ID, array( 'ingredients', 'recipe_ingredients', 'recipeIngredient' ) ); if ( $val ) { return self::normalise_list( $val ); } // 2. Heading "Ingredients" followed by a

. if ( preg_match( '/]*>\s*ingredients?\s*<\/h[2-4]>\s*<(ul|ol)[^>]*>(.*?)<\/\1>/is', $html, $m ) ) { $list_html = $m[2]; preg_match_all( '/]*>(.*?)<\/li>/is', $list_html, $items ); if ( ! empty( $items[1] ) ) { $cleaned = array_map( static function ( $i ) { return trim( wp_strip_all_tags( $i ) ); }, $items[1] ); $cleaned = array_values( array_filter( $cleaned ) ); if ( $cleaned ) { return $cleaned; } } } return null; } /** * Extract step-by-step instructions. * * @return array|null */ public static function instructions( WP_Post $post, string $text, string $html ): ?array { // ACF / meta. $val = self::meta_value( $post->ID, array( 'instructions', 'recipe_instructions', 'directions', 'steps' ) ); if ( $val ) { $items = self::normalise_list( $val ); $out = array(); foreach ( $items as $i => $step ) { // translators: %d: step number in a how-to or recipe. $aeomatic_step_name = sprintf( __( 'Step %d', 'aeomatic' ), $i + 1 ); $out[] = array( '@type' => 'HowToStep', 'name' => $aeomatic_step_name, 'text' => $step, ); } return $out; } // "Instructions" or "Directions" heading followed by ordered list. if ( preg_match( '/]*>\s*(instructions?|directions?|steps?|method)\s*<\/h[2-4]>\s*<(ol|ul)[^>]*>(.*?)<\/\2>/is', $html, $m ) ) { preg_match_all( '/]*>(.*?)<\/li>/is', $m[3], $items ); if ( ! empty( $items[1] ) ) { $out = array(); foreach ( $items[1] as $i => $step ) { $plain = trim( wp_strip_all_tags( $step ) ); if ( $plain ) { // translators: %d: step number in a how-to or recipe. $aeomatic_step_name = sprintf( __( 'Step %d', 'aeomatic' ), $i + 1 ); $out[] = array( '@type' => 'HowToStep', 'name' => $aeomatic_step_name, 'text' => $plain, ); } } if ( $out ) { return $out; } } } // Fall back: H2/H3 headings starting with "Step" build the steps. preg_match_all( '/]*>(.*?)<\/h[2-3]>/is', $html, $hm ); $steps = array(); foreach ( ( $hm[1] ?? array() ) as $h ) { $plain = trim( wp_strip_all_tags( $h ) ); if ( preg_match( '/^\s*(step\s+\d+|[\d]+[\.\)])/i', $plain ) ) { $steps[] = array( '@type' => 'HowToStep', 'name' => $plain, 'text' => $plain, ); } } return $steps ?: null; } /** * Extract an ISO 8601 duration ("PT30M" etc.) from natural-language phrases. * * @return string|null */ public static function duration( string $text, string $label ): ?string { // Look for "{label} time: 30 minutes" or "{label}: 1 hour 30 minutes". $pattern = '/' . preg_quote( $label, '/' ) . '\s*(?:time)?\s*[:\-]?\s*((?:\d+\s*(?:hours?|h|hrs?|minutes?|mins?|m)\s*)+)/i'; if ( preg_match( $pattern, $text, $m ) ) { return self::phrase_to_iso8601_duration( $m[1] ); } return null; } private static function phrase_to_iso8601_duration( string $phrase ): ?string { $h = 0; $min = 0; if ( preg_match( '/(\d+)\s*(hours?|h|hrs?)/i', $phrase, $m1 ) ) { $h = (int) $m1[1]; } if ( preg_match( '/(\d+)\s*(minutes?|mins?|m)/i', $phrase, $m2 ) ) { $min = (int) $m2[1]; } if ( ! $h && ! $min ) { return null; } $out = 'PT'; if ( $h ) $out .= $h . 'H'; if ( $min ) $out .= $min . 