[Feature] Html reader improvements (#884)

* Extract character set, so we can convert to UTF-8 if required

* Set column width and row height when defined on tr/td

* Parse align and valign on td

* Specify number format of cell via html attribute

* Formatting of b, strong, i and em tags

* Inserting image in cell when using img tag in html

* Add applying inline styles: border, fonts, alignment, dimensions

* Add tests for applying inline styles

CHANGELOG.md

@@ -5,6 +5,12 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com)
 and this project adheres to [Semantic Versioning](https://semver.org).
 
+## [Unreleased] - 
+
+### Added
+
+- Added support for inline styles in Html reader (borders, alignment, width, height)
+
 ## [1.6.0] - 2019-01-02
 
 ### Added

src/PhpSpreadsheet/Reader/Html.php

@@ -12,6 +12,9 @@ use PhpOffice\PhpSpreadsheet\Spreadsheet;
 use PhpOffice\PhpSpreadsheet\Style\Border;
 use PhpOffice\PhpSpreadsheet\Style\Color;
 use PhpOffice\PhpSpreadsheet\Style\Fill;
+use PhpOffice\PhpSpreadsheet\Style\Font;
+use PhpOffice\PhpSpreadsheet\Style\Style;
+use PhpOffice\PhpSpreadsheet\Worksheet\Drawing;
 use PhpOffice\PhpSpreadsheet\Worksheet\Worksheet;
 
 /** PhpSpreadsheet root directory */
@@ -96,6 +99,26 @@ class Html extends BaseReader
                 ],
             ],
         ], //    Bottom border
+        'strong' => [
+            'font' => [
+                'bold' => true,
+            ],
+        ], //    Bold
+        'b' => [
+            'font' => [
+                'bold' => true,
+            ],
+        ], //    Bold
+        'i' => [
+            'font' => [
+                'italic' => true,
+            ],
+        ], //    Italic
+        'em' => [
+            'font' => [
+                'italic' => true,
+            ],
+        ], //    Italic
     ];
 
     protected $rowspan = [];
@@ -295,11 +318,9 @@ class Html extends BaseReader
                 switch ($child->nodeName) {
                     case 'meta':
                         foreach ($attributeArray as $attributeName => $attributeValue) {
-                            switch ($attributeName) {
-                                case 'content':
-                                    //    TODO
                             // Extract character set, so we can convert to UTF-8 if required
-                                    break;
+                            if ($attributeName === 'charset') {
+                                $this->setInputEncoding($attributeValue);
                             }
                         }
                         $this->processDomElement($child, $sheet, $row, $column, $cellContent);
@@ -334,6 +355,10 @@ class Html extends BaseReader
                             $cellContent .= ' ';
                         }
 
+                        if (isset($this->formats[$child->nodeName])) {
+                            $sheet->getStyle($column . $row)->applyFromArray($this->formats[$child->nodeName]);
+                        }
+
                         break;
                     case 'hr':
                         $this->flushCell($sheet, $column, $row, $cellContent);
@@ -423,6 +448,10 @@ class Html extends BaseReader
                             $column = 'A';
                         }
 
+                        break;
+                    case 'img':
+                        $this->insertImage($sheet, $column, $row, $attributeArray);
+
                         break;
                     case 'table':
                         $this->flushCell($sheet, $column, $row, $cellContent);
@@ -448,6 +477,11 @@ class Html extends BaseReader
                         $column = $this->getTableStartColumn();
                         $cellContent = '';
                         $this->processDomElement($child, $sheet, $row, $column, $cellContent);
+
+                        if (isset($attributeArray['height'])) {
+                            $sheet->getRowDimension($row)->setRowHeight($attributeArray['height']);
+                        }
+
                         ++$row;
 
                         break;
@@ -501,6 +535,27 @@ class Html extends BaseReader
                                 ]
                             );
                         }
+
+                        if (isset($attributeArray['width'])) {
+                            $sheet->getColumnDimension($column)->setWidth($attributeArray['width']);
+                        }
+
+                        if (isset($attributeArray['height'])) {
+                            $sheet->getRowDimension($row)->setRowHeight($attributeArray['height']);
+                        }
+
+                        if (isset($attributeArray['align'])) {
+                            $sheet->getStyle($column . $row)->getAlignment()->setHorizontal($attributeArray['align']);
+                        }
+
+                        if (isset($attributeArray['valign'])) {
+                            $sheet->getStyle($column . $row)->getAlignment()->setVertical($attributeArray['valign']);
+                        }
+
+                        if (isset($attributeArray['data-format'])) {
+                            $sheet->getStyle($column . $row)->getNumberFormat()->setFormatCode($attributeArray['data-format']);
+                        }
+
                         ++$column;
 
