Word2007 Reader : Added support for Comments

PHPOffice · Sep 14, 2023 · 0ea59e7 · 0ea59e7
1 parent 34db7cf
commit 0ea59e7
Show file tree

Hide file tree

Showing 13 changed files with 216 additions and 279 deletions.
diff --git a/docs/changes/1.x/1.2.0.md b/docs/changes/1.x/1.2.0.md
@@ -11,6 +11,7 @@
 - HTML Writer : Added border-spacing to default styles for table by [@kernusr](https://github.com/kernusr) in GH-2451
 - Word2007 Reader : Support for table cell borders and margins by [@kernusr](https://github.com/kernusr) in GH-2454
 - PDF Writer : Add config for defining the default font by [@MikeMaldini](https://github.com/MikeMaldini) in [#2262](https://github.com/PHPOffice/PHPWord/pull/2262) & [#2468](https://github.com/PHPOffice/PHPWord/pull/2468)
+- Word2007 Reader : Added support for Comments by [@shaedrich](https://github.com/shaedrich) in [#2161](https://github.com/PHPOffice/PHPWord/pull/2161) & [#2469](https://github.com/PHPOffice/PHPWord/pull/2469)
 
 ### Bug fixes
 

diff --git a/docs/index.md b/docs/index.md
@@ -95,7 +95,7 @@ Below are the supported features for each file formats.
 | | Footer | :material-check: | | | | |
 | | Footnote | :material-check: | | | | |
 | | Endnote | :material-check: | | | | |
-| | Comments |   | | | | |
+| | Comments | :material-check: | | | | |
 | **Graphs** | 2D basic graphs | | | | | |
 | | 2D advanced graphs | | | | | |
 | | 3D graphs | | | | | |

diff --git a/phpstan-baseline.neon b/phpstan-baseline.neon
@@ -230,11 +230,6 @@ parameters:
  count: 2
  path: src/PhpWord/Reader/Word2007/AbstractPart.php
 
- -
- message: "#^Call to method setChangeInfo\\(\\) on an unknown class PhpOffice\\\\PhpWord\\\\Reader\\\\Word2007\\\\AbstractElement\\.$#"
- count: 1
- path: src/PhpWord/Reader/Word2007/AbstractPart.php
-
  -
  message: "#^Method PhpOffice\\\\PhpWord\\\\Reader\\\\Word2007\\\\AbstractPart\\:\\:getHeadingDepth\\(\\) never returns float so it can be removed from the return type\\.$#"
  count: 1
@@ -250,11 +245,6 @@ parameters:
  count: 1
  path: src/PhpWord/Reader/Word2007/AbstractPart.php
 
- -
- message: "#^PHPDoc tag @var for variable \\$element contains unknown class PhpOffice\\\\PhpWord\\\\Reader\\\\Word2007\\\\AbstractElement\\.$#"
- count: 1
- path: src/PhpWord/Reader/Word2007/AbstractPart.php
-
  -
  message: "#^Parameter \\#1 \\$count of method PhpOffice\\\\PhpWord\\\\Element\\\\AbstractContainer\\:\\:addTextBreak\\(\\) expects int, null given\\.$#"
  count: 1

diff --git a/src/PhpWord/Collection/AbstractCollection.php b/src/PhpWord/Collection/AbstractCollection.php
@@ -79,7 +79,7 @@ public function setItem($index, $item): void
  */
  public function addItem($item)
  {
- $index = $this->countItems() + 1;
+ $index = $this->countItems();
  $this->items[$index] = $item;
 
  return $index;

diff --git a/src/PhpWord/PhpWord.php b/src/PhpWord/PhpWord.php
@@ -18,8 +18,6 @@
 namespace PhpOffice\PhpWord;
 
 use BadMethodCallException;
-use InvalidArgumentException;
-use PhpOffice\PhpWord\Element\AbstractElement;
 use PhpOffice\PhpWord\Element\Section;
 use PhpOffice\PhpWord\Exception\Exception;
 
@@ -70,14 +68,7 @@ class PhpWord
  private $metadata = [];
 
  /**
- * Comment reference cache
- *
- * @var array
- */
- private $commentReferenceCache = [];
-
- /**
- * Create new instance
+ * Create new instance.
  *
  * Collections are created dynamically
  */
@@ -336,7 +327,7 @@ public function save($filename, $format = 'Word2007', $download = false)
  }
 
  /**
- * Create new section
+ * Create new section.
  *
  * @deprecated 0.10.0
  *
@@ -352,7 +343,7 @@ public function createSection($settings = null)
  }
 
  /**
- * Get document properties object
+ * Get document properties object.
  *
  * @deprecated 0.12.0
  *
@@ -366,7 +357,7 @@ public function getDocumentProperties()
  }
 
