| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* Google Search Console Client Class |
| 5 |
* |
| 6 |
* Handles communication with Google Search Console API for search performance |
| 7 |
* data retrieval and site verification. Extends the base Google API client with |
| 8 |
* Search Console-specific functionality and rate limiting. |
| 9 |
* |
| 10 |
* @package ThinkRank\Integrations |
| 11 |
* @since 1.0.0 |
| 12 |
*/ |
| 13 |
|
| 14 |
declare(strict_types=1); |
| 15 |
|
| 16 |
namespace ThinkRank\Integrations; |
| 17 |
|
| 18 |
// Prevent direct access |
| 19 |
if (!defined('ABSPATH')) { |
| 20 |
exit; |
| 21 |
} |
| 22 |
|
| 23 |
/** |
| 24 |
* Google Search Console Client Class |
| 25 |
* |
| 26 |
* Single Responsibility: Handle Google Search Console API communication |
| 27 |
* Following ThinkRank HTTP client patterns from Claude_Client and OpenAI_Client |
| 28 |
* |
| 29 |
* @since 1.0.0 |
| 30 |
*/ |
| 31 |
class Google_Search_Console_Client extends Google_API_Base_Client { |
| 32 |
|
| 33 |
/** |
| 34 |
* Google Search Console API base URL |
| 35 |
*/ |
| 36 |
private const API_BASE_URL = 'https://www.googleapis.com/webmasters/v3'; |
| 37 |
|
| 38 |
/** |
| 39 |
* Rate limit transient key prefix |
| 40 |
* Following ThinkRank option naming patterns |
| 41 |
*/ |
| 42 |
private const RATE_LIMIT_KEY = 'thinkrank_gsc_rate_limit'; |
| 43 |
|
| 44 |
/** |
| 45 |
* Maximum requests per day (Google standard quota) |
| 46 |
*/ |
| 47 |
private const MAX_REQUESTS_PER_DAY = 2000; |
| 48 |
|
| 49 |
/** |
| 50 |
* Test API connection |
| 51 |
* Following ThinkRank test_connection patterns from AI clients |
| 52 |
* |
| 53 |
* @return array Connection test results |
| 54 |
*/ |
| 55 |
public function test_connection(): array { |
| 56 |
try { |
| 57 |
$result = $this->list_sites(); |
| 58 |
|
| 59 |
return [ |
| 60 |
'success' => true, |
| 61 |
'message' => 'Google Search Console API connection successful', |
| 62 |
'sites_count' => count($result['siteEntry'] ?? []) |
| 63 |
]; |
| 64 |
} catch (\Exception $e) { |
| 65 |
// `message` as well as `error`: the success branch above reports |
| 66 |
// its outcome as `message`, and every consumer reads that key, so |
| 67 |
// a failure that only set `error` was reported to the admin as an |
| 68 |
// error with no reason at all (#852). `error` is kept for any |
| 69 |
// caller that already reads it. |
| 70 |
return [ |
| 71 |
'success' => false, |
| 72 |
'message' => $e->getMessage(), |
| 73 |
'error' => $e->getMessage() |
| 74 |
]; |
| 75 |
} |
| 76 |
} |
| 77 |
|
| 78 |
/** |
| 79 |
* List verified sites in Search Console |
| 80 |
* |
| 81 |
* @return array List of verified sites |
| 82 |
* @throws \Exception If API request fails |
| 83 |
*/ |
| 84 |
public function list_sites(): array { |
| 85 |
$endpoint = '/sites'; |
| 86 |
// The API key (when no OAuth token) is sent via the x-goog-api-key |
| 87 |
// header by the base client — not the query string. |
| 88 |
$full_url = self::API_BASE_URL . $endpoint; |
| 89 |
return $this->make_request($full_url, [], 'GET'); |
| 90 |
} |
| 91 |
|
| 92 |
/** |
| 93 |
* Verify site ownership in Search Console |
| 94 |
* |
| 95 |
* @param string $site_url Site URL to verify |
| 96 |
* @param string $verification_method Verification method used |
| 97 |
* @return array Verification results |
| 98 |
*/ |
| 99 |
public function verify_site(string $site_url, string $verification_method = 'meta'): array { |
