thinkrank / includes / seo / class-performance-data-collector.php

class-performance-data-collector.php in ThinkRank AI SEO – AI SEO Plugin for WordPress: Schema, XML Sitemaps, Meta Tags, Search Console & Local SEO 1.1.0, at includes/seo/class-performance-data-collector.php

355 lines 11.5 KB

Raw Download Zip

1	<?php
2	/**
3	* Performance Data Collector Class
4	*
5	* Automated system for collecting and storing performance data from Google PageSpeed Insights.
6	* Runs on WordPress cron to build historical performance tracking over time.
7	*
8	* @package ThinkRank\SEO
9	* @since 1.0.0
10	*/
11
12	declare(strict_types=1);
13
14	namespace ThinkRank\SEO;
15
16	use ThinkRank\Integrations\Google_PageSpeed_Client;
17
18	/**
19	* Performance Data Collector Class
20	*
21	* Handles automated collection and storage of performance data for historical tracking.
22	* Integrates with WordPress cron system for scheduled data collection.
23	*
24	* @since 1.0.0
25	*/
26	class Performance_Data_Collector {
27
28	/**
29	* Performance Monitoring Manager instance
30	*
31	* @var Performance_Monitoring_Manager
32	*/
33	private Performance_Monitoring_Manager $performance_manager;
34
35	/**
36	* Google PageSpeed Client instance
37	*
38	* @var Google_PageSpeed_Client\|null
39	*/
40	private ?Google_PageSpeed_Client $pagespeed_client;
41
42	/**
43	* Cron hook name for data collection
44	*/
45	private const CRON_HOOK = 'thinkrank_collect_performance_data';
46
47	/**
48	* Constructor
49	*/
50	public function __construct() {
51	$this->performance_manager = new Performance_Monitoring_Manager();
52
53	// Initialize PageSpeed client with proper class loading
54	$this->initialize_pagespeed_client();
55
56	// Register cron hooks
57	add_action(self::CRON_HOOK, [$this, 'collect_performance_data']);
58
59	// Schedule cron if not already scheduled
60	if (!wp_next_scheduled(self::CRON_HOOK)) {
61	wp_schedule_event(time(), 'daily', self::CRON_HOOK);
62	}
63	}
64
65	/**
66	* Initialize PageSpeed client with proper error handling
67	*
68	* @return void
69	*/
70	private function initialize_pagespeed_client(): void {
71	// Ensure the Google PageSpeed Client class is loaded
72	if (!class_exists('ThinkRank\\Integrations\\Google_PageSpeed_Client')) {
73	$pagespeed_file = THINKRANK_PLUGIN_DIR . 'includes/integrations/class-google-pagespeed-client.php';
74	if (file_exists($pagespeed_file)) {
75	require_once $pagespeed_file;
76	}
77	}
78
79	// Ensure the base client is loaded
80	if (!class_exists('ThinkRank\\Integrations\\Google_API_Base_Client')) {
81	$base_client_file = THINKRANK_PLUGIN_DIR . 'includes/integrations/class-google-api-base-client.php';
82	if (file_exists($base_client_file)) {
83	require_once $base_client_file;
84	}
85	}
86
87	try {
88	// Get Google PageSpeed API key from settings
89	$api_key = $this->get_pagespeed_api_key();
90
91	if (empty($api_key)) {
92	$this->pagespeed_client = null;
93	return;
94	}
95
96	$this->pagespeed_client = new Google_PageSpeed_Client($api_key);
97	} catch (\Exception $e) {
98	$this->pagespeed_client = null;
99	}
100	}
101
102	/**
103	* Get Google PageSpeed API key from settings
104	*
105	* @return string API key or empty string if not configured
106	*/
107	private function get_pagespeed_api_key(): string {
108	// Get the API key from integrations settings
109	$integrations_settings = get_option('thinkrank_integrations_settings', []);
110	return $integrations_settings['google_pagespeed_api_key'] ?? '';
111	}
112
113	/**
114	* Collect performance data for the site
115	*
116	* @return bool Success status
117	*/
118	public function collect_performance_data(): bool {
119	try {
120	$home_url = home_url();
121
122	// Test both mobile and desktop
123	$devices = ['mobile', 'desktop'];
124	$success = true;
125
126	foreach ($devices as $device) {
127	$device_success = $this->collect_device_performance_data($home_url, $device);
128	if (!$device_success) {
129	$success = false;
130	}
131	}
132
133	// Run data cleanup (keep 1 year of data)
134	$this->cleanup_old_data(365);
135
136	return $success;
137
138	} catch (\Exception $e) {
139	return false;
140	}
141	}
142
143	/**
144	* Collect performance data for specific device type
145	*
146	* @param string $url URL to test
147	* @param string $device_type Device type (mobile/desktop)
148	* @return bool Success status
149	*/
150	private function collect_device_performance_data(string $url, string $device_type): bool {
151	try {
152	// Check if PageSpeed client is available
153	if (!$this->pagespeed_client) {
154	return false;
155	}
156
157	// Get Core Web Vitals data
158	$core_web_vitals = $this->pagespeed_client->get_core_web_vitals($url, $device_type);
159
160	// Run full PageSpeed test to get performance score
161	$pagespeed_result = $this->pagespeed_client->run_pagespeed_test($url, $device_type, ['performance']);
162
163	// Extract performance score
164	$performance_score = 0;
165	if (isset($pagespeed_result['lighthouseResult']['categories']['performance']['score'])) {
166	$performance_score = $pagespeed_result['lighthouseResult']['categories']['performance']['score'] * 100;
167	}
168
169	// Prepare data for storage
170	$performance_data = $core_web_vitals;
