Add support for comments to reader

docs/intro.rst

@@ -164,7 +164,7 @@ Readers
 +---------------------------+----------------------+--------+-------+-------+-------+-------+
 |                           | Endnote              | ✓      |       |       |       |       |
 +---------------------------+----------------------+--------+-------+-------+-------+-------+
-|                           | Comments             |        |       |       |       |       |
+|                           | Comments             | ✓      |       |       |       |       |
 +---------------------------+----------------------+--------+-------+-------+-------+-------+
 | **Graphs**                | 2D basic graphs      |        |       |       |       |       |
 +---------------------------+----------------------+--------+-------+-------+-------+-------+

src/PhpWord/PhpWord.php

@@ -17,6 +17,8 @@
 
 namespace PhpOffice\PhpWord;
 
+use InvalidArgumentException;
+use PhpOffice\PhpWord\Element\AbstractElement;
 use PhpOffice\PhpWord\Element\Section;
 use PhpOffice\PhpWord\Exception\Exception;
 
@@ -87,6 +89,14 @@ class PhpWord
      */
     private $metadata = array();
 
+    /**
+     * Comment reference cache
+     *
+     * @var array
+     * @since 0.18.3
+     */
+    private $commentReferenceCache = array();
+
     /**
      * Create new instance
      *
@@ -422,4 +432,40 @@ public function setDocumentProperties($documentProperties)
 
         return $this;
     }
+
+    /**
+     * Cache commentReference (as well as commentRangeStart and commentRangeEnd) for later use
+     * 
+     * @param 'start'|'end' $type
+     * @param string $id,
+     * @param $element
+     *
+     * @return self
+     */
+    public function cacheCommentReference(string $type, string $id, AbstractElement $element)
+    {
+        //dump('cacheCommentReference', func_get_args(), array_key_exists($id, $this->commentReferenceCache));
+        if (!in_array($type, [ 'start', 'end' ])) {
+            throw new InvalidArgumentException('Type must be "start" or "end"');
+        }
+
+        if (!array_key_exists($id, $this->commentReferenceCache)) {
+            $this->commentReferenceCache[$id] = (object)[
+                "start" => null,
+                "end" => null
+            ];
+        }
+        $this->commentReferenceCache[$id]->{$type} = $element;
+
+        return $this;
+    }
+
+    public function getCommentReference(string $id)
+    {
+        if (!array_key_exists($id, $this->commentReferenceCache)) {
+            //dd($this->commentReferenceCache);
+            throw new InvalidArgumentException('Comment with id '.$id.' isn\'t referenced in document');
+        }
+        return $this->commentReferenceCache[$id];
+    }
 }

src/PhpWord/Reader/Word2007.php

@@ -56,6 +56,7 @@ public function load($docFile)
                 'endnotes'  => 'Endnotes',
                 'footnotes' => 'Footnotes',
                 'settings'  => 'Settings',
+                'comments' => 'Comments'
             )),
         );
 