'M'; return $out; } /** * Recipe yield ("Serves 4" / "Yield: 12 cookies"). */ public static function recipe_yield( WP_Post $post, string $text ): ?string { $val = self::meta_value( $post->ID, array( 'yield', 'recipe_yield', 'servings' ) ); if ( $val ) { return is_string( $val ) ? trim( $val ) : (string) $val; } if ( preg_match( '/(?:serves|servings|yield|makes)\s*[:\-]?\s*([\d]+(?:\s*\w+)?)/i', $text, $m ) ) { return trim( $m[1] ); } return null; } /* ----- Review / rating extractors ----- */ /** * Extract a rating value as a float (0–5 scale normalised). * * @return array{value:float,best:int}|null */ public static function rating( WP_Post $post, string $text ): ?array { // ACF / meta. $val = self::meta_value( $post->ID, array( 'rating', 'review_rating', 'star_rating' ) ); if ( is_numeric( $val ) ) { $f = (float) $val; if ( $f > 0 && $f <= 5 ) return array( 'value' => $f, 'best' => 5 ); if ( $f > 5 && $f <= 10 ) return array( 'value' => $f, 'best' => 10 ); } // Pattern: "4.5 out of 5", "4.8/5", "8/10". if ( preg_match( '/(\d(?:\.\d)?)\s*(?:out of|\/)\s*(5|10)\b/i', $text, $m ) ) { return array( 'value' => (float) $m[1], 'best' => (int) $m[2] ); } // Pattern: "Rating: 4.5". if ( preg_match( '/(?:rating|score)\s*[:\-]\s*(\d(?:\.\d)?)/i', $text, $m ) ) { return array( 'value' => (float) $m[1], 'best' => 5 ); } return null; } /** * Pull pros and cons lists. * * @return array{pros:array,cons:array}|null */ public static function pros_and_cons( WP_Post $post, string $html ): ?array { $pros = self::list_after_heading( $html, '(pros|advantages|the good|what we liked)' ); $cons = self::list_after_heading( $html, '(cons|disadvantages|the bad|what we didn\'t like|drawbacks)' ); // ACF. $pros_meta = self::meta_value( $post->ID, array( 'pros', 'positives', 'advantages' ) ); $cons_meta = self::meta_value( $post->ID, array( 'cons', 'negatives', 'disadvantages' ) ); if ( $pros_meta ) { $pros = array_merge( (array) $pros, self::normalise_list( $pros_meta ) ); } if ( $cons_meta ) { $cons = array_merge( (array) $cons, self::normalise_list( $cons_meta ) ); } if ( empty( $pros ) && empty( $cons ) ) { return null; } return array( 'pros' => array_values( array_filter( $pros ?: array() ) ), 'cons' => array_values( array_filter( $cons ?: array() ) ), ); } /* ----- Event extractors ----- */ /** * Extract event start date as ISO 8601. */ public static function event_start( WP_Post $post, string $text ): ?string { $val = self::meta_value( $post->ID, array( 'event_date', 'start_date', 'event_start_date', '_EventStartDate' ) ); if ( $val ) { $ts = is_numeric( $val ) ? (int) $val : strtotime( (string) $val ); if ( $ts ) { return gmdate( 'c', $ts ); } } // Pattern: "Date: 2026-04-28" or "Apr 28, 2026 at 7pm". if ( preg_match( '/\b(\d{4}-\d{2}-\d{2}(?:T\d{2}:\d{2})?)