| 100 |
try { |
| 101 |
// Check if the site is already verified by listing sites |
| 102 |
$sites = $this->list_sites(); |
| 103 |
$site_verified = false; |
| 104 |
|
| 105 |
// A domain property is compared in its punycode form, so an IDN |
| 106 |
// matches whichever form the account lists it in. |
| 107 |
$wanted = \ThinkRank\Core\Url_Validator::search_console_domain_property($site_url) ?? $site_url; |
| 108 |
|
| 109 |
foreach ($sites['siteEntry'] ?? [] as $site) { |
| 110 |
$listed = (string) ($site['siteUrl'] ?? ''); |
| 111 |
if ($listed === $site_url || (\ThinkRank\Core\Url_Validator::search_console_domain_property($listed) ?? $listed) === $wanted) { |
| 112 |
$site_verified = true; |
| 113 |
break; |
| 114 |
} |
| 115 |
} |
| 116 |
|
| 117 |
return [ |
| 118 |
'success' => $site_verified, |
| 119 |
'message' => $site_verified ? 'Site is verified in Search Console' : 'Site not found in Search Console', |
| 120 |
'site_url' => $site_url, |
| 121 |
'verification_method' => $verification_method |
| 122 |
]; |
| 123 |
} catch (\Exception $e) { |
| 124 |
return [ |
| 125 |
'success' => false, |
| 126 |
'error' => $e->getMessage() |
| 127 |
]; |
| 128 |
} |
| 129 |
} |
| 130 |
|
| 131 |
/** |
| 132 |
* Get search performance data with flexible dimensions |
| 133 |
* |
| 134 |
* @param string $site_url Site URL to get data for |
| 135 |
* @param string $date_range Date range ('7d', '30d', '90d') |
| 136 |
* @param array $dimensions Dimensions to group by (query, page, country, device, searchAppearance) |
| 137 |
* @param int $row_limit Maximum number of rows to return |
| 138 |
* @return array Search performance data |
| 139 |
* @throws \Exception If API request fails |
| 140 |
*/ |
| 141 |
public function get_search_performance(string $site_url, string $date_range = '30d', array $dimensions = ['query'], int $row_limit = 1000): array { |
| 142 |
// GSC data for the current day is never complete; use yesterday as the end date |
| 143 |
// so the N-day window matches exactly what the GSC dashboard shows. |
| 144 |
$days = (int) str_replace('d', '', $date_range); |
| 145 |
$end_date = gmdate('Y-m-d', strtotime('-2 days')); |
| 146 |
$start_date = gmdate('Y-m-d', strtotime('-' . ($days - 1) . ' days', strtotime($end_date))); |
| 147 |
|
| 148 |
$endpoint = '/sites/' . rawurlencode($site_url) . '/searchAnalytics/query'; |
| 149 |
|
| 150 |
$request_body = [ |
| 151 |
'startDate' => $start_date, |
| 152 |
'endDate' => $end_date, |
| 153 |
'dimensions' => $dimensions, |
| 154 |
'rowLimit' => $row_limit, |
| 155 |
'dataState' => 'all', |
| 156 |
]; |
| 157 |
|
| 158 |
$full_url = self::API_BASE_URL . $endpoint; |
| 159 |
return $this->make_request($full_url, $request_body, 'POST'); |
| 160 |
} |
| 161 |
|
| 162 |
/** |
| 163 |
* Get aggregated search totals (clicks, impressions, ctr, position) |
| 164 |
* |
| 165 |
* @param string $site_url Site URL to get data for |
| 166 |
* @param string $date_range Date range ('7d', '30d', '90d') |
| 167 |
* @return array Aggregated totals |
| 168 |
* @throws \Exception If API request fails |
| 169 |
*/ |
| 170 |
public function get_search_totals(string $site_url, string $date_range = '30d'): array { |
| 171 |
// GSC data has a 2-day delay; use D-2 as end_date to match the GSC dashboard. |
| 172 |
$days = (int) str_replace('d', '', $date_range); |
| 173 |
$end_date = gmdate('Y-m-d', strtotime('-2 days')); |
