← All changes
|
includes/seo/class-email-report-data-provider.php
+502
-29
1.27.0
→
2.14.1
View file →
| @@ -6,12 +6,17 @@ | ||
| 6 | 6 | * preceding period of equal length, then hands both to sections so they |
| 7 | 7 | * can diff and rank. Sections never call Analytics_Manager directly — |
| 8 | 8 | * one fetch per period, shared across the report. |
| 9 | 9 | * |
| 10 | - * If Analytics_Manager isn't available (no Google connection, etc.) the | |
| 11 | - * provider returns an `available => false` result; sections then render | |
| 12 | - * their fallback HTML per PRD's graceful-degradation requirement. | |
| 10 | + * readiness() answers, before any data is pulled, whether there is a report | |
| 11 | + * to build: Search Console is required, Google Analytics 4 and the AI | |
| 12 | + * traffic tracker each add a card when present (#742). The generator asks | |
| 13 | + * this first so a disconnected site is paused with a reason rather than | |
| 14 | + * fetched, rendered and sent as a column of blanks. | |
| 13 | 15 | * |
| 16 | + * If Analytics_Manager isn't available the provider returns an | |
| 17 | + * `available => false` result and sections collect nothing. | |
| 18 | + * | |
| 14 | 19 | * @package ThinkRank |
| 15 | 20 | * @subpackage SEO |
| 16 | 21 | * @since 1.9.0 |
| 17 | 22 | */ |
| @@ -33,8 +38,153 @@ | ||
| 33 | 38 | */ |
| 34 | 39 | final class Email_Report_Data_Provider { |
| 35 | 40 | |
| 36 | 41 | /** |
| 42 | + * Rows fetched per window for the query and page comparisons. | |
| 43 | + * | |
| 44 | + * Search Console orders rows by clicks and stops here, so a list that | |
| 45 | + * comes back this long may be truncated: a row missing from it may sit | |
| 46 | + * just below the cut rather than have no traffic (#906). | |
| 47 | + */ | |
| 48 | + private const COMPARISON_ROW_LIMIT = 1000; | |
| 49 | + | |
| 50 | + /** | |
| 51 | + * How many apparent drop-outs are confirmed against the current window. | |
| 52 | + * | |
| 53 | + * The losing cards show five rows, so the biggest previous-period | |
| 54 | + * candidates are all that can reach them. Anything past this cap is left | |
| 55 | + * out of the comparison rather than presumed lost. | |
| 56 | + */ | |
| 57 | + private const DROPOUT_CONFIRM_LIMIT = 25; | |
| 58 | + | |
| 59 | + /** | |
| 60 | + * Longest regular expression sent in one confirmation request. Search | |
| 61 | + * Console rejects longer expressions, so candidates are batched under it. | |
| 62 | + */ | |
| 63 | + private const DROPOUT_REGEX_MAX_LENGTH = 3500; | |
| 64 | + | |
| 65 | + /** | |
| 66 | + * Which data sources the report can draw on right now. | |
| 67 | + * | |
| 68 | + * Cheap on purpose — no dashboard fetch, no Search Console query. It | |
| 69 | + * runs on every hourly tick while a site is paused and on every load of | |
| 70 | + * the Email Reporting panel. | |
| 71 | + * | |
| 72 | + * @return array{ | |
| 73 | + * ready:bool, | |
| 74 | + * search_console:bool, | |
| 75 | + * analytics:bool, | |
| 76 | + * ai_traffic:bool, | |
| 77 | + * reason:?string | |
| 78 | + * } | |
| 79 | + */ | |
| 80 | + public function readiness(): array { | |
| 81 | + // Credentials, not client objects. Analytics_Manager constructs a | |
| 82 | + // Search Console client whether or not a token exists, so "is there | |
| 83 | + // a client?" is always yes. The predicates below are the ones | |
| 84 | + // get-integrations-status reports, so the panel, the ability and | |
| 85 | + // the report can never disagree about the same site. | |
| 86 | + $settings = $this->settings(); | |
| 87 | + | |
| 88 | + $oauth_present = '' !== (string) $settings->get('google_access_token', ''); | |
| 89 | + $search_console = $oauth_present | |
| 90 | + || '' !== (string) $settings->get('google_search_console_api_key', ''); | |
| 91 | + | |
| 92 | + // GA4: the Data API client is built only when a token AND a selected | |
