Merge pull request #145 from box/speed_up

Various speed improvements

src/Spout/Reader/XLSX/Helper/CellHelper.php

@@ -12,6 +12,14 @@ use Box\Spout\Common\Exception\InvalidArgumentException;
  */
 class CellHelper
 {
+    // Using ord() is super slow... Using a pre-computed hash table instead.
+    private static $columnLetterToIndexMapping = [
+        'A' => 0, 'B' => 1, 'C' => 2, 'D' => 3, 'E' => 4, 'F' => 5, 'G' => 6,
+        'H' => 7, 'I' => 8, 'J' => 9, 'K' => 10, 'L' => 11, 'M' => 12, 'N' => 13,
+        'O' => 14, 'P' => 15, 'Q' => 16, 'R' => 17, 'S' => 18, 'T' => 19, 'U' => 20,
+        'V' => 21, 'W' => 22, 'X' => 23, 'Y' => 24, 'Z' => 25,
+    ];
+
     /**
      * Fills the missing indexes of an array with a given value.
      * For instance, $dataArray = []; $a[1] = 1; $a[3] = 3;
@@ -50,34 +58,31 @@ class CellHelper
         }
 
         $columnIndex = 0;
-        $capitalAAsciiValue = ord('A');
-        $capitalZAsciiValue = ord('Z');
-        $step = $capitalZAsciiValue - $capitalAAsciiValue + 1;
 
         // Remove row information
-        $column = preg_replace('/\d/', '', $cellIndex);
+        $columnLetters = preg_replace('/\d/', '', $cellIndex);
-        $columnLength = strlen($column);
 
-        /*
+        // strlen() is super slow too... Using isset() is way faster and not too unreadable,
-         * This is how the following loop will process the data:
+        // since we checked before that there are between 1 and 3 letters.
-         * A   => 0
+        $columnLength = isset($columnLetters[1]) ? (isset($columnLetters[2]) ? 3 : 2) : 1;
-         * Z   => 25
-         * AA  => 26   : (26^(2-1) * (0+1)) + 0
-         * AB  => 27   : (26^(2-1) * (0+1)) + 1
-         * BC  => 54   : (26^(2-1) * (1+1)) + 2
-         * BCZ => 1455 : (26^(3-1) * (1+1)) + (26^(2-1) * (2+1)) + 25
-         */
-        foreach (str_split($column) as $single_cell_index)
-        {
-            $currentColumnIndex = ord($single_cell_index) - $capitalAAsciiValue;
 
-            if ($columnLength === 1) {
+        // Looping over the different letters of the column is slower than this method.
-                $columnIndex += $currentColumnIndex;
+        // Also, not using the pow() function because it's slooooow...
-            } else {
+        switch ($columnLength) {
-                $columnIndex += pow($step, ($columnLength - 1)) * ($currentColumnIndex + 1);
+            case 1:
-            }
+                $columnIndex = (self::$columnLetterToIndexMapping[$columnLetters]);
-
+                break;
-            $columnLength--;
+            case 2:
+                $firstLetterIndex = (self::$columnLetterToIndexMapping[$columnLetters[0]] + 1) * 26;
+                $secondLetterIndex = self::$columnLetterToIndexMapping[$columnLetters[1]];
+                $columnIndex = $firstLetterIndex + $secondLetterIndex;
+                break;
+            case 3:
+                $firstLetterIndex = (self::$columnLetterToIndexMapping[$columnLetters[0]] + 1) * 676;
+                $secondLetterIndex = (self::$columnLetterToIndexMapping[$columnLetters[1]] + 1) * 26;
+                $thirdLetterIndex = self::$columnLetterToIndexMapping[$columnLetters[2]];
+                $columnIndex = $firstLetterIndex + $secondLetterIndex + $thirdLetterIndex;
+                break;
         }
 
         return $columnIndex;
@@ -86,12 +91,13 @@ class CellHelper
     /**
      * Returns whether a cell index is valid, in an Excel world.
      * To be valid, the cell index should start with capital letters and be followed by numbers.
+     * There can only be 3 letters, as there can only be 16,384 rows, which is equivalent to 'XFE'.
      *
      * @param string $cellIndex The Excel cell index ('A1', 'BC13', ...)
      * @return bool
      */
     protected static function isValidCellIndex($cellIndex)
     {
-        return (preg_match('/^[A-Z]+\d+$/', $cellIndex) === 1);
+        return (preg_match('/^[A-Z]{1,3}\d+$/', $cellIndex) === 1);
     }
 }

src/Spout/Reader/XLSX/Helper/SharedStringsCaching/FileBasedStrategy.php

@@ -143,7 +143,9 @@ class FileBasedStrategy implements CachingStrategyInterface
         }
 
