25414441731

Committed 06 May 2026 03:12AM UTC coverage: 94.696% (-2.1%) from 96.757%

Build # 25414441731

Build Type

Pull #2874

github

Committed by

web-flow

Commit Message

Merge 3fa85d6e3 into 0ab0b4940

Pull Request Pull Request #2874: Addresses issue #12

Coverage Stats

135 of 420 new or added lines in 30 files covered. (32.14%)

1 existing line in 1 file now uncovered.

12588 of 13293 relevant lines covered (94.7%)

34.84 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

71.68

/src/PhpWord/Reader/ODText/Content.php

<?php

/**
 * This file is part of PHPWord - A pure PHP library for reading and writing
 * word processing documents.
 *
 * PHPWord is free software distributed under the terms of the GNU Lesser
 * General Public License version 3 as published by the Free Software Foundation.
 *
 * For the full copyright and license information, please read the LICENSE
 * file that was distributed with this source code. For the full list of
 * contributors, visit https://github.com/PHPOffice/PHPWord/contributors.
 *
 * @see         https://github.com/PHPOffice/PHPWord
 *
 * @license     http://www.gnu.org/licenses/lgpl.txt LGPL version 3
 */

namespace PhpOffice\PhpWord\Reader\ODText;

use DateTime;
use DOMElement;
use DOMNodeList;
use PhpOffice\Math\Reader\MathML;
use PhpOffice\PhpWord\Element\Section;
use PhpOffice\PhpWord\Element\TrackChange;
use PhpOffice\PhpWord\PhpWord;
use PhpOffice\PhpWord\Shared\XMLReader;

/**
 * Content reader.
 *
 * @since 0.10.0
 */
class Content extends AbstractPart
{
    /** @var ?Section */
    private $section;

    /**
     * Read content.xml.
     */
    public function read(PhpWord $phpWord): void
    {
        $xmlReader = new XMLReader();
        $xmlReader->getDomFromZip($this->docFile, $this->xmlFile);

        $nodes = $xmlReader->getElements('office:body/office:text/*');
        $this->section = null;
        $this->processNodes($nodes, $xmlReader, $phpWord);
        $this->section = null;
    }

    /** @param DOMNodeList<DOMElement> $nodes */
    public function processNodes(DOMNodeList $nodes, XMLReader $xmlReader, PhpWord $phpWord): void
    {
        if ($nodes->length > 0) {
            foreach ($nodes as $node) {
                // $styleName = $xmlReader->getAttribute('text:style-name', $node);
                switch ($node->nodeName) {
                    case 'text:h': // Heading
                        $depth = $xmlReader->getAttribute('text:outline-level', $node);
                        $this->getSection($phpWord)->addTitle($node->nodeValue, $depth);

                        break;
                    case 'text:p': // Paragraph
                        $styleName = $xmlReader->getAttribute('text:style-name', $node);                                                
                        if (substr($styleName, 0, 2) === 'SB') {
                            break;                                                                                                      
                        }           

                        // Track page number via soft page break hints (LibreOffice equivalent of lastRenderedPageBreak)                
                        foreach ($node->childNodes as $pageBreakChild) {
                            if ($pageBreakChild->nodeName === 'text:soft-page-break') {                                                 
                                ++$this->currentPage;
                                break;                                                                                                  
                            }       
                        }

                        // Caption: paragraph contains a text:sequence element (ODT auto-numbered figure/table labels)                  
                        $sequenceNode = $xmlReader->getElement('text:sequence', $node);
                        if ($sequenceNode !== null) {
                            $label = $sequenceNode->getAttribute('text:name') ?: 'Figure';
                            $text = trim($node->nodeValue);
                            if ($text !== '') {
                                $this->getSection($phpWord)->addCaption($label, $text, null, null, $this->currentPage);
                            }
                            break;
                        }
                                    
                        $element = $xmlReader->getElement('draw:frame/draw:object', $node);  
                        if ($element) {
                            $mathFile = str_replace('./', '', $element->getAttribute('xlink:href')) . '/content.xml';

                            $xmlReaderObject = new XMLReader();
                            $mathElement = $xmlReaderObject->getDomFromZip($this->docFile, $mathFile);
                            if ($mathElement) {
                                $mathXML = $mathElement->saveXML($mathElement);

                                if (is_string($mathXML)) {
                                    $reader = new MathML();
                                    $math = $reader->read($mathXML);

