BookStack Code Mirror - bookstack/blob - app/Entities/Tools/PageIncludeParser.php

   1 <?php
   2
   3 namespace BookStack\Entities\Tools;
   4
   5 use BookStack\Util\HtmlDocument;
   6 use Closure;
   7 use DOMDocument;
   8 use DOMElement;
   9 use DOMNode;
  10 use DOMText;
  11
  12 class PageIncludeParser
  13 {
  14     protected static string $includeTagRegex = "/{{@\s?([0-9].*?)}}/";
  15
  16     public function __construct(
  17         protected string $pageHtml,
  18         protected Closure $pageContentForId,
  19     ) {
  20     }
  21
  22     public function parse(): string
  23     {
  24         $doc = new HtmlDocument($this->pageHtml);
  25
  26         $tags = $this->locateAndIsolateIncludeTags($doc);
  27         $topLevel = [...$doc->getBodyChildren()];
  28
  29         foreach ($tags as $tag) {
  30             $htmlContent = $this->pageContentForId->call($this, $tag->getPageId());
  31             $content = new PageIncludeContent($htmlContent, $tag);
  32
  33             if (!$content->isInline()) {
  34                 $isParentTopLevel = in_array($tag->domNode->parentNode, $topLevel, true);
  35                 if ($isParentTopLevel) {
  36                     $this->splitNodeAtChildNode($tag->domNode->parentNode, $tag->domNode);
  37                 } else {
  38                     $this->promoteTagNodeToBody($tag, $doc->getBody());
  39                 }
  40             }
  41
  42             $this->replaceNodeWithNodes($tag->domNode, $content->toDomNodes());
  43         }
  44
  45         // TODO Notes: May want to eventually parse through backwards, which should avoid issues
  46         //   in changes affecting the next tag, where tags may be in the same/adjacent nodes.
  47
  48         return $doc->getBodyInnerHtml();
  49     }
  50
  51     /**
  52      * Locate include tags within the given document, isolating them to their
  53      * own nodes in the DOM for future targeted manipulation.
  54      * @return PageIncludeTag[]
  55      */
  56     protected function locateAndIsolateIncludeTags(HtmlDocument $doc): array
  57     {
  58         $includeHosts = $doc->queryXPath("//body//*[contains(text(), '{{@')]");
  59         $includeTags = [];
  60
  61         /** @var DOMNode $node */
  62         /** @var DOMNode $childNode */
  63         foreach ($includeHosts as $node) {
  64             foreach ($node->childNodes as $childNode) {
  65                 if ($childNode->nodeName === '#text') {
  66                     array_push($includeTags, ...$this->splitTextNodesAtTags($childNode));
  67                 }
  68             }
  69         }
  70
  71         return $includeTags;
  72     }
  73
  74     /**
  75      * Takes a text DOMNode and splits its text content at include tags
  76      * into multiple text nodes within the original parent.
  77      * Returns found PageIncludeTag references.
  78      * @return PageIncludeTag[]
  79      */
  80     protected function splitTextNodesAtTags(DOMNode $textNode): array
  81     {
  82         $includeTags = [];
  83         $text = $textNode->textContent;
  84         preg_match_all(static::$includeTagRegex, $text, $matches, PREG_OFFSET_CAPTURE);
  85
  86         $currentOffset = 0;
  87         foreach ($matches[0] as $index => $fullTagMatch) {
  88             $tagOuterContent = $fullTagMatch[0];
  89             $tagInnerContent = $matches[1][$index][0];
  90             $tagStartOffset = $fullTagMatch[1];
  91
  92             if ($currentOffset < $tagStartOffset) {
  93                 $previousText = substr($text, $currentOffset, $tagStartOffset - $currentOffset);
  94                 $textNode->parentNode->insertBefore(new DOMText($previousText), $textNode);
  95             }
  96
  97             $node = $textNode->parentNode->insertBefore(new DOMText($tagOuterContent), $textNode);
  98             $includeTags[] = new PageIncludeTag($tagInnerContent, $node);
  99             $currentOffset = $tagStartOffset + strlen($tagOuterContent);
 100         }
 101
 102         if ($currentOffset > 0) {
 103             $textNode->textContent = substr($text, $currentOffset);
 104         }
 105
 106         return $includeTags;
 107     }
 108
 109     /**
 110      * @param DOMNode[] $replacements
 111      */
 112     protected function replaceNodeWithNodes(DOMNode $toReplace, array $replacements): void
 113     {
 114         /** @var DOMDocument $targetDoc */
 115         $targetDoc = $toReplace->ownerDocument;
 116
 117         foreach ($replacements as $replacement) {
 118             if ($replacement->ownerDocument !== $targetDoc) {
 119                 $replacement = $targetDoc->adoptNode($replacement);
 120             }
 121
 122             $toReplace->parentNode->insertBefore($replacement, $toReplace);
 123         }
 124
 125         $toReplace->parentNode->removeChild($toReplace);
 126     }
 127
 128     protected function promoteTagNodeToBody(PageIncludeTag $tag, DOMNode $body): void
 129     {
 130         /** @var DOMNode $topParent */
 131         $topParent = $tag->domNode->parentNode;
 132         while ($topParent->parentNode !== $body) {
 133             $topParent = $topParent->parentNode;
 134         }
 135
 136         $parentText = $topParent->textContent;
 137         $tagPos = strpos($parentText, $tag->tagContent);
 138         $before = $tagPos < (strlen($parentText) / 2);
 139
 140         if ($before) {
 141             $body->insertBefore($tag->domNode, $topParent);
 142         } else {
 143             $body->insertBefore($tag->domNode, $topParent->nextSibling);
 144         }
 145     }
 146
 147     protected function splitNodeAtChildNode(DOMElement $parentNode, DOMNode $domNode): void
 148     {
 149         $children = [...$parentNode->childNodes];
 150         $splitPos = array_search($domNode, $children, true) ?: count($children);
 151         $parentClone = $parentNode->cloneNode();
 152         $parentClone->removeAttribute('id');
 153
 154         /** @var DOMNode $child */
 155         for ($i = 0; $i < $splitPos; $i++) {
 156             $child = $children[0];
 157             $parentClone->appendChild($child);
 158         }
 159
 160         if ($parentClone->hasChildNodes()) {
 161             $parentNode->parentNode->insertBefore($parentClone, $parentNode);
 162         }
 163
 164         $parentNode->parentNode->insertBefore($domNode, $parentNode);
 165
 166         $parentClone->normalize();
 167         $parentNode->normalize();
 168         if (!$parentNode->hasChildNodes()) {
 169             $parentNode->remove();
 170         }
 171         if (!$parentClone->hasChildNodes()) {
 172             $parentClone->remove();
 173         }
 174     }
 175 }