  /**
- * Set document properties object
+ * Set document properties object.
  *
  * @deprecated 0.12.0
  *
@@ -382,40 +373,4 @@ 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];
- }
 }
diff --git a/src/PhpWord/Reader/Word2007.php b/src/PhpWord/Reader/Word2007.php
@@ -17,7 +17,10 @@
 
 namespace PhpOffice\PhpWord\Reader;
 
+use Exception;
+use PhpOffice\PhpWord\Element\AbstractElement;
 use PhpOffice\PhpWord\PhpWord;
+use PhpOffice\PhpWord\Reader\Word2007\AbstractPart;
 use PhpOffice\PhpWord\Shared\XMLReader;
 use PhpOffice\PhpWord\Shared\ZipArchive;
 
@@ -42,24 +45,34 @@ public function load($docFile)
  {
  $phpWord = new PhpWord();
  $relationships = $this->readRelationships($docFile);
+ $commentRefs = [];
 
  $steps = [
- ['stepPart' => 'document', 'stepItems' => [
- 'styles' => 'Styles',
- 'numbering' => 'Numbering',
- ]],
- ['stepPart' => 'main', 'stepItems' => [
- 'officeDocument' => 'Document',
- 'core-properties' => 'DocPropsCore',
- 'extended-properties' => 'DocPropsApp',
- 'custom-properties' => 'DocPropsCustom',
- ]],
- ['stepPart' => 'document', 'stepItems' => [
- 'endnotes' => 'Endnotes',
- 'footnotes' => 'Footnotes',
- 'settings' => 'Settings',
- 'comments' => 'Comments'
- ]],
+ [
+ 'stepPart' => 'document',
+ 'stepItems' => [
+ 'styles' => 'Styles',
+ 'numbering' => 'Numbering',
+ ],
+ ],
+ [
+ 'stepPart' => 'main',
+ 'stepItems' => [
+ 'officeDocument' => 'Document',
+ 'core-properties' => 'DocPropsCore',
+ 'extended-properties' => 'DocPropsApp',
+ 'custom-properties' => 'DocPropsCustom',
+ ],
+ ],
+ [
+ 'stepPart' => 'document',
+ 'stepItems' => [
+ 'endnotes' => 'Endnotes',
+ 'footnotes' => 'Footnotes',
+ 'settings' => 'Settings',
+ 'comments' => 'Comments',
+ ],
+ ],
  ];
 
  foreach ($steps as $step) {
@@ -73,7 +86,8 @@ public function load($docFile)
  if (isset($stepItems[$relType])) {
  $partName = $stepItems[$relType];
  $xmlFile = $relItem['target'];
- $this->readPart($phpWord, $relationships, $partName, $docFile, $xmlFile);
+ $part = $this->readPart($phpWord, $relationships, $commentRefs, $partName, $docFile, $xmlFile);
+ $commentRefs = $part->getCommentReferences();
  }
  }
  }
@@ -84,21 +98,23 @@ public function load($docFile)
  /**
  * Read document part.
  *
- * @param array $relationships
- * @param string $partName
- * @param string $docFile
- * @param string $xmlFile
+ * @param array<string, array<string, null|AbstractElement>> $commentRefs
  */
- private function readPart(PhpWord $phpWord, $relationships, $partName, $docFile, $xmlFile): void
+ private function readPart(PhpWord $phpWord, array $relationships, array $commentRefs, string $partName, string $docFile, string $xmlFile): AbstractPart
  {
  $partClass = "PhpOffice\\PhpWord\\Reader\\Word2007\\{$partName}";
- if (class_exists($partClass)) {
- /** @var \PhpOffice\PhpWord\Reader\Word2007\AbstractPart $part Type hint */
- $part = new $partClass($docFile, $xmlFile);
- $part->setImageLoading($this->hasImageLoading());
- $part->setRels($relationships);
- $part->read($phpWord);
+ if (!class_exists($partClass)) {
+ throw new Exception(sprintf('The part "%s" doesn\'t exist', $partClass));
  }
+
+ /** @var AbstractPart $part Type hint */
+ $part = new $partClass($docFile, $xmlFile);
+ $part->setImageLoading($this->hasImageLoading());
+ $part->setRels($relationships);
+ $part->setCommentReferences($commentRefs);
+ $part->read($phpWord);
+
+ return $part;
  }
 
  /**

diff --git a/src/PhpWord/Reader/Word2007/AbstractPart.php b/src/PhpWord/Reader/Word2007/AbstractPart.php
@@ -19,8 +19,10 @@
 
 use DateTime;
 use DOMElement;
+use InvalidArgumentException;
 use PhpOffice\PhpWord\ComplexType\TblWidth as TblWidthComplexType;
 use PhpOffice\PhpWord\Element\AbstractContainer;
+use PhpOffice\PhpWord\Element\AbstractElement;
 use PhpOffice\PhpWord\Element\TextRun;
 use PhpOffice\PhpWord\Element\TrackChange;
 use PhpOffice\PhpWord\PhpWord;
@@ -67,6 +69,13 @@ abstract class AbstractPart
  */
  protected $rels = [];
 
