MDL-56012 lib: Update spout to version 2.6.0

lib/spout/src/Spout/Common/Escaper/ODS.php

@@ -2,6 +2,8 @@
 
 namespace Box\Spout\Common\Escaper;
 
+use Box\Spout\Common\Singleton;
+
 /**
  * Class ODS
  * Provides functions to escape and unescape data for ODS files
@@ -10,6 +12,8 @@
  */
 class ODS implements EscaperInterface
 {
+    use Singleton;
+
     /**
      * Escapes the given string to make it compatible with XLSX
      *

lib/spout/src/Spout/Common/Escaper/XLSX.php

@@ -2,6 +2,8 @@
 
 namespace Box\Spout\Common\Escaper;
 
+use Box\Spout\Common\Singleton;
+
 /**
  * Class XLSX
  * Provides functions to escape and unescape data for XLSX files
@@ -10,15 +12,25 @@
  */
 class XLSX implements EscaperInterface
 {
-    /** @var string[] Control characters to be escaped */
+    use Singleton;
+
+    /** @var string Regex pattern to detect control characters that need to be escaped */
+    protected $escapableControlCharactersPattern;
+
+    /** @var string[] Map containing control characters to be escaped (key) and their escaped value (value) */
     protected $controlCharactersEscapingMap;
 
+    /** @var string[] Map containing control characters to be escaped (value) and their escaped value (key) */
+    protected $controlCharactersEscapingReverseMap;
+
     /**
-     *
+     * Initializes the singleton instance
      */
-    public function __construct()
+    protected function init()
     {
+        $this->escapableControlCharactersPattern = $this->getEscapableControlCharactersPattern();
         $this->controlCharactersEscapingMap = $this->getControlCharactersEscapingMap();
+        $this->controlCharactersEscapingReverseMap = array_flip($this->controlCharactersEscapingMap);
     }
 
     /**
@@ -49,6 +61,20 @@ public function unescape($string)
         return $unescapedString;
     }
 
+    /**
+     * @return string Regex pattern containing all escapable control characters
+     */
+    protected function getEscapableControlCharactersPattern()
+    {
+        // control characters values are from 0 to 1F (hex values) in the ASCII table
+        // some characters should not be escaped though: "\t", "\r" and "\n".
+        return '[\x00-\x08' .
+                // skipping "\t" (0x9) and "\n" (0xA)
+                '\x0B-\x0C' .
+                // skipping "\r" (0xD)
+                '\x0E-\x1F]';
+    }
+
     /**
      * Builds the map containing control characters to be escaped
      * mapped to their escaped values.
@@ -62,14 +88,14 @@ public function unescape($string)
     protected function getControlCharactersEscapingMap()
     {
         $controlCharactersEscapingMap = [];
-        $whitelistedControlCharacters = ["\t", "\r", "\n"];
 
         // control characters values are from 0 to 1F (hex values) in the ASCII table
-        for ($charValue = 0x0; $charValue <= 0x1F; $charValue++) {
-            if (!in_array(chr($charValue), $whitelistedControlCharacters)) {
+        for ($charValue = 0x00; $charValue <= 0x1F; $charValue++) {
+            $character = chr($charValue);
+            if (preg_match("/{$this->escapableControlCharactersPattern}/", $character)) {
                 $charHexValue = dechex($charValue);
                 $escapedChar = '_x' . sprintf('%04s' , strtoupper($charHexValue)) . '_';
-                $controlCharactersEscapingMap[$escapedChar] = chr($charValue);
+                $controlCharactersEscapingMap[$escapedChar] = $character;
             }
         }
 
@@ -92,7 +118,15 @@ protected function getControlCharactersEscapingMap()
     protected function escapeControlCharacters($string)
     {
         $escapedString = $this->escapeEscapeCharacter($string);
-        return str_replace(array_values($this->controlCharactersEscapingMap), array_keys($this->controlCharactersEscapingMap), $escapedString);
+
+        // if no control characters
+        if (!preg_match("/{$this->escapableControlCharactersPattern}/", $escapedString)) {
+            return $escapedString;
+        }
+
+        return preg_replace_callback("/({$this->escapableControlCharactersPattern})/", function($matches) {
+            return $this->controlCharactersEscapingReverseMap[$matches[0]];
+        }, $escapedString);
     }
 
     /**
@@ -122,6 +156,7 @@ protected function escapeEscapeCharacter($string)
     protected function unescapeControlCharacters($string)
     {
         $unescapedString = $string;
+
         foreach ($this->controlCharactersEscapingMap as $escapedCharValue => $charValue) {
             // only unescape characters that don't contain the escaped escape character for now
             $unescapedString = preg_replace("/(?<!_x005F)($escapedCharValue)/", $charValue, $unescapedString);

lib/spout/src/Spout/Common/Helper/EncodingHelper.php

@@ -59,7 +59,7 @@ public function getBytesOffsetToSkipBOM($filePointer, $encoding)
     {
         $byteOffsetToSkipBom = 0;
 
-        if ($this->hasBom($filePointer, $encoding)) {
+        if ($this->hasBOM($filePointer, $encoding)) {
             $bomUsed = $this->supportedEncodingsWithBom[$encoding];
 