                         break;
@@ -608,36 +663,271 @@ class Html extends BaseReader
             return;
         }
 
-        $supported_styles = ['background-color', 'color'];
+        $cellStyle = $sheet->getStyle($column . $row);
 
         // add color styles (background & text) from dom element,currently support : td & th, using ONLY inline css style with RGB color
         $styles = explode(';', $attributeArray['style']);
         foreach ($styles as $st) {
             $value = explode(':', $st);
+            $styleName = isset($value[0]) ? trim($value[0]) : null;
+            $styleValue = isset($value[1]) ? trim($value[1]) : null;
 
-            if (empty(trim($value[0])) || !in_array(trim($value[0]), $supported_styles)) {
+            if (!$styleName) {
                 continue;
             }
 
-            //check if has #, so we can get clean hex
+            switch ($styleName) {
-            if (substr(trim($value[1]), 0, 1) == '#') {
+                case 'background':
-                $style_color = substr(trim($value[1]), 1);
-            }
-
-            if (empty($style_color)) {
-                continue;
-            }
-
-            switch (trim($value[0])) {
                 case 'background-color':
-                    $sheet->getStyle($column . $row)->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => "{$style_color}"]]]);
+                    $styleColor = $this->getStyleColor($styleValue);
+
+                    if (!$styleColor) {
+                        continue 2;
+                    }
+
+                    $cellStyle->applyFromArray(['fill' => ['fillType' => Fill::FILL_SOLID, 'color' => ['rgb' => $styleColor]]]);
 
                     break;
                 case 'color':
-                    $sheet->getStyle($column . $row)->applyFromArray(['font' => ['color' => ['rgb' => "{$style_color}"]]]);
+                    $styleColor = $this->getStyleColor($styleValue);
+
+                    if (!$styleColor) {
+                        continue 2;
+                    }
+
+                    $cellStyle->applyFromArray(['font' => ['color' => ['rgb' => $styleColor]]]);
+
+                    break;
+
+                case 'border':
+                    $this->setBorderStyle($cellStyle, $styleValue, 'allBorders');
+
+                    break;
+
+                case 'border-top':
+                    $this->setBorderStyle($cellStyle, $styleValue, 'top');
+
+                    break;
+
+                case 'border-bottom':
+                    $this->setBorderStyle($cellStyle, $styleValue, 'bottom');
+
+                    break;
+
+                case 'border-left':
+                    $this->setBorderStyle($cellStyle, $styleValue, 'left');
+
+                    break;
+
+                case 'border-right':
+                    $this->setBorderStyle($cellStyle, $styleValue, 'right');
+
+                    break;
+
+                case 'font-size':
+                    $cellStyle->getFont()->setSize(
+                        (float) $styleValue
+                    );
+
+                    break;
+
+                case 'font-weight':
+                    if ($styleValue === 'bold' || $styleValue >= 500) {
+                        $cellStyle->getFont()->setBold(true);
+                    }
+
+                    break;
+
+                case 'font-style':
+                    if ($styleValue === 'italic') {
+                        $cellStyle->getFont()->setItalic(true);
+                    }
+
+                    break;
+
+                case 'font-family':
+                    $cellStyle->getFont()->setName(str_replace('\'', '', $styleValue));
+
+                    break;
+
+                case 'text-decoration':
+                    switch ($styleValue) {
+                        case 'underline':
+                            $cellStyle->getFont()->setUnderline(Font::UNDERLINE_SINGLE);
+
+                            break;
+                        case 'line-through':
+                            $cellStyle->getFont()->setStrikethrough(true);
+
+                            break;
+                    }
+
+                    break;
+
+                case 'text-align':
+                    $cellStyle->getAlignment()->setHorizontal($styleValue);
+
+                    break;
+
+                case 'vertical-align':
+                    $cellStyle->getAlignment()->setVertical($styleValue);
+
+                    break;
+
+                case 'width':
+                    $sheet->getColumnDimension($column)->setWidth(
+                        str_replace('px', '', $styleValue)
+                    );
+
+                    break;
+
+                case 'height':
+                    $sheet->getRowDimension($row)->setRowHeight(
+                        str_replace('px', '', $styleValue)
+                    );
+
+                    break;
+
+                case 'word-wrap':
+                    $cellStyle->getAlignment()->setWrapText(
+                        $styleValue === 'break-word'
+                    );
+
+                    break;
+
+                case 'text-indent':
+                    $cellStyle->getAlignment()->setIndent(
+                        (int) str_replace(['px'], '', $styleValue)
+                    );
 