                                    $this->getSection($phpWord)->addFormula($math);
                                }
                            }
                        } else {
                            $children = $node->childNodes;
                            $spans = false;
                            /** @var DOMElement $child */
                            foreach ($children as $child) {
                                switch ($child->nodeName) {
                                    case 'text:change-start':
                                        $changeId = $child->getAttribute('text:change-id');
                                        if (isset($trackedChanges[$changeId])) {
                                            $changed = $trackedChanges[$changeId];
                                        }

                                        break;
                                    case 'text:change-end':
                                        unset($changed);

                                        break;
                                    case 'text:change':
                                        $changeId = $child->getAttribute('text:change-id');
                                        if (isset($trackedChanges[$changeId])) {
                                            $changed = $trackedChanges[$changeId];
                                        }

                                        break;
                                    case 'text:span':
                                        $spans = true;

                                        break;
                                }
                            }

                            if ($spans) {
                                $element = $this->getSection($phpWord)->addTextRun();
                                foreach ($children as $child) {
                                    switch ($child->nodeName) {
                                        case 'text:span':
                                            /** @var DOMElement $child2 */
                                            foreach ($child->childNodes as $child2) {
                                                switch ($child2->nodeName) {
                                                    case '#text':
                                                        $element->addText($child2->nodeValue);

                                                        break;
                                                    case 'text:tab':
                                                        $element->addText("\t");

                                                        break;
                                                    case 'text:s':
                                                        $spaces = (int) $child2->getAttribute('text:c') ?: 1;
                                                        $element->addText(str_repeat(' ', $spaces));

                                                        break;
                                                }
                                            }

                                            break;
                                    }
                                }
                            } else {
                                $element = $this->getSection($phpWord)->addText($node->nodeValue);
                            }
                            if (isset($changed) && is_array($changed)) {
                                $element->setTrackChange($changed['changed']);
                                if (isset($changed['textNodes'])) {
                                    foreach ($changed['textNodes'] as $changedNode) {
                                        $element = $this->getSection($phpWord)->addText($changedNode->nodeValue);
                                        $element->setTrackChange($changed['changed']);
                                    }
                                }
                            }
                        }

                        break;
                    case 'text:list': // List
                        $listItems = $xmlReader->getElements('text:list-item/text:p', $node);
                        foreach ($listItems as $listItem) {
                            // $listStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
                            $this->getSection($phpWord)->addListItem($listItem->nodeValue, 0);
                        }

                        break;
                    case 'text:tracked-changes':
                        $changedRegions = $xmlReader->getElements('text:changed-region', $node);
                        foreach ($changedRegions as $changedRegion) {
                            $type = ($changedRegion->firstChild->nodeName == 'text:insertion') ? TrackChange::INSERTED : TrackChange::DELETED;
                            $creatorNode = $xmlReader->getElements('office:change-info/dc:creator', $changedRegion->firstChild);
                            $author = $creatorNode[0]->nodeValue;
                            $dateNode = $xmlReader->getElements('office:change-info/dc:date', $changedRegion->firstChild);
                            $date = $dateNode[0]->nodeValue;
                            $date = preg_replace('/\.\d+$/', '', $date);
                            $date = DateTime::createFromFormat('Y-m-d\TH:i:s', $date);
                            $changed = new TrackChange($type, $author, $date);
                            $textNodes = $xmlReader->getElements('text:deletion/text:p', $changedRegion);
                            $trackedChanges[$changedRegion->getAttribute('text:id')] = ['changed' => $changed, 'textNodes' => $textNodes];
                        }

                        break;
                    case 'text:section': // Section
                        // $sectionStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
                        $this->section = $phpWord->addSection();
                        /** @var DOMNodeList<DOMElement> $children */
                        $children = $node->childNodes;
                        $this->processNodes($children, $xmlReader, $phpWord);