+ /**
+ * Comment references.
+ *
+ * @var array<string, array<string, AbstractElement>>
+ */
+ protected $commentRefs = [];
+
  /**
  * Image Loading.
  *
@@ -113,6 +122,62 @@ public function hasImageLoading(): bool
  return $this->imageLoading;
  }
 
+ /**
+ * Get comment references.
+ *
+ * @return array<string, array<string, null|AbstractElement>>
+ */
+ public function getCommentReferences(): array
+ {
+ return $this->commentRefs;
+ }
+
+ /**
+ * Set comment references.
+ *
+ * @param array<string, array<string, null|AbstractElement>> $commentRefs
+ */
+ public function setCommentReferences(array $commentRefs): self
+ {
+ $this->commentRefs = $commentRefs;
+
+ return $this;
+ }
+
+ /**
+ * Set comment reference.
+ */
+ private function setCommentReference(string $type, string $id, AbstractElement $element): self
+ {
+ if (!in_array($type, ['start', 'end'])) {
+ throw new InvalidArgumentException('Type must be "start" or "end"');
+ }
+
+ if (!array_key_exists($id, $this->commentRefs)) {
+ $this->commentRefs[$id] = [
+ 'start' => null,
+ 'end' => null,
+ ];
+ }
+ $this->commentRefs[$id][$type] = $element;
+
+ return $this;
+ }
+
+ /**
+ * Get comment reference.
+ *
+ * @return array<string, null|AbstractElement>
+ */
+ protected function getCommentReference(string $id): array
+ {
+ if (!array_key_exists($id, $this->commentRefs)) {
+ throw new InvalidArgumentException(sprintf('Comment with id %s isn\'t referenced in document', $id));
+ }
+
+ return $this->commentRefs[$id];
+ }
+
  /**
  * Read w:p.
  *
@@ -126,10 +191,18 @@ protected function readParagraph(XMLReader $xmlReader, DOMElement $domNode, $par
  // Paragraph style
  $paragraphStyle = null;
  $headingDepth = null;
- if ($xmlReader->elementExists('w:commentReference', $domNode) || $xmlReader->elementExists('w:commentRangeStart', $domNode) || $xmlReader->elementExists('w:commentRangeEnd', $domNode)) {
+ if ($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 = $node->attributes->getNamedItem('id')->value;
+ if ($node) {
+ $attributeIdentifier = $node->attributes->getNamedItem('id');
+ if ($attributeIdentifier) {
+ $id = $attributeIdentifier->nodeValue;
+ }
+ }
  }
  if ($xmlReader->elementExists('w:pPr', $domNode)) {
  $paragraphStyle = $this->readParagraphStyle($xmlReader, $domNode);
@@ -235,7 +308,7 @@ private function getHeadingDepth(?array $paragraphStyle = null)
  */
  protected function readRun(XMLReader $xmlReader, DOMElement $domNode, $parent, $docPart, $paragraphStyle = null): void
  {
- if (in_array($domNode->nodeName, array('w:ins', 'w:del', 'w:smartTag', 'w:hyperlink', 'w:commentReference'))) {
+ if (in_array($domNode->nodeName, ['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);
@@ -248,13 +321,15 @@ protected function readRun(XMLReader $xmlReader, DOMElement $domNode, $parent, $
  }
  }
 
- 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));
+ if ($xmlReader->elementExists('.//*["commentReference"=local-name()]', $domNode)) {
+ $node = iterator_to_array($xmlReader->getElements('.//*["commentReference"=local-name()]', $domNode))[0];
+ $attributeIdentifier = $node->attributes->getNamedItem('id');
+ if ($attributeIdentifier) {
+ $id = $attributeIdentifier->nodeValue;
+
+ $this->setCommentReference('start', $id, $parent->getElement($parent->countElements() - 1));
+ $this->setCommentReference('end', $id, $parent->getElement($parent->countElements() - 1));
+ }
  }
  }
 
@@ -353,6 +428,7 @@ protected function readRunChild(XMLReader $xmlReader, DOMElement $node, Abstract
  $type = ($runParent->nodeName == 'w:del') ? TrackChange::DELETED : TrackChange::INSERTED;
  $author = $runParent->getAttribute('w:author');
  $date = DateTime::createFromFormat('Y-m-d\TH:i:s\Z', $runParent->getAttribute('w:date'));
+ $date = $date instanceof DateTime ? $date : null;
  $element->setChangeInfo($type, $author, $date);
  }
  }