                     break;
             }
         }
     }
+
+    /**
+     * Check if has #, so we can get clean hex.
+     *
+     * @param $value
+     *
+     * @return null|string
+     */
+    public function getStyleColor($value)
+    {
+        if (strpos($value, '#') === 0) {
+            return substr($value, 1);
+        }
+
+        return null;
+    }
+
+    /**
+     * @param Worksheet $sheet
+     * @param string    $column
+     * @param int       $row
+     * @param array     $attributes
+     *
+     * @throws \PhpOffice\PhpSpreadsheet\Exception
+     */
+    private function insertImage(Worksheet $sheet, $column, $row, array $attributes)
+    {
+        if (!isset($attributes['src'])) {
+            return;
+        }
+
+        $src = urldecode($attributes['src']);
+        $width = isset($attributes['width']) ? (float) $attributes['width'] : null;
+        $height = isset($attributes['height']) ? (float) $attributes['height'] : null;
+        $name = isset($attributes['alt']) ? (float) $attributes['alt'] : null;
+
+        $drawing = new Drawing();
+        $drawing->setPath($src);
+        $drawing->setWorksheet($sheet);
+        $drawing->setCoordinates($column . $row);
+        $drawing->setOffsetX(0);
+        $drawing->setOffsetY(10);
+        $drawing->setResizeProportional(true);
+
+        if ($name) {
+            $drawing->setName($name);
+        }
+
+        if ($width) {
+            $drawing->setWidth((int) $width);
+        }
+
+        if ($height) {
+            $drawing->setHeight((int) $height);
+        }
+
+        $sheet->getColumnDimension($column)->setWidth(
+            $drawing->getWidth() / 6
+        );
+
+        $sheet->getRowDimension($row)->setRowHeight(
+            $drawing->getHeight() * 0.9
+        );
+    }
+
+    /**
+     * Map html border style to PhpSpreadsheet border style.
+     *
+     * @param  string $style
+     *
+     * @return null|string
+     */
+    public function getBorderStyle($style)
+    {
+        switch ($style) {
+            case 'solid':
+                return Border::BORDER_THIN;
+            case 'dashed':
+                return Border::BORDER_DASHED;
+            case 'dotted':
+                return Border::BORDER_DOTTED;
+            case 'medium':
+                return Border::BORDER_MEDIUM;
+            case 'thick':
+                return Border::BORDER_THICK;
+            case 'none':
+                return Border::BORDER_NONE;
+            case 'dash-dot':
+                return Border::BORDER_DASHDOT;
+            case 'dash-dot-dot':
+                return Border::BORDER_DASHDOTDOT;
+            case 'double':
+                return Border::BORDER_DOUBLE;
+            case 'hair':
+                return Border::BORDER_HAIR;
+            case 'medium-dash-dot':
+                return Border::BORDER_MEDIUMDASHDOT;
+            case 'medium-dash-dot-dot':
+                return Border::BORDER_MEDIUMDASHDOTDOT;
+            case 'medium-dashed':
+                return Border::BORDER_MEDIUMDASHED;
+            case 'slant-dash-dot':
+                return Border::BORDER_SLANTDASHDOT;
+        }
+
+        return null;
+    }
+
+    /**
+     * @param Style  $cellStyle
+     * @param string $styleValue
+     * @param string $type
+     */
+    private function setBorderStyle(Style $cellStyle, $styleValue, $type)
+    {
+        list(, $borderStyle, $color) = explode(' ', $styleValue);
+
+        $cellStyle->applyFromArray([
+            'borders' => [
+                $type => [
+                    'borderStyle' => $this->getBorderStyle($borderStyle),
+                    'color' => ['rgb' => $this->getStyleColor($color)],
+                ],
+            ],
+        ]);
+    }
 }

tests/PhpSpreadsheetTests/Reader/HtmlTest.php

@@ -3,6 +3,10 @@
 namespace PhpOffice\PhpSpreadsheetTests\Reader;
 
 use PhpOffice\PhpSpreadsheet\Reader\Html;
+use PhpOffice\PhpSpreadsheet\Style\Alignment;
+use PhpOffice\PhpSpreadsheet\Style\Border;
+use PhpOffice\PhpSpreadsheet\Style\Font;
+use PhpOffice\PhpSpreadsheet\Worksheet\Drawing;
 use PHPUnit\Framework\TestCase;
 
 class HtmlTest extends TestCase
@@ -34,14 +38,13 @@ class HtmlTest extends TestCase
      */
     public function testCanReadVerySmallFile($expected, $content)
     {
-        $filename = tempnam(sys_get_temp_dir(), 'html');
+        $filename = $this->createHtml($content);
-        file_put_contents($filename, $content);
-
         $reader = new Html();
         $actual = $reader->canRead($filename);
-        unlink($filename);
 
         self::assertSame($expected, $actual);
+
+        unlink($filename);
     }
 
     public function testBackgroundColorInRanding()
@@ -51,14 +54,238 @@ class HtmlTest extends TestCase
                         <td style="background-color: #000000;color: #FFFFFF">Blue background</td>
                     </tr>
                 </table>';
-        $filename = tempnam(sys_get_temp_dir(), 'html');
+        $filename = $this->createHtml($html);
-        file_put_contents($filename, $html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
-        $reader = new Html();
-        $spreadsheet = $reader->load($filename);
         $firstSheet = $spreadsheet->getSheet(0);
         $style = $firstSheet->getCell('A1')->getStyle();
 
         self::assertEquals('FFFFFF', $style->getFont()->getColor()->getRGB());
+
         unlink($filename);
     }
+
+    public function testCanApplyInlineBordersStyles()
+    {
+        $html = '<table>
+                    <tr>
+                        <td style="border: 1px solid #333333;">Thin border</td>
+                        <td style="border-bottom: 1px solid #333333;">Border bottom</td>
+                        <td style="border-top: 1px solid #333333;">Border top</td>
+                        <td style="border-left: 1px solid #333333;">Border left</td>
+                        <td style="border-right: 1px solid #333333;">Border right</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+        $style = $firstSheet->getCell('A1')->getStyle();
+        $borders = $style->getBorders();
+
+        /** @var Border $border */
+        foreach ([$borders->getTop(), $borders->getBottom(), $borders->getLeft(), $borders->getRight()] as $border) {
+            self::assertEquals('333333', $border->getColor()->getRGB());
+            self::assertEquals(Border::BORDER_THIN, $border->getBorderStyle());
+        }
+
+        $style = $firstSheet->getCell('B1')->getStyle();
+        $border = $style->getBorders()->getBottom();
+        self::assertEquals('333333', $border->getColor()->getRGB());
+        self::assertEquals(Border::BORDER_THIN, $border->getBorderStyle());
+
+        $style = $firstSheet->getCell('C1')->getStyle();
+        $border = $style->getBorders()->getTop();
+        self::assertEquals('333333', $border->getColor()->getRGB());
+        self::assertEquals(Border::BORDER_THIN, $border->getBorderStyle());
+
+        $style = $firstSheet->getCell('D1')->getStyle();
+        $border = $style->getBorders()->getLeft();
+        self::assertEquals('333333', $border->getColor()->getRGB());
+        self::assertEquals(Border::BORDER_THIN, $border->getBorderStyle());
+
+        $style = $firstSheet->getCell('E1')->getStyle();
+        $border = $style->getBorders()->getRight();
+        self::assertEquals('333333', $border->getColor()->getRGB());
+        self::assertEquals(Border::BORDER_THIN, $border->getBorderStyle());
+
+        unlink($filename);
+    }
+
+    public function testCanApplyInlineFontStyles()
+    {
+        $html = '<table>
+                    <tr>
+                        <td style="font-size: 16px;">16px</td>
+                        <td style="font-family: \'Times New Roman\'">Times New Roman</td>
+                        <td style="font-weight: bold;">Bold</td>
+                        <td style="font-style: italic;">Italic</td>
+                        <td style="text-decoration: underline;">Underline</td>
+                        <td style="text-decoration: line-through;">Line through</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        $style = $firstSheet->getCell('A1')->getStyle();
+        self::assertEquals(16, $style->getFont()->getSize());
+
+        $style = $firstSheet->getCell('B1')->getStyle();
+        self::assertEquals('Times New Roman', $style->getFont()->getName());
+
+        $style = $firstSheet->getCell('C1')->getStyle();
+        self::assertTrue($style->getFont()->getBold());
+
+        $style = $firstSheet->getCell('D1')->getStyle();
+        self::assertTrue($style->getFont()->getItalic());
+
+        $style = $firstSheet->getCell('E1')->getStyle();