| 174 |
$start_date = gmdate('Y-m-d', strtotime('-' . ($days - 1) . ' days', strtotime($end_date))); |
| 175 |
|
| 176 |
$endpoint = '/sites/' . rawurlencode($site_url) . '/searchAnalytics/query'; |
| 177 |
|
| 178 |
// diverse from get_search_performance: no dimensions, just totals |
| 179 |
$request_body = [ |
| 180 |
'startDate' => $start_date, |
| 181 |
'endDate' => $end_date, |
| 182 |
'dimensions' => [], // Empty dimensions for aggregation |
| 183 |
'rowLimit' => 1, // We only need the totals, but API might require at least 1 |
| 184 |
'dataState' => 'all', |
| 185 |
]; |
| 186 |
|
| 187 |
$full_url = self::API_BASE_URL . $endpoint; |
| 188 |
$response = $this->make_request($full_url, $request_body, 'POST'); |
| 189 |
|
| 190 |
// The API returns rows even if we don't ask for dimensions? |
| 191 |
// Actually, without dimensions, it returns one row with aggregated values if successful. |
| 192 |
// Or sometimes it returns just the aggregates if available. |
| 193 |
// Let's inspect the response format for GSC API v3. |
| 194 |
// "If no dimensions are requested, the response will contain a single row with the aggregated values." |
| 195 |
|
| 196 |
if (!empty($response['rows'])) { |
| 197 |
$row = $response['rows'][0]; |
| 198 |
return [ |
| 199 |
'clicks' => $row['clicks'] ?? 0, |
| 200 |
'impressions' => $row['impressions'] ?? 0, |
| 201 |
'ctr' => round(($row['ctr'] ?? 0) * 100, 2), |
| 202 |
'position' => round($row['position'] ?? 0, 1) |
| 203 |
]; |
| 204 |
} |
| 205 |
|
| 206 |
return [ |
| 207 |
'clicks' => 0, |
| 208 |
'impressions' => 0, |
| 209 |
'ctr' => 0, |
| 210 |
'position' => 0 |
| 211 |
]; |
| 212 |
} |
| 213 |
|
| 214 |
/** |
| 215 |
* Get aggregated search totals for explicit start/end dates |
| 216 |
* |
| 217 |
* @param string $site_url Site URL to get data for |
| 218 |
* @param string $start_date Start date (Y-m-d) |
| 219 |
* @param string $end_date End date (Y-m-d) |
| 220 |
* @return array Aggregated totals |
| 221 |
* @throws \Exception If API request fails |
| 222 |
*/ |
| 223 |
public function get_search_totals_by_dates(string $site_url, string $start_date, string $end_date): array { |
| 224 |
$endpoint = '/sites/' . rawurlencode($site_url) . '/searchAnalytics/query'; |
| 225 |
|
| 226 |
$request_body = [ |
| 227 |
'startDate' => $start_date, |
| 228 |
'endDate' => $end_date, |
| 229 |
'dimensions' => [], |
| 230 |
'rowLimit' => 1, |
| 231 |
'dataState' => 'all', |
| 232 |
]; |
| 233 |
|
| 234 |
$full_url = self::API_BASE_URL . $endpoint; |
| 235 |
$response = $this->make_request($full_url, $request_body, 'POST'); |
| 236 |
|
| 237 |
if (!empty($response['rows'])) { |
| 238 |
$row = $response['rows'][0]; |
| 239 |
return [ |
| 240 |
'clicks' => $row['clicks'] ?? 0, |
| 241 |
'impressions' => $row['impressions'] ?? 0, |
| 242 |
'ctr' => round(($row['ctr'] ?? 0) * 100, 2), |
| 243 |
'position' => round($row['position'] ?? 0, 1), |
| 244 |
]; |
| 245 |
} |
| 246 |
|
| 247 |
return ['clicks' => 0, 'impressions' => 0, 'ctr' => 0, 'position' => 0]; |
| 248 |
} |
| 249 |
|
| 250 |
/** |
| 251 |
* Get query-level search performance for explicit start/end dates |
| 252 |
* |
| 253 |
* @param string $site_url Site URL to get data for |
| 254 |
* @param string $start_date Start date (Y-m-d) |
| 255 |
* @param string $end_date End date (Y-m-d) |
| 256 |
* @param int $row_limit Maximum rows to return |
| 257 |
