<?php

/**
This script will find and report broken internal shelf/book/chapter/page links
within the HTML of shelf/book/chapter descriptions or page content.

This is very simplistic and used just as quick helper.
This script is unsupported, may have issues, and may break with future BookStack versions.

== Usage ==
Download the script to your system, then run it via artisan tinker on the BookStack instance.
The output will show scan progress, then list all detected links by shelf/book/chapter/page.
== Example ==
cd /var/www/bookstack
wget https://codeberg.org/bookstack/devops/raw/branch/main/tinker-scripts/find-broken-internal-references.php
php artisan tinker ./find-broken-internal-references.php --no-interaction
*/

use BookStack\Entities\Models\Book;
use BookStack\Entities\Models\Bookshelf;
use BookStack\Entities\Models\Chapter;
use BookStack\Entities\Models\Entity;
use BookStack\Entities\Models\Page;
use BookStack\Entities\Queries\ChapterQueries;
use BookStack\Entities\Queries\PageQueries;
use Illuminate\Database\Eloquent\Builder;

function findLinksInContent(string $content): array
{
    $pattern = '(' . preg_quote(url('/'), '/') . '.+?)([\s"#?<]|$)';
    $matches = [];
    preg_match_all("/$pattern/i", $content, $matches);
    return array_unique($matches[1]);
}

function linkTargetExists(string $link): bool|null
{
    static $pageQueries = null;
    static $chapterQueries = null;

    if (is_null($pageQueries)) {
        $pageQueries = app()->make(PageQueries::class);
        $chapterQueries = app()->make(ChapterQueries::class);
    }

    // Page
    $pageMatch = preg_match('/\/books\/(.+?)\/page\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
    if ($pageMatch) {
        return $pageQueries->usingSlugs($matches[1], $matches[2])->exists();
    }

    // Page permalink
    $pageMatch = preg_match('/\/link\/(\d+?)([\/?#\s$]|$)/i', $link, $matches);
    if ($pageMatch) {
        return Page::query()->where('id', '=', intval($matches[1]))->exists();
    }

    // Chapter
    $chapterMatch = preg_match('/\/books\/(.+?)\/chapter\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
    if ($chapterMatch) {
        return $chapterQueries->usingSlugs($matches[1], $matches[2])->exists();
    }

    // Book
    $bookMatch = preg_match('/\/books\/(.+?)([\/?#\s$]|$)/i', $link, $matches);
    if ($bookMatch) {
        return Book::query()->where('slug', '=', $matches[1])->exists();
    }

    // Shelf
    $shelfMatch = preg_match('/\/shelves\/(.*?)([\/?#\s$]|$)/i', $link, $matches);
    if ($shelfMatch) {
        return Bookshelf::query()->where('slug', '=', $matches[1])->exists();
    }

    return null;
}
function isLinkBroken(string $link): bool
{
    static $linkCache = [];

    if (isset($linkCache[$link])) {
        return $linkCache[$link];
    }

    $exists = linkTargetExists($link);
    $isBroken = $exists === false;

    $linkCache[$link] = $isBroken;
    return $isBroken;
}

function findBrokenLinksInEntity(Entity $entity): array
{
    $allLinks = findLinksInContent($entity->getAttribute($entity->htmlField));
    return array_filter($allLinks, fn($link) => isLinkBroken($link));
}

function reportForEntityType(Builder $baseQuery, string $entityType): array
{
    $total = $baseQuery->clone()->count();
    $current = 0;
    $count = 500;
    $logLines = [];
    $baseQuery->clone()->chunk($count, function ($entities) use ($count, &$current, $total, $entityType, &$logLines) {
        $rangeEnd = min($count + $current, $total);
        echo "Checking {$entityType}s [{$current}-{$rangeEnd} of {$total}]\n";
        foreach ($entities as $entity) {
            $brokenLinks = findBrokenLinksInEntity($entity);
            if (count($brokenLinks) > 0) {
                $logLines[] = "------------------------------------------------------";
                $logLines[] = "Broken/Old URLs found in {$entityType} {$entity->getUrl()}";
                $logLines[] = "|";
            }
            foreach ($brokenLinks as $brokenLink) {
                 $logLines[] = "|- $brokenLink";
            }
        }
        $current = $rangeEnd;
    });

    return $logLines;
}

function findBrokenLinks(): void
{
    $logLines = [
        ...reportForEntityType(Bookshelf::query(), 'shelf'),
        ...reportForEntityType(Book::query(), 'book'),
        ...reportForEntityType(Chapter::query(), 'chapter'),
        ...reportForEntityType(Page::query(), 'page'),
    ];

    echo "\n=======================\nSearch Complete\n=======================\n\n";

    echo implode("\n", $logLines) . "\n";
}

findBrokenLinks();