refactor search functionality; remove deprecated services and implement new action and task classes for improved structure and maintainability

This commit is contained in:
F4ilji
2025-07-18 14:14:10 +05:00
parent 0874d3e6b5
commit b8235404b8
20 changed files with 525 additions and 440 deletions
@@ -0,0 +1,87 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Str;
use RecursiveDirectoryIterator;
use RecursiveIteratorIterator;
use App\Containers\Search\Tasks\ExtractHtmlContentTask;
use App\Containers\Search\Tasks\GetBreadcrumbFromHtmlTask;
class BuildStaticFileIndexTask
{
const CACHE_KEY = 'static_html_index_v2';
const FILES_DIR = 'sveden';
const CACHE_TTL = 86400; // 24 часа
public function run(): array
{
return Cache::remember(self::CACHE_KEY, self::CACHE_TTL, function() {
try {
$index = [];
$directory = public_path(self::FILES_DIR);
if (!is_dir($directory)) {
Log::error("Directory not found: {$directory}");
return [];
}
$iterator = new RecursiveIteratorIterator(
new RecursiveDirectoryIterator($directory, RecursiveDirectoryIterator::SKIP_DOTS),
RecursiveIteratorIterator::SELF_FIRST
);
foreach ($iterator as $file) {
if ($file->isFile() && $this->isHtmlFile($file)) {
$html = file_get_contents($file->getPathname());
[$content, $breadcrumb] = app(ExtractHtmlContentTask::class)->run($html);
if ($content !== false) {
$breadcrumb = app(GetBreadcrumbFromHtmlTask::class)->run($breadcrumb);
$text = $this->normalizeText(strip_tags($content));
$index[$file->getPathname()] = [
'title' => $this->getFirstH1Content($content),
'content' => $text,
'category' => $breadcrumb ?? null,
];
}
}
}
return $index;
} catch (\Exception $e) {
Log::error('Index creation error: ' . $e->getMessage());
return [];
}
});
}
protected function isHtmlFile(\SplFileInfo $file): bool
{
$extension = strtolower($file->getExtension());
return in_array($extension, ['html', 'htm']);
}
protected function normalizeText(string $text): string
{
$text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
$text = preg_replace('/\s+/u', ' ', $text);
$text = trim($text);
return Str::lower($text);
}
public function getFirstH1Content(string $content): ?string
{
try {
if (preg_match('/<h1[^>]*>(.*?)<\/h1>/is', $content, $matches)) {
return $this->normalizeText($matches[1]);
}
return null;
} catch (\Exception $e) {
Log::error('Failed to get H1: ' . $e->getMessage());
return null;
}
}
}
@@ -0,0 +1,24 @@
<?php
namespace App\Containers\Search\Tasks;
use DiDom\Document;
class ExtractHtmlContentTask
{
public function run(string $html): ?array
{
$document = new Document($html);
$content = $document->first('.vikon-content');
if ($content === null) {
return null;
}
$breadcrumb = $content->first('.row');
if ($breadcrumb !== null) {
$content->firstInDocument('.row')->remove();
}
return [$content->html(), $breadcrumb->html() ?? null];
}
}
@@ -0,0 +1,17 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Support\Collection;
class FilterSearchResultsByCategoryTask
{
public function run(Collection $resources, ?string $category): Collection
{
if ($category === null || $category === "All") {
return $resources;
} else {
return $resources->where('type', $category);
}
}
}
@@ -0,0 +1,13 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Support\Collection;
class GetAvailableSearchCategoriesTask
{
public function run(Collection $resources): array
{
return $resources->pluck('type')->unique()->values()->all();
}
}
@@ -0,0 +1,22 @@
<?php
namespace App\Containers\Search\Tasks;
use DiDom\Document;
class GetBreadcrumbFromHtmlTask
{
public function run(string $html): ?string
{
$document = new Document($html);
$breadcrumbs = $document->first('ol.breadcrumb')?->find('li') ?? [];
foreach ($breadcrumbs as $index => $breadcrumb) {
if ($index === 2) { // Индексация с 0 → третий элемент = 2
return trim($breadcrumb->text());
}
}
return null;
}
}