\b/', $text, $m ) ) { $ts = strtotime( $m[1] ); if ( $ts ) return gmdate( 'c', $ts ); } return null; } public static function event_end( WP_Post $post ): ?string { $val = self::meta_value( $post->ID, array( 'event_end_date', 'end_date', '_EventEndDate' ) ); if ( $val ) { $ts = is_numeric( $val ) ? (int) $val : strtotime( (string) $val ); if ( $ts ) { return gmdate( 'c', $ts ); } } return null; } public static function event_location( WP_Post $post ): ?array { $venue = self::meta_value( $post->ID, array( 'venue', 'event_venue', 'location_name', '_EventVenue' ) ); $addr = self::meta_value( $post->ID, array( 'address', 'event_address', 'location_address', '_EventAddress' ) ); if ( ! $venue && ! $addr ) { return null; } $loc = array( '@type' => 'Place' ); if ( $venue ) $loc['name'] = is_string( $venue ) ? $venue : (string) $venue; if ( $addr ) $loc['address'] = is_string( $addr ) ? $addr : (string) $addr; return $loc; } /* ----- Job posting extractors ----- */ public static function job_salary( WP_Post $post, string $text ): ?array { $val = self::meta_value( $post->ID, array( 'salary', 'job_salary', 'compensation' ) ); if ( $val ) { if ( is_array( $val ) && isset( $val['min'], $val['max'] ) ) { return array( '@type' => 'MonetaryAmount', 'currency' => $val['currency'] ?? 'USD', 'value' => array( '@type' => 'QuantitativeValue', 'minValue' => (float) $val['min'], 'maxValue' => (float) $val['max'], 'unitText' => $val['unit'] ?? 'YEAR', ), ); } if ( is_string( $val ) || is_numeric( $val ) ) { return array( '@type' => 'MonetaryAmount', 'currency' => 'USD', 'value' => (string) $val, ); } } // Pattern: "$80,000 – $120,000 per year". if ( preg_match( '/(?:[\$€£₹]|usd|inr|eur|gbp)\s*([\d,]+)\s*[–\-to]+\s*(?:[\$€£₹]|usd|inr|eur|gbp)?\s*([\d,]+)\s*(?:per|\/)?\s*(year|month|hour|annum)?/i', $text, $m ) ) { $min = (float) str_replace( ',', '', $m[1] ); $max = (float) str_replace( ',', '', $m[2] ); $unit_raw = strtolower( $m[3] ?? 'year' ); $unit = 'YEAR'; if ( false !== strpos( $unit_raw, 'month' ) ) $unit = 'MONTH'; elseif ( false !== strpos( $unit_raw, 'hour' ) ) $unit = 'HOUR'; return array( '@type' => 'MonetaryAmount', 'currency' => 'USD', 'value' => array( '@type' => 'QuantitativeValue', 'minValue' => $min, 'maxValue' => $max, 'unitText' => $unit, ), ); } return null; } public static function employment_type( WP_Post $post, string $text ): ?string { $val = self::meta_value( $post->ID, array( 'employment_type', 'job_type' ) ); if ( $val && is_string( $val ) ) { return strtoupper( str_replace( array( ' ', '-' ), '_', $val ) ); } if ( preg_match( '/\b(full[\s-]?time)\b/i', $text ) ) return 'FULL_TIME'; if ( preg_match( '/\b(part[\s-]?time)\b/i', $text ) ) return 'PART_TIME'; if ( preg_match( '/\bcontract(or)?\b/i', $text ) ) return 'CONTRACTOR'; if ( preg_match( '/\binternship\b/i', $text ) ) return 'INTERN'; if ( preg_match( '/\btemporary\b/i', $text ) ) return 'TEMPORARY'; return null; } /* ----- Local business extractors ----- */ public static function address( WP_Post $post, string $text ): ?array { $street = self::meta_value( $post->ID, array( 'street_address', 'address_line_1', 'address' ) ); $city = self::meta_value( $post->ID, array( 'city', 'locality' ) ); $region = self::meta_value( $post->ID, array( 'state', 'region' ) ); $postal = self::meta_value( $post->ID, array( 'postcode', 'zip', 'postal_code' ) ); $country = self::meta_value( $post->ID, array( 'country' ) ); if ( $street || $city ) { $out = array( '@type' => 'PostalAddress' ); if ( $street ) $out['streetAddress'] = (string) ( is_array( $street ) ? implode( ', ', $street ) : $street ); if ( $city ) $out['addressLocality'] = (string) $city; if ( $region ) $out['addressRegion'] = (string) $region; if ( $postal ) $out['postalCode'] = (string) $postal; if ( $country ) $out['addressCountry'] = (string) $country; return $out; } // Best-effort pattern match. if ( preg_match( '/(\d+\s+[A-Z][A-Za-z\s]+(?:Street|St\.?