BookStack/app/Entities/Tools/ExportFormatter.php

323 lines
9.4 KiB
PHP
Raw Normal View History

2021-06-26 11:23:15 -04:00
<?php
namespace BookStack\Entities\Tools;
2016-01-20 17:13:13 -05:00
use BookStack\Entities\Models\Book;
use BookStack\Entities\Models\Chapter;
use BookStack\Entities\Models\Page;
use BookStack\Entities\Tools\Markdown\HtmlToMarkdown;
use BookStack\Uploads\ImageService;
use DOMDocument;
use DOMElement;
use DOMXPath;
use Exception;
use Throwable;
2016-01-20 17:13:13 -05:00
class ExportFormatter
2016-01-20 17:13:13 -05:00
{
protected $imageService;
protected $pdfGenerator;
/**
* ExportService constructor.
*/
public function __construct(ImageService $imageService, PdfGenerator $pdfGenerator)
{
$this->imageService = $imageService;
$this->pdfGenerator = $pdfGenerator;
}
2016-01-20 17:13:13 -05:00
/**
* Convert a page to a self-contained HTML file.
* Includes required CSS & image content. Images are base64 encoded into the HTML.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
2016-01-20 17:13:13 -05:00
*/
public function pageToContainedHtml(Page $page)
{
$page->html = (new PageContent($page))->render();
$pageHtml = view('pages.export', [
2021-06-26 11:23:15 -04:00
'page' => $page,
'format' => 'html',
])->render();
2021-06-26 11:23:15 -04:00
return $this->containHtml($pageHtml);
}
/**
* Convert a chapter to a self-contained HTML file.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
*/
public function chapterToContainedHtml(Chapter $chapter)
{
$pages = $chapter->getVisiblePages();
$pages->each(function ($page) {
$page->html = (new PageContent($page))->render();
});
$html = view('chapters.export', [
'chapter' => $chapter,
2021-06-26 11:23:15 -04:00
'pages' => $pages,
'format' => 'html',
])->render();
2021-06-26 11:23:15 -04:00
return $this->containHtml($html);
}
/**
* Convert a book to a self-contained HTML file.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
*/
public function bookToContainedHtml(Book $book)
{
$bookTree = (new BookContents($book))->getTree(false, true);
$html = view('books.export', [
2021-06-26 11:23:15 -04:00
'book' => $book,
'bookChildren' => $bookTree,
2021-06-26 11:23:15 -04:00
'format' => 'html',
])->render();
2021-06-26 11:23:15 -04:00
return $this->containHtml($html);
}
/**
* Convert a page to a PDF file.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
*/
public function pageToPdf(Page $page)
2016-01-20 17:13:13 -05:00
{
$page->html = (new PageContent($page))->render();
$html = view('pages.export', [
2021-06-26 11:23:15 -04:00
'page' => $page,
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
])->render();
2021-06-26 11:23:15 -04:00
return $this->htmlToPdf($html);
}
/**
* Convert a chapter to a PDF file.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
*/
public function chapterToPdf(Chapter $chapter)
{
$pages = $chapter->getVisiblePages();
$pages->each(function ($page) {
$page->html = (new PageContent($page))->render();
});
$html = view('chapters.export', [
'chapter' => $chapter,
2021-06-26 11:23:15 -04:00
'pages' => $pages,
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
])->render();
return $this->htmlToPdf($html);
}
/**
* Convert a book to a PDF file.
2021-06-26 11:23:15 -04:00
*
* @throws Throwable
*/
public function bookToPdf(Book $book)
{
$bookTree = (new BookContents($book))->getTree(false, true);
$html = view('books.export', [
2021-06-26 11:23:15 -04:00
'book' => $book,
'bookChildren' => $bookTree,
2021-06-26 11:23:15 -04:00
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
])->render();
2021-06-26 11:23:15 -04:00
return $this->htmlToPdf($html);
}
/**
* Convert normal web-page HTML to a PDF.
2021-06-26 11:23:15 -04:00
*
* @throws Exception
*/
protected function htmlToPdf(string $html): string
{
$html = $this->containHtml($html);
$html = $this->replaceIframesWithLinks($html);
return $this->pdfGenerator->fromHtml($html);
}
/**
* Within the given HTML content, replace any iframe elements
* with anchor links within paragraph blocks.
*/
protected function replaceIframesWithLinks(string $html): string
{
libxml_use_internal_errors(true);
$doc = new DOMDocument();
$doc->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
$xPath = new DOMXPath($doc);
$iframes = $xPath->query('//iframe');
/** @var DOMElement $iframe */
foreach ($iframes as $iframe) {
$link = $iframe->getAttribute('src');
if (strpos($link, '//') === 0) {
$link = 'https:' . $link;
}
$anchor = $doc->createElement('a', $link);
$anchor->setAttribute('href', $link);
$paragraph = $doc->createElement('p');
$paragraph->appendChild($anchor);
$iframe->parentNode->replaceChild($paragraph, $iframe);
}
2021-06-26 11:23:15 -04:00
return $doc->saveHTML();
}
2016-01-20 17:13:13 -05:00
/**
* Bundle of the contents of a html file to be self-contained.
