image.php
1 year ago
message-builder.php
10 months ago
model-environment.php
11 months ago
response-id-manager.php
1 year ago
session.php
11 months ago
usage-stats.php
10 months ago
usage-stats.php
340 lines
| 1 | <?php |
| 2 | |
| 3 | class Meow_MWAI_Services_UsageStats { |
| 4 | private $core; |
| 5 | private $tiktoken_encoders = []; |
| 6 | private $encoder_provider = null; |
| 7 | |
| 8 | public function __construct( $core ) { |
| 9 | $this->core = $core; |
| 10 | } |
| 11 | |
| 12 | /** |
| 13 | * Get the cl100k_base tiktoken encoder |
| 14 | * Note: We always use cl100k_base regardless of model since it's the standard for all modern OpenAI models |
| 15 | * @return object|null The tiktoken encoder or null if not available |
| 16 | */ |
| 17 | private function get_tiktoken_encoder() { |
| 18 | // Return cached encoder if available |
| 19 | if ( isset( $this->tiktoken_encoders['cl100k_base'] ) ) { |
| 20 | return $this->tiktoken_encoders['cl100k_base']; |
| 21 | } |
| 22 | |
| 23 | try { |
| 24 | // Check if class exists |
| 25 | if ( !class_exists( 'Yethee\Tiktoken\EncoderProvider' ) ) { |
| 26 | error_log( '[AI Engine Tiktoken] EncoderProvider class not found. Check if composer autoload is working.' ); |
| 27 | return null; |
| 28 | } |
| 29 | |
| 30 | // Initialize encoder provider if needed |
| 31 | if ( $this->encoder_provider === null ) { |
| 32 | $this->encoder_provider = new \Yethee\Tiktoken\EncoderProvider(); |
| 33 | } |
| 34 | |
| 35 | // Get the cl100k_base encoder (standard for modern OpenAI models) |
| 36 | $encoder = $this->encoder_provider->get( 'cl100k_base' ); |
| 37 | $this->tiktoken_encoders['cl100k_base'] = $encoder; |
| 38 | |
| 39 | // Removed success log to reduce noise |
| 40 | return $encoder; |
| 41 | } |
| 42 | catch ( \Exception $e ) { |
| 43 | error_log( '[AI Engine Tiktoken] Failed to initialize encoder: ' . $e->getMessage() ); |
| 44 | return null; |
| 45 | } |
| 46 | } |
| 47 | |
| 48 | public function estimate_tokens( ...$args ) { |
| 49 | // Handle multiple argument formats for backward compatibility |
| 50 | $text = ''; |
| 51 | $model = null; |
| 52 | |
| 53 | // If first argument is an array, process messages |
| 54 | if ( !empty( $args[0] ) && is_array( $args[0] ) ) { |
| 55 | foreach ( $args[0] as $message ) { |
| 56 | $text .= isset( $message['content']['text'] ) ? $message['content']['text'] : ''; |
| 57 | $text .= isset( $message['content'] ) && is_string( $message['content'] ) ? $message['content'] : ''; |
| 58 | } |
| 59 | $model = $args[1] ?? null; |
| 60 | } |
| 61 | // Otherwise treat first argument as text |
| 62 | else { |
| 63 | $text = $args[0] ?? ''; |
| 64 | $model = $args[1] ?? null; |
| 65 | } |
| 66 | |
| 67 | // Convert to string if needed |
| 68 | if ( !is_string( $text ) ) { |
| 69 | // Handle arrays that weren't caught by the first condition |
| 70 | if ( is_array( $text ) ) { |
| 71 | $text = json_encode( $text ); |
| 72 | } |
| 73 | // Handle objects |
| 74 | elseif ( is_object( $text ) ) { |
| 75 | $text = method_exists( $text, '__toString' ) ? (string) $text : json_encode( $text ); |
| 76 | } |
| 77 | // Handle other types (int, float, bool, null) |
| 78 | else { |
| 79 | $text = (string) $text; |
| 80 | } |
| 81 | } |
| 82 | |
| 83 | // Apply filters for customization |
| 84 | $text = apply_filters( 'mwai_estimate_tokens_text', $text, $model ); |
| 85 | $tokens = apply_filters( 'mwai_estimate_tokens', null, $text, $model ); |