* @param array $dimensions Dimensions to group by |
| 258 |
* @param array $filters Optional dimension filters (ANDed), each |
| 259 |
* {dimension, operator, expression}. @since 2.15.0 |
| 260 |
* @return array Raw rows from GSC API |
| 261 |
* @throws \Exception If API request fails |
| 262 |
*/ |
| 263 |
public function get_search_performance_by_dates(string $site_url, string $start_date, string $end_date, int $row_limit = 500, array $dimensions = ['query'], array $filters = []): array { |
| 264 |
$endpoint = '/sites/' . rawurlencode($site_url) . '/searchAnalytics/query'; |
| 265 |
|
| 266 |
$request_body = [ |
| 267 |
'startDate' => $start_date, |
| 268 |
'endDate' => $end_date, |
| 269 |
'dimensions' => $dimensions, |
| 270 |
'rowLimit' => $row_limit, |
| 271 |
// 'all' includes both finalised data and fresh (still-processing) |
| 272 |
// data — matches what the Search Console web UI displays, so the |
| 273 |
// last 2-4 days aren't missing. |
| 274 |
'dataState' => 'all', |
| 275 |
]; |
| 276 |
|
| 277 |
// Optional dimension filters, ANDed, e.g. one page's queries: |
| 278 |
// [['dimension' => 'page', 'operator' => 'equals', 'expression' => $url]]. |
| 279 |
if (!empty($filters)) { |
| 280 |
$request_body['dimensionFilterGroups'] = [['filters' => array_values($filters)]]; |
| 281 |
} |
| 282 |
|
| 283 |
$full_url = self::API_BASE_URL . $endpoint; |
| 284 |
$response = $this->make_request($full_url, $request_body, 'POST'); |
| 285 |
|
| 286 |
return $response['rows'] ?? []; |
| 287 |
} |
| 288 |
|
| 289 |
/** |
| 290 |
* Get top search queries |
| 291 |
* |
| 292 |
* @param string $site_url Site URL to get data for |
| 293 |
* @param int $limit Number of queries to retrieve |
| 294 |
* @return array Top search queries |
| 295 |
* @throws \Exception If API request fails |
| 296 |
*/ |
| 297 |
public function get_top_queries(string $site_url, int $limit = 10): array { |
| 298 |
try { |
| 299 |
$performance_data = $this->get_search_performance($site_url, '30d'); |
| 300 |
$queries = []; |
| 301 |
|
| 302 |
foreach ($performance_data['rows'] ?? [] as $row) { |
| 303 |
if (count($queries) >= $limit) { |
| 304 |
break; |
| 305 |
} |
| 306 |
|
| 307 |
$queries[] = [ |
| 308 |
'query' => $row['keys'][0] ?? '', |
| 309 |
'clicks' => $row['clicks'] ?? 0, |
| 310 |
'impressions' => $row['impressions'] ?? 0, |
| 311 |
'ctr' => $row['ctr'] ?? 0, |
| 312 |
'position' => $row['position'] ?? 0 |
| 313 |
]; |
| 314 |
} |
| 315 |
|
| 316 |
return [ |
| 317 |
'queries' => $queries, |
| 318 |
'site_url' => $site_url, |
| 319 |
'limit' => $limit |
| 320 |
]; |
| 321 |
} catch (\Exception $e) { |
| 322 |
return [ |
| 323 |
'queries' => [], |
| 324 |
'site_url' => $site_url, |
| 325 |
'limit' => $limit, |
| 326 |
'error' => $e->getMessage() |
| 327 |
]; |
| 328 |
} |
| 329 |
} |
| 330 |
|
| 331 |
/** |
| 332 |
* Get device performance breakdown for mobile SEO insights |
| 333 |
* |
| 334 |
* @param string $site_url Site URL to get data for |
| 335 |
* @param string $date_range Date range for data |
| 336 |
* @return array Device performance data |
| 337 |
* @throws \Exception If API request fails |
| 338 |
*/ |
| 339 |
public function get_device_performance(string $site_url, string $date_range = '30d'): array { |
| 340 |
$result = $this->get_search_performance($site_url, $date_range, ['device'], 10); |
| 341 |
|
| 342 |
$devices = []; |
| 343 |
$rows = $result['rows'] ?? []; |
| 344 |