             // we skip the N first bytes

lib/spout/src/Spout/Common/Helper/GlobalFunctionsHelper.php

@@ -252,6 +252,19 @@ public function header($string)
         header($string);
     }
 
+    /**
+     * Wrapper around global function ob_end_clean()
+     * @see ob_end_clean()
+     *
+     * @return void
+     */
+    public function ob_end_clean()
+    {
+        if (ob_get_length() > 0) {
+            ob_end_clean();
+        }
+    }
+
     /**
      * Wrapper around global function iconv()
      * @see iconv()

lib/spout/src/Spout/Common/Singleton.php

@@ -0,0 +1,41 @@
+<?php
+
+namespace Box\Spout\Common;
+
+/**
+ * Class Singleton
+ * Defines a class as a singleton.
+ *
+ * @package Box\Spout\Common
+ */
+trait Singleton
+{
+    protected static $instance;
+
+    /**
+     * @return static
+     */
+    final public static function getInstance()
+    {
+        return isset(static::$instance)
+            ? static::$instance
+            : static::$instance = new static;
+    }
+
+    /**
+     * Singleton constructor.
+     */
+    final private function __construct()
+    {
+        $this->init();
+    }
+
+    /**
+     * Initializes the singleton
+     * @return void
+     */
+    protected function init() {}
+
+    final private function __wakeup() {}
+    final private function __clone() {}
+}

lib/spout/src/Spout/Reader/AbstractReader.php

@@ -19,6 +19,9 @@ abstract class AbstractReader implements ReaderInterface
     /** @var \Box\Spout\Common\Helper\GlobalFunctionsHelper Helper to work with global functions */
     protected $globalFunctionsHelper;
 
+    /** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
+    protected $shouldFormatDates = false;
+
     /**
      * Returns whether stream wrappers are supported
      *
@@ -49,7 +52,7 @@ abstract public function getConcreteSheetIterator();
     abstract protected function closeReader();
 
     /**
-     * @param $globalFunctionsHelper
+     * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
      * @return AbstractReader
      */
     public function setGlobalFunctionsHelper($globalFunctionsHelper)
@@ -58,6 +61,18 @@ public function setGlobalFunctionsHelper($globalFunctionsHelper)
         return $this;
     }
 
+    /**
+     * Sets whether date/time values should be returned as PHP objects or be formatted as strings.
+     *
+     * @param bool $shouldFormatDates
+     * @return AbstractReader
+     */
+    public function setShouldFormatDates($shouldFormatDates)
+    {
+        $this->shouldFormatDates = $shouldFormatDates;
+        return $this;
+    }
+
     /**
      * Prepares the reader to read the given file. It also makes sure
      * that the file exists and is readable.

lib/spout/src/Spout/Reader/CSV/RowIterator.php

@@ -57,6 +57,7 @@ class RowIterator implements IteratorInterface
      * @param string $fieldDelimiter Character that delimits fields
      * @param string $fieldEnclosure Character that enclose fields
      * @param string $encoding Encoding of the CSV file to be read
+     * @param string $endOfLineDelimiter End of line delimiter
      * @param \Box\Spout\Common\Helper\GlobalFunctionsHelper $globalFunctionsHelper
      */
     public function __construct($filePointer, $fieldDelimiter, $fieldEnclosure, $encoding, $endOfLineDelimiter, $globalFunctionsHelper)
@@ -152,7 +153,7 @@ public function next()
      */
     protected function getNextUTF8EncodedRow()
     {
-        $encodedRowData = fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
+        $encodedRowData = $this->globalFunctionsHelper->fgetcsv($this->filePointer, self::MAX_READ_BYTES_PER_LINE, $this->fieldDelimiter, $this->fieldEnclosure);
         if (false === $encodedRowData) {
             return false;
         }

lib/spout/src/Spout/Reader/ODS/Helper/CellValueFormatter.php

@@ -23,6 +23,8 @@ class CellValueFormatter
     /** Definition of XML nodes names used to parse data */
     const XML_NODE_P = 'p';
     const XML_NODE_S = 'text:s';
+    const XML_NODE_A = 'text:a';
+    const XML_NODE_SPAN = 'text:span';
 
     /** Definition of XML attribute used to parse data */
     const XML_ATTRIBUTE_TYPE = 'office:value-type';
@@ -33,16 +35,21 @@ class CellValueFormatter
     const XML_ATTRIBUTE_CURRENCY = 'office:currency';
     const XML_ATTRIBUTE_C = 'text:c';
 
+    /** @var bool Whether date/time values should be returned as PHP objects or be formatted as strings */
+    protected $shouldFormatDates;
+
     /** @var \Box\Spout\Common\Escaper\ODS Used to unescape XML data */
     protected $escaper;
 
     /**
-     *
+     * @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
      */
-    public function __construct()
+    public function __construct($shouldFormatDates)
     {
+        $this->shouldFormatDates = $shouldFormatDates;
+
         /** @noinspection PhpUnnecessaryFullyQualifiedNameInspection */
-        $this->escaper = new \Box\Spout\Common\Escaper\ODS();
+        $this->escaper = \Box\Spout\Common\Escaper\ODS::getInstance();
     }
 