| 93 | + // property both exist — mirror that exactly. | |
| 94 | + $ga_property = (string) $settings->get('seo_analytics_google_analytics_property_id', ''); | |
| 95 | + $analytics = ($oauth_present && '' !== $ga_property) | |
| 96 | + || '' !== (string) $settings->get('google_analytics_api_key', ''); | |
| 97 | + | |
| 98 | + $readiness = [ | |
| 99 | + 'ready' => $search_console, | |
| 100 | + 'search_console' => $search_console, | |
| 101 | + 'analytics' => $analytics, | |
| 102 | + 'ai_traffic' => $this->ai_tracker_has_data(), | |
| 103 | + 'reason' => $search_console ? null : 'search_console_not_connected', | |
| 104 | + ]; | |
| 105 | + | |
| 106 | + /** | |
| 107 | + * Filter the report's readiness. | |
| 108 | + * | |
| 109 | + * Lets a host that feeds the report from somewhere other than the | |
| 110 | + * Google integrations declare itself ready, and lets tests force | |
| 111 | + * either state. | |
| 112 | + * | |
| 113 | + * @since 2.8.0 | |
| 114 | + * | |
| 115 | + * @param array $readiness See readiness(). | |
| 116 | + */ | |
| 117 | + $filtered = apply_filters('thinkrank_email_report_readiness', $readiness); | |
| 118 | + | |
| 119 | + return is_array($filtered) ? array_merge($readiness, $filtered) : $readiness; | |
| 120 | + } | |
| 121 | + | |
| 122 | + /** | |
| 123 | + * Subject-line tokens for a fetched report (#742). | |
| 124 | + * | |
| 125 | + * `%headline%` is the sentence the default subject is built from: | |
| 126 | + * "12,480 Google clicks (+12.4%) in the last 30 days" when both windows | |
| 127 | + * have totals, "Your SEO report for Aug 19 – Sep 17" when there is | |
| 128 | + * nothing to compare. The parts are exposed as their own tokens so a | |
| 129 | + * custom subject template can rebuild it differently. | |
| 130 | + * | |
| 131 | + * @param array $shared Output of fetch(). | |
| 132 | + * @param int $frequency_days Report window in days. | |
| 133 | + * @return array<string,string> Token => value. | |
| 134 | + */ | |
| 135 | + public static function subject_tokens(array $shared, int $frequency_days): array { | |
| 136 | + $days = max(1, $frequency_days); | |
| 137 | + $label = (string) ($shared['period_label'] ?? ''); | |
| 138 | + $totals = $shared['comparison']['totals'] ?? []; | |
| 139 | + $current = is_array($totals['current'] ?? null) ? $totals['current'] : []; | |
| 140 | + $previous = is_array($totals['previous'] ?? null) ? $totals['previous'] : []; | |
| 141 | + | |
| 142 | + $clicks = (int) ($current['clicks'] ?? 0); | |
| 143 | + $impressions = (int) ($current['impressions'] ?? 0); | |
| 144 | + if ($clicks === 0 && $impressions === 0) { | |
| 145 | + $dash = $shared['current']['search_performance']['totals'] ?? []; | |
| 146 | + $clicks = (int) ($dash['clicks'] ?? 0); | |
| 147 | + $impressions = (int) ($dash['impressions'] ?? 0); | |
| 148 | + } | |
| 149 | + | |
| 150 | + $change = null; | |
| 151 | + if ((int) ($previous['clicks'] ?? 0) > 0 && class_exists(Email_Report_Sections\Email_Report_Html::class)) { | |
| 152 | + $change = Email_Report_Sections\Email_Report_Html::pct_change((float) $clicks, (float) ($previous['clicks'] ?? 0)); | |
| 153 | + } | |
| 154 | + $signed = $change !== null && $change['direction'] !== 'flat' | |
| 155 | + ? ($change['direction'] === 'up' ? '+' : '−') . $change['text'] | |
| 156 | + : ''; | |
| 157 | + | |
| 158 | + if ($clicks > 0 || $impressions > 0) { | |
| 159 | + $headline = sprintf( | |
| 160 | + /* translators: 1: number of clicks, 2: change in parentheses or empty, 3: number of days. */ | |
| 161 | + _n('%1$s Google clicks%2$s in the last %3$d day', '%1$s Google clicks%2$s in the last %3$d days', $days, 'thinkrank'), | |
| 162 | + number_format_i18n($clicks), | |
| 163 | + $signed !== '' ? ' (' . $signed . ')' : '', | |
| 164 | + $days | |
| 165 | + ); | |
| 166 | + } else { | |
| 167 | + $headline = $label !== '' | |
