Fetch XML file paths from Workbook Relationships

src/Spout/Reader/Wrapper/XMLReader.php

@@ -45,7 +45,10 @@ class XMLReader extends \XMLReader
      */
     public function getRealPathURIForFileInZip($zipFilePath, $fileInsideZipPath)
     {
-        return (self::ZIP_WRAPPER . realpath($zipFilePath) . '#' . $fileInsideZipPath);
+        // The file path should not start with a '/', otherwise it won't be found
+        $fileInsideZipPathWithoutLeadingSlash = ltrim($fileInsideZipPath, '/');
+
+        return (self::ZIP_WRAPPER . realpath($zipFilePath) . '#' . $fileInsideZipPathWithoutLeadingSlash);
     }
 
     /**

src/Spout/Reader/XLSX/Creator/ManagerFactory.php

@@ -6,6 +6,7 @@ use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyFactory;
 use Box\Spout\Reader\XLSX\Manager\SharedStringsManager;
 use Box\Spout\Reader\XLSX\Manager\SheetManager;
 use Box\Spout\Reader\XLSX\Manager\StyleManager;
+use Box\Spout\Reader\XLSX\Manager\WorkbookRelationshipsManager;
 
 /**
  * Class ManagerFactory
@@ -19,6 +20,9 @@ class ManagerFactory
     /** @var CachingStrategyFactory */
     private $cachingStrategyFactory;
 
+    /** @var WorkbookRelationshipsManager */
+    private $cachedWorkbookRelationshipsManager;
+
     /**
      * @param HelperFactory $helperFactory Factory to create helpers
      * @param CachingStrategyFactory $cachingStrategyFactory Factory to create shared strings caching strategies
@@ -37,7 +41,30 @@ class ManagerFactory
      */
     public function createSharedStringsManager($filePath, $tempFolder, $entityFactory)
     {
-        return new SharedStringsManager($filePath, $tempFolder, $entityFactory, $this->helperFactory, $this->cachingStrategyFactory);
+        $workbookRelationshipsManager = $this->createWorkbookRelationshipsManager($filePath, $entityFactory);
+
+        return new SharedStringsManager(
+            $filePath,
+            $tempFolder,
+            $workbookRelationshipsManager,
+            $entityFactory,
+            $this->helperFactory,
+            $this->cachingStrategyFactory
+        );
+    }
+
+    /**
+     * @param string $filePath Path of the XLSX file being read
+     * @param EntityFactory $entityFactory Factory to create entities
+     * @return WorkbookRelationshipsManager
+     */
+    private function createWorkbookRelationshipsManager($filePath, $entityFactory)
+    {
+        if (!isset($this->cachedWorkbookRelationshipsManager)) {
+            $this->cachedWorkbookRelationshipsManager = new WorkbookRelationshipsManager($filePath, $entityFactory);
+        }
+
+        return $this->cachedWorkbookRelationshipsManager;
     }
 
     /**
@@ -61,6 +88,8 @@ class ManagerFactory
      */
     public function createStyleManager($filePath, $entityFactory)
     {
-        return new StyleManager($filePath, $entityFactory);
+        $workbookRelationshipsManager = $this->createWorkbookRelationshipsManager($filePath, $entityFactory);
+
+        return new StyleManager($filePath, $workbookRelationshipsManager, $entityFactory);
     }
 }

src/Spout/Reader/XLSX/Manager/SharedStringsManager.php

@@ -9,6 +9,7 @@ use Box\Spout\Reader\XLSX\Creator\EntityFactory;
 use Box\Spout\Reader\XLSX\Creator\HelperFactory;
 use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyFactory;
 use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyInterface;
+use Box\Spout\Writer\Common\Entity\Workbook;
 
 /**
  * Class SharedStringsManager
@@ -16,9 +17,6 @@ use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyInterface;
  */
 class SharedStringsManager
 {
-    /** Path of sharedStrings XML file inside the XLSX file */
-    const SHARED_STRINGS_XML_FILE_PATH = 'xl/sharedStrings.xml';
-
     /** Main namespace for the sharedStrings.xml file */
     const MAIN_NAMESPACE_FOR_SHARED_STRINGS_XML = 'http://schemas.openxmlformats.org/spreadsheetml/2006/main';
 
@@ -40,6 +38,9 @@ class SharedStringsManager
     /** @var string Temporary folder where the temporary files to store shared strings will be stored */
     protected $tempFolder;
 
+    /** @var WorkbookRelationshipsManager Helps retrieving workbook relationships */
+    protected $workbookRelationshipsManager;
+
     /** @var EntityFactory Factory to create entities */
     protected $entityFactory;
 
@@ -55,14 +56,22 @@ class SharedStringsManager
     /**
      * @param string $filePath Path of the XLSX file being read
      * @param string $tempFolder Temporary folder where the temporary files to store shared strings will be stored
+     * @param WorkbookRelationshipsManager $workbookRelationshipsManager Helps retrieving workbook relationships
      * @param EntityFactory $entityFactory Factory to create entities
      * @param HelperFactory $helperFactory Factory to create helpers
      * @param CachingStrategyFactory $cachingStrategyFactory Factory to create shared strings caching strategies
      */
-    public function __construct($filePath, $tempFolder, $entityFactory, $helperFactory, $cachingStrategyFactory)
-    {
+    public function __construct(
+        $filePath,
+        $tempFolder,
+        $workbookRelationshipsManager,
+        $entityFactory,
+        $helperFactory,
+        $cachingStrategyFactory
+    ) {
         $this->filePath = $filePath;
         $this->tempFolder = $tempFolder;
+        $this->workbookRelationshipsManager = $workbookRelationshipsManager;
         $this->entityFactory = $entityFactory;
         $this->helperFactory = $helperFactory;
         $this->cachingStrategyFactory = $cachingStrategyFactory;
@@ -75,15 +84,7 @@ class SharedStringsManager
      */
     public function hasSharedStrings()
     {
-        $hasSharedStrings = false;
-        $zip = $this->entityFactory->createZipArchive();
-
-        if ($zip->open($this->filePath) === true) {
-            $hasSharedStrings = ($zip->locateName(self::SHARED_STRINGS_XML_FILE_PATH) !== false);
-            $zip->close();
-        }
-
-        return $hasSharedStrings;
+        return $this->workbookRelationshipsManager->hasSharedStringsXMLFile();
     }
 
     /**
@@ -96,16 +97,17 @@ class SharedStringsManager
      * The XML file can be really big with sheets containing a lot of data. That is why
      * we need to use a XML reader that provides streaming like the XMLReader library.
      *
-     * @throws \Box\Spout\Common\Exception\IOException If sharedStrings.xml can't be read
+     * @throws \Box\Spout\Common\Exception\IOException If shared strings XML file can't be read
      * @return void
      */
     public function extractSharedStrings()
     {
+        $sharedStringsXMLFilePath = $this->workbookRelationshipsManager->getSharedStringsXMLFilePath();
         $xmlReader = $this->entityFactory->createXMLReader();
         $sharedStringIndex = 0;
 
-        if ($xmlReader->openFileInZip($this->filePath, self::SHARED_STRINGS_XML_FILE_PATH) === false) {
-            throw new IOException('Could not open "' . self::SHARED_STRINGS_XML_FILE_PATH . '".');
+        if ($xmlReader->openFileInZip($this->filePath, $sharedStringsXMLFilePath) === false) {
+            throw new IOException('Could not open "' . $sharedStringsXMLFilePath . '".');
         }
 
         try {

src/Spout/Reader/XLSX/Manager/StyleManager.php

@@ -10,9 +10,6 @@ use Box\Spout\Reader\XLSX\Creator\EntityFactory;
  */
 class StyleManager
 {
-    /** Paths of XML files relative to the XLSX file root */
-    const STYLES_XML_FILE_PATH = 'xl/styles.xml';
-
     /** Nodes used to find relevant information in the styles XML file */
     const XML_NODE_NUM_FMTS = 'numFmts';
     const XML_NODE_NUM_FMT = 'numFmt';
@@ -51,6 +48,9 @@ class StyleManager
     /** @var string Path of the XLSX file being read */
     protected $filePath;
 
+    /** @var string Path of the styles XML file */
+    protected $stylesXMLFilePath;
+
     /** @var EntityFactory Factory to create entities */
     protected $entityFactory;
 
@@ -68,13 +68,15 @@ class StyleManager
 
     /**
      * @param string $filePath Path of the XLSX file being read
+     * @param WorkbookRelationshipsManager $workbookRelationshipsManager Helps retrieving workbook relationships
      * @param EntityFactory $entityFactory Factory to create entities
      */
-    public function __construct($filePath, $entityFactory)
+    public function __construct($filePath, $workbookRelationshipsManager, $entityFactory)
     {
         $this->filePath = $filePath;
         $this->entityFactory = $entityFactory;
         $this->builtinNumFmtIdIndicatingDates = array_keys(self::$builtinNumFmtIdToNumFormatMapping);
+        $this->stylesXMLFilePath = $workbookRelationshipsManager->getStylesXMLFilePath();
     }
 
     /**
@@ -112,7 +114,7 @@ class StyleManager
 
         $xmlReader = $this->entityFactory->createXMLReader();
 
