webapp/bookstack
1
0
Fork 0
mirror of https://github.com/BookStackApp/BookStack.git synced 2025-07-31 15:24:31 +03:00
Code Activity

Merge branch 'portazips' into development

Browse Source
This commit is contained in:
Dan Brown
2024-12-01 13:06:43 +00:00
parent 0a07b0d162 bdca9fc1ce
commit fdbbcf2b8a
84 changed files with 5269 additions and 717 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
app/Entities/Tools/Cloner.php
View File

@ -18,17 +18,12 @@ use Illuminate\Http\UploadedFile;
class Cloner
{
protected PageRepo $pageRepo;
protected ChapterRepo $chapterRepo;
protected BookRepo $bookRepo;
protected ImageService $imageService;
public function __construct(PageRepo $pageRepo, ChapterRepo $chapterRepo, BookRepo $bookRepo, ImageService $imageService)
{
$this->pageRepo = $pageRepo;
$this->chapterRepo = $chapterRepo;
$this->bookRepo = $bookRepo;
$this->imageService = $imageService;
public function __construct(
protected PageRepo $pageRepo,
protected ChapterRepo $chapterRepo,
protected BookRepo $bookRepo,
protected ImageService $imageService,
) {
}
/**

343
app/Entities/Tools/ExportFormatter.php
View File

@ -1,343 +0,0 @@
<?php
namespace BookStack\Entities\Tools;
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 BookStack\Util\CspService;
use BookStack\Util\HtmlDocument;
use DOMElement;
use Exception;
use Throwable;
class ExportFormatter
{
public function __construct(
protected ImageService $imageService,
protected PdfGenerator $pdfGenerator,
protected CspService $cspService
) {
}
/**
* Convert a page to a self-contained HTML file.
* Includes required CSS & image content. Images are base64 encoded into the HTML.
*
* @throws Throwable
*/
public function pageToContainedHtml(Page $page): string
{
$page->html = (new PageContent($page))->render();
$pageHtml = view('exports.page', [
'page' => $page,
'format' => 'html',
'cspContent' => $this->cspService->getCspMetaTagValue(),
'locale' => user()->getLocale(),
])->render();
return $this->containHtml($pageHtml);
}
/**
* Convert a chapter to a self-contained HTML file.
*
* @throws Throwable
*/
public function chapterToContainedHtml(Chapter $chapter): string
{
$pages = $chapter->getVisiblePages();
$pages->each(function ($page) {
$page->html = (new PageContent($page))->render();
});
$html = view('exports.chapter', [
'chapter' => $chapter,
'pages' => $pages,
'format' => 'html',
'cspContent' => $this->cspService->getCspMetaTagValue(),
'locale' => user()->getLocale(),
])->render();
return $this->containHtml($html);
}
/**
* Convert a book to a self-contained HTML file.
*
* @throws Throwable
*/
public function bookToContainedHtml(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, true);
$html = view('exports.book', [
'book' => $book,
'bookChildren' => $bookTree,
'format' => 'html',
'cspContent' => $this->cspService->getCspMetaTagValue(),
'locale' => user()->getLocale(),
])->render();
return $this->containHtml($html);
}
/**
* Convert a page to a PDF file.
*
* @throws Throwable
*/
public function pageToPdf(Page $page): string
{
$page->html = (new PageContent($page))->render();
$html = view('exports.page', [
'page' => $page,
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
'locale' => user()->getLocale(),
])->render();
return $this->htmlToPdf($html);
}
/**
* Convert a chapter to a PDF file.
*
* @throws Throwable
*/
public function chapterToPdf(Chapter $chapter): string
{
$pages = $chapter->getVisiblePages();
$pages->each(function ($page) {
$page->html = (new PageContent($page))->render();
});
$html = view('exports.chapter', [
'chapter' => $chapter,
'pages' => $pages,
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
'locale' => user()->getLocale(),
])->render();
return $this->htmlToPdf($html);
}
/**
* Convert a book to a PDF file.
*
* @throws Throwable
*/
public function bookToPdf(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, true);
$html = view('exports.book', [
'book' => $book,
'bookChildren' => $bookTree,
'format' => 'pdf',
'engine' => $this->pdfGenerator->getActiveEngine(),
'locale' => user()->getLocale(),
])->render();
return $this->htmlToPdf($html);
}
/**
* Convert normal web-page HTML to a PDF.
*
* @throws Exception
*/
protected function htmlToPdf(string $html): string
{
$html = $this->containHtml($html);
$doc = new HtmlDocument();
$doc->loadCompleteHtml($html);
$this->replaceIframesWithLinks($doc);
$this->openDetailElements($doc);
$cleanedHtml = $doc->getHtml();
return $this->pdfGenerator->fromHtml($cleanedHtml);
}
/**
* Within the given HTML content, Open any detail blocks.
