From f8baef148f0ffb64ad4270e85c5932ff6e63114c Mon Sep 17 00:00:00 2001
From: Salavat Salakhutdinov <salahutdinov.salavat@gmail.com>
Date: Fri, 30 Oct 2020 18:47:08 +0300
Subject: [PATCH] update xlsx reader

---
 .../Reader/XLSX/Helper/CellValueFormatter.php | 16 +++++++++++
 src/Spout/Reader/XLSX/RowIterator.php         |  2 ++
 tests/Spout/Reader/XLSX/ReaderTest.php        | 28 +++++++++++++++++--
 3 files changed, 44 insertions(+), 2 deletions(-)

diff --git a/src/Spout/Reader/XLSX/Helper/CellValueFormatter.php b/src/Spout/Reader/XLSX/Helper/CellValueFormatter.php
index 169c395..baf39a2 100644
--- a/src/Spout/Reader/XLSX/Helper/CellValueFormatter.php
+++ b/src/Spout/Reader/XLSX/Helper/CellValueFormatter.php
@@ -23,6 +23,7 @@ class CellValueFormatter
 
     /** Definition of XML nodes names used to parse data */
     const XML_NODE_VALUE = 'v';
+    const XML_NODE_FORMULA = 'f';
     const XML_NODE_INLINE_STRING_VALUE = 't';
 
     /** Definition of XML attributes used to parse data */
@@ -107,6 +108,21 @@ class CellValueFormatter
         }
     }
 
+    /**
+     * Returns the cell formula associated to the given XML node.
+     *
+     * @param \DOMNode $node
+     * @return string The formula associated with the cell
+     */
+    public function extractNodeFormula($node)
+    {
+        // for cell types having a "f" tag containing the formula.
+        // if not, the returned formula should be empty string.
+        $vNode = $node->getElementsByTagName(self::XML_NODE_FORMULA)->item(0);
+
+        return ($vNode !== null) ? $vNode->nodeValue : '';
+    }
+
     /**
      * Returns the cell's string value from a node's nested value node
      *
diff --git a/src/Spout/Reader/XLSX/RowIterator.php b/src/Spout/Reader/XLSX/RowIterator.php
index 4af4530..4395f2b 100644
--- a/src/Spout/Reader/XLSX/RowIterator.php
+++ b/src/Spout/Reader/XLSX/RowIterator.php
@@ -359,7 +359,9 @@ class RowIterator implements IteratorInterface
     {
         try {
             $cellValue = $this->cellValueFormatter->extractAndFormatNodeValue($node);
+            $cellFormula = $this->cellValueFormatter->extractNodeFormula($node);
             $cell = $this->entityFactory->createCell($cellValue);
+            $cell->setFormula($cellFormula);
         } catch (InvalidValueException $exception) {
             $cell = $this->entityFactory->createCell($exception->getInvalidValue());
             $cell->setType(Cell::TYPE_ERROR);
diff --git a/tests/Spout/Reader/XLSX/ReaderTest.php b/tests/Spout/Reader/XLSX/ReaderTest.php
index 51a7b0e..2e892c6 100644
--- a/tests/Spout/Reader/XLSX/ReaderTest.php
+++ b/tests/Spout/Reader/XLSX/ReaderTest.php
@@ -574,6 +574,29 @@ class ReaderTest extends TestCase
         $this->assertEquals($expectedRows, $allRows);
     }
 
+    /**
+     * @return void
+     */
+    public function testReadShouldReadFormulas()
+    {
+        $allRows = $this->getAllRowsForFile('sheet_with_formulas.xlsx', false, false, false);
+
+        $expectedRows = [
+            ['', '', '', ''],
+            ['', '', 'A2+B2', 'SUM(A:A)'],
+            ['', '', 'A3+B3', 'SUM(B:B)'],
+        ];
+
+        $rowsWithFormulas = array_map(function($row) {
+            $cells = $row->getCells();
+            return array_map(function($cell) {
+                return $cell->getFormula();
+            }, $cells);
+        }, $allRows);
+
+        $this->assertEquals($expectedRows, $rowsWithFormulas);
+    }
+
     /**
      * @return void
      */
@@ -703,9 +726,10 @@ class ReaderTest extends TestCase
      * @param string $fileName
      * @param bool $shouldFormatDates
      * @param bool $shouldPreserveEmptyRows
+     * @param bool $shouldConvertToArray
      * @return array All the read rows the given file
      */
-    private function getAllRowsForFile($fileName, $shouldFormatDates = false, $shouldPreserveEmptyRows = false)
+    private function getAllRowsForFile($fileName, $shouldFormatDates = false, $shouldPreserveEmptyRows = false, $shouldConvertToArray = true)
     {
         $allRows = [];
         $resourcePath = $this->getResourcePath($fileName);
@@ -717,7 +741,7 @@ class ReaderTest extends TestCase
 
         foreach ($reader->getSheetIterator() as $sheetIndex => $sheet) {
             foreach ($sheet->getRowIterator() as $rowIndex => $row) {
-                $allRows[] = $row->toArray();
+                $allRows[] = $shouldConvertToArray ? $row->toArray() : $row;
             }
         }