|Road|Rd\.?|Avenue|Ave\.?|Lane|Ln\.?|Drive|Dr\.?))/i', $text, $m ) ) { return array( '@type' => 'PostalAddress', 'streetAddress' => trim( $m[1] ), ); } return null; } public static function telephone( WP_Post $post, string $text ): ?string { $val = self::meta_value( $post->ID, array( 'phone', 'telephone', 'contact_phone' ) ); if ( $val && is_string( $val ) ) return $val; if ( preg_match( '/(\+\d{1,3}[\s-]?\d{2,4}[\s-]?\d{3,4}[\s-]?\d{3,4})/', $text, $m ) ) { return trim( $m[1] ); } if ( preg_match( '/(\d{3}[\s-]\d{3}[\s-]\d{4})/', $text, $m ) ) { return trim( $m[1] ); } return null; } public static function opening_hours( WP_Post $post, string $text ): ?array { $val = self::meta_value( $post->ID, array( 'opening_hours', 'hours' ) ); if ( is_array( $val ) ) { return $val; } if ( is_string( $val ) ) { return array_filter( array_map( 'trim', preg_split( '/\r\n|\r|\n/', $val ) ) ?: array() ); } return null; } /* ----- Course extractors ----- */ public static function course_provider( WP_Post $post ): ?array { $provider_name = self::meta_value( $post->ID, array( 'provider', 'course_provider', 'instructor' ) ); $name = is_string( $provider_name ) ? $provider_name : ( is_array( $provider_name ) ? ( $provider_name['name'] ?? '' ) : '' ); if ( ! $name ) { $name = get_bloginfo( 'name' ); } return array( '@type' => 'Organization', 'name' => $name, 'url' => home_url( '/' ), ); } /* ----- Generic helpers ----- */ /** * Normalise a value into an array of trimmed strings. * * @param mixed $val * @return string[] */ private static function normalise_list( $val ): array { if ( is_array( $val ) ) { $out = array(); foreach ( $val as $row ) { if ( is_string( $row ) ) { $s = trim( wp_strip_all_tags( $row ) ); if ( $s ) $out[] = $s; } elseif ( is_array( $row ) ) { // ACF repeater: pick the first text-ish field. foreach ( $row as $sub ) { if ( is_string( $sub ) && '' !== trim( $sub ) ) { $out[] = trim( wp_strip_all_tags( $sub ) ); break; } } } } return $out; } if ( is_string( $val ) ) { $lines = preg_split( '/\r\n|\r|\n/', $val ); $out = array(); foreach ( $lines as $line ) { $line = trim( wp_strip_all_tags( $line ) ); if ( $line ) $out[] = $line; } return $out; } return array(); } /** * Find a heading matching $heading_pattern, then return the items * of the first
or
that follows it. * * @return string[]|null */ private static function list_after_heading( string $html, string $heading_pattern ): ?array { $re = '/]*>\s*' . $heading_pattern . '\s*<\/h[2-4]>\s*<(ul|ol)[^>]*>(.*?)<\/\2>/is'; if ( preg_match( $re, $html, $m ) ) { $inner = $m[3] ?? $m[2]; preg_match_all( '/]*>(.*?)<\/li>/is', $inner, $items ); if ( ! empty( $items[1] ) ) { $out = array_map( static function ( $i ) { return trim( wp_strip_all_tags( $i ) ); }, $items[1] ); return array_values( array_filter( $out ) ); } } return null; } }