Merge pull request #4035 from oleibman/issue3982a

Xlsx Reader Optionally Ignore Rows With No Cells

Author: oleibman

CHANGELOG.md

@@ -9,7 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org).
 
 ### Added
 
-- Nothing
+- Xlsx Reader Optionally Ignore Rows With No Cells. [Issue #3982](https://github.com/PHPOffice/PhpSpreadsheet/issues/3982) [PR #4035](https://github.com/PHPOffice/PhpSpreadsheet/pull/4035)
 
 ### Changed
 

src/PhpSpreadsheet/Reader/BaseReader.php

@@ -39,6 +39,13 @@ abstract class BaseReader implements IReader
      */
     protected ?array $loadSheetsOnly = null;
 
+    /**
+     * Ignore rows with no cells?
+     * Identifies whether the Reader should ignore rows with no cells.
+     *        Currently implemented only for Xlsx.
+     */
+    protected bool $ignoreRowsWithNoCells = false;
+
     /**
      * IReadFilter instance.
      */
@@ -78,6 +85,18 @@ public function setReadEmptyCells(bool $readEmptyCells): self
         return $this;
     }
 
+    public function getIgnoreRowsWithNoCells(): bool
+    {
+        return $this->ignoreRowsWithNoCells;
+    }
+
+    public function setIgnoreRowsWithNoCells(bool $ignoreRowsWithNoCells): self
+    {
+        $this->ignoreRowsWithNoCells = $ignoreRowsWithNoCells;
+
+        return $this;
+    }
+
     public function getIncludeCharts(): bool
     {
         return $this->includeCharts;
@@ -150,6 +169,9 @@ protected function processFlags(int $flags): void
         if (((bool) ($flags & self::SKIP_EMPTY_CELLS) || (bool) ($flags & self::IGNORE_EMPTY_CELLS)) === true) {
             $this->setReadEmptyCells(false);
         }
+        if (((bool) ($flags & self::IGNORE_ROWS_WITH_NO_CELLS)) === true) {
+            $this->setIgnoreRowsWithNoCells(true);
+        }
     }
 
     protected function loadSpreadsheetFromFile(string $filename): Spreadsheet

src/PhpSpreadsheet/Reader/IReader.php

@@ -13,6 +13,8 @@ interface IReader
     public const SKIP_EMPTY_CELLS = 4;
     public const IGNORE_EMPTY_CELLS = 4;
 
+    public const IGNORE_ROWS_WITH_NO_CELLS = 8;
+
     public function __construct();
 
     /**

src/PhpSpreadsheet/Reader/Xlsx.php

@@ -796,10 +796,10 @@ protected function loadSpreadsheetFromFile(string $filename): Spreadsheet
                                 }
 
                                 $sheetViewOptions = new SheetViewOptions($docSheet, $xmlSheetNS);
-                                $sheetViewOptions->load($this->getReadDataOnly(), $this->styleReader);
+                                $sheetViewOptions->load($this->readDataOnly, $this->styleReader);
 
                                 (new ColumnAndRowAttributes($docSheet, $xmlSheetNS))
-                                    ->load($this->getReadFilter(), $this->getReadDataOnly());
+                                    ->load($this->getReadFilter(), $this->readDataOnly, $this->ignoreRowsWithNoCells);
                             }
 
                             $holdSelectedCells = $docSheet->getSelectedCells();

src/PhpSpreadsheet/Reader/Xlsx/ColumnAndRowAttributes.php

@@ -72,7 +72,7 @@ private function setRowAttributes(int $rowNumber, array $rowAttributes): void
         }
     }
 
-    public function load(?IReadFilter $readFilter = null, bool $readDataOnly = false): void
+    public function load(?IReadFilter $readFilter = null, bool $readDataOnly = false, bool $ignoreRowsWithNoCells = false): void
     {
         if ($this->worksheetXml === null) {
             return;
@@ -85,7 +85,7 @@ public function load(?IReadFilter $readFilter = null, bool $readDataOnly = false
         }
 
         if ($this->worksheetXml->sheetData && $this->worksheetXml->sheetData->row) {
-            $rowsAttributes = $this->readRowAttributes($this->worksheetXml->sheetData->row, $readDataOnly);
+            $rowsAttributes = $this->readRowAttributes($this->worksheetXml->sheetData->row, $readDataOnly, $ignoreRowsWithNoCells);
         }
 
         if ($readFilter !== null && $readFilter::class === DefaultReadFilter::class) {
@@ -189,13 +189,13 @@ private function isFilteredRow(IReadFilter $readFilter, int $rowCoordinate, arra
         return false;
     }
 
-    private function readRowAttributes(SimpleXMLElement $worksheetRow, bool $readDataOnly): array
+    private function readRowAttributes(SimpleXMLElement $worksheetRow, bool $readDataOnly, bool $ignoreRowsWithNoCells): array
     {
         $rowAttributes = [];
 
         foreach ($worksheetRow as $rowx) {
             $row = $rowx->attributes();
-            if ($row !== null) {
+            if ($row !== null && (!$ignoreRowsWithNoCells || isset($rowx->c))) {
                 if (isset($row['ht']) && !$readDataOnly) {
                     $rowAttributes[(int) $row['r']]['rowHeight'] = (float) $row['ht'];
                 }

tests/PhpSpreadsheetTests/Reader/Xlsx/Issue3982Test.php

@@ -0,0 +1,41 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PhpOffice\PhpSpreadsheetTests\Reader\Xlsx;
+
+use PhpOffice\PhpSpreadsheet\IOFactory;
+use PhpOffice\PhpSpreadsheet\Reader\IReader;
+use PhpOffice\PhpSpreadsheet\Reader\Xlsx as XlsxReader;
+
+class Issue3982Test extends \PHPUnit\Framework\TestCase
+{
+    private static string $testbook = 'tests/data/Reader/XLSX/issue.3982.xlsx';
+
+    public function testLoadAllRows(): void
+    {
+        $spreadsheet = IOFactory::load(self::$testbook);
+        $sheet = $spreadsheet->getActiveSheet();
+        $data = $sheet->toArray(null, true, false, true);
+        self::assertCount(1048576, $data);
+        $spreadsheet->disconnectWorksheets();
+    }
+
+    public function testIgnoreCellsWithNoRows(): void
+    {
+        $spreadsheet = IOFactory::load(self::$testbook, IReader::IGNORE_ROWS_WITH_NO_CELLS);
+        $sheet = $spreadsheet->getActiveSheet();
+        $data = $sheet->toArray(null, true, false, true);
+        self::assertSame([1, 2, 3, 4, 5, 6], array_keys($data));
+        $spreadsheet->disconnectWorksheets();
+    }
+
+    public function testDefaultSetting(): void
+    {
+        $reader = new XlsxReader();
+        self::assertFalse($reader->getIgnoreRowsWithNoCells());
+        self::assertFalse($reader->getReadDataOnly());
+        self::assertFalse($reader->getIncludeCharts());
+        self::assertTrue($reader->getReadEmptyCells());
+    }
+}