|
| 345 |
foreach ($rows as $row) { |
| 346 |
$device = $row['keys'][0] ?? ''; |
| 347 |
$devices[$device] = [ |
| 348 |
'clicks' => $row['clicks'] ?? 0, |
| 349 |
'impressions' => $row['impressions'] ?? 0, |
| 350 |
'ctr' => round(($row['ctr'] ?? 0) * 100, 2), |
| 351 |
'position' => round($row['position'] ?? 0, 1) |
| 352 |
]; |
| 353 |
} |
| 354 |
|
| 355 |
return [ |
| 356 |
'devices' => $devices, |
| 357 |
'site_url' => $site_url, |
| 358 |
'date_range' => $date_range |
| 359 |
]; |
| 360 |
} |
| 361 |
|
| 362 |
/** |
| 363 |
* Get search appearance data for rich results tracking |
| 364 |
* |
| 365 |
* @param string $site_url Site URL to get data for |
| 366 |
* @param string $date_range Date range for data |
| 367 |
* @return array Search appearance data |
| 368 |
* @throws \Exception If API request fails |
| 369 |
*/ |
| 370 |
public function get_search_appearance(string $site_url, string $date_range = '30d'): array { |
| 371 |
$result = $this->get_search_performance($site_url, $date_range, ['searchAppearance'], 20); |
| 372 |
|
| 373 |
$appearances = []; |
| 374 |
$rows = $result['rows'] ?? []; |
| 375 |
|
| 376 |
foreach ($rows as $row) { |
| 377 |
$appearance = $row['keys'][0] ?? ''; |
| 378 |
$appearances[$appearance] = [ |
| 379 |
'clicks' => $row['clicks'] ?? 0, |
| 380 |
'impressions' => $row['impressions'] ?? 0, |
| 381 |
'ctr' => round(($row['ctr'] ?? 0) * 100, 2), |
| 382 |
'position' => round($row['position'] ?? 0, 1) |
| 383 |
]; |
| 384 |
} |
| 385 |
|
| 386 |
return [ |
| 387 |
'appearances' => $appearances, |
| 388 |
'site_url' => $site_url, |
| 389 |
'date_range' => $date_range |
| 390 |
]; |
| 391 |
} |
| 392 |
|
| 393 |
/** |
| 394 |
* Get keyword opportunities for SEO insights |
| 395 |
* Identifies queries with high impressions but low CTR or position |
| 396 |
* |
| 397 |
* @param string $site_url Site URL to analyze |
| 398 |
* @param string $date_range Date range for analysis |
| 399 |
* @param int $min_impressions Minimum impressions threshold |
| 400 |
* @return array Keyword opportunities |
| 401 |
* @throws \Exception If API request fails |
| 402 |
*/ |
| 403 |
public function get_keyword_opportunities(string $site_url, string $date_range = '30d', int $min_impressions = 100): array { |
| 404 |
$result = $this->get_search_performance($site_url, $date_range, ['query'], 500); |
| 405 |
|
| 406 |
$opportunities = []; |
| 407 |
$rows = $result['rows'] ?? []; |
| 408 |
|
| 409 |
foreach ($rows as $row) { |
| 410 |
$impressions = $row['impressions'] ?? 0; |
| 411 |
$ctr = $row['ctr'] ?? 0; |
| 412 |
$position = $row['position'] ?? 0; |
| 413 |
$clicks = $row['clicks'] ?? 0; |
| 414 |
|
| 415 |
// Identify opportunities: high impressions, low CTR, or position 4-10 |
| 416 |
if ($impressions >= $min_impressions) { |
| 417 |
$opportunity_score = 0; |
| 418 |
$opportunity_reasons = []; |
| 419 |
|
| 420 |
// Low CTR opportunity |
| 421 |
if ($ctr < 0.05 && $position <= 10) { // Less than 5% CTR in top 10 |
| 422 |
$opportunity_score += 30; |
| 423 |
$opportunity_reasons[] = 'Low CTR for top 10 position'; |
| 424 |
} |
| 425 |
|
| 426 |
// Position 4-10 opportunity (could reach top 3) |
| 427 |
if ($position >= 4 && $position <= 10) { |
| 428 |
$opportunity_score += 40; |
| 429 |
$opportunity_reasons[] = 'Ranking 4-10, potential for top 3'; |
| 430 |
} |
| 431 |
|
| 432 |
// High impressions, low clicks |
| 433 |
if ($impressions > 500 && $clicks < 25) { |