| 86 | if ( $tokens !== null ) { |
| 87 | return $tokens; |
| 88 | } |
| 89 | |
| 90 | // Try to use tiktoken for accurate counting (cl100k_base encoder) |
| 91 | $encoder = $this->get_tiktoken_encoder(); |
| 92 | if ( $encoder ) { |
| 93 | try { |
| 94 | $encoded = $encoder->encode( $text ); |
| 95 | $token_count = count( $encoded ); |
| 96 | |
| 97 | // Comparison logging removed - tiktoken is working correctly |
| 98 | |
| 99 | return $token_count; |
| 100 | } |
| 101 | catch ( Exception $e ) { |
| 102 | error_log( '[AI Engine Tiktoken] Encoding failed, falling back to estimation: ' . $e->getMessage() ); |
| 103 | } |
| 104 | } |
| 105 | else { |
| 106 | error_log( '[AI Engine Tiktoken] Encoder not available, using fallback' ); |
| 107 | } |
| 108 | |
| 109 | // Fallback to old estimation method |
| 110 | return $this->fallback_estimate_tokens( $text ); |
| 111 | } |
| 112 | |
| 113 | /** |
| 114 | * Fallback token estimation method (the original implementation) |
| 115 | */ |
| 116 | private function fallback_estimate_tokens( $text ) { |
| 117 | $multiplier = 4; |
| 118 | $hasChineseChars = preg_match( '/[\x{4e00}-\x{9fa5}]/u', $text ); |
| 119 | $hasJapaneseChars = preg_match( '/[\x{3040}-\x{309f}\x{30a0}-\x{30ff}]/u', $text ); |
| 120 | $hasKoreanChars = preg_match( '/[\x{ac00}-\x{d7af}]/u', $text ); |
| 121 | if ( $hasChineseChars || $hasJapaneseChars || $hasKoreanChars ) { |
| 122 | $multiplier = 2; |
| 123 | } |
| 124 | $tokens = (int) ( ( function_exists( 'mb_strlen' ) ? mb_strlen( $text ) : strlen( $text ) ) / $multiplier ); |
| 125 | return $tokens; |
| 126 | } |
| 127 | |
| 128 | public function record_tokens_usage( $model, $in_tokens, $out_tokens = 0, $returned_price = null ) { |
| 129 | if ( !is_numeric( $in_tokens ) ) { |
| 130 | $in_tokens = 0; |
| 131 | } |
| 132 | if ( !is_numeric( $out_tokens ) ) { |
| 133 | $out_tokens = 0; |
| 134 | } |
| 135 | |
| 136 | // Record monthly usage |
| 137 | $usage = $this->core->get_option( 'ai_usage' ); |
| 138 | $month = date( 'Y-m' ); |
| 139 | if ( !isset( $usage[$month] ) ) { |
| 140 | $usage[$month] = []; |
| 141 | } |
| 142 | if ( !isset( $usage[$month][$model] ) ) { |
| 143 | $usage[$month][$model] = [ |
| 144 | 'prompt_tokens' => 0, |
| 145 | 'completion_tokens' => 0, |
| 146 | 'total_tokens' => 0, |
| 147 | 'returned_price' => 0, |
| 148 | 'queries' => 0 |
| 149 | ]; |
| 150 | } |
| 151 | // Ensure queries field exists for existing data |
| 152 | if ( !isset( $usage[$month][$model]['queries'] ) ) { |
| 153 | $usage[$month][$model]['queries'] = 0; |
| 154 | } |
| 155 | $usage[$month][$model]['prompt_tokens'] += $in_tokens; |
| 156 | $usage[$month][$model]['completion_tokens'] += $out_tokens; |
| 157 | $usage[$month][$model]['total_tokens'] += $in_tokens + $out_tokens; |
| 158 | $usage[$month][$model]['queries'] += 1; |
| 159 | if ( !empty( $returned_price ) ) { |
| 160 | // Ensure returned_price is numeric to avoid type errors |
| 161 | $price_value = is_array( $returned_price ) ? |
| 162 | ( isset( $returned_price['price'] ) ? $returned_price['price'] : 0 ) : |
| 163 | ( is_numeric( $returned_price ) ? $returned_price : 0 ); |
| 164 | $usage[$month][$model]['returned_price'] += $price_value; |
| 165 | } |
| 166 | |
| 167 | // Clean up old monthly data (keep only last 2 years) |
| 168 | $this->cleanup_old_monthly_data( $usage ); |
| 169 | $this->core->update_option( 'ai_usage', $usage ); |
| 170 | |
| 171 | // Record daily usage |
