BookStack Code Mirror - devops/blob - tinker-scripts/find-broken-internal-references.php

   1 <?php
   2
   3 /**
   4 This script will find and report broken internal shelf/book/chapter/page links
   5 within the HTML of shelf/book/chapter descriptions or page content.
   6
   7 This is very simplistic and used just as quick helper.
   8 This script is unsupported, may have issues, and may break with future BookStack versions.
   9
  10 == Usage ==
  11 Download the script to your system, then run it via artisan tinker on the BookStack instance.
  12 The output will show scan progress, then list all detected links by shelf/book/chapter/page.
  13 == Example ==
  14 cd /var/www/bookstack
  15 wget https://codeberg.org/bookstack/devops/raw/branch/main/tinker-scripts/find-broken-internal-references.php
  16 php artisan tinker ./find-broken-internal-references.php --no-interaction
  17 */
  18
  19 use BookStack\Entities\Models\Book;
  20 use BookStack\Entities\Models\Bookshelf;
  21 use BookStack\Entities\Models\Chapter;
  22 use BookStack\Entities\Models\Entity;
  23 use BookStack\Entities\Models\Page;
  24 use BookStack\Entities\Queries\ChapterQueries;
  25 use BookStack\Entities\Queries\PageQueries;
  26 use Illuminate\Database\Eloquent\Builder;
  27
  28 function findLinksInContent(string $content): array
  29 {
  30     $pattern = '(' . preg_quote(url('/'), '/') . '.+?)([\s"#?<]|$)';
  31     $matches = [];
  32     preg_match_all("/$pattern/i", $content, $matches);
  33     return array_unique($matches[1]);
  34 }
  35
  36 function linkTargetExists(string $link): bool|null
  37 {
  38     static $pageQueries = null;
  39     static $chapterQueries = null;
  40
  41     if (is_null($pageQueries)) {
  42         $pageQueries = app()->make(PageQueries::class);
  43         $chapterQueries = app()->make(ChapterQueries::class);
  44     }
  45
  46     // Page
  47     $pageMatch = preg_match('/\/books\/(.+?)\/page\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
  48     if ($pageMatch) {
  49         return $pageQueries->usingSlugs($matches[1], $matches[2])->exists();
  50     }
  51
  52     // Page permalink
  53     $pageMatch = preg_match('/\/link\/(\d+?)([\/?#\s$]|$)/i', $link, $matches);
  54     if ($pageMatch) {
  55         return Page::query()->where('id', '=', intval($matches[1]))->exists();
  56     }
  57
  58     // Chapter
  59     $chapterMatch = preg_match('/\/books\/(.+?)\/chapter\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
  60     if ($chapterMatch) {
  61         return $chapterQueries->usingSlugs($matches[1], $matches[2])->exists();
  62     }
  63
  64     // Book
  65     $bookMatch = preg_match('/\/books\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
  66     if ($bookMatch) {
  67         return Book::query()->where('slug', '=', $matches[1])->exists();
  68     }
  69
  70     // Shelf
  71     $shelfMatch = preg_match('/\/shelves\/(.*?)([\/?#\s$]|$)/i', $link, $matches);
  72     if ($shelfMatch) {
  73         return Bookshelf::query()->where('slug', '=', $matches[1])->exists();
  74     }
  75
  76     return null;
  77 }
  78 function isLinkBroken(string $link): bool
  79 {
  80     static $linkCache = [];
  81
  82     if (isset($linkCache[$link])) {
  83         return $linkCache[$link];
  84     }
  85
  86     $exists = linkTargetExists($link);
  87     $isBroken = $exists === false;
  88
  89     $linkCache[$link] = $isBroken;
  90     return $isBroken;
  91 }
  92
  93 function findBrokenLinksInEntity(Entity $entity): array
  94 {
  95     $allLinks = findLinksInContent($entity->getAttribute($entity->htmlField));
  96     return array_filter($allLinks, fn($link) => isLinkBroken($link));
  97 }
  98
  99 function reportForEntityType(Builder $baseQuery, string $entityType): array
 100 {
 101     $total = $baseQuery->clone()->count();
 102     $current = 0;
 103     $count = 500;
 104     $logLines = [];
 105     $baseQuery->clone()->chunk($count, function ($entities) use ($count, &$current, $total, $entityType, &$logLines) {
 106         $rangeEnd = min($count + $current, $total);
 107         echo "Checking {$entityType}s [{$current}-{$rangeEnd} of {$total}]\n";
 108         foreach ($entities as $entity) {
 109             $brokenLinks = findBrokenLinksInEntity($entity);
 110             if (count($brokenLinks) > 0) {
 111                 $logLines[] = "------------------------------------------------------";
 112                 $logLines[] = "Broken/Old URLs found in {$entityType} {$entity->getUrl()}";
 113                 $logLines[] = "|";
 114             }
 115             foreach ($brokenLinks as $brokenLink) {
 116                  $logLines[] = "|- $brokenLink";
 117             }
 118         }
 119         $current = $rangeEnd;
 120     });
 121
 122     return $logLines;
 123 }
 124
 125 function findBrokenLinks(): void
 126 {
 127     $logLines = [
 128         ...reportForEntityType(Bookshelf::query(), 'shelf'),
 129         ...reportForEntityType(Book::query(), 'book'),
 130         ...reportForEntityType(Chapter::query(), 'chapter'),
 131         ...reportForEntityType(Page::query(), 'page'),
 132     ];
 133
 134     echo "\n=======================\nSearch Complete\n=======================\n\n";
 135
 136     echo implode("\n", $logLines) . "\n";
 137 }
 138
 139 findBrokenLinks();