| 168 | + ? sprintf( | |
| 169 | + /* translators: %s: period label, e.g. "Aug 19 – Sep 17, 2026". */ | |
| 170 | + __('Your SEO report for %s', 'thinkrank'), | |
| 171 | + $label | |
| 172 | + ) | |
| 173 | + : __('Your SEO report', 'thinkrank'); | |
| 174 | + } | |
| 175 | + | |
| 176 | + return [ | |
| 177 | + '%period%' => $label, | |
| 178 | + '%period_days%' => (string) $days, | |
| 179 | + '%clicks%' => number_format_i18n($clicks), | |
| 180 | + '%clicks_change%' => $signed, | |
| 181 | + '%impressions%' => number_format_i18n($impressions), | |
| 182 | + '%headline%' => $headline, | |
| 183 | + ]; | |
| 184 | + } | |
| 185 | + | |
| 186 | + /** | |
| 37 | 187 | * Pull current + prior period dashboard data. |
| 38 | 188 | * |
| 39 | 189 | * @param int $frequency_days Reporting frequency in days. |
| 40 | 190 | * @return array{ |
| @@ -48,11 +198,20 @@ | ||
| 48 | 198 | * } |
| 49 | 199 | */ |
| 50 | 200 | public function fetch(int $frequency_days): array { |
| 51 | 201 | $frequency_days = max(1, $frequency_days); |
| 52 | - $today = current_time('Y-m-d'); | |
| 53 | - $period_end = $today; | |
| 54 | - $period_start = wp_date('Y-m-d', strtotime("-{$frequency_days} days", strtotime($today))); | |
| 202 | + | |
| 203 | + // One canonical window for the whole report. Search Console lags | |
| 204 | + // ~2 days, so it ends on the last date that actually has data — | |
| 205 | + // the same anchor Analytics_Manager::get_dashboard_data() uses for | |
| 206 | + // Key Metrics and Position Summary. Previously the header label ran | |
| 207 | + // through today and the comparison through yesterday, so a reader | |
| 208 | + // was handed three windows and told they were one. | |
| 209 | + $period_end = gmdate('Y-m-d', strtotime('-2 days')); | |
| 210 | + $period_start = gmdate( | |
| 211 | + 'Y-m-d', | |
| 212 | + strtotime('-' . ($frequency_days - 1) . ' days', strtotime($period_end)) | |
| 213 | + ); | |
| 55 | 214 | $period_label = $this->format_period_label($period_start, $period_end); |
| 56 | 215 | |
| 57 | 216 | $manager = $this->get_analytics_manager(); |
| 58 | 217 | if ($manager === null) { |
| @@ -74,14 +233,16 @@ | ||
| 74 | 233 | // Real period-over-period comparison: pull query- and page-level |
| 75 | 234 | // metrics for the current window AND the immediately preceding |
| 76 | 235 | // window of equal length straight from Search Console, then key |
| 77 | 236 | // them so winning/losing sections can compute true deltas. |
| 78 | - $comparison = $this->build_comparison($manager, $frequency_days); | |
| 237 | + $comparison = $this->build_comparison($manager, $frequency_days, $period_start, $period_end); | |
| 79 | 238 | |
| 80 | 239 | return [ |
| 81 | 240 | 'available' => true, |
| 241 | + 'readiness' => $this->readiness(), | |
| 82 | 242 | 'current' => is_array($current) ? $current : [], |
| 83 | 243 | 'comparison' => $comparison, |
| 244 | + 'ai' => $this->ai_summary($frequency_days), | |
| 84 | 245 | 'period_start' => $period_start, |
| 85 | 246 | 'period_end' => $period_end, |
| 86 | 247 | 'period_label' => $period_label, |
| 87 | 248 | ]; |
| @@ -103,18 +264,23 @@ | ||
| 103 | 264 | * |
| 104 | 265 | * Uses the Search Console client's arbitrary date-range API |
| 105 | 266 | * (`get_search_performance_by_dates`) — the same one the Rank Tracker |
| 106 | 267 | * and the Pro winning/losing endpoint use — to fetch query- and |
| 107 | - * page-level rows for two equal, adjacent windows. GSC lags ~2 days, so | |
| 108 | - * the current window ends yesterday and the previous window is the N | |
| 109 | - * days before it. | |
| 268 | + * page-level rows for two equal, adjacent windows. | |
| 110 | 269 | * |
| 270 | + * The current window is handed in by fetch() rather than recomputed | |
| 271 | + * here, so the deltas describe exactly the period the report's header | |