                        break;
                }
            }
        }
    }

    private function getSection(PhpWord $phpWord): Section
    {
        $section = $this->section;
        if ($section === null) {
            $section = $this->section = $phpWord->addSection();
        }

        return $section;
    }
}

1	<?php
2
3	/**
4	* This file is part of PHPWord - A pure PHP library for reading and writing
5	* word processing documents.
6	*
7	* PHPWord is free software distributed under the terms of the GNU Lesser
8	* General Public License version 3 as published by the Free Software Foundation.
9	*
10	* For the full copyright and license information, please read the LICENSE
11	* file that was distributed with this source code. For the full list of
12	* contributors, visit https://github.com/PHPOffice/PHPWord/contributors.
13	*
14	* @see https://github.com/PHPOffice/PHPWord
15	*
16	* @license http://www.gnu.org/licenses/lgpl.txt LGPL version 3
17	*/
18
19	namespace PhpOffice\PhpWord\Reader\ODText;
20
21	use DateTime;
22	use DOMElement;
23	use DOMNodeList;
24	use PhpOffice\Math\Reader\MathML;
25	use PhpOffice\PhpWord\Element\Section;
26	use PhpOffice\PhpWord\Element\TrackChange;
27	use PhpOffice\PhpWord\PhpWord;
28	use PhpOffice\PhpWord\Shared\XMLReader;
29
30	/**
31	* Content reader.
32	*
33	* @since 0.10.0
34	*/
35	class Content extends AbstractPart
36	{
37	/** @var ?Section */
38	private $section;
39
40	/**
41	* Read content.xml.
42	*/
43	public function read(PhpWord $phpWord): void
44	{
45	$xmlReader = new XMLReader();	5✔
46	$xmlReader->getDomFromZip($this->docFile, $this->xmlFile);	5✔
47
48	$nodes = $xmlReader->getElements('office:body/office:text/*');	5✔
49	$this->section = null;	5✔
50	$this->processNodes($nodes, $xmlReader, $phpWord);	5✔
51	$this->section = null;	5✔
52	}
53
54	/** @param DOMNodeList<DOMElement> $nodes */
55	public function processNodes(DOMNodeList $nodes, XMLReader $xmlReader, PhpWord $phpWord): void
56	{
57	if ($nodes->length > 0) {	5✔
58	foreach ($nodes as $node) {	5✔
59	// $styleName = $xmlReader->getAttribute('text:style-name', $node);
60	switch ($node->nodeName) {	5✔
61	case 'text:h': // Heading	5✔
62	$depth = $xmlReader->getAttribute('text:outline-level', $node);	1✔
63	$this->getSection($phpWord)->addTitle($node->nodeValue, $depth);	1✔
64
65	break;	1✔
66	case 'text:p': // Paragraph	5✔
67	$styleName = $xmlReader->getAttribute('text:style-name', $node);	5✔
68	if (substr($styleName, 0, 2) === 'SB') {	5✔
69	break;	2✔
70	}
71
72	// Track page number via soft page break hints (LibreOffice equivalent of lastRenderedPageBreak)
73	foreach ($node->childNodes as $pageBreakChild) {	5✔
74	if ($pageBreakChild->nodeName === 'text:soft-page-break') {	5✔
NEW 75	++$this->currentPage;	×
NEW 76	break;	×
77	}
78	}
79
80	// Caption: paragraph contains a text:sequence element (ODT auto-numbered figure/table labels)
81	$sequenceNode = $xmlReader->getElement('text:sequence', $node);	5✔
82	if ($sequenceNode !== null) {	5✔
NEW 83	$label = $sequenceNode->getAttribute('text:name') ?: 'Figure';	×
NEW 84	$text = trim($node->nodeValue);	×
NEW 85	if ($text !== '') {	×
NEW 86	$this->getSection($phpWord)->addCaption($label, $text, null, null, $this->currentPage);	×
87	}
UNCOV 88	break;	×
89	}
90
91	$element = $xmlReader->getElement('draw:frame/draw:object', $node);	5✔
92	if ($element) {	5✔
93	$mathFile = str_replace('./', '', $element->getAttribute('xlink:href')) . '/content.xml';	1✔
94
95	$xmlReaderObject = new XMLReader();	1✔
96	$mathElement = $xmlReaderObject->getDomFromZip($this->docFile, $mathFile);	1✔
97	if ($mathElement) {	1✔
98	$mathXML = $mathElement->saveXML($mathElement);	1✔
99
100	if (is_string($mathXML)) {	1✔
101	$reader = new MathML();	1✔
102	$math = $reader->read($mathXML);	1✔
103
104	$this->getSection($phpWord)->addFormula($math);	1✔
105	}
106	}
107	} else {