| 172 | $daily_usage = $this->core->get_option( 'ai_usage_daily', [] ); |
| 173 | $day = date( 'Y-m-d' ); |
| 174 | if ( !isset( $daily_usage[$day] ) ) { |
| 175 | $daily_usage[$day] = []; |
| 176 | } |
| 177 | if ( !isset( $daily_usage[$day][$model] ) ) { |
| 178 | $daily_usage[$day][$model] = [ |
| 179 | 'prompt_tokens' => 0, |
| 180 | 'completion_tokens' => 0, |
| 181 | 'total_tokens' => 0, |
| 182 | 'returned_price' => 0, |
| 183 | 'queries' => 0 |
| 184 | ]; |
| 185 | } |
| 186 | // Ensure queries field exists for existing data |
| 187 | if ( !isset( $daily_usage[$day][$model]['queries'] ) ) { |
| 188 | $daily_usage[$day][$model]['queries'] = 0; |
| 189 | } |
| 190 | $daily_usage[$day][$model]['prompt_tokens'] += $in_tokens; |
| 191 | $daily_usage[$day][$model]['completion_tokens'] += $out_tokens; |
| 192 | $daily_usage[$day][$model]['total_tokens'] += $in_tokens + $out_tokens; |
| 193 | $daily_usage[$day][$model]['queries'] += 1; |
| 194 | if ( !empty( $returned_price ) ) { |
| 195 | $daily_usage[$day][$model]['returned_price'] += $returned_price; |
| 196 | } |
| 197 | |
| 198 | // Clean up old daily data (keep only last 30 days) |
| 199 | $this->cleanup_old_daily_data( $daily_usage ); |
| 200 | $this->core->update_option( 'ai_usage_daily', $daily_usage ); |
| 201 | |
| 202 | // Return the usage data for this specific request |
| 203 | return [ |
| 204 | 'prompt_tokens' => $in_tokens, |
| 205 | 'completion_tokens' => $out_tokens, |
| 206 | 'total_tokens' => $in_tokens + $out_tokens, |
| 207 | 'price' => $returned_price, |
| 208 | 'queries' => 1 |
| 209 | ]; |
| 210 | } |
| 211 | |
| 212 | public function record_audio_usage( $model, $seconds ) { |
| 213 | // Record monthly usage |
| 214 | $usage = $this->core->get_option( 'ai_usage' ); |
| 215 | $month = date( 'Y-m' ); |
| 216 | if ( !isset( $usage[$month] ) ) { |
| 217 | $usage[$month] = []; |
| 218 | } |
| 219 | if ( !isset( $usage[$month][$model] ) ) { |
| 220 | $usage[$month][$model] = [ 'seconds' => 0, 'queries' => 0 ]; |
| 221 | } |
| 222 | if ( !isset( $usage[$month][$model]['seconds'] ) ) { |
| 223 | $usage[$month][$model]['seconds'] = 0; |
| 224 | } |
| 225 | if ( !isset( $usage[$month][$model]['queries'] ) ) { |
| 226 | $usage[$month][$model]['queries'] = 0; |
| 227 | } |
| 228 | $usage[$month][$model]['seconds'] += $seconds; |
| 229 | $usage[$month][$model]['queries'] += 1; |
| 230 | $this->cleanup_old_monthly_data( $usage ); |
| 231 | $this->core->update_option( 'ai_usage', $usage ); |
| 232 | |
| 233 | // Record daily usage |
| 234 | $daily_usage = $this->core->get_option( 'ai_usage_daily', [] ); |
| 235 | $day = date( 'Y-m-d' ); |
| 236 | if ( !isset( $daily_usage[$day] ) ) { |
| 237 | $daily_usage[$day] = []; |
| 238 | } |
| 239 | if ( !isset( $daily_usage[$day][$model] ) ) { |
| 240 | $daily_usage[$day][$model] = [ 'seconds' => 0, 'queries' => 0 ]; |
| 241 | } |
| 242 | if ( !isset( $daily_usage[$day][$model]['seconds'] ) ) { |
| 243 | $daily_usage[$day][$model]['seconds'] = 0; |
| 244 | } |
| 245 | if ( !isset( $daily_usage[$day][$model]['queries'] ) ) { |
| 246 | $daily_usage[$day][$model]['queries'] = 0; |
| 247 | } |
| 248 | $daily_usage[$day][$model]['seconds'] += $seconds; |
| 249 | $daily_usage[$day][$model]['queries'] += 1; |
| 250 | $this->cleanup_old_daily_data( $daily_usage ); |
| 251 | $this->core->update_option( 'ai_usage_daily', $daily_usage ); |
| 252 | |
| 253 | // Return the usage data for this specific request |
| 254 | return [ |