| 272 | + * advertises. The previous window is the same length, immediately | |
| 273 | + * before it, with no gap or overlap. | |
| 274 | + * | |
| 111 | 275 | * @param object $manager Analytics_Manager instance. |
| 112 | 276 | * @param int $frequency_days Window length in days. |
| 277 | + * @param string $cur_start Current window start (Y-m-d). | |
| 278 | + * @param string $cur_end Current window end (Y-m-d). | |
| 113 | 279 | * @return array{available:bool,queries:array,pages:array} |
| 114 | 280 | */ |
| 115 | - private function build_comparison($manager, int $frequency_days): array { | |
| 116 | - $empty = ['available' => false, 'queries' => [], 'pages' => []]; | |
| 281 | + private function build_comparison($manager, int $frequency_days, string $cur_start, string $cur_end): array { | |
| 282 | + $empty = ['available' => false, 'queries' => [], 'pages' => [], 'totals' => []]; | |
| 117 | 283 | |
| 118 | 284 | if (!method_exists($manager, 'get_search_console_client')) { |
| 119 | 285 | return $empty; |
| 120 | 286 | } |
| @@ -126,35 +292,281 @@ | ||
| 126 | 292 | if ($site_url === '') { |
| 127 | 293 | return $empty; |
| 128 | 294 | } |
| 129 | 295 | |
| 130 | - $cur_end = gmdate('Y-m-d', strtotime('-1 day')); | |
| 131 | - $cur_start = gmdate('Y-m-d', strtotime('-' . $frequency_days . ' days')); | |
| 132 | - $prev_end = gmdate('Y-m-d', strtotime('-' . ($frequency_days + 1) . ' days')); | |
| 133 | - $prev_start = gmdate('Y-m-d', strtotime('-' . ($frequency_days * 2) . ' days')); | |
| 296 | + // Previous window: the same number of days, ending the day before | |
| 297 | + // the current window opens. | |
| 298 | + $prev_end = gmdate('Y-m-d', strtotime('-1 day', strtotime($cur_start))); | |
| 299 | + $prev_start = gmdate('Y-m-d', strtotime('-' . ($frequency_days - 1) . ' days', strtotime($prev_end))); | |
| 134 | 300 | |
| 135 | - $cur_q = $sc->get_search_performance_by_dates($site_url, $cur_start, $cur_end, 1000, ['query']); | |
| 136 | - $prev_q = $sc->get_search_performance_by_dates($site_url, $prev_start, $prev_end, 1000, ['query']); | |
| 137 | - $cur_p = $sc->get_search_performance_by_dates($site_url, $cur_start, $cur_end, 1000, ['page']); | |
| 138 | - $prev_p = $sc->get_search_performance_by_dates($site_url, $prev_start, $prev_end, 1000, ['page']); | |
| 301 | + $limit = self::COMPARISON_ROW_LIMIT; | |
| 302 | + $cur_q = $sc->get_search_performance_by_dates($site_url, $cur_start, $cur_end, $limit, ['query']); | |
| 303 | + $prev_q = $sc->get_search_performance_by_dates($site_url, $prev_start, $prev_end, $limit, ['query']); | |
| 304 | + $cur_p = $sc->get_search_performance_by_dates($site_url, $cur_start, $cur_end, $limit, ['page']); | |
| 305 | + $prev_p = $sc->get_search_performance_by_dates($site_url, $prev_start, $prev_end, $limit, ['page']); | |
| 139 | 306 | |
| 307 | + // A row missing from a full current list may only have fallen below | |
| 308 | + // the cut. Ask Search Console about the likely losers directly before | |
| 309 | + // any of them is reported as having lost everything (#906). | |
| 310 | + [$found_q, $dropouts_q] = $this->confirm_dropouts($sc, $site_url, $cur_start, $cur_end, $cur_q, $prev_q, 'query'); | |
| 311 | + [$found_p, $dropouts_p] = $this->confirm_dropouts($sc, $site_url, $cur_start, $cur_end, $cur_p, $prev_p, 'page'); | |
| 312 | + | |
| 313 | + // Whole-property totals for both windows. A query with no | |
| 314 | + // dimensions returns one aggregated row, so the hero's clicks, | |
| 315 | + // impressions, CTR and position — and their change — are exact | |
| 316 | + // rather than summed from the 1,000-row query lists above. | |
| 317 | + $cur_t = $sc->get_search_performance_by_dates($site_url, $cur_start, $cur_end, 1, []); | |
| 318 | + $prev_t = $sc->get_search_performance_by_dates($site_url, $prev_start, $prev_end, 1, []); | |
| 319 | + | |
| 140 | 320 | return [ |
| 141 | 321 | 'available' => true, |