src/PhpWord/Reader/Word2007/AbstractPart.php

@@ -107,6 +107,11 @@ protected function readParagraph(XMLReader $xmlReader, \DOMElement $domNode, $pa
         // Paragraph style
         $paragraphStyle = null;
         $headingDepth = null;
+        if ($xmlReader->elementExists('w:commentReference', $domNode) || $xmlReader->elementExists('w:commentRangeStart', $domNode) || $xmlReader->elementExists('w:commentRangeEnd', $domNode)) {
+            $nodes = $xmlReader->getElements('w:commentReference|w:commentRangeStart|w:commentRangeEnd', $domNode);
+            $node = current(iterator_to_array($nodes));
+            $id = $node->attributes->getNamedItem('id')->value;
+        }
         if ($xmlReader->elementExists('w:pPr', $domNode)) {
             $paragraphStyle = $this->readParagraphStyle($xmlReader, $domNode);
             $headingDepth = $this->getHeadingDepth($paragraphStyle);
@@ -163,7 +168,7 @@ protected function readParagraph(XMLReader $xmlReader, \DOMElement $domNode, $pa
             $parent->addTitle($textContent, $headingDepth);
         } else {
             // Text and TextRun
-            $textRunContainers = $xmlReader->countElements('w:r|w:ins|w:del|w:hyperlink|w:smartTag', $domNode);
+            $textRunContainers = $xmlReader->countElements('w:r|w:ins|w:del|w:hyperlink|w:smartTag|w:commentReference|w:commentRangeStart|w:commentRangeEnd', $domNode);
             if (0 === $textRunContainers) {
                 $parent->addTextBreak(null, $paragraphStyle);
             } else {
@@ -212,7 +217,7 @@ private function getHeadingDepth(array $paragraphStyle = null)
      */
     protected function readRun(XMLReader $xmlReader, \DOMElement $domNode, $parent, $docPart, $paragraphStyle = null)
     {
-        if (in_array($domNode->nodeName, array('w:ins', 'w:del', 'w:smartTag', 'w:hyperlink'))) {
+        if (in_array($domNode->nodeName, array('w:ins', 'w:del', 'w:smartTag', 'w:hyperlink', 'w:commentReference'))) {
             $nodes = $xmlReader->getElements('*', $domNode);
             foreach ($nodes as $node) {
                 $this->readRun($xmlReader, $node, $parent, $docPart, $paragraphStyle);
@@ -224,6 +229,15 @@ protected function readRun(XMLReader $xmlReader, \DOMElement $domNode, $parent,
                 $this->readRunChild($xmlReader, $node, $parent, $docPart, $paragraphStyle, $fontStyle);
             }
         }
+
+        if($xmlReader->elementExists('.//*["commentReference"=local-name()]', $domNode)) {
+            $curEl = iterator_to_array($xmlReader->getElements('.//*["commentReference"=local-name()]', $domNode))[0];
+            $id = $curEl->attributes->getNamedItem('id')->value;
+            //$path = './/*[("commentRangeStart"=local-name() or "commentRangeEnd"=local-name()) and @*[local-name()="id" and .="'.$id.'"]]';
+            //$range = $xmlReader->getElements($path);
+            $this->phpWord->cacheCommentReference('start', $id, $parent->getElement($parent->countElements() - 1));
+            $this->phpWord->cacheCommentReference('end', $id, $parent->getElement($parent->countElements() - 1));
+        }
     }
 