-        if ($xmlReader->openFileInZip($this->filePath, self::STYLES_XML_FILE_PATH)) {
+        if ($xmlReader->openFileInZip($this->filePath, $this->stylesXMLFilePath)) {
             while ($xmlReader->read()) {
                 if ($xmlReader->isPositionedOnStartingNode(self::XML_NODE_NUM_FMTS)) {
                     $this->extractNumberFormats($xmlReader);

src/Spout/Reader/XLSX/Manager/WorkbookRelationshipsManager.php

@@ -0,0 +1,136 @@
+<?php
+
+namespace Box\Spout\Reader\XLSX\Manager;
+
+use Box\Spout\Common\Exception\IOException;
+use Box\Spout\Reader\Wrapper\XMLReader;
+use Box\Spout\Reader\XLSX\Creator\EntityFactory;
+
+/**
+ * Class WorkbookRelationshipsManager
+ * This class manages the workbook relationships defined in the associated XML file
+ */
+class WorkbookRelationshipsManager
+{
+    const BASE_PATH = 'xl/';
+
+    /** Path of workbook relationships XML file inside the XLSX file */
+    const WORKBOOK_RELS_XML_FILE_PATH = 'xl/_rels/workbook.xml.rels';
+
+    /** Relationships types */
+    const RELATIONSHIP_TYPE_SHARED_STRINGS = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/sharedStrings';
+    const RELATIONSHIP_TYPE_STYLES = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/styles';
+    const RELATIONSHIP_TYPE_WORKSHEET = 'http://schemas.openxmlformats.org/officeDocument/2006/relationships/worksheet';
+
+    /** Nodes and attributes used to find relevant information in the workbook relationships XML file */
+    const XML_NODE_RELATIONSHIP = 'Relationship';
+    const XML_ATTRIBUTE_TYPE = 'Type';
+    const XML_ATTRIBUTE_TARGET = 'Target';
+
+    /** @var string Path of the XLSX file being read */
+    private $filePath;
+
+    /** @var EntityFactory Factory to create entities */
+    private $entityFactory;
+
+    /** @var array Cache of the already read workbook relationships: [TYPE] => [FILE_NAME] */
+    private $cachedWorkbookRelationships;
+
+    /**
+     * @param string $filePath Path of the XLSX file being read
+     * @param EntityFactory $entityFactory Factory to create entities
+     */
+    public function __construct($filePath, $entityFactory)
+    {
+        $this->filePath = $filePath;
+        $this->entityFactory = $entityFactory;
+    }
+
+    /**
+     * @return string The path of the shared string XML file
+     */
+    public function getSharedStringsXMLFilePath()
+    {
+        $workbookRelationships = $this->getWorkbookRelationships();
+        $sharedStringsXMLFilePath = $workbookRelationships[self::RELATIONSHIP_TYPE_SHARED_STRINGS];
+
+        // the file path can be relative (e.g. "styles.xml") or absolute (e.g. "/xl/styles.xml")
+        $doesContainBasePath = (strpos($sharedStringsXMLFilePath, self::BASE_PATH) !== false);
+        if (!$doesContainBasePath) {
+            // make sure we return an absolute file path
+            $sharedStringsXMLFilePath = self::BASE_PATH . $sharedStringsXMLFilePath;
+        }
+
+        return $sharedStringsXMLFilePath;
+    }
+
+    /**
+     * @return bool Whether the XLSX file contains a shared string XML file
+     */
+    public function hasSharedStringsXMLFile()
+    {
+        $workbookRelationships = $this->getWorkbookRelationships();
+
+        return isset($workbookRelationships[self::RELATIONSHIP_TYPE_SHARED_STRINGS]);
+    }
+
+    /**
+     * @return string|null The path of the styles XML file
+     */
+    public function getStylesXMLFilePath()
+    {
+        $workbookRelationships = $this->getWorkbookRelationships();
+        $stylesXMLFilePath = $workbookRelationships[self::RELATIONSHIP_TYPE_STYLES];
+
+        // the file path can be relative (e.g. "styles.xml") or absolute (e.g. "/xl/styles.xml")
+        $doesContainBasePath = (strpos($stylesXMLFilePath, self::BASE_PATH) !== false);
+        if (!$doesContainBasePath) {
+            // make sure we return a full path
+            $stylesXMLFilePath = self::BASE_PATH . $stylesXMLFilePath;
+        }
+
+        return $stylesXMLFilePath;
+    }
+
+    /**
+     * Reads the workbook.xml.rels and extracts the filename associated to the different types.
+     * It caches the result so that the file is read only once.
+     *
+     * @throws \Box\Spout\Common\Exception\IOException If workbook.xml.rels can't be read
+     * @return array
+     */
+    private function getWorkbookRelationships()
+    {
+        if (!isset($this->cachedWorkbookRelationships)) {
+            $xmlReader = $this->entityFactory->createXMLReader();
+
+            if ($xmlReader->openFileInZip($this->filePath, self::WORKBOOK_RELS_XML_FILE_PATH) === false) {
+                throw new IOException('Could not open "' . self::WORKBOOK_RELS_XML_FILE_PATH . '".');
+            }
+
+            $this->cachedWorkbookRelationships = [];
+
+            while ($xmlReader->readUntilNodeFound(self::XML_NODE_RELATIONSHIP)) {
+                $this->processWorkbookRelationship($xmlReader);
+            }
+        }
+
+        return $this->cachedWorkbookRelationships;
+    }
+
+    /**
+     * Extracts and store the data of the current workbook relationship.
+     *
+     * @param XMLReader $xmlReader
+     * @return void
+     */
+    private function processWorkbookRelationship($xmlReader)
+    {
+        $type = $xmlReader->getAttribute(self::XML_ATTRIBUTE_TYPE);
+        $target = $xmlReader->getAttribute(self::XML_ATTRIBUTE_TARGET);
+
+        // @NOTE: if a type is defined more than once, we overwrite the previous value
+        // To be changed if we want to get the file paths of sheet XML files for instance.
+        $this->cachedWorkbookRelationships[$type] = $target;
+    }
+}