| 142 | - 'queries' => $this->merge_periods($cur_q, $prev_q, true), | |
| 143 | - 'pages' => $this->merge_periods($cur_p, $prev_p, false), | |
| 322 | + 'queries' => $this->merge_periods(array_merge($cur_q, $found_q), $prev_q, true, $dropouts_q), | |
| 323 | + 'pages' => $this->merge_periods(array_merge($cur_p, $found_p), $prev_p, false, $dropouts_p), | |
| 324 | + 'totals' => [ | |
| 325 | + 'current' => $this->totals_row($cur_t), | |
| 326 | + 'previous' => $this->totals_row($prev_t), | |
| 327 | + ], | |
| 144 | 328 | ]; |
| 145 | 329 | } |
| 146 | 330 | |
| 147 | 331 | /** |
| 332 | + * Sort out which previous-window rows missing from the current list | |
| 333 | + * really lost all their traffic. | |
| 334 | + * | |
| 335 | + * Both windows are capped lists ordered by clicks. Absence from a | |
| 336 | + * current list shorter than the cap is real: Search Console returned | |
| 337 | + * everything it has. Absence from a full list is not evidence of | |
| 338 | + * anything, because growth elsewhere pushes unchanged rows below the | |
| 339 | + * cut. Those were synthesised as total losses and led the losing cards, | |
| 340 | + * labelled "No impressions this period", with the same clicks in both | |
| 341 | + * windows (#906). | |
| 342 | + * | |
| 343 | + * So when the current list is full, the biggest candidates by previous | |
| 344 | + * clicks are looked up in the current window by exact key. A candidate | |
| 345 | + * that comes back has its real current row returned, so its change is | |
| 346 | + * measured rather than presumed. Only one that comes back empty is a | |
| 347 | + * drop-out. Candidates past DROPOUT_CONFIRM_LIMIT, or in a batch whose | |
| 348 | + * request failed, are neither: they are left out of the comparison, | |
| 349 | + * since saying nothing is better than reporting a loss nobody measured. | |
| 350 | + * | |
| 351 | + * @since 2.15.0 | |
| 352 | + * | |
| 353 | + * @param object $sc Search Console client. | |
| 354 | + * @param string $site_url Property URL. | |
| 355 | + * @param string $cur_start Current window start (Y-m-d). | |
| 356 | + * @param string $cur_end Current window end (Y-m-d). | |
| 357 | + * @param array $current Current-window rows, as fetched. | |
| 358 | + * @param array $previous Previous-window rows, as fetched. | |
| 359 | + * @param string $dimension 'query' or 'page'. | |
| 360 | + * @return array{0: array, 1: ?array<string,bool>} Current rows found | |
| 361 | + * below the cut, and the confirmed drop-out keys, or null when | |
| 362 | + * the current list is complete and every absence is real. | |
| 363 | + */ | |
| 364 | + private function confirm_dropouts($sc, string $site_url, string $cur_start, string $cur_end, array $current, array $previous, string $dimension): array { | |
| 365 | + if (count($current) < self::COMPARISON_ROW_LIMIT) { | |
| 366 | + return [[], null]; | |
| 367 | + } | |
| 368 | + | |
| 369 | + $is_query = 'query' === $dimension; | |
| 370 | + | |
| 371 | + $present = []; | |
| 372 | + foreach ($current as $row) { | |
| 373 | + $raw = (string) ($row['keys'][0] ?? ''); | |
| 374 | + if ('' !== $raw) { | |
| 375 | + $present[$this->normalize_key($raw, $is_query)] = true; | |
| 376 | + } | |
| 377 | + } | |
| 378 | + | |
| 379 | + $candidates = []; | |
| 380 | + foreach ($previous as $row) { | |
| 381 | + $raw = (string) ($row['keys'][0] ?? ''); | |
| 382 | + if ('' === $raw) { | |
| 383 | + continue; | |
| 384 | + } | |
| 385 | + $key = $this->normalize_key($raw, $is_query); | |
| 386 | + if (!isset($present[$key]) && !isset($candidates[$key])) { | |
| 387 | + $candidates[$key] = ['raw' => $raw, 'clicks' => (int) ($row['clicks'] ?? 0)]; | |
| 388 | + } | |
| 389 | + } | |
| 390 | + | |
| 391 | + if ([] === $candidates) { | |
| 392 | + return [[], []]; | |
| 393 | + } | |
| 394 | + | |
| 395 | + uasort($candidates, static fn($a, $b) => $b['clicks'] <=> $a['clicks']); | |