*/
protected function openDetailElements(HtmlDocument $doc): void
{
$details = $doc->queryXPath('//details');
/** @var DOMElement $detail */
foreach ($details as $detail) {
$detail->setAttribute('open', 'open');
}
}
/**
* Within the given HTML document, replace any iframe elements
* with anchor links within paragraph blocks.
*/
protected function replaceIframesWithLinks(HtmlDocument $doc): void
{
$iframes = $doc->queryXPath('//iframe');
/** @var DOMElement $iframe */
foreach ($iframes as $iframe) {
$link = $iframe->getAttribute('src');
if (str_starts_with($link, '//')) {
$link = 'https:' . $link;
}
$anchor = $doc->createElement('a', $link);
$anchor->setAttribute('href', $link);
$paragraph = $doc->createElement('p');
$paragraph->appendChild($anchor);
$iframe->parentNode->replaceChild($paragraph, $iframe);
}
}
/**
* Bundle of the contents of a html file to be self-contained.
*
* @throws Exception
*/
protected function containHtml(string $htmlContent): string
{
$imageTagsOutput = [];
preg_match_all("/\<img.*?src\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $imageTagsOutput);
// Replace image src with base64 encoded image strings
if (isset($imageTagsOutput[0]) && count($imageTagsOutput[0]) > 0) {
foreach ($imageTagsOutput[0] as $index => $imgMatch) {
$oldImgTagString = $imgMatch;
$srcString = $imageTagsOutput[2][$index];
$imageEncoded = $this->imageService->imageUrlToBase64($srcString);
if ($imageEncoded === null) {
$imageEncoded = $srcString;
}
$newImgTagString = str_replace($srcString, $imageEncoded, $oldImgTagString);
$htmlContent = str_replace($oldImgTagString, $newImgTagString, $htmlContent);
}
}
$linksOutput = [];
preg_match_all("/\<a.*href\=(\'|\")(.*?)(\'|\").*?\>/i", $htmlContent, $linksOutput);
// Update relative links to be absolute, with instance url
if (isset($linksOutput[0]) && count($linksOutput[0]) > 0) {
foreach ($linksOutput[0] as $index => $linkMatch) {
$oldLinkString = $linkMatch;
$srcString = $linksOutput[2][$index];
if (!str_starts_with(trim($srcString), 'http')) {
$newSrcString = url($srcString);
$newLinkString = str_replace($srcString, $newSrcString, $oldLinkString);
$htmlContent = str_replace($oldLinkString, $newLinkString, $htmlContent);
}
}
}
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, bool $pageRendered = false, bool $fromParent = false): string
{
$html = $pageRendered ? $page->html : (new PageContent($page))->render();
// Add proceeding spaces before tags so spaces remain between
// text within elements after stripping tags.
$html = str_replace('<', " <", $html);
$text = trim(strip_tags($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 = $page->name . ($fromParent ? "\n" : "\n\n") . $text;
return $text;
}
/**
* Convert a chapter into a plain text string.
*/
public function chapterToPlainText(Chapter $chapter): string
{
$text = $chapter->name . "\n" . $chapter->description;
$text = trim($text) . "\n\n";
$parts = [];
foreach ($chapter->getVisiblePages() as $page) {
$parts[] = $this->pageToPlainText($page, false, true);
}
return $text . implode("\n\n", $parts);
}
/**
* Convert a book into a plain text string.
*/
public function bookToPlainText(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, true);
$text = $book->name . "\n" . $book->description;
$text = rtrim($text) . "\n\n";
$parts = [];
foreach ($bookTree as $bookChild) {
if ($bookChild->isA('chapter')) {
$parts[] = $this->chapterToPlainText($bookChild);
} else {
$parts[] = $this->pageToPlainText($bookChild, true, true);
}
}
return $text . implode("\n\n", $parts);
}
/**
* Convert a page to a Markdown file.
*/
public function pageToMarkdown(Page $page): string
{
if ($page->markdown) {
return '# ' . $page->name . "\n\n" . $page->markdown;
}
return '# ' . $page->name . "\n\n" . (new HtmlToMarkdown($page->html))->convert();
}
/**
* Convert a chapter to a Markdown file.
*/
public function chapterToMarkdown(Chapter $chapter): string
{
$text = '# ' . $chapter->name . "\n\n";
$text .= $chapter->description . "\n\n";
foreach ($chapter->pages as $page) {
$text .= $this->pageToMarkdown($page) . "\n\n";
}
return trim($text);
}
/**
* Convert a book into a plain text string.
*/
public function bookToMarkdown(Book $book): string
{
$bookTree = (new BookContents($book))->getTree(false, true);
$text = '# ' . $book->name . "\n\n";
foreach ($bookTree as $bookChild) {
if ($bookChild instanceof Chapter) {
$text .= $this->chapterToMarkdown($bookChild) . "\n\n";
} else {
$text .= $this->pageToMarkdown($bookChild) . "\n\n";
}
}
return trim($text);
}
}