     /**
@@ -98,6 +105,8 @@ protected function formatStringCellValue($node)
                     $spaceAttribute = $childNode->getAttribute(self::XML_ATTRIBUTE_C);
                     $numSpaces = (!empty($spaceAttribute)) ? intval($spaceAttribute) : 1;
                     $currentPValue .= str_repeat(' ', $numSpaces);
+                } else if ($childNode->nodeName === self::XML_NODE_A || $childNode->nodeName === self::XML_NODE_SPAN) {
+                    $currentPValue .= $childNode->nodeValue;
                 }
             }
 
@@ -119,6 +128,7 @@ protected function formatFloatCellValue($node)
     {
         $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_VALUE);
         $nodeIntValue = intval($nodeValue);
+        // The "==" is intentionally not a "===" because only the value matters, not the type
         $cellValue = ($nodeIntValue == $nodeValue) ? $nodeIntValue : floatval($nodeValue);
         return $cellValue;
     }
@@ -141,31 +151,55 @@ protected function formatBooleanCellValue($node)
      * Returns the cell Date value from the given node.
      *
      * @param \DOMNode $node
-     * @return \DateTime|null The value associated with the cell or NULL if invalid date value
+     * @return \DateTime|string|null The value associated with the cell or NULL if invalid date value
      */
     protected function formatDateCellValue($node)
     {
-        try {
-            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_DATE_VALUE);
-            return new \DateTime($nodeValue);
-        } catch (\Exception $e) {
-            return null;
+        // The XML node looks like this:
+        // <table:table-cell calcext:value-type="date" office:date-value="2016-05-19T16:39:00" office:value-type="date">
+        //   <text:p>05/19/16 04:39 PM</text:p>
+        // </table:table-cell>
+
+        if ($this->shouldFormatDates) {
+            // The date is already formatted in the "p" tag
+            $nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
+            return $nodeWithValueAlreadyFormatted->nodeValue;
+        } else {
+            // otherwise, get it from the "date-value" attribute
+            try {
+                $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_DATE_VALUE);
+                return new \DateTime($nodeValue);
+            } catch (\Exception $e) {
+                return null;
+            }
         }
     }
 
     /**
      * Returns the cell Time value from the given node.
      *
      * @param \DOMNode $node
-     * @return \DateInterval|null The value associated with the cell or NULL if invalid time value
+     * @return \DateInterval|string|null The value associated with the cell or NULL if invalid time value
      */
     protected function formatTimeCellValue($node)
     {
-        try {
-            $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_TIME_VALUE);
-            return new \DateInterval($nodeValue);
-        } catch (\Exception $e) {
-            return null;
+        // The XML node looks like this:
+        // <table:table-cell calcext:value-type="time" office:time-value="PT13H24M00S" office:value-type="time">
+        //   <text:p>01:24:00 PM</text:p>
+        // </table:table-cell>
+
+        if ($this->shouldFormatDates) {
+            // The date is already formatted in the "p" tag
+            $nodeWithValueAlreadyFormatted = $node->getElementsByTagName(self::XML_NODE_P)->item(0);
+            return $nodeWithValueAlreadyFormatted->nodeValue;
+        } else {
+            // otherwise, get it from the "time-value" attribute
+            try {
+                $nodeValue = $node->getAttribute(self::XML_ATTRIBUTE_TIME_VALUE);
+                return new \DateInterval($nodeValue);
+            } catch (\Exception $e) {
+                return null;
+            }
         }
     }
 

lib/spout/src/Spout/Reader/ODS/Reader.php

@@ -42,7 +42,7 @@ protected function openReader($filePath)
         $this->zip = new \ZipArchive();
 
         if ($this->zip->open($filePath) === true) {
-            $this->sheetIterator = new SheetIterator($filePath);
+            $this->sheetIterator = new SheetIterator($filePath, $this->shouldFormatDates);
         } else {
             throw new IOException("Could not open $filePath for reading.");
         }

lib/spout/src/Spout/Reader/ODS/RowIterator.php

@@ -45,11 +45,12 @@ class RowIterator implements IteratorInterface
 
     /**
      * @param XMLReader $xmlReader XML Reader, positioned on the "<table:table>" element
+     * @param bool $shouldFormatDates Whether date/time values should be returned as PHP objects or be formatted as strings
      */
-    public function __construct($xmlReader)
+    public function __construct($xmlReader, $shouldFormatDates)
     {
         $this->xmlReader = $xmlReader;
-        $this->cellValueFormatter = new CellValueFormatter();
+        $this->cellValueFormatter = new CellValueFormatter($shouldFormatDates);
     }
 
     /**
@@ -186,7 +187,7 @@ protected function getCellValue($node)
     /**
      * empty() replacement that honours 0 as a valid value
      *
-     * @param $value The cell value
+     * @param string|int|float|bool|\DateTime|\DateInterval|null $value The cell value
      * @return bool
      */
     protected function isEmptyCellValue($value)