| 396 | + $candidates = array_slice($candidates, 0, self::DROPOUT_CONFIRM_LIMIT, true); | |
| 397 | + | |
| 398 | + // Batch the exact-match lookups into as few requests as the | |
| 399 | + // expression length allows: one per dimension in practice. | |
| 400 | + $batches = []; | |
| 401 | + $batch = []; | |
| 402 | + $length = 0; | |
| 403 | + foreach ($candidates as $key => $candidate) { | |
| 404 | + $part = $this->re2_quote($candidate['raw']); | |
| 405 | + if ([] !== $batch && $length + strlen($part) + 1 > self::DROPOUT_REGEX_MAX_LENGTH) { | |
| 406 | + $batches[] = $batch; | |
| 407 | + $batch = []; | |
| 408 | + $length = 0; | |
| 409 | + } | |
| 410 | + $batch[$key] = $part; | |
| 411 | + $length += strlen($part) + 1; | |
| 412 | + } | |
| 413 | + $batches[] = $batch; | |
| 414 | + | |
| 415 | + $found = []; | |
| 416 | + $dropouts = []; | |
| 417 | + foreach ($batches as $batch) { | |
| 418 | + try { | |
| 419 | + $rows = $sc->get_search_performance_by_dates( | |
| 420 | + $site_url, | |
| 421 | + $cur_start, | |
| 422 | + $cur_end, | |
| 423 | + count($batch), | |
| 424 | + [$dimension], | |
| 425 | + [[ | |
| 426 | + 'dimension' => $dimension, | |
| 427 | + 'operator' => 'includingRegex', | |
| 428 | + 'expression' => '^(?:' . implode('|', $batch) . ')$', | |
| 429 | + ]] | |
| 430 | + ); | |
| 431 | + } catch (Throwable $e) { | |
| 432 | + // Unconfirmed: leave these out rather than call them lost. | |
| 433 | + continue; | |
| 434 | + } | |
| 435 | + | |
| 436 | + $seen = []; | |
| 437 | + foreach ((array) $rows as $row) { | |
| 438 | + $raw = (string) ($row['keys'][0] ?? ''); | |
| 439 | + $key = '' === $raw ? '' : $this->normalize_key($raw, $is_query); | |
| 440 | + if (isset($batch[$key])) { | |
| 441 | + $seen[$key] = true; | |
| 442 | + $found[] = $row; | |
| 443 | + } | |
| 444 | + } | |
| 445 | + | |
| 446 | + foreach (array_keys($batch) as $key) { | |
| 447 | + if (!isset($seen[$key])) { | |
| 448 | + $dropouts[$key] = true; | |
| 449 | + } | |
| 450 | + } | |
| 451 | + } | |
| 452 | + | |
| 453 | + return [$found, $dropouts]; | |
| 454 | + } | |
| 455 | + | |
| 456 | + /** | |
| 457 | + * Escape a literal for a Search Console (RE2) regular expression. | |
| 458 | + * | |
| 459 | + * RE2 accepts a backslash before any ASCII punctuation as that literal, | |
| 460 | + * so every punctuation character is escaped, not only the ones that are | |
| 461 | + * special today. | |
| 462 | + * | |
| 463 | + * @since 2.15.0 | |
| 464 | + * | |
| 465 | + * @param string $literal Text to match exactly. | |
| 466 | + * @return string | |
| 467 | + */ | |
| 468 | + private function re2_quote(string $literal): string { | |
| 469 | + return (string) preg_replace('/[[:punct:]]/', '\\\\$0', $literal); | |
| 470 | + } | |
| 471 | + | |
| 472 | + /** | |
| 473 | + * Normalise the single aggregate row Search Console returns for a | |
| 474 | + * dimensionless query. An empty result (a property with no traffic in | |
| 475 | + * the window) yields zeroes, which the hero treats as "no comparison". | |
| 476 | + * | |
| 477 | + * @param array $rows API rows. | |
| 478 | + * @return array{clicks:int,impressions:int,ctr:float,position:float} | |
| 479 | + */ | |
| 480 | + private function totals_row(array $rows): array { | |
| 481 | + $row = is_array($rows[0] ?? null) ? $rows[0] : []; | |
| 482 | + return [ | |
| 483 | + 'clicks' => (int) ($row['clicks'] ?? 0), | |
| 484 | + 'impressions' => (int) ($row['impressions'] ?? 0), | |
| 485 | + 'ctr' => (float) ($row['ctr'] ?? 0.0), | |
| 486 | + 'position' => (float) ($row['position'] ?? 0.0), | |
| 487 | + ]; | |
| 488 | + } | |
| 489 | + | |
| 490 | + /** | |
| 491 | + * AI-assistant traffic for the current window and the one before it, | |
| 492 | + * from the first-party tracker AI Insights already runs. Null when the | |
| 493 | + * tracker is absent or has recorded nothing. | |
