators) {
         if (firstSheetIndex < 0) {
@@ -41,7 +41,7 @@ final class SheetRangeEvaluator implements SheetRange {
     public SheetRangeEvaluator(int onlySheetIndex, SheetRefEvaluator sheetEvaluator) {
         this(onlySheetIndex, onlySheetIndex, new SheetRefEvaluator[] {sheetEvaluator});
     }
-    
+
     public SheetRefEvaluator getSheetEvaluator(int sheetIndex) {
         if (sheetIndex < _firstSheetIndex || sheetIndex > _lastSheetIndex) {
             throw new IllegalArgumentException("Invalid SheetIndex: " + sheetIndex +
@@ -49,7 +49,7 @@ final class SheetRangeEvaluator implements SheetRange {
         }
         return _sheetEvaluators[sheetIndex-_firstSheetIndex];
     }
-    
+
     public int getFirstSheetIndex() {
         return _firstSheetIndex;
     }
@@ -73,4 +73,26 @@ final class SheetRangeEvaluator implements SheetRange {
     public ValueEval getEvalForCell(int sheetIndex, int rowIndex, int columnIndex) {
         return getSheetEvaluator(sheetIndex).getEvalForCell(rowIndex, columnIndex);
     }
+
+       /**
+        * This method returns a lower row-number if it would lie outside the row-boundaries of
+        * any sheet.
+        *
+        * This is used to optimize cases where very high number of rows would be checked otherwise
+        * without any benefit as no such row exists anyway.
+        *
+        * @param rowIndex The 0-based row-index to check
+        * @return If the given index lies withing the max row number across all sheets, it is returned.
+        *              Otherwise, the highest used row number across all sheets is returned.
+        */
+       public int adjustRowNumber(int rowIndex) {
+               int maxRowNum = rowIndex;
+
+               for (int i = _firstSheetIndex; i < _lastSheetIndex; i++) {
+                       maxRowNum = Math.max(maxRowNum, _sheetEvaluators[i].getLastRowNum());
+               }
+
+               // do not try to evaluate further than there are rows in any sheet
+               return Math.min(rowIndex, maxRowNum);
+       }
 }

diff --git a/poi/src/main/java/org/apache/poi/ss/formula/SheetRefEvaluator.java b/poi/src/main/java/org/apache/poi/ss/formula/SheetRefEvaluator.java
index eb411bd6642e17452af446a51f47223890fb7004..e765af3bf85135e28a574b2244d4818650df292d 100644 (file)
--- a/poi/src/main/java/org/apache/poi/ss/formula/SheetRefEvaluator.java
+++ b/poi/src/main/java/org/apache/poi/ss/formula/SheetRefEvaluator.java
@@ -56,8 +56,8 @@ final class SheetRefEvaluator {
     }
 
     /**
-     * @param rowIndex 
-     * @param columnIndex 
+     * @param rowIndex The 0-based row-index to check
+     * @param columnIndex The 0-based column-index to check
      * @return  whether cell at rowIndex and columnIndex is a subtotal
      * @see org.apache.poi.ss.formula.functions.Subtotal
      */
@@ -83,10 +83,17 @@ final class SheetRefEvaluator {
      * Used by functions that calculate differently depending on row visibility, like some
      * variations of SUBTOTAL()
      * @see org.apache.poi.ss.formula.functions.Subtotal
-     * @param rowIndex
+     * @param rowIndex The 0-based row-index to check
      * @return true if the row is hidden
      */
     public boolean isRowHidden(int rowIndex) {
         return getSheet().isRowHidden(rowIndex);
     }
+
+       /**
+        * @return The last used row in this sheet
+        */
+       public int getLastRowNum() {
+               return getSheet().getLastRowNum();
+       }
 }