| 434 |
$opportunity_score += 20; |
| 435 |
$opportunity_reasons[] = 'High impressions but low clicks'; |
| 436 |
} |
| 437 |
|
| 438 |
if ($opportunity_score > 0) { |
| 439 |
$opportunities[] = [ |
| 440 |
'query' => $row['keys'][0] ?? '', |
| 441 |
'clicks' => $clicks, |
| 442 |
'impressions' => $impressions, |
| 443 |
'ctr' => round($ctr * 100, 2), |
| 444 |
'position' => round($position, 1), |
| 445 |
'opportunity_score' => $opportunity_score, |
| 446 |
'reasons' => $opportunity_reasons |
| 447 |
]; |
| 448 |
} |
| 449 |
} |
| 450 |
} |
| 451 |
|
| 452 |
// Sort by opportunity score (highest first) |
| 453 |
usort($opportunities, function ($a, $b) { |
| 454 |
return $b['opportunity_score'] <=> $a['opportunity_score']; |
| 455 |
}); |
| 456 |
|
| 457 |
return [ |
| 458 |
'opportunities' => array_slice($opportunities, 0, 50), // Top 50 opportunities |
| 459 |
'site_url' => $site_url, |
| 460 |
'date_range' => $date_range, |
| 461 |
'total_opportunities' => count($opportunities) |
| 462 |
]; |
| 463 |
} |
| 464 |
|
| 465 |
/** |
| 466 |
* Return branded vs non-branded click/impression split that matches the GSC platform. |
| 467 |
* |
| 468 |
* Uses two server-side aggregate calls per period (empty dimensions + dimensionFilterGroups) |
| 469 |
* so the totals are exact — not limited by the 1 000-row query cap: |
| 470 |
* |
| 471 |
* • Call A: no filter → real site total clicks |
| 472 |
* • Call B: query contains brand → branded clicks |
| 473 |
* • Non-branded = A − B |
| 474 |
* |
| 475 |
* Also fetches the equivalent previous period so the frontend can render trend arrows. |
| 476 |
* |
| 477 |
* @param string $site_url Registered GSC property URL |
| 478 |
* @param string $date_range '7d' | '30d' | '90d' |
| 479 |
* @param string $brand_name Comma-separated brand keywords. Auto-derived from domain when empty. |
| 480 |
* @return array { |
| 481 |
* branded, non_branded, previous: { branded, non_branded }, |
| 482 |
* brand_terms, total_clicks, site_url, date_range |
| 483 |
* } |
| 484 |
*/ |
| 485 |
public function get_branded_performance(string $site_url, string $date_range = '30d', string $brand_name = ''): array { |
| 486 |
$days = max(1, (int) str_replace('d', '', $date_range)); |
| 487 |
$end = gmdate('Y-m-d', strtotime('-2 days')); |
| 488 |
$start = gmdate('Y-m-d', strtotime('-' . ($days - 1) . ' days', strtotime($end))); |
| 489 |
|
| 490 |
$prev_end = gmdate('Y-m-d', strtotime('-1 day', strtotime($start))); |
| 491 |
$prev_start = gmdate('Y-m-d', strtotime('-' . ($days - 1) . ' days', strtotime($prev_end))); |
| 492 |
|
| 493 |
// Auto-derive brand from domain when not provided. |
| 494 |
// Handles both URL-prefix (https://example.com) and domain (sc-domain:example.com) formats. |
| 495 |
if (empty($brand_name)) { |
| 496 |
$stripped = preg_replace('#^sc-domain:#i', '', $site_url); |
| 497 |
$host = wp_parse_url($stripped, PHP_URL_HOST) ?? wp_parse_url('https://' . $stripped, PHP_URL_HOST) ?? $stripped; |
| 498 |
$host = preg_replace('/^www\./i', '', (string) $host); |
| 499 |
$brand_name = strtolower(explode('.', $host)[0]); |
| 500 |
} |
| 501 |
$terms = array_values(array_filter(array_map('trim', explode(',', strtolower($brand_name))))); |
| 502 |
|
| 503 |
// For hyphenated brands (e.g. "essential-blocks") also match the space variant |
| 504 |
// ("essential blocks") since users type both forms in Google searches. |
| 505 |
$extra = []; |
| 506 |
foreach ($terms as $t) { |
| 507 |
if (str_contains($t, '-')) { |
| 508 |
$spaced = str_replace('-', ' ', $t); |
| 509 |
if (!in_array($spaced, $terms, true)) { |
| 510 |
$extra[] = $spaced; |
| 511 |
} |
| 512 |
} |
| 513 |
} |
| 514 |
$terms = array_values(array_merge($terms, $extra)); |
| 515 |
|
| 516 |
$split_cur = $this->gsc_split_by_brand($site_url, $start, $end, $terms); |
| 517 |
$split_prev = $this->gsc_split_by_brand($site_url, $prev_start, $prev_end, $terms); |
| 518 |
|
| 519 |
return [ |
| 520 |
'branded' => $split_cur['branded'], |
| 521 |
'non_branded' => $split_cur['non_branded'], |
| 522 |
'previous' => [ |
| 523 |
'branded' => $split_prev['branded'], |
| 524 |
'non_branded' => $split_prev['non_branded'], |
| 525 |
], |
| 526 |
'brand_terms' => $terms, |
| 527 |
'total_clicks' => $split_cur['total_clicks'], |
| 528 |
'site_url' => $site_url, |
| 529 |
'date_range' => $date_range, |
| 530 |
]; |
| 531 |
} |
| 532 |
|
| 533 |
/** |
| 534 |
* Fetch all web query rows for a date window and split into branded / non-branded |
| 535 |
* using a single API call. Both totals come from the same data set so the |
| 536 |
* percentages always add up to 100 %. |
| 537 |
* |
| 538 |
* @param string $site_url GSC property URL |
| 539 |
* @param string $start Start date (Y-m-d) |
| 540 |
* @param string $end End date (Y-m-d) |
| 541 |
* @param string[] $brand_terms Brand keywords to match (substring, case-insensitive) |
| 542 |
* @return array { branded: {...}, non_branded: {...}, total_clicks: int } |
| 543 |
*/ |
| 544 |
private function gsc_split_by_brand(string $site_url, string $start, string $end, array $brand_terms): array { |
| 545 |
$url = self::API_BASE_URL . '/sites/' . rawurlencode($site_url) . '/searchAnalytics/query'; |
| 546 |
|
| 547 |
$page_size = 25000; |
| 548 |
$start_row = 0; |
| 549 |
$total_clicks = 0; |
| 550 |
$branded_clicks = 0; |
| 551 |
$total_impr = 0; |
| 552 |
$branded_impr = 0; |
| 553 |
|
| 554 |
// Safety cap so a runaway query can never loop unbounded. With a 25k |
| 555 |
// page size this stops after ~100k rows (4 pages), which is far beyond |
| 556 |
// the query volume of any real site for a single date window. |
| 557 |
$max_pages = 4; |
| 558 |
$pages_fetched = 0; |
| 559 |
|
| 560 |
do { |
| 561 |
$response = $this->make_request($url, [ |
| 562 |
'startDate' => $start, |
| 563 |
'endDate' => $end, |
| 564 |
'type' => 'web', |
| 565 |
'dimensions' => ['query'], |
| 566 |
'rowLimit' => $page_size, |
| 567 |
'startRow' => $start_row, |
| 568 |
'dataState' => 'all', |
| 569 |
], 'POST'); |
| 570 |
|
| 571 |
$rows = $response['rows'] ?? []; |
| 572 |
foreach ($rows as $row) { |
| 573 |
$query = strtolower($row['keys'][0] ?? ''); |
| 574 |
$clicks = (int) ($row['clicks'] ?? 0); |
| 575 |
$impr = (int) ($row['impressions'] ?? 0); |
| 576 |
|
| 577 |
$total_clicks += $clicks; |
| 578 |
$total_impr += $impr; |
| 579 |
|
| 580 |
foreach ($brand_terms as $term) { |
| 581 |
if (str_contains($query, $term)) { |
| 582 |
$branded_clicks += $clicks; |
| 583 |
$branded_impr += $impr; |
| 584 |
break; |
| 585 |
} |
| 586 |
} |
| 587 |
} |
| 588 |
|
| 589 |
$fetched = count($rows); |
| 590 |
$start_row += $fetched; |
| 591 |
$pages_fetched++; |
| 592 |
} while ($fetched === $page_size && $pages_fetched < $max_pages); |
| 593 |
|
| 594 |
$non_branded_clicks = max(0, $total_clicks - $branded_clicks); |
