Adding open_file_in_zip() helper function to XMLReader (#238)

src/Spout/Reader/ODS/SheetIterator.php

@@ -15,6 +15,8 @@ use Box\Spout\Reader\Wrapper\XMLReader;
  */
 class SheetIterator implements IteratorInterface
 {
+    const CONTENT_XML_FILE_PATH = 'content.xml';
+
     /** Definition of XML nodes name and attribute used to parse sheet data */
     const XML_NODE_TABLE = 'table:table';
     const XML_ATTRIBUTE_TABLE_NAME = 'table:name';
@@ -63,8 +65,8 @@ class SheetIterator implements IteratorInterface
     {
         $this->xmlReader->close();
 
-        $contentXmlFilePath = $this->filePath . '#content.xml';
-        if ($this->xmlReader->open('zip://' . $contentXmlFilePath) === false) {
+        if ($this->xmlReader->openFileInZip($this->filePath, self::CONTENT_XML_FILE_PATH) === false) {
+            $contentXmlFilePath = $this->filePath . '#' . self::CONTENT_XML_FILE_PATH;
             throw new IOException("Could not open \"{$contentXmlFilePath}\".");
         }
 

src/Spout/Reader/Wrapper/XMLReader.php

@@ -14,66 +14,44 @@ class XMLReader extends \XMLReader
 {
     use XMLInternalErrorsHelper;
 
+    const ZIP_WRAPPER = 'zip://';
+
     /**
-     * Set the URI containing the XML to parse
-     * @see \XMLReader::open
+     * Opens the XML Reader to read a file located inside a ZIP file.
      *
-     * @param string $URI URI pointing to the document
-   	 * @param string|null|void $encoding The document encoding
-   	 * @param int $options A bitmask of the LIBXML_* constants
+     * @param string $zipFilePath Path to the ZIP file
+     * @param string $fileInsideZipPath Relative or absolute path of the file inside the zip
      * @return bool TRUE on success or FALSE on failure
      */
-    public function open($URI, $encoding = null, $options = 0)
+    public function openFileInZip($zipFilePath, $fileInsideZipPath)
     {
         $wasOpenSuccessful = false;
-        $realPathURI = $this->convertURIToUseRealPath($URI);
+        $realPathURI = $this->getRealPathURIForFileInZip($zipFilePath, $fileInsideZipPath);
 
         // HHVM does not check if file exists within zip file
         // @link https://github.com/facebook/hhvm/issues/5779
-        if ($this->isRunningHHVM() && $this->isZipStream($realPathURI)) {
+        if ($this->isRunningHHVM()) {
             if ($this->fileExistsWithinZip($realPathURI)) {
-                $wasOpenSuccessful = parent::open($realPathURI, $encoding, $options|LIBXML_NONET);
+                $wasOpenSuccessful = $this->open($realPathURI, null, LIBXML_NONET);
             }
         } else {
-            $wasOpenSuccessful = parent::open($realPathURI, $encoding, $options|LIBXML_NONET);
+            $wasOpenSuccessful = $this->open($realPathURI, null, LIBXML_NONET);
         }
 
         return $wasOpenSuccessful;
     }
 
     /**
-     * Updates the given URI to use a real path.
-     * This is to avoid issues on some Windows setup.
+     * Returns the real path for the given path components.
+     * This is useful to avoid issues on some Windows setup.
      *
-     * @param string $URI URI
-     * @return string The URI using a real path
+     * @param string $zipFilePath Path to the ZIP file
+     * @param string $fileInsideZipPath Relative or absolute path of the file inside the zip
+     * @return string The real path URI
      */
-    protected function convertURIToUseRealPath($URI)
+    public function getRealPathURIForFileInZip($zipFilePath, $fileInsideZipPath)
     {
-        $realPathURI = $URI;
-
-        if ($this->isZipStream($URI)) {
-            if (preg_match('/zip:\/\/(.*)#(.*)/', $URI, $matches)) {
-                $documentPath = $matches[1];
-                $documentInsideZipPath = $matches[2];
-                $realPathURI = 'zip://' . realpath($documentPath) . '#' . $documentInsideZipPath;
-            }
-        } else {
-            $realPathURI = realpath($URI);
-        }
-
-        return $realPathURI;
-    }
-
-    /**
-     * Returns whether the given URI is a zip stream.
-     *
-     * @param string $URI URI pointing to a document
-     * @return bool TRUE if URI is a zip stream, FALSE otherwise
-     */
-    protected function isZipStream($URI)
-    {
-        return (strpos($URI, 'zip://') === 0);
+        return (self::ZIP_WRAPPER . realpath($zipFilePath) . '#' . $fileInsideZipPath);
     }
 
     /**

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

@@ -55,7 +55,7 @@ class SheetHelper
         $sheetIndex = 0;
 
         $xmlReader = new XMLReader();
-        if ($xmlReader->open('zip://' . $this->filePath . '#' . self::WORKBOOK_XML_FILE_PATH)) {
+        if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_XML_FILE_PATH)) {
             while ($xmlReader->read()) {
                 if ($xmlReader->isPositionedOnStartingNode('sheet')) {
                     $sheets[] = $this->getSheetFromSheetXMLNode($xmlReader, $sheetIndex);
@@ -105,7 +105,7 @@ class SheetHelper
 
         // find the file path of the sheet, by looking at the "workbook.xml.res" file
         $xmlReader = new XMLReader();
-        if ($xmlReader->open('zip://' . $this->filePath . '#' . self::WORKBOOK_XML_RELS_FILE_PATH)) {
+        if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_XML_RELS_FILE_PATH)) {
             while ($xmlReader->read()) {
                 if ($xmlReader->isPositionedOnStartingNode('Relationship')) {
                     $relationshipSheetId = $xmlReader->getAttribute('Id');

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

@@ -76,10 +76,9 @@ class StyleHelper
         $this->customNumberFormats = [];
         $this->stylesAttributes = [];
 
-        $stylesXmlFilePath = $this->filePath .'#' . self::STYLES_XML_FILE_PATH;
         $xmlReader = new XMLReader();
 
-        if ($xmlReader->open('zip://' . $stylesXmlFilePath)) {
+        if ($xmlReader->openFileInZip($this->filePath, self::STYLES_XML_FILE_PATH)) {
             while ($xmlReader->read()) {
                 if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_NUM_FMTS)) {
                     $numFmtsNode = new SimpleXMLElement($xmlReader->readOuterXml());