     /**

src/PhpWord/Reader/Word2007/Comments.php

@@ -0,0 +1,109 @@
+<?php
+
+namespace PhpOffice\PhpWord\Reader\Word2007;
+
+use DateTime;
+use PhpOffice\PhpWord\Element\Comment;
+use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Reader\Word2007\AbstractPart;
+use PhpOffice\PhpWord\Shared\XMLReader;
+
+class Comments extends AbstractPart
+{
+    /**
+     * Collection name comments
+     *
+     * @var string
+     */
+    protected $collection = 'comments';
+
+    /**
+     * Read settings.xml.
+     *
+     * @param \PhpOffice\PhpWord\PhpWord $phpWord
+     */
+    public function read(PhpWord $phpWord)
+    {
+        $xmlReader = new XMLReader();
+        $xmlReader->getDomFromZip($this->docFile, $this->xmlFile);
+
+        //$xmlReader2 = new XMLReader();
+        //$xmlReader2->getDomFromZip($this->docFile, 'word/document.xml');
+        //dd($xmlReader2);
+
+        $comments = $phpWord->getComments();
+
+        $nodes = $xmlReader->getElements('*');
+        if ($nodes->length > 0) {
+            foreach ($nodes as $node) {
+                $name = str_replace('w:', '', $node->nodeName);
+                $value = $xmlReader->getAttribute('w:author', $node);
+                $author = $xmlReader->getAttribute('w:author', $node);
+                $date = $xmlReader->getAttribute('w:date', $node);
+                $initials = $xmlReader->getAttribute('w:initials', $node);
+                $id = $xmlReader->getAttribute('w:id', $node);
+                $element = new Comment($author, new DateTime($date), $initials);//$this->getElement($phpWord, $id);
+                //$element->set
+                // $range = $xmlReader2->getElements('.//*[("commentRangeStart"=local-name() or "commentRangeEnd"=local-name()) and @*[local-name()="id" and .="'.$id.'"]]');
+                try {
+                    unset($range);
+                    $range = $phpWord->getCommentReference($id);
+                    $range->start->setCommentRangeStart($element);
+                    $range->end->setCommentRangeEnd($element);
+                } catch(\Exception $e) {
+                    //dd('range', [$element, $id, $node, $node->C14N(), $range ?? null, $e]);
+                }
+                //dd($startElement, $endElement, current(current($phpWord->getSections())->getElements()));
+                //dump($element, $range);
+                //dd($element, $node, $id, $node->C14N());
+                $method = 'set' . $name;
+                //dump([$element, $id, $name, $value, $author, $date, $initials, $method, $xmlReader->getElements('w:p/w:r/w:t', $node)]);
+                //dd('dsf');
+                $pNodes = $xmlReader->getElements('w:p/w:r', $node);
+                foreach ($pNodes as $pNode) {
+                    //dump(['>', $xmlReader, $pNode, $node, $this->collection, '<']);
+                    $this->readRun($xmlReader, $pNode, $element, $this->collection);
+                }
+
+                /*if (in_array($name, $this::$booleanProperties)) {
+                    if ($value == 'false') {
+                        $comments->$method(false);
+                    } else {
+                        $comments->$method(true);
+                    }
+                } else*/if (method_exists($this, $method)) {
+                    $this->$method($xmlReader, $phpWord, $node);
+                } elseif (method_exists($comments, $method)) {
+                    $comments->$method($value);
+                } elseif (method_exists($phpWord, $method)) {
+                    $phpWord->$method($value);
+                } elseif (method_exists($comments, 'addItem')) {
+                    $comments->addItem($element);
+                }
+            }
+        }
+	}
+
+    /**
+     * Searches for the element with the given relationId
+     *
+     * @param PhpWord $phpWord
+     * @param int $relationId
+     * @return \PhpOffice\PhpWord\Element\AbstractContainer|null
+     */
+    private function getElement(PhpWord $phpWord, $relationId)
+    {
+        $getMethod = "get{$this->collection}";
+        //$getMethod = "getTrackChange";
+        $collection = $phpWord->$getMethod();//->getItems();
+
+        //not found by key, looping to search by relationId
+        foreach ($collection as $collectionElement) {
+            if ($collectionElement->getRelationId() == $relationId) {
+                return $collectionElement;
+            }
+        }
+
+        return null;
+    }
+}

src/PhpWord/Reader/Word2007/Document.php

@@ -34,7 +34,7 @@ class Document extends AbstractPart
      *
      * @var \PhpOffice\PhpWord\PhpWord
      */
-    private $phpWord;
+    protected $phpWord;
 
     /**
      * Read document.xml.
@@ -180,4 +180,9 @@ private function readWSectPrNode(XMLReader $xmlReader, \DOMElement $node, Sectio
         $section->setStyle($style);
         $this->readHeaderFooter($style, $section);
     }
+
+    protected function cacheCommentReference(string $type, string $id, $element)
+    {
+        $this->phpWord->cacheCommentReference($type, $id, $element);
+    }
 }

tests/PhpWord/Reader/Word2007Test.php

@@ -17,6 +17,8 @@
 
 namespace PhpOffice\PhpWord\Reader;
 
+use DateTime;
+use PhpOffice\PhpWord\Element\Comment;
 use PhpOffice\PhpWord\IOFactory;
 use PhpOffice\PhpWord\TestHelperDOCX;
 
@@ -78,4 +80,18 @@ public function testLoadWord2011()
         $doc = TestHelperDOCX::getDocument($phpWord);
         $this->assertTrue($doc->elementExists('/w:document/w:body/w:p[3]/w:r/w:pict/v:shape/v:imagedata'));
     }
+
+    public function testLoadComments()
+    {
+        $filename = __DIR__ . '/../_files/documents/reader-ooxml-comments.docx';
+        $phpWord = IOFactory::load($filename);
+
+        $this->assertInstanceOf('PhpOffice\\PhpWord\\PhpWord', $phpWord);
+
+        //$doc = TestHelperDOCX::getDocument($phpWord);
+        $comment = new Comment('shaedrich', new DateTime('2021-10-28T13:56:00Z'), 'SH');
+        $comment2 = $phpWord->getComments()[0];
+        $this->assertEquals($comment->getAuthor(), $comment2->getAuthor());
+        $this->assertEquals($comment->getInitials(), $comment2->getInitials());
+    }
 }