108	$children = $node->childNodes;	4✔
109	$spans = false;	4✔
110	/** @var DOMElement $child */
111	foreach ($children as $child) {	4✔
112	switch ($child->nodeName) {	4✔
113	case 'text:change-start':	4✔
114	$changeId = $child->getAttribute('text:change-id');	×
115	if (isset($trackedChanges[$changeId])) {	×
116	$changed = $trackedChanges[$changeId];	×
117	}
118
119	break;	×
120	case 'text:change-end':	4✔
121	unset($changed);	×
122
123	break;	×
124	case 'text:change':	4✔
125	$changeId = $child->getAttribute('text:change-id');	×
126	if (isset($trackedChanges[$changeId])) {	×
127	$changed = $trackedChanges[$changeId];	×
128	}
129
130	break;	×
131	case 'text:span':	4✔
132	$spans = true;	3✔
133
134	break;	3✔
135	}
136	}
137
138	if ($spans) {	4✔
139	$element = $this->getSection($phpWord)->addTextRun();	3✔
140	foreach ($children as $child) {	3✔
141	switch ($child->nodeName) {	3✔
142	case 'text:span':	3✔
143	/** @var DOMElement $child2 */
144	foreach ($child->childNodes as $child2) {	3✔
145	switch ($child2->nodeName) {	3✔
146	case '#text':	3✔
147	$element->addText($child2->nodeValue);	3✔
148
149	break;	3✔
150	case 'text:tab':	1✔
151	$element->addText("\t");	1✔
152
153	break;	1✔
154	case 'text:s':	1✔
155	$spaces = (int) $child2->getAttribute('text:c') ?: 1;	1✔
156	$element->addText(str_repeat(' ', $spaces));	1✔
157
158	break;	1✔
159	}
160	}
161
162	break;	3✔
163	}
164	}
165	} else {
166	$element = $this->getSection($phpWord)->addText($node->nodeValue);	2✔
167	}
168	if (isset($changed) && is_array($changed)) {	4✔
169	$element->setTrackChange($changed['changed']);	×
170	if (isset($changed['textNodes'])) {	×
171	foreach ($changed['textNodes'] as $changedNode) {	×
172	$element = $this->getSection($phpWord)->addText($changedNode->nodeValue);	×
173	$element->setTrackChange($changed['changed']);	×
174	}
175	}
176	}
177	}
178
179	break;	5✔
180	case 'text:list': // List	4✔
181	$listItems = $xmlReader->getElements('text:list-item/text:p', $node);	1✔
182	foreach ($listItems as $listItem) {	1✔
183	// $listStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
184	$this->getSection($phpWord)->addListItem($listItem->nodeValue, 0);	1✔
185	}
186
187	break;	1✔
188	case 'text:tracked-changes':	4✔
189	$changedRegions = $xmlReader->getElements('text:changed-region', $node);	2✔
190	foreach ($changedRegions as $changedRegion) {	2✔
191	$type = ($changedRegion->firstChild->nodeName == 'text:insertion') ? TrackChange::INSERTED : TrackChange::DELETED;	×
192	$creatorNode = $xmlReader->getElements('office:change-info/dc:creator', $changedRegion->firstChild);	×
193	$author = $creatorNode[0]->nodeValue;	×
194	$dateNode = $xmlReader->getElements('office:change-info/dc:date', $changedRegion->firstChild);	×
195	$date = $dateNode[0]->nodeValue;	×
196	$date = preg_replace('/\.\d+$/', '', $date);	×
197	$date = DateTime::createFromFormat('Y-m-d\TH:i:s', $date);	×
198	$changed = new TrackChange($type, $author, $date);	×
199	$textNodes = $xmlReader->getElements('text:deletion/text:p', $changedRegion);	×
200	$trackedChanges[$changedRegion->getAttribute('text:id')] = ['changed' => $changed, 'textNodes' => $textNodes];	×
201	}
202
203	break;	2✔
204	case 'text:section': // Section	4✔
205	// $sectionStyleName = $xmlReader->getAttribute('text:style-name', $listItem);
206	$this->section = $phpWord->addSection();	2✔
207	/** @var DOMNodeList<DOMElement> $children */
208	$children = $node->childNodes;	2✔
209	$this->processNodes($children, $xmlReader, $phpWord);	2✔
210
211	break;	2✔
212	}
213	}
214	}
215	}
216
217	private function getSection(PhpWord $phpWord): Section
218	{
219	$section = $this->section;	5✔
220	if ($section === null) {	5✔
221	$section = $this->section = $phpWord->addSection();	3✔
222	}
223
224	return $section;	5✔
225	}
226	}

PHPOffice / PHPWord / 25414441731

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous