yatra / app / Sitemap / SitemapService.php

SitemapService.php in Yatra – Travel Booking & Tour Operator Software 3.0.14.2, at app/Sitemap/SitemapService.php

286 lines 8.3 KB

Raw Download Zip

1	<?php
2
3	declare(strict_types=1);
4
5	namespace Yatra\Sitemap;
6
7	use Yatra\Database\Tables\ClassificationsTable;
8	use Yatra\Database\Tables\TripsTable;
9
10	/**
11	* Single source of truth for every public Yatra URL that belongs in a sitemap.
12	*
13	* Yatra stores trips and classifications (destinations / activities /
14	* categories) in custom tables, not as WordPress posts, so no sitemap
15	* generator (WP core, Yoast, Rank Math, AIOSEO) can discover them on its own.
16	* This service enumerates every such URL with its last-modified date.
17	*
18	* Everything is exposed as ONE consolidated list (getAllEntries) so Yatra
19	* publishes a single, clean /yatra-sitemap.xml. Per-type access is kept
20	* internal/available but the public sitemap is one file.
21	*
22	* Each entry is: ['loc' => absolute URL, 'lastmod' => W3C datetime\|''].
23	*/
24	class SitemapService
25	{
26	public const TYPE_ARCHIVE = 'archive';
27	public const TYPE_TRIP = 'trip';
28	public const TYPE_DESTINATION = 'destination';
29	public const TYPE_ACTIVITY = 'activity';
30	public const TYPE_CATEGORY = 'category';
31
32	/** Order entries appear in the consolidated sitemap. */
33	private const TYPES = [
34	self::TYPE_ARCHIVE,
35	self::TYPE_TRIP,
36	self::TYPE_DESTINATION,
37	self::TYPE_ACTIVITY,
38	self::TYPE_CATEGORY,
39	];
40
41	private const CACHE_KEY = 'yatra_sitemap_entries_all';
42	private const CACHE_TTL = HOUR_IN_SECONDS;
43
44	/** sitemaps.org caps a single sitemap file at 50,000 URLs. */
45	private const MAX_URLS = 50000;
46
47	/**
48	* Every public Yatra URL, merged into one de-duplicated, cached list.
49	*
50	* @return array<int, array{loc: string, lastmod: string}>
51	*/
52	public function getAllEntries(): array
53	{
54	static $runtime = null;
55	if ($runtime !== null) {
56	return $runtime;
57	}
58
59	$cached = get_transient(self::CACHE_KEY);
60	if (is_array($cached)) {
61	$runtime = $cached;
62	return $cached;
63	}
64
65	$entries = [];
66	foreach (self::TYPES as $type) {
67	foreach ($this->entriesForType($type) as $entry) {
68	$entries[] = $entry;
69	}
70	}
71	$entries = $this->dedupe($entries);
72
73	/**
74	* Final say over the sitemap's entries — lets integrations drop
75	* noindex URLs, add multilingual alternates, etc.
76	*
77	* @param array<int, array{loc: string, lastmod: string}> $entries
78	*/
79	$entries = (array) apply_filters('yatra_sitemap_entries', $entries);
80
81	// Never emit more than a single sitemap file may legally hold. Realistic
82	// travel catalogues are orders of magnitude below this; the cap is a
83	// safety valve so an extreme catalogue can't produce an invalid file.
84	$max = (int) apply_filters('yatra_sitemap_max_urls', self::MAX_URLS);
85	if ($max > 0 && count($entries) > $max) {
86	$entries = array_slice($entries, 0, $max);
87	}
88
89	set_transient(self::CACHE_KEY, $entries, self::CACHE_TTL);
90	$runtime = $entries;
91
92	return $entries;
93	}
94
95	public function getCount(): int
96	{
97	return count($this->getAllEntries());
98	}
99
100	/**
101	* Newest lastmod across all entries (for index <lastmod> in SEO plugins).
102	*/
103	public function getLatestLastmod(): string
104	{
105	$latest = '';
106	foreach ($this->getAllEntries() as $entry) {
107	if ($entry['lastmod'] !== '' && $entry['lastmod'] > $latest) {
108	$latest = $entry['lastmod'];
109	}
110	}
111
112	return $latest;
113	}
114
115	/**
116	* Drop the cached list. Call when a trip or classification changes.
117	*/
118	public static function flushCache(): void
119	{
120	delete_transient(self::CACHE_KEY);
121	}
122
123	/**
124	* @return array<int, array{loc: string, lastmod: string}>
125	*/
126	private function entriesForType(string $type): array
127	{
128	switch ($type) {
129	case self::TYPE_ARCHIVE:
130	return $this->archiveEntries();
131	case self::TYPE_TRIP:
132	return $this->tripEntries();
133	case self::TYPE_DESTINATION:
134	case self::TYPE_ACTIVITY:
135	case self::TYPE_CATEGORY:
136	return $this->classificationEntries($type);
137	default:
138	return [];
139	}
140	}
141
142	/**
143	* The trip listing page plus each taxonomy index page.
144	*
145	* @return array<int, array{loc: string, lastmod: string}>
146	*/
147	private function archiveEntries(): array
148	{
149	$entries = [];
150
151	$listing = yatra_get_trip_listing_url();
152	if ($listing !== '') {
153	$entries[] = ['loc' => $listing, 'lastmod' => ''];
154	}
155
156	if (function_exists('yatra_get_taxonomy_listing_url')) {
157	foreach (['destination', 'activity', 'trip-category'] as $listingType) {
158	$url = yatra_get_taxonomy_listing_url($listingType);
159	if ($url !== '' && $url !== home_url('/')) {
160	$entries[] = ['loc' => $url, 'lastmod' => ''];
161	}
162	}
163	}
164
165	return $entries;
166	}
167
168	/**
169	* @return array<int, array{loc: string, lastmod: string}>
170	*/
171	private function tripEntries(): array
172	{
173	global $wpdb;
174
175	$table = TripsTable::getTableName();
176	$rows = $wpdb->get_results(
177	"SELECT id, slug, updated_at
178	FROM `{$table}`
179	WHERE status IN ('publish', 'published')
180	AND (deleted_at IS NULL OR deleted_at = '0000-00-00 00:00:00')
181	AND slug <> ''
182	ORDER BY updated_at DESC"
183	);
184
185	$entries = [];
186	foreach ((array) $rows as $row) {
187	$loc = yatra_get_trip_permalink($row);
188	if ($loc === '') {
189	continue;
190	}
191	$entries[] = ['loc' => $loc, 'lastmod' => $this->toW3c((string) ($row->updated_at ?? ''))];
192	}
193
194	return $entries;
195	}
196
197	/**
198	* @return array<int, array{loc: string, lastmod: string}>
199	*/
200	private function classificationEntries(string $type): array
201	{
202	global $wpdb;
203
204	$table = ClassificationsTable::getTableName();
205	$rows = $wpdb->get_results(
206	$wpdb->prepare(
207	"SELECT id, slug, updated_at
208	FROM `{$table}`
209	WHERE type = %s
210	AND status = 'publish'
211	AND slug <> ''
212	ORDER BY updated_at DESC",
213	$type
214	)
215	);
216
217	$resolver = $this->permalinkResolver($type);
218
219	$entries = [];
220	foreach ((array) $rows as $row) {
221	$loc = $resolver($row);
222	if ($loc === '') {
223	continue;
224	}
225	$entries[] = ['loc' => $loc, 'lastmod' => $this->toW3c((string) ($row->updated_at ?? ''))];
226	}
227
228	return $entries;
229	}
230
231	/**
232	* @return callable(object): string
233	*/
234	private function permalinkResolver(string $type): callable
235	{
236	switch ($type) {
237	case self::TYPE_DESTINATION:
238	return static fn ($row) => yatra_get_destination_permalink($row);
239	case self::TYPE_ACTIVITY:
240	return static fn ($row) => yatra_get_activity_permalink($row);
241	case self::TYPE_CATEGORY:
242	return static fn ($row) => yatra_get_category_permalink($row);
243	default:
244	return static fn ($row) => '';
245	}
246	}
247
248	/**
249	* Convert a MySQL datetime to a W3C / ISO-8601 string for <lastmod>.
250	*/
251	private function toW3c(string $mysqlDate): string
252	{
253	$mysqlDate = trim($mysqlDate);
254	if ($mysqlDate === '' \|\| strpos($mysqlDate, '0000-00-00') === 0) {
255	return '';
256	}
257
258	$timestamp = strtotime($mysqlDate);
259	if ($timestamp === false) {
260	return '';
261	}
262
263	return gmdate('Y-m-d\TH:i:s+00:00', $timestamp);
264	}
265
266	/**
267	* @param array<int, array{loc: string, lastmod: string}> $entries
268	* @return array<int, array{loc: string, lastmod: string}>
269	*/
270	private function dedupe(array $entries): array
271	{
272	$seen = [];
273	$out = [];
274	foreach ($entries as $entry) {
275	$loc = $entry['loc'];
276	if ($loc === '' \|\| isset($seen[$loc])) {
277	continue;
278	}
279	$seen[$loc] = true;
280	$out[] = $entry;
281	}
282
283	return $out;
284	}
285	}
286