Improved masking for number format handling, particularly for datetime masks

src/PhpSpreadsheet/Style/NumberFormat.php

@@ -425,26 +425,43 @@ class NumberFormat extends Supervisor implements \PHPExcel\IComparable
             'h'  => 'g'
         );
 
+    private static function setLowercaseCallback($matches) {
+        return mb_strtolower($matches[0]);
+    }
+
+    private static function escapeQuotesCallback($matches) {
+        return '\\' . implode('\\', str_split($matches[1]));
+    }
+
     private static function formatAsDate(&$value, &$format)
     {
-        // dvc: convert Excel formats to PHP date formats
-
         // strip off first part containing e.g. [$-F800] or [$USD-409]
         // general syntax: [$<Currency string>-<language info>]
         // language info is in hexadecimal
         $format = preg_replace('/^(\[\$[A-Z]*-[0-9A-F]*\])/i', '', $format);
 
-        // OpenOffice.org uses upper-case number formats, e.g. 'YYYY', convert to lower-case
+        // OpenOffice.org uses upper-case number formats, e.g. 'YYYY', convert to lower-case;
-        $format = strtolower($format);
+        //    but we don't want to change any quoted strings
+        $format = preg_replace_callback('/(?:^|")([^"]*)(?:$|")/', ['self', 'setLowercaseCallback'], $format);
 
-        $format = strtr($format, self::$dateFormatReplacements);
+        // Only process the non-quoted blocks for date format characters
-        if (!strpos($format, 'A')) {
+        $blocks = explode('"', $format);
-            // 24-hour time format
+        foreach($blocks as $key => &$block) {
-            $format = strtr($format, self::$dateFormatReplacements24);
+            if ($key % 2 == 0) {
-        } else {
+                $block = strtr($block, self::$dateFormatReplacements);
-            // 12-hour time format
+                if (!strpos($block, 'A')) {
-            $format = strtr($format, self::$dateFormatReplacements12);
+                    // 24-hour time format
+                    $block = strtr($block, self::$dateFormatReplacements24);
+                } else {
+                    // 12-hour time format
+                    $block = strtr($block, self::$dateFormatReplacements12);
+                }
+            }
         }
+        $format = implode('"', $blocks);
+
+        // escape any quoted characters so that DateTime format() will render them correctly
+        $format = preg_replace_callback('/"(.*)"/U', ['self', 'escapeQuotesCallback'], $format);
 
         $dateObj = \PHPExcel\Shared\Date::ExcelToPHPObject($value);
         $value = $dateObj->format($format);
@@ -553,10 +570,12 @@ class NumberFormat extends Supervisor implements \PHPExcel\IComparable
             return $value;
         }
 
-        // Get the sections, there can be up to four sections
+        // Convert any escaped characters to quoted strings, e.g. (\T to "T")
-        $sections = explode(';', $format);
+        $format = preg_replace('/(\\\(.))(?=(?:[^"]|"[^"]*")*$)/u', '"${2}"', $format);
+        // Get the sections, there can be up to four sections, separated with a semi-colon (but only if not a quoted literal)
+        $sections = preg_split('/(;)(?=(?:[^"]|"[^"]*")*$)/u', $format);
 
-        // Fetch the relevant section depending on whether number is positive, negative, or zero?
+        // Extract the relevant section depending on whether number is positive, negative, or zero?
         // Text not supported yet.
         // Here is how the sections apply to various values in Excel:
         //   1 section:   [POSITIVE/NEGATIVE/ZERO/TEXT]
@@ -597,9 +616,13 @@ class NumberFormat extends Supervisor implements \PHPExcel\IComparable
         $format = preg_replace($color_regex, '', $format);
 
         // Let's begin inspecting the format and converting the value to a formatted string
-        if (preg_match('/^(\[\$[A-Z]*-[0-9A-F]*\])*[hmsdy]/i', $format)) { // datetime format
+
+        //  Check for date/time characters (not inside quotes)
+        if (preg_match('/(\[\$[A-Z]*-[0-9A-F]*\])*[hmsdy](?=(?:[^"]|"[^"]*")*$)/miu', $format, $matches)) {
+            // datetime format
             self::formatAsDate($value, $format);
-        } elseif (preg_match('/%$/', $format)) { // % number format
+        } elseif (preg_match('/%$/', $format)) {
+            // % number format
             self::formatAsPercentage($value, $format);
         } else {
             if ($format === self::FORMAT_CURRENCY_EUR_SIMPLE) {
@@ -687,7 +710,7 @@ class NumberFormat extends Supervisor implements \PHPExcel\IComparable
                     }
                 }
                 if (preg_match('/\[\$(.*)\]/u', $format, $m)) {
-                    //    Currency or Accounting
+                    //  Currency or Accounting
                     $currencyFormat = $m[0];
                     $currencyCode = $m[1];
                     list($currencyCode) = explode('-', $currencyCode);

unitTests/Classes/src/Style/NumberFormatDateTest.php

@@ -0,0 +1,35 @@
+<?php
+
+
+require_once 'testDataFileIterator.php';
+
+class NumberFormatDateTest extends PHPUnit_Framework_TestCase
+{
+
+    public function setUp()
+    {
+        if (!defined('PHPEXCEL_ROOT')) {
+            define('PHPEXCEL_ROOT', APPLICATION_PATH . '/');
+        }
+        require_once(PHPEXCEL_ROOT . 'PHPExcel/Autoloader.php');
+
+        PHPExcel_Shared_String::setDecimalSeparator('.');
+        PHPExcel_Shared_String::setThousandsSeparator(',');
+    }
+
+    /**
+     * @dataProvider providerNumberFormat
+     */
+    public function testFormatValueWithMask()
+    {
+        $args = func_get_args();
+        $expectedResult = array_pop($args);
+        $result = call_user_func_array(array('PHPExcel_Style_NumberFormat','toFormattedString'), $args);
+        $this->assertEquals($expectedResult, $result);
+    }
+
+    public function providerNumberFormat()
+    {
+        return new testDataFileIterator('rawTestData/Style/NumberFormatDates.data');
+    }
+}

unitTests/rawTestData/Style/NumberFormatDates.data

@@ -0,0 +1,6 @@
+22269.0625, 'dd-mm-yyyy hh:mm:ss',                  "19-12-1960 01:30:00"
+22269.0625, 'MM/DD/YYYY HH:MM:SS',                  "12/19/1960 01:30:00"     // Oasis uses upper-case
+22269.0625, 'yyyy-mm-dd\Thh:mm:ss',                 "1960-12-19T01:30:00"     // Date with plaintext escaped with a \
+22269.0625, 'yyyy-mm-dd"T"hh:mm:ss \Z',             "1960-12-19T01:30:00 Z"   // Date with plaintext in quotes
+22269.0625, '"y-m-d" yyyy-mm-dd "h:m:s" hh:mm:ss',  "y-m-d 1960-12-19 h:m:s 01:30:00"  // Date with quoted formatting characters
+22269.0625, '"y-m-d "yyyy-mm-dd" h:m:s "hh:mm:ss',  "y-m-d 1960-12-19 h:m:s 01:30:00"  // Date with quoted formatting characters