+        self::assertEquals(Font::UNDERLINE_SINGLE, $style->getFont()->getUnderline());
+
+        $style = $firstSheet->getCell('F1')->getStyle();
+        self::assertTrue($style->getFont()->getStrikethrough());
+
+        unlink($filename);
+    }
+
+    public function testCanApplyInlineWidth()
+    {
+        $html = '<table>
+                    <tr>
+                        <td width="50">50px</td>
+                        <td style="width: 100px;">100px</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        $dimension = $firstSheet->getColumnDimension('A');
+        self::assertEquals(50, $dimension->getWidth());
+
+        $dimension = $firstSheet->getColumnDimension('B');
+        self::assertEquals(100, $dimension->getWidth());
+
+        unlink($filename);
+    }
+
+    public function testCanApplyInlineHeight()
+    {
+        $html = '<table>
+                    <tr>
+                        <td height="50">1</td>
+                    </tr>
+                    <tr>
+                        <td style="height: 100px;">2</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        $dimension = $firstSheet->getRowDimension(1);
+        self::assertEquals(50, $dimension->getRowHeight());
+
+        $dimension = $firstSheet->getRowDimension(2);
+        self::assertEquals(100, $dimension->getRowHeight());
+
+        unlink($filename);
+    }
+
+    public function testCanApplyAlignment()
+    {
+        $html = '<table>
+                    <tr>
+                        <td align="center">Center align</td>
+                        <td valign="center">Center valign</td>
+                        <td style="text-align: center;">Center align</td>
+                        <td style="vertical-align: center;">Center valign</td>
+                        <td style="text-indent: 10px;">Text indent</td>
+                        <td style="word-wrap: break-word;">Wraptext</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        $style = $firstSheet->getCell('A1')->getStyle();
+        self::assertEquals(Alignment::HORIZONTAL_CENTER, $style->getAlignment()->getHorizontal());
+
+        $style = $firstSheet->getCell('B1')->getStyle();
+        self::assertEquals(Alignment::VERTICAL_CENTER, $style->getAlignment()->getVertical());
+
+        $style = $firstSheet->getCell('C1')->getStyle();
+        self::assertEquals(Alignment::HORIZONTAL_CENTER, $style->getAlignment()->getHorizontal());
+
+        $style = $firstSheet->getCell('D1')->getStyle();
+        self::assertEquals(Alignment::VERTICAL_CENTER, $style->getAlignment()->getVertical());
+
+        $style = $firstSheet->getCell('E1')->getStyle();
+        self::assertEquals(10, $style->getAlignment()->getIndent());
+
+        $style = $firstSheet->getCell('F1')->getStyle();
+        self::assertTrue($style->getAlignment()->getWrapText());
+
+        unlink($filename);
+    }
+
+    public function testCanApplyInlineDataFormat()
+    {
+        $html = '<table>
+                    <tr>
+                        <td data-format="mmm-yy">2019-02-02 12:34:00</td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        $style = $firstSheet->getCell('A1')->getStyle();
+        self::assertEquals('mmm-yy', $style->getNumberFormat()->getFormatCode());
+
+        unlink($filename);
+    }
+
+    public function testCanInsertImage()
+    {
+        $imagePath = realpath(__DIR__ . '/../../data/Reader/HTML/image.jpg');
+
+        $html = '<table>
+                    <tr>
+                        <td><img src="' . $imagePath . '" alt=""></td>
+                    </tr>
+                </table>';
+        $filename = $this->createHtml($html);
+        $spreadsheet = $this->loadHtmlIntoSpreadsheet($filename);
+        $firstSheet = $spreadsheet->getSheet(0);
+
+        /** @var Drawing $drawing */
+        $drawing = $firstSheet->getDrawingCollection()[0];
+        self::assertEquals($imagePath, $drawing->getPath());
+        self::assertEquals('A1', $drawing->getCoordinates());
+
+        unlink($filename);
+    }
+
+    /**
+     * @param string $html
+     *
+     * @return string
+     */
+    private function createHtml($html)
+    {
+        $filename = tempnam(sys_get_temp_dir(), 'html');
+        file_put_contents($filename, $html);
+
+        return $filename;
+    }
+
+    /**
+     * @param $filename
+     *
+     * @return \PhpOffice\PhpSpreadsheet\Spreadsheet
+     */
+    private function loadHtmlIntoSpreadsheet($filename)
+    {
+        return (new Html())->load($filename);
+    }
 }