171	$performance_data['performance_score'] = $performance_score;
172
173	// Store in database
174	$stored = $this->performance_manager->store_historical_performance_data(
175	$performance_data,
176	'site',
177	null,
178	$device_type
179	);
180
181
182
183	return $stored;
184
185	} catch (\Exception $e) {
186	return false;
187	}
188	}
189
190	/**
191	* Manually trigger data collection (for testing or immediate collection)
192	*
193	* @return array Collection results
194	*/
195	public function manual_collect(): array {
196	$results = [
197	'success' => false,
198	'message' => '',
199	'data_collected' => false,
200	'errors' => []
201	];
202
203	try {
204	$success = $this->collect_performance_data();
205
206	if ($success) {
207	$results['success'] = true;
208	$results['data_collected'] = true;
209	$results['message'] = __('Performance data collected successfully', 'thinkrank');
210	} else {
211	$results['message'] = __('Failed to collect performance data', 'thinkrank');
212	$results['errors'][] = 'Data collection failed';
213	}
214
215	} catch (\Exception $e) {
216	$results['message'] = __('Error during data collection', 'thinkrank');
217	$results['errors'][] = $e->getMessage();
218	}
219
220	return $results;
221	}
222
223	/**
224	* Get collection schedule information
225	*
226	* @return array Schedule information
227	*/
228	public function get_schedule_info(): array {
229	$next_scheduled = wp_next_scheduled(self::CRON_HOOK);
230
231	return [
232	'is_scheduled' => $next_scheduled !== false,
233	'next_run' => $next_scheduled ? gmdate('Y-m-d H:i:s', $next_scheduled) : null,
234	'next_run_human' => $next_scheduled ? human_time_diff($next_scheduled) : null,
235	'cron_hook' => self::CRON_HOOK,
236	'frequency' => 'daily'
237	];
238	}
239
240	/**
241	* Clean up old performance data (data retention policy)
242	*
243	* @param int $days Number of days to keep (default: 365 days = 1 year)
244	* @return int Number of records deleted
245	*/
246	public function cleanup_old_data(int $days = 365): int {
247	global $wpdb;
248
249	$table_name = $wpdb->prefix . 'thinkrank_seo_performance';
250	$cutoff_date = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));
251
252	$sql = sprintf(
253	"DELETE FROM %s WHERE measured_at < %%s",
254	$table_name
255	);
256
257	// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Performance data cleanup requires direct database access
258	$deleted = $wpdb->query(
259	// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- SQL is properly prepared with placeholders
260	$wpdb->prepare($sql, $cutoff_date)
261	);
262
263	return $deleted !== false ? (int) $deleted : 0;
264	}
265
266	/**
267	* Reschedule data collection
268	*
269	* @param string $frequency Cron frequency (hourly, daily, weekly)
270	* @return bool Success status
271	*/
272	public function reschedule_collection(string $frequency = 'daily'): bool {
273	try {
274	// Clear existing schedule
275	wp_clear_scheduled_hook(self::CRON_HOOK);
276
277	// Schedule new collection
278	$scheduled = wp_schedule_event(time(), $frequency, self::CRON_HOOK);
279
280	return $scheduled !== false;
281
282	} catch (\Exception $e) {
283	return false;
284	}
285	}
286
287	/**
288	* Clear scheduled data collection
289	*
290	* @return bool Success status
291	*/
292	public function clear_schedule(): bool {
293	try {
294	wp_clear_scheduled_hook(self::CRON_HOOK);
295	return true;
296	} catch (\Exception $e) {
297	return false;
298	}
299	}
300
301	/**
302	* Get recent collection statistics
303	*
304	* @param int $days Number of days to check
305	* @return array Collection statistics
306	*/
307	public function get_collection_stats(int $days = 7): array {
308	global $wpdb;
309
310	$table_name = $wpdb->prefix . 'thinkrank_seo_performance';
311	$start_date = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));
312
313	try {
314	$sql = sprintf("
315	SELECT
316	COUNT(*) as total_records,
317	COUNT(DISTINCT DATE(measured_at)) as days_with_data,
318	COUNT(DISTINCT device_type) as device_types,
319	MIN(measured_at) as first_measurement,
320	MAX(measured_at) as last_measurement
321	FROM %s
322	WHERE measured_at >= %%s
323	AND context_type = 'site'
324	AND measured_by = 'google_pagespeed'
325	", $table_name);
326
327	// phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Performance statistics retrieval requires direct database access
328	$stats = $wpdb->get_row(
329	// phpcs:ignore WordPress.DB.PreparedSQL.NotPrepared -- SQL is properly prepared with placeholders
330	$wpdb->prepare($sql, $start_date),
331	ARRAY_A
332	);
333
334	return [
335	'total_records' => (int) ($stats['total_records'] ?? 0),
336	'days_with_data' => (int) ($stats['days_with_data'] ?? 0),
337	'device_types' => (int) ($stats['device_types'] ?? 0),
338	'first_measurement' => $stats['first_measurement'] ?? null,
339	'last_measurement' => $stats['last_measurement'] ?? null,
340	'collection_rate' => $stats['days_with_data'] ? round(($stats['days_with_data'] / $days) * 100, 1) : 0
341	];
342
343	} catch (\Exception $e) {
344	return [
345	'total_records' => 0,
346	'days_with_data' => 0,
347	'device_types' => 0,
348	'first_measurement' => null,
349	'last_measurement' => null,
350	'collection_rate' => 0
351	];
352	}
353	}
354	}
355