| 595 |
$non_branded_impr = max(0, $total_impr - $branded_impr); |
| 596 |
|
| 597 |
return [ |
| 598 |
'branded' => [ |
| 599 |
'clicks' => $branded_clicks, |
| 600 |
'impressions' => $branded_impr, |
| 601 |
'percentage' => $total_clicks > 0 ? round($branded_clicks / $total_clicks * 100) : 0, |
| 602 |
], |
| 603 |
'non_branded' => [ |
| 604 |
'clicks' => $non_branded_clicks, |
| 605 |
'impressions' => $non_branded_impr, |
| 606 |
'percentage' => $total_clicks > 0 ? round($non_branded_clicks / $total_clicks * 100) : 0, |
| 607 |
], |
| 608 |
'total_clicks' => $total_clicks, |
| 609 |
]; |
| 610 |
} |
| 611 |
|
| 612 |
/** |
| 613 |
* Get top countries by clicks from Search Console. |
| 614 |
* |
| 615 |
* Queries with the `country` dimension and returns rows sorted by clicks |
| 616 |
* descending, each enriched with a percentage share of the total clicks. |
| 617 |
* |
| 618 |
* @param string $site_url Site URL to query |
| 619 |
* @param string $date_range Date range ('7d', '30d', '90d') |
| 620 |
* @param int $row_limit Maximum countries to return (default 10) |
| 621 |
* @return array { countries: array, total_clicks: int, site_url: string, date_range: string } |
| 622 |
*/ |
| 623 |
public function get_country_performance(string $site_url, string $date_range = '30d', int $row_limit = 10): array { |
| 624 |
$result = $this->get_search_performance($site_url, $date_range, ['country'], $row_limit); |
| 625 |
$rows = $result['rows'] ?? []; |
| 626 |
|
| 627 |
$total_clicks = 0; |
| 628 |
foreach ($rows as $row) { |
| 629 |
$total_clicks += (int) ($row['clicks'] ?? 0); |
| 630 |
} |
| 631 |
|
| 632 |
$countries = []; |
| 633 |
foreach ($rows as $row) { |
| 634 |
$clicks = (int) ($row['clicks'] ?? 0); |
| 635 |
$countries[] = [ |
| 636 |
'country' => strtolower($row['keys'][0] ?? ''), |
| 637 |
'clicks' => $clicks, |
| 638 |
'impressions' => (int) ($row['impressions'] ?? 0), |
| 639 |
'ctr' => round(($row['ctr'] ?? 0) * 100, 2), |
| 640 |
'position' => round($row['position'] ?? 0, 1), |
| 641 |
'percentage' => $total_clicks > 0 ? round(($clicks / $total_clicks) * 100) : 0, |
| 642 |
]; |
| 643 |
} |
| 644 |
|
| 645 |
return [ |
| 646 |
'countries' => $countries, |
| 647 |
'total_clicks' => $total_clicks, |
| 648 |
'site_url' => $site_url, |
| 649 |
'date_range' => $date_range, |
| 650 |
]; |
| 651 |
} |
| 652 |
|
| 653 |
/** |
| 654 |
* Get rate limit configuration |
| 655 |
* Following ThinkRank rate limiting patterns |
| 656 |
* |
| 657 |
* @return array Rate limit configuration |
| 658 |
*/ |
| 659 |
protected function get_rate_limits(): array { |
| 660 |
return [ |
| 661 |
'max_requests_per_day' => self::MAX_REQUESTS_PER_DAY, |
| 662 |
'reset_time' => get_transient(self::RATE_LIMIT_KEY . '_reset') ?: strtotime('tomorrow') |
| 663 |
]; |
| 664 |
} |
| 665 |
|
| 666 |
/** |
| 667 |
* Get rate limit transient key |
| 668 |
* Following ThinkRank option naming patterns |
| 669 |
* |
| 670 |
* @return string Rate limit key |
| 671 |
*/ |
| 672 |
protected function get_rate_limit_key(): string { |
| 673 |
return self::RATE_LIMIT_KEY; |
| 674 |
} |
| 675 |
|
| 676 |
/** |
| 677 |
* Get rate limit error message |
| 678 |
* |
| 679 |
* @return string Error message |
| 680 |
*/ |
| 681 |
protected function get_rate_limit_error_message(): string { |
| 682 |
return 'Google Search Console API rate limit exceeded. Try again tomorrow.'; |
| 683 |
} |
| 684 |
} |
| 685 |
|