| 494 | + * | |
| 495 | + * The tracker summarises a trailing window, so the previous period is | |
| 496 | + * the double window minus the current one. | |
| 497 | + * | |
| 498 | + * @return array{current:array,previous:array}|null | |
| 499 | + */ | |
| 500 | + private function ai_summary(int $frequency_days): ?array { | |
| 501 | + $tracker = $this->get_ai_tracker(); | |
| 502 | + if ($tracker === null) { | |
| 503 | + return null; | |
| 504 | + } | |
| 505 | + try { | |
| 506 | + $current = (array) $tracker->summary($frequency_days); | |
| 507 | + if ((int) ($current['ai_sessions'] ?? 0) === 0 && (int) ($current['baseline'] ?? 0) === 0) { | |
| 508 | + return null; | |
| 509 | + } | |
| 510 | + $double = (array) $tracker->summary($frequency_days * 2); | |
| 511 | + $previous = [ | |
| 512 | + 'ai_sessions' => max(0, (int) ($double['ai_sessions'] ?? 0) - (int) ($current['ai_sessions'] ?? 0)), | |
| 513 | + 'baseline' => max(0, (int) ($double['baseline'] ?? 0) - (int) ($current['baseline'] ?? 0)), | |
| 514 | + ]; | |
| 515 | + return ['current' => $current, 'previous' => $previous]; | |
| 516 | + } catch (Throwable $e) { | |
| 517 | + return null; | |
| 518 | + } | |
| 519 | + } | |
| 520 | + | |
| 521 | + private function ai_tracker_has_data(): bool { | |
| 522 | + $tracker = $this->get_ai_tracker(); | |
| 523 | + if ($tracker === null) { | |
| 524 | + return false; | |
| 525 | + } | |
| 526 | + try { | |
| 527 | + $summary = (array) $tracker->summary(30); | |
| 528 | + return (int) ($summary['ai_sessions'] ?? 0) > 0 || (int) ($summary['baseline'] ?? 0) > 0; | |
| 529 | + } catch (Throwable $e) { | |
| 530 | + return false; | |
| 531 | + } | |
| 532 | + } | |
| 533 | + | |
| 534 | + private function get_ai_tracker() { | |
| 535 | + $cls = '\\ThinkRank\\SEO\\Ai_Traffic_Tracker'; | |
| 536 | + if (!class_exists($cls) || !method_exists($cls, 'summary')) { | |
| 537 | + return null; | |
| 538 | + } | |
| 539 | + try { | |
| 540 | + return new $cls(); | |
| 541 | + } catch (Throwable $e) { | |
| 542 | + return null; | |
| 543 | + } | |
| 544 | + } | |
| 545 | + | |
| 546 | + /** | |
| 148 | 547 | * Merge current + previous GSC rows into one keyed map carrying both |
| 149 | 548 | * periods' clicks and (for queries) average position. |
| 150 | 549 | * |
| 151 | - * @param array $current Current-window rows. | |
| 152 | - * @param array $previous Previous-window rows. | |
| 153 | - * @param bool $is_query True for query rows, false for page rows. | |
| 550 | + * The key set is the union of both windows. Search Console omits rows | |
| 551 | + * with no activity in a window, so a page or query that dropped to zero | |
| 552 | + * clicks has no current row at all — keying off `$current` alone would | |
| 553 | + * silently discard exactly the biggest losers the losing sections exist | |
| 554 | + * to surface. | |
| 555 | + * | |
| 556 | + * Which absent rows count as drop-outs is up to the caller. Absence from | |
| 557 | + * a truncated list proves nothing, so build_comparison() passes the keys | |
| 558 | + * it confirmed with Search Console; any other absent row is left out | |
| 559 | + * (#906). Null means the current list was complete and every absence is | |
| 560 | + * a drop-out. | |
| 561 | + * | |
| 562 | + * @param array $current Current-window rows. | |
| 563 | + * @param array $previous Previous-window rows. | |
| 564 | + * @param bool $is_query True for query rows, false for page rows. | |
| 565 | + * @param array<string,bool>|null $dropout_keys Confirmed drop-out keys, or null for all. | |
| 154 | 566 | * @return array<string,array> |
| 155 | 567 | */ |
| 156 | - private function merge_periods(array $current, array $previous, bool $is_query): array { | |
| 568 | + private function merge_periods(array $current, array $previous, bool $is_query, ?array $dropout_keys = null): array { | |
| 157 | 569 | $prev_map = []; |