tests/Spout/Reader/XLSX/Manager/SharedStringsManagerTest.php

@@ -51,8 +51,16 @@ class SharedStringsManagerTest extends \PHPUnit_Framework_TestCase
         $helperFactory = new HelperFactory();
         $managerFactory = new ManagerFactory($helperFactory, $cachingStrategyFactory);
         $entityFactory = new EntityFactory($managerFactory, $helperFactory);
+        $workbookRelationshipsManager = new WorkbookRelationshipsManager($resourcePath, $entityFactory);
 
-        $this->sharedStringsManager = new SharedStringsManager($resourcePath, $tempFolder, $entityFactory, $helperFactory, $cachingStrategyFactory);
+        $this->sharedStringsManager = new SharedStringsManager(
+            $resourcePath,
+            $tempFolder,
+            $workbookRelationshipsManager,
+            $entityFactory,
+            $helperFactory,
+            $cachingStrategyFactory
+        );
 
         return $this->sharedStringsManager;
     }

tests/Spout/Reader/XLSX/Manager/StyleManagerTest.php

@@ -3,9 +3,6 @@
 namespace Box\Spout\Reader\XLSX\Manager;
 
 use Box\Spout\Reader\XLSX\Creator\EntityFactory;
-use Box\Spout\Reader\XLSX\Creator\HelperFactory;
-use Box\Spout\Reader\XLSX\Creator\ManagerFactory;
-use Box\Spout\Reader\XLSX\Manager\SharedStringsCaching\CachingStrategyFactory;
 
 /**
  * Class StyleManagerTest
@@ -19,13 +16,12 @@ class StyleManagerTest extends \PHPUnit_Framework_TestCase
      */
     private function getStyleManagerMock($styleAttributes = [], $customNumberFormats = [])
     {
-        $helperFactory = new HelperFactory();
-        $managerFactory = new ManagerFactory($helperFactory, new CachingStrategyFactory());
-        $entityFactory = new EntityFactory($managerFactory, $helperFactory);
+        $entityFactory = $this->createMock(EntityFactory::class);
+        $workbookRelationshipsManager = $this->createMock(WorkbookRelationshipsManager::class);
 
         /** @var StyleManager $styleManager */
         $styleManager = $this->getMockBuilder('\Box\Spout\Reader\XLSX\Manager\StyleManager')
-                             ->setConstructorArgs(['/path/to/file.xlsx', $entityFactory])
+                             ->setConstructorArgs(['/path/to/file.xlsx', $workbookRelationshipsManager, $entityFactory])
                              ->setMethods(['getCustomNumberFormats', 'getStylesAttributes'])
                              ->getMock();
 

tests/Spout/Reader/XLSX/ReaderTest.php

@@ -169,6 +169,23 @@ class ReaderTest extends \PHPUnit_Framework_TestCase
         $this->assertEquals($expectedRows, $allRows);
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldSupportFilesWithCapitalSharedStringsFileName()
+    {
+        $allRows = $this->getAllRowsForFile('one_sheet_with_capital_shared_strings_filename.xlsx');
+
+        $expectedRows = [
+            ['s1--A1', 's1--B1', 's1--C1', 's1--D1', 's1--E1'],
+            ['s1--A2', 's1--B2', 's1--C2', 's1--D2', 's1--E2'],
+            ['s1--A3', 's1--B3', 's1--C3', 's1--D3', 's1--E3'],
+            ['s1--A4', 's1--B4', 's1--C4', 's1--D4', 's1--E4'],
+            ['s1--A5', 's1--B5', 's1--C5', 's1--D5', 's1--E5'],
+        ];
+        $this->assertEquals($expectedRows, $allRows);
+    }
+
     /**
      * @return void
      */