Merge pull request #3482 from PHPOffice/Xlsx-Reader_Windows-Folder-Separator-in-Zip

Handle zips with Windoze directory separators (`\`) rather than the standard linux (`/`)

Author: MarkBaker

src/PhpSpreadsheet/Reader/Xlsx.php

@@ -252,6 +252,7 @@ public function listWorksheetInfo($filename)
                 $xmlWorkbook = $this->loadZip($relTarget, $mainNS);
                 if ($xmlWorkbook->sheets) {
                     $dir = dirname($relTarget);
+
                     /** @var SimpleXMLElement $eleSheet */
                     foreach ($xmlWorkbook->sheets->sheet as $eleSheet) {
                         $tmpInfo = [
@@ -267,8 +268,8 @@ public function listWorksheetInfo($filename)
 
                         $xml = new XMLReader();
                         $xml->xml(
-                            $this->getSecurityScannerOrThrow()->scanFile(
-                                'zip://' . File::realpath($filename) . '#' . $fileWorksheetPath
+                            $this->getSecurityScannerOrThrow()->scan(
+                                $this->getFromZipArchive($this->zip, $fileWorksheetPath)
                             ),
                             null,
                             Settings::getLibXmlLoaderOptions()
@@ -403,12 +404,18 @@ private function getFromZipArchive(ZipArchive $archive, $fileName = '')
         // Sadly, some 3rd party xlsx generators don't use consistent case for filenaming
         //    so we need to load case-insensitively from the zip file
 
-        // Apache POI fixes
         $contents = $archive->getFromName($fileName, 0, ZipArchive::FL_NOCASE);
+
+        // Apache POI fixes
         if ($contents === false) {
             $contents = $archive->getFromName(substr($fileName, 1), 0, ZipArchive::FL_NOCASE);
         }
 
+        // Has the file been saved with Windoze directory separators rather than unix?
+        if ($contents === false) {
+            $contents = $archive->getFromName(str_replace('/', '\\', $fileName), 0, ZipArchive::FL_NOCASE);
+        }
+
         return ($contents === false) ? '' : $contents;
     }
 

src/PhpSpreadsheet/Shared/File.php

@@ -156,7 +156,11 @@ public static function assertFile(string $filename, string $zipMember = ''): voi
         if ($zipMember !== '') {
             $zipfile = "zip://$filename#$zipMember";
             if (!self::fileExists($zipfile)) {
-                throw new ReaderException("Could not find zip member $zipfile");
+                // Has the file been saved with Windoze directory separators rather than unix?
+                $zipfile = "zip://$filename#" . str_replace('/', '\\', $zipMember);
+                if (!self::fileExists($zipfile)) {
+                    throw new ReaderException("Could not find zip member $zipfile");
+                }
             }
         }
     }
@@ -180,6 +184,14 @@ public static function testFileNoThrow(string $filename, ?string $zipMember = nu
             return self::validateZipFirst4($filename);
         }
 
-        return self::fileExists("zip://$filename#$zipMember");
+        $zipfile = "zip://$filename#$zipMember";
+        if (self::fileExists($zipfile)) {
+            return true;
+        }
+
+        // Has the file been saved with Windoze directory separators rather than unix?
+        $zipfile = "zip://$filename#" . str_replace('/', '\\', $zipMember);
+
+        return self::fileExists($zipfile);
     }
 }

tests/PhpSpreadsheetTests/Reader/Xlsx/DirectorySeparatorTest.php

@@ -0,0 +1,71 @@
+<?php
+
+namespace PhpOffice\PhpSpreadsheetTests\Reader\Xlsx;
+
+use PhpOffice\PhpSpreadsheet\IOFactory;
+use PhpOffice\PhpSpreadsheet\Reader\Xlsx;
+use PHPUnit\Framework\TestCase;
+
+class DirectorySeparatorTest extends TestCase
+{
+    /**
+     * @dataProvider providerDirectorySeparator
+     */
+    public function testDirectorySeparatorIdentify(string $fileName): void
+    {
+        $filename = "tests/data/Reader/XLSX/{$fileName}";
+        $reader = IOFactory::identify($filename);
+
+        self::assertSame('Xlsx', $reader);
+    }
+
+    /**
+     * @dataProvider providerDirectorySeparator
+     */
+    public function testDirectorySeparatorWorksheetNames(string $fileName): void
+    {
+        $filename = "tests/data/Reader/XLSX/{$fileName}";
+        $reader = new Xlsx();
+        $sheetList = $reader->listWorksheetNames($filename);
+
+        self::assertCount(1, $sheetList);
+        self::assertSame('Sheet', $sheetList[0]);
+    }
+
+    /**
+     * @dataProvider providerDirectorySeparator
+     */
+    public function testDirectorySeparatorWorksheetInfo(string $fileName): void
+    {
+        $filename = "tests/data/Reader/XLSX/{$fileName}";
+        $reader = new Xlsx();
+        $sheetData = $reader->listWorksheetInfo($filename);
+
+        self::assertCount(1, $sheetData);
+        self::assertSame('Sheet', $sheetData[0]['worksheetName']);
+        self::assertSame(3, (int) $sheetData[0]['totalRows']);
+        self::assertSame(21, (int) $sheetData[0]['totalColumns']);
+    }
+
+    /**
+     * @dataProvider providerDirectorySeparator
+     */
+    public function testDirectorySeparatorLoad(string $fileName): void
+    {
+        $filename = "tests/data/Reader/XLSX/{$fileName}";
+        $reader = new Xlsx();
+        $spreadsheet = $reader->load($filename);
+
+        $cellValue = $spreadsheet->getActiveSheet()->getCell('A1')->getValue();
+
+        self::assertSame('Key ID', $cellValue);
+    }
+
+    public function providerDirectorySeparator(): array
+    {
+        return [
+            ['Zip-Linux-Directory-Separator.xlsx'],
+            ['Zip-Windows-Directory-Separator.xlsx'],
+        ];
+    }
+}