139
app/Entities/Tools/PdfGenerator.php
View File

@ -1,139 +0,0 @@
<?php
namespace BookStack\Entities\Tools;
use BookStack\Exceptions\PdfExportException;
use Knp\Snappy\Pdf as SnappyPdf;
use Dompdf\Dompdf;
use Symfony\Component\Process\Exception\ProcessTimedOutException;
use Symfony\Component\Process\Process;
class PdfGenerator
{
const ENGINE_DOMPDF = 'dompdf';
const ENGINE_WKHTML = 'wkhtml';
const ENGINE_COMMAND = 'command';
/**
* Generate PDF content from the given HTML content.
* @throws PdfExportException
*/
public function fromHtml(string $html): string
{
return match ($this->getActiveEngine()) {
self::ENGINE_COMMAND => $this->renderUsingCommand($html),
self::ENGINE_WKHTML => $this->renderUsingWkhtml($html),
default => $this->renderUsingDomPdf($html)
};
}
/**
* Get the currently active PDF engine.
* Returns the value of an `ENGINE_` const on this class.
*/
public function getActiveEngine(): string
{
if (config('exports.pdf_command')) {
return self::ENGINE_COMMAND;
}
if ($this->getWkhtmlBinaryPath() && config('app.allow_untrusted_server_fetching') === true) {
return self::ENGINE_WKHTML;
}
return self::ENGINE_DOMPDF;
}
protected function getWkhtmlBinaryPath(): string
{
$wkhtmlBinaryPath = config('exports.snappy.pdf_binary');
if (file_exists(base_path('wkhtmltopdf'))) {
$wkhtmlBinaryPath = base_path('wkhtmltopdf');
}
return $wkhtmlBinaryPath ?: '';
}
protected function renderUsingDomPdf(string $html): string
{
$options = config('exports.dompdf');
$domPdf = new Dompdf($options);
$domPdf->setBasePath(base_path('public'));
$domPdf->loadHTML($this->convertEntities($html));
$domPdf->render();
return (string) $domPdf->output();
}
/**
* @throws PdfExportException
*/
protected function renderUsingCommand(string $html): string
{
$command = config('exports.pdf_command');
$inputHtml = tempnam(sys_get_temp_dir(), 'bs-pdfgen-html-');
$outputPdf = tempnam(sys_get_temp_dir(), 'bs-pdfgen-output-');
$replacementsByPlaceholder = [
'{input_html_path}' => $inputHtml,
'{output_pdf_path}' => $outputPdf,
];
foreach ($replacementsByPlaceholder as $placeholder => $replacement) {
$command = str_replace($placeholder, escapeshellarg($replacement), $command);
}
file_put_contents($inputHtml, $html);
$timeout = intval(config('exports.pdf_command_timeout'));
$process = Process::fromShellCommandline($command);
$process->setTimeout($timeout);
try {
$process->run();
} catch (ProcessTimedOutException $e) {
throw new PdfExportException("PDF Export via command failed due to timeout at {$timeout} second(s)");
}
if (!$process->isSuccessful()) {
throw new PdfExportException("PDF Export via command failed with exit code {$process->getExitCode()}, stdout: {$process->getOutput()}, stderr: {$process->getErrorOutput()}");
}
$pdfContents = file_get_contents($outputPdf);
unlink($outputPdf);
if ($pdfContents === false) {
throw new PdfExportException("PDF Export via command failed, unable to read PDF output file");
} else if (empty($pdfContents)) {
throw new PdfExportException("PDF Export via command failed, PDF output file is empty");
}
return $pdfContents;
}
protected function renderUsingWkhtml(string $html): string
{
$snappy = new SnappyPdf($this->getWkhtmlBinaryPath());
$options = config('exports.snappy.options');
return $snappy->getOutputFromHtml($html, $options);
}
/**
* Taken from https://github.com/barryvdh/laravel-dompdf/blob/v2.1.1/src/PDF.php
* Copyright (c) 2021 barryvdh, MIT License
* https://github.com/barryvdh/laravel-dompdf/blob/v2.1.1/LICENSE
*/
protected function convertEntities(string $subject): string
{
$entities = [
'€' => '&euro;',
'£' => '&pound;',
];
foreach ($entities as $search => $replace) {
$subject = str_replace($search, $replace, $subject);
}
return $subject;
}
}