@@ -0,0 +1,45 @@
<?php
namespace App\Containers\Search\Tasks;
use DiDom\Document;
use Illuminate\Support\Facades\Log;
class GetStaticFileCategoriesTask
{
const FILES_DIR = 'sveden';
public function run(): array
{
try {
$file = public_path(self::FILES_DIR . '/' . 'index.html');
if (!file_exists($file)) {
Log::error("Category index file not found: {$file}");
return [];
}
$html = file_get_contents($file);
$document = new Document($html);
$dropdownMenu = $document->first('ul.dropdown-menu');
if ($dropdownMenu === null) {
Log::warning("Dropdown menu not found in category index file: {$file}");
return [];
}
$links = $dropdownMenu->find('a');
$categories = [];
foreach ($links as $link) {
$category = trim($link->text());
$categories[] = $category;
}
return $categories;
} catch (\Exception $e) {
Log::error('Error extracting categories: ' . $e->getMessage());
return [];
}
}
}
@@ -0,0 +1,22 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Support\Facades\Cache;
class GetStaticSearchCacheStatusTask
{
const CACHE_KEY = 'static_html_index_v2';
const FILES_DIR = 'sveden';
public function run(): array
{
return [
'exists' => Cache::has(self::CACHE_KEY),
'ttl' => Cache::get(self::CACHE_KEY.'_ttl', null),
'driver' => config('cache.default'),
'path' => public_path(self::FILES_DIR),
'directory_exists' => is_dir(public_path(self::FILES_DIR))
];
}
}
@@ -0,0 +1,33 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Http\Request;
use Illuminate\Pagination\LengthAwarePaginator;
use Illuminate\Support\Collection;
class PaginateCollectionTask
{
public function run(Collection $resources, Request $request, int $perPage = 10): array
{
$currentPage = LengthAwarePaginator::resolveCurrentPage();
// Отрезаем нужные элементы для текущей страницы
$currentItems = $resources->slice(($currentPage - 1) * $perPage, $perPage)->all();
// Создаем экземпляр LengthAwarePaginator
$paginator = new LengthAwarePaginator($currentItems, count($resources), $perPage, $currentPage, [
'path' => $request->url(),
'query' => $request->query(), // Changed from $request->query to $request->query()
]);
$nextPage = $paginator->hasMorePages() ? $paginator->currentPage() + 1 : null;
$prevPage = $paginator->onFirstPage() ? null : $paginator->currentPage() - 1;
return [
'paginator' => $paginator,
'next_page' => $nextPage,
'prev_page' => $prevPage
];
}
}
@@ -0,0 +1,27 @@
<?php
namespace App\Containers\Search\Tasks;
class PaginateSearchResultsTask
{
public function run(array $results, int $page, int $perPage, ?array $categories): array
{
$total = count($results);
$lastPage = max(1, ceil($total / $perPage));
$page = max(1, min($page, $lastPage));
$offset = ($page - 1) * $perPage;
$paginatedResults = array_slice($results, $offset, $perPage);
return [
'data' => $paginatedResults,
'meta' => [
'current_page' => $page,
'total' => $total,
'per_page' => $perPage,
'last_page' => $lastPage
],
'categories' => $categories
];
}
}
@@ -0,0 +1,42 @@
<?php
namespace App\Containers\Search\Tasks;
use Illuminate\Support\Str;
use Illuminate\Support\Collection;
class SortAndHighlightSearchResultsTask
{
public function run(Collection $data, string $searchRequest): array
{
$sortedData = [];
foreach ($data as $item) {
$searchData = $item['search_data'] ?? '';
$matches = $this->getMatches($searchData, $searchRequest);
$sortedData[$item['type']][] = [
'data' => $item,
'matches' => $matches,
'tag' => $item['type']
];
}
return $sortedData;
}
private function getMatches(string $haystack, string $needle): array
{
$matches = [];
$offset = 0;
if (($offset = mb_strpos($haystack, $needle, $offset, 'UTF-8')) !== false) {
for ($i = 0; 1 > count($matches); $i++) {
$left = max(0, $offset - 50);
$right = min(mb_strlen($haystack, 'UTF-8'), $offset + 100);
$excerpt = mb_substr($haystack, $left, $right - $left, 'UTF-8');
$matches[] = $excerpt;
}
}
return $matches;
}
}