| 255 | 'seconds' => $seconds, |
| 256 | 'queries' => 1 |
| 257 | ]; |
| 258 | } |
| 259 | |
| 260 | public function record_images_usage( $model, $resolution, $images ) { |
| 261 | // Record monthly usage |
| 262 | $usage = $this->core->get_option( 'ai_usage' ); |
| 263 | $month = date( 'Y-m' ); |
| 264 | if ( !isset( $usage[$month] ) ) { |
| 265 | $usage[$month] = []; |
| 266 | } |
| 267 | if ( !isset( $usage[$month][$model] ) ) { |
| 268 | $usage[$month][$model] = [ 'resolution' => [], 'images' => 0, 'queries' => 0 ]; |
| 269 | } |
| 270 | if ( !isset( $usage[$month][$model]['images'] ) ) { |
| 271 | $usage[$month][$model]['images'] = 0; |
| 272 | } |
| 273 | if ( !isset( $usage[$month][$model]['resolution'] ) ) { |
| 274 | $usage[$month][$model]['resolution'] = []; |
| 275 | } |
| 276 | if ( !isset( $usage[$month][$model]['resolution'][$resolution] ) ) { |
| 277 | $usage[$month][$model]['resolution'][$resolution] = 0; |
| 278 | } |
| 279 | if ( !isset( $usage[$month][$model]['queries'] ) ) { |
| 280 | $usage[$month][$model]['queries'] = 0; |
| 281 | } |
| 282 | $usage[$month][$model]['images'] += $images; |
| 283 | $usage[$month][$model]['resolution'][$resolution] += $images; |
| 284 | $usage[$month][$model]['queries'] += 1; |
| 285 | $this->cleanup_old_monthly_data( $usage ); |
| 286 | $this->core->update_option( 'ai_usage', $usage ); |
| 287 | |
| 288 | // Record daily usage |
| 289 | $daily_usage = $this->core->get_option( 'ai_usage_daily', [] ); |
| 290 | $day = date( 'Y-m-d' ); |
| 291 | if ( !isset( $daily_usage[$day] ) ) { |
| 292 | $daily_usage[$day] = []; |
| 293 | } |
| 294 | if ( !isset( $daily_usage[$day][$model] ) ) { |
| 295 | $daily_usage[$day][$model] = [ 'resolution' => [], 'images' => 0, 'queries' => 0 ]; |
| 296 | } |
| 297 | if ( !isset( $daily_usage[$day][$model]['images'] ) ) { |
| 298 | $daily_usage[$day][$model]['images'] = 0; |
| 299 | } |
| 300 | if ( !isset( $daily_usage[$day][$model]['resolution'] ) ) { |
| 301 | $daily_usage[$day][$model]['resolution'] = []; |
| 302 | } |
| 303 | if ( !isset( $daily_usage[$day][$model]['resolution'][$resolution] ) ) { |
| 304 | $daily_usage[$day][$model]['resolution'][$resolution] = 0; |
| 305 | } |
| 306 | if ( !isset( $daily_usage[$day][$model]['queries'] ) ) { |
| 307 | $daily_usage[$day][$model]['queries'] = 0; |
| 308 | } |
| 309 | $daily_usage[$day][$model]['images'] += $images; |
| 310 | $daily_usage[$day][$model]['resolution'][$resolution] += $images; |
| 311 | $daily_usage[$day][$model]['queries'] += 1; |
| 312 | $this->cleanup_old_daily_data( $daily_usage ); |
| 313 | $this->core->update_option( 'ai_usage_daily', $daily_usage ); |
| 314 | |
| 315 | // Return the usage data for this specific request |
| 316 | return [ |
| 317 | 'images' => $images, |
| 318 | 'queries' => 1 |
| 319 | ]; |
| 320 | } |
| 321 | |
| 322 | private function cleanup_old_monthly_data( &$usage ) { |
| 323 | $two_years_ago = date( 'Y-m', strtotime( '-2 years' ) ); |
| 324 | foreach ( $usage as $month => $data ) { |
| 325 | if ( $month < $two_years_ago ) { |
| 326 | unset( $usage[$month] ); |
| 327 | } |
| 328 | } |
| 329 | } |
| 330 | |
| 331 | private function cleanup_old_daily_data( &$usage ) { |
| 332 | $thirty_days_ago = date( 'Y-m-d', strtotime( '-30 days' ) ); |
| 333 | foreach ( $usage as $day => $data ) { |
| 334 | if ( $day < $thirty_days_ago ) { |
| 335 | unset( $usage[$day] ); |
| 336 | } |
| 337 | } |
| 338 | } |
| 339 | } |
| 340 |