         $sharedString = null;
-        if (array_key_exists($indexInFile, $this->inMemoryTempFileContents)) {
+
+        // Using isset here because it is way faster than array_key_exists...
+        if (isset($this->inMemoryTempFileContents[$indexInFile])) {
             $escapedSharedString = $this->inMemoryTempFileContents[$indexInFile];
             $sharedString = $this->unescapeLineFeed($escapedSharedString);
         }

src/Spout/Reader/XLSX/Helper/StyleHelper.php

@@ -148,7 +148,8 @@ class StyleHelper
 
         // Default style (0) does not format numeric values as timestamps. Only custom styles do.
         // Also if the style ID does not exist in the styles.xml file, format as numeric value.
-        if ($styleId === self::DEFAULT_STYLE_ID || !array_key_exists($styleId, $stylesAttributes)) {
+        // Using isset here because it is way faster than array_key_exists...
+        if ($styleId === self::DEFAULT_STYLE_ID || !isset($stylesAttributes[$styleId])) {
             return false;
         }
 
@@ -193,7 +194,8 @@ class StyleHelper
     {
         $customNumberFormats = $this->getCustomNumberFormats();
 
-        if (!array_key_exists($numFmtId, $customNumberFormats)) {
+        // Using isset here because it is way faster than array_key_exists...
+        if (!isset($customNumberFormats[$numFmtId])) {
             return false;
         }
 

src/Spout/Reader/XLSX/SheetIterator.php

@@ -67,7 +67,8 @@ class SheetIterator implements IteratorInterface
      */
     public function next()
     {
-        if (array_key_exists($this->currentSheetIndex, $this->sheets)) {
+        // Using isset here because it is way faster than array_key_exists...
+        if (isset($this->sheets[$this->currentSheetIndex])) {
             $currentSheet = $this->sheets[$this->currentSheetIndex];
             $currentSheet->getRowIterator()->end();
 

src/Spout/Writer/Common/Helper/AbstractStyleHelper.php

@@ -56,7 +56,9 @@ abstract class AbstractStyleHelper
     protected function hasStyleAlreadyBeenRegistered($style)
     {
         $serializedStyle = $style->serialize();
-        return array_key_exists($serializedStyle, $this->serializedStyleToStyleIdMappingTable);
+
+        // Using isset here because it is way faster than array_key_exists...
+        return isset($this->serializedStyleToStyleIdMappingTable[$serializedStyle]);
     }
 
     /**

src/Spout/Writer/Common/Helper/CellHelper.php

@@ -10,6 +10,9 @@ namespace Box\Spout\Writer\Common\Helper;
  */
 class CellHelper
 {
+    /** @var array Cache containing the mapping column index => cell index */
+    private static $columnIndexToCellIndexCache = [];
+
     /**
      * Returns the cell index (base 26) associated to the base 10 column index.
      * Excel uses A to Z letters for column indexing, where A is the 1st column,
@@ -21,6 +24,10 @@ class CellHelper
      */
     public static function getCellIndexFromColumnIndex($columnIndex)
     {
+        $originalColumnIndex = $columnIndex;
+
+        // Using isset here because it is way faster than array_key_exists...
+        if (!isset(self::$columnIndexToCellIndexCache[$originalColumnIndex])) {
             $cellIndex = '';
             $capitalAAsciiValue = ord('A');
 
@@ -30,9 +37,13 @@ class CellHelper
 
                 // substracting 1 because it's zero-based
                 $columnIndex = intval($columnIndex / 26) - 1;
+
             } while ($columnIndex >= 0);
 
-        return $cellIndex;
+            self::$columnIndexToCellIndexCache[$originalColumnIndex] = $cellIndex;
+        }
+
+        return self::$columnIndexToCellIndexCache[$originalColumnIndex];
     }
 
     /**

src/Spout/Writer/ODS/Internal/Worksheet.php

@@ -146,7 +146,8 @@ class Worksheet implements WorksheetInterface
         for ($i = 0; $i < $cellsCount; $i++) {
             $currentCellValue = $dataRow[$currentCellIndex];
 
-            if (!array_key_exists($nextCellIndex, $dataRow) || $currentCellValue !== $dataRow[$nextCellIndex]) {
+            // Using isset here because it is way faster than array_key_exists...
+            if (!isset($dataRow[$nextCellIndex]) || $currentCellValue !== $dataRow[$nextCellIndex]) {
                 $numTimesValueRepeated = ($nextCellIndex - $currentCellIndex);
                 $data .= $this->getCellContent($currentCellValue, $styleIndex, $numTimesValueRepeated);