2021-06-26 11:23:15 -04:00
*
* @throws Exception
*/
protected function containHtml(string $htmlContent): string
{
2016-01-20 17:13:13 -05:00
$imageTagsOutput = [];
preg_match_all("/\<img.*?src\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $imageTagsOutput);
2016-01-20 17:13:13 -05:00
// Replace image src with base64 encoded image strings
if (isset($imageTagsOutput[0]) && count($imageTagsOutput[0]) > 0) {
foreach ($imageTagsOutput[0] as $index => $imgMatch) {
$oldImgTagString = $imgMatch;
2016-01-20 17:13:13 -05:00
$srcString = $imageTagsOutput[2][$index];
$imageEncoded = $this->imageService->imageUriToBase64($srcString);
if ($imageEncoded === null) {
$imageEncoded = $srcString;
}
$newImgTagString = str_replace($srcString, $imageEncoded, $oldImgTagString);
$htmlContent = str_replace($oldImgTagString, $newImgTagString, $htmlContent);
2016-01-20 17:13:13 -05:00
}
}
$linksOutput = [];
preg_match_all("/\<a.*href\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $linksOutput);
2016-01-20 17:13:13 -05:00
// Replace image src with base64 encoded image strings
if (isset($linksOutput[0]) && count($linksOutput[0]) > 0) {
foreach ($linksOutput[0] as $index => $linkMatch) {
$oldLinkString = $linkMatch;
$srcString = $linksOutput[2][$index];
if (strpos(trim($srcString), 'http') !== 0) {
$newSrcString = url($srcString);
$newLinkString = str_replace($srcString, $newSrcString, $oldLinkString);
$htmlContent = str_replace($oldLinkString, $newLinkString, $htmlContent);
2016-01-20 17:13:13 -05:00
}
}
}
// Replace any relative links with system domain
return $htmlContent;
}
/**
* Converts the page contents into simple plain text.
* This method filters any bad looking content to provide a nice final output.
*/
public function pageToPlainText(Page $page): string
{
$html = (new PageContent($page))->render();
$text = strip_tags($html);
// Replace multiple spaces with single spaces
$text = preg_replace('/\ {2,}/', ' ', $text);
// Reduce multiple horrid whitespace characters.
$text = preg_replace('/(\x0A|\xA0|\x0A|\r|\n){2,}/su', "\n\n", $text);
$text = html_entity_decode($text);
// Add title
$text = $page->name . "\n\n" . $text;
2021-06-26 11:23:15 -04:00
return $text;
2016-01-20 17:13:13 -05:00
}
/**
* Convert a chapter into a plain text string.
*/
public function chapterToPlainText(Chapter $chapter): string
{
$text = $chapter->name . "\n\n";
$text .= $chapter->description . "\n\n";
foreach ($chapter->getVisiblePages() as $page) {
$text .= $this->pageToPlainText($page);
}
2021-06-26 11:23:15 -04:00
return $text;
}
/**
* Convert a book into a plain text string.
*/
public function bookToPlainText(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, false);
$text = $book->name . "\n\n";
foreach ($bookTree as $bookChild) {
if ($bookChild->isA('chapter')) {
$text .= $this->chapterToPlainText($bookChild);
} else {
$text .= $this->pageToPlainText($bookChild);
}
}
2021-06-26 11:23:15 -04:00
return $text;
}
2020-05-13 00:12:26 -04:00
/**
* Convert a page to a Markdown file.
*/
public function pageToMarkdown(Page $page): string
2020-05-13 00:12:26 -04:00
{
if ($page->markdown) {
2021-06-26 11:23:15 -04:00
return '# ' . $page->name . "\n\n" . $page->markdown;
2020-05-13 00:12:26 -04:00
}
2021-06-26 11:23:15 -04:00
return '# ' . $page->name . "\n\n" . (new HtmlToMarkdown($page->html))->convert();
2020-05-13 00:12:26 -04:00
}
/**
* Convert a chapter to a Markdown file.
*/
public function chapterToMarkdown(Chapter $chapter): string
2020-05-13 00:12:26 -04:00
{
2021-06-26 11:23:15 -04:00
$text = '# ' . $chapter->name . "\n\n";
2020-05-13 00:12:26 -04:00
$text .= $chapter->description . "\n\n";
foreach ($chapter->pages as $page) {
$text .= $this->pageToMarkdown($page) . "\n\n";
2020-05-13 00:12:26 -04:00
}
2021-06-26 11:23:15 -04:00
2020-05-13 00:12:26 -04:00
return $text;
}
/**
* Convert a book into a plain text string.
*/
public function bookToMarkdown(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, true);
2021-06-26 11:23:15 -04:00
$text = '# ' . $book->name . "\n\n";
2020-05-13 00:12:26 -04:00
foreach ($bookTree as $bookChild) {
if ($bookChild instanceof Chapter) {
2020-05-13 00:12:26 -04:00
$text .= $this->chapterToMarkdown($bookChild);
} else {
$text .= $this->pageToMarkdown($bookChild);
}
}
2021-06-26 11:23:15 -04:00
2020-05-13 00:12:26 -04:00
return $text;
}
}