| 158 | 570 | foreach ($previous as $row) { |
| 159 | 571 | $key = (string) ($row['keys'][0] ?? ''); |
| 160 | 572 | if ($key === '') { |
| @@ -185,8 +597,37 @@ | ||
| 185 | 597 | } |
| 186 | 598 | $merged[$key] = $entry; |
| 187 | 599 | } |
| 188 | 600 | |
| 601 | + // Total drop-outs: present last period, absent now. Synthesize them | |
| 602 | + // from the previous window with the current metrics zeroed. Position | |
| 603 | + // stays null rather than 0 — "no data" is not "ranked first". | |
| 604 | + foreach ($prev_map as $key => $prev_row) { | |
| 605 | + if (isset($merged[$key])) { | |
| 606 | + continue; | |
| 607 | + } | |
| 608 | + if (null !== $dropout_keys && !isset($dropout_keys[$key])) { | |
| 609 | + continue; | |
| 610 | + } | |
| 611 | + $raw = (string) ($prev_row['keys'][0] ?? ''); | |
| 612 | + if ($raw === '') { | |
| 613 | + continue; | |
| 614 | + } | |
| 615 | + | |
| 616 | + $entry = [ | |
| 617 | + 'cur_clicks' => 0, | |
| 618 | + 'prev_clicks' => (int) ($prev_row['clicks'] ?? 0), | |
| 619 | + ]; | |
| 620 | + if ($is_query) { | |
| 621 | + $entry['query'] = $raw; | |
| 622 | + $entry['cur_pos'] = null; | |
| 623 | + $entry['prev_pos'] = round((float) ($prev_row['position'] ?? 0), 1); | |
| 624 | + } else { | |
| 625 | + $entry['url'] = $raw; | |
| 626 | + } | |
| 627 | + $merged[$key] = $entry; | |
| 628 | + } | |
| 629 | + | |
| 189 | 630 | return $merged; |
| 190 | 631 | } |
| 191 | 632 | |
| 192 | 633 | private function normalize_key(string $key, bool $is_query): string { |
| @@ -192,8 +633,29 @@ | ||
| 192 | 633 | private function normalize_key(string $key, bool $is_query): string { |
| 193 | 634 | return $is_query ? trim(strtolower($key)) : $key; |
| 194 | 635 | } |
| 195 | 636 | |
| 637 | + /** | |
| 638 | + * The plugin settings store, or a null-object when it is not loaded | |
| 639 | + * (unit tests without the core classes), which reads as "nothing | |
| 640 | + * configured". | |
| 641 | + */ | |
| 642 | + private function settings() { | |
| 643 | + $cls = '\\ThinkRank\\Core\\Settings'; | |
| 644 | + if (class_exists($cls) && method_exists($cls, 'instance')) { | |
| 645 | + try { | |
| 646 | + return $cls::instance(); | |
| 647 | + } catch (Throwable $e) { | |
| 648 | + // Fall through to the null object. | |
| 649 | + } | |
| 650 | + } | |
| 651 | + return new class() { | |
| 652 | + public function get(string $key, $fallback = null) { | |
| 653 | + return $fallback; | |
| 654 | + } | |
| 655 | + }; | |
| 656 | + } | |
| 657 | + | |
| 196 | 658 | private function get_analytics_manager() { |
| 197 | 659 | $cls = '\\ThinkRank\\SEO\\Analytics_Manager'; |
| 198 | 660 | if (!class_exists($cls)) { |
| 199 | 661 | return null; |
| @@ -206,9 +668,20 @@ | ||
| 206 | 668 | } |
| 207 | 669 | |
| 208 | 670 | private function format_period_label(string $start, string $end): string { |
| 209 | 671 | $fmt = (string) get_option('date_format', 'M j, Y'); |
| 210 | - $a = wp_date($fmt, strtotime($start) ?: time()); | |
| 211 | - $b = wp_date($fmt, strtotime($end) ?: time()); | |
| 212 | - return sprintf('%s – %s', $a, $b); | |
| 672 | + return sprintf('%s – %s', $this->format_day($start, $fmt), $this->format_day($end, $fmt)); | |
| 673 | + } | |
| 674 | + | |
| 675 | + /** | |
| 676 | + * Render a bare Y-m-d as a localized date. | |
| 677 | + * | |
| 678 | + * Anchored at midday UTC on purpose: wp_date() shifts the timestamp into | |
| 679 | + * the site timezone, and a date parsed at midnight would render as the | |
| 680 | + * day before on any negative offset. Midday leaves the calendar date | |
| 681 | + * intact across every real-world offset. | |
| 682 | + */ | |
| 683 | + private function format_day(string $date, string $format): string { | |
| 684 | + $ts = strtotime($date . ' 12:00:00 UTC'); | |
| 685 | + return wp_date($format, $ts ?: time()); | |
| 213 | 686 | } |
| 214 | 687 | } |