| @@ -1,249 +1,249 @@ | ||
| 1 | -<?php | |
| 2 | - | |
| 3 | -namespace LearnPress\AI\Assistant; | |
| 4 | - | |
| 5 | -/** | |
| 6 | - * ResponseNormalizer — decodes JSON model output and extracts human-readable text. | |
| 7 | - * | |
| 8 | - * Handles the common pattern where OpenAI returns JSON-wrapped text such as | |
| 9 | - * {"message":"..."} and extracts the best text candidate for frontend rendering. | |
| 10 | - * | |
| 11 | - * @package LearnPress\AI\Assistant | |
| 12 | - * @since 4.3.5 | |
| 13 | - */ | |
| 14 | -class ResponseNormalizer { | |
| 15 | - | |
| 16 | - /** | |
| 17 | - * Decode JSON content safely without throwing exceptions. | |
| 18 | - * | |
| 19 | - * Accepts either pure JSON or content containing a JSON object substring. | |
| 20 | - * Returns empty array when content is plain text or invalid JSON. | |
| 21 | - * | |
| 22 | - * @param string $content Raw model content. | |
| 23 | - * | |
| 24 | - * @return array | |
| 25 | - */ | |
| 26 | - public function decode_json( string $content ): array { | |
| 27 | - $content = trim( $content ); | |
| 28 | - if ( '' === $content ) { | |
| 29 | - return array(); | |
| 30 | - } | |
| 31 | - | |
| 32 | - $decoded = json_decode( $content, true ); | |
| 33 | - if ( is_array( $decoded ) && JSON_ERROR_NONE === json_last_error() ) { | |
| 34 | - return $decoded; | |
| 35 | - } | |
| 36 | - | |
| 37 | - $first_brace = strpos( $content, '{' ); | |
| 38 | - $last_brace = strrpos( $content, '}' ); | |
| 39 | - if ( false === $first_brace || false === $last_brace || $last_brace <= $first_brace ) { | |
| 40 | - return array(); | |
| 41 | - } | |
| 42 | - | |
| 43 | - $json_slice = substr( $content, $first_brace, $last_brace - $first_brace + 1 ); | |
| 44 | - if ( ! is_string( $json_slice ) || '' === $json_slice ) { | |
| 45 | - return array(); | |
| 46 | - } | |
| 47 | - | |
| 48 | - $decoded = json_decode( $json_slice, true ); | |
| 49 | - | |
| 50 | - return ( is_array( $decoded ) && JSON_ERROR_NONE === json_last_error() ) ? $decoded : array(); | |
| 51 | - } | |
| 52 | - | |
| 53 | - /** | |
| 54 | - * Convert model JSON output into readable assistant text. | |
| 55 | - * | |
| 56 | - * @param string $content Raw model content. | |
| 57 | - * | |
| 58 | - * @return string | |
| 59 | - */ | |
| 60 | - public function normalize( string $content ): string { | |
| 61 | - $content = trim( $content ); | |
| 62 | - if ( '' === $content ) { | |
| 63 | - return ''; | |
| 64 | - } | |
| 65 | - | |
| 66 | - $decoded = $this->decode_json( $content ); | |
| 67 | - if ( empty( $decoded ) ) { | |
| 68 | - if ( $this->looks_like_metadata_value( $content ) ) { | |
| 69 | - return __( 'I could not generate a complete answer. Please try rephrasing your request.', 'learnpress' ); | |
| 70 | - } | |
| 71 | - | |
| 72 | - return $content; | |
| 73 | - } | |
| 74 | - | |
| 75 | - $extracted = $this->extract_text_from_array( $decoded ); | |
| 76 | - | |
| 77 | - if ( '' !== $extracted ) { | |
| 78 | - return $extracted; | |
| 79 | - } | |
| 80 | - | |
| 81 | - return $this->looks_like_metadata_value( $content ) | |
| 82 | - ? __( 'I could not generate a complete answer. Please try rephrasing your request.', 'learnpress' ) | |
| 83 | - : $content; | |
| 84 | - } | |
| 85 | - | |
| 86 | - /** | |
| 87 | - * Build normalized response from assistant content. | |
| 88 | - * | |
| 89 | - * @param string $content The raw assistant text content. | |
| 90 | - * | |
| 91 | - * @return array{type: string, message: string, quiz: array|null} | |
| 92 | - */ | |
| 93 | - public function build_response( string $content ): array { | |
| 94 | - return array( | |
| 95 | - 'type' => 'text', | |
| 96 | - 'message' => $this->normalize( $content ), | |
| 97 | - 'quiz' => null, | |
| 98 | - ); | |
| 99 | - } | |
| 100 | - | |
| 101 | - /** | |
| 102 | - * Extract first meaningful text value from a decoded JSON object. | |
| 103 | - * | |
| 104 | - * @param array $data Decoded JSON. | |
| 105 | - * | |
| 106 | - * @return string | |
| 107 | - */ | |
| 108 | - private function extract_text_from_array( array $data ): string { | |
| 109 | - $preferred_keys = array( | |
| 110 | - 'message', | |
| 111 | - 'answer', | |
| 112 | - 'content', | |
| 113 | - 'summary', | |
| 114 | - 'explanation', | |
| 115 | - 'text', | |
| 116 | - 'response', | |
| 117 | - 'result', | |
| 118 | - 'key_points', | |
| 119 | - 'main_points', | |
| 120 | - 'bullet_points', | |
| 121 | - ); | |
| 122 | - | |
| 123 | - foreach ( $preferred_keys as $key ) { | |
| 124 | - if ( empty( $data[ $key ] ) ) { | |
| 125 | - continue; | |
| 126 | - } | |
| 127 | - | |
| 128 | - $value = $data[ $key ]; | |
| 129 | - if ( is_string( $value ) && '' !== trim( $value ) ) { | |
| 130 | - return trim( $value ); | |
| 131 | - } | |
| 132 | - | |
| 133 | - if ( is_array( $value ) ) { | |
| 134 | - $nested = $this->extract_text_from_array( $value ); | |
| 135 | - if ( '' !== $nested ) { | |
| 136 | - return $nested; | |
| 137 | - } | |
| 138 | - } | |
| 139 | - } | |
| 140 | - | |
| 141 | - foreach ( $data as $key => $value ) { | |
| 142 | - if ( $this->is_metadata_text_key( (string) $key ) ) { | |
| 143 | - continue; | |
| 144 | - } | |
| 145 | - | |
| 146 | - if ( is_array( $value ) ) { | |
| 147 | - $nested = $this->extract_text_from_array( $value ); | |
| 148 | - if ( '' !== $nested ) { | |
| 149 | - return $nested; | |
| 150 | - } | |
| 151 | - } | |
| 152 | - } | |
| 153 | - | |
| 154 | - foreach ( $data as $key => $value ) { | |
| 155 | - if ( $this->is_metadata_text_key( (string) $key ) || ! is_string( $value ) ) { | |
| 156 | - continue; | |
| 157 | - } | |
| 158 | - | |
| 159 | - $candidate = trim( $value ); | |
| 160 | - if ( '' === $candidate || $this->looks_like_metadata_value( $candidate ) ) { | |
| 161 | - continue; | |
| 162 | - } | |
| 163 | - | |
| 164 | - if ( $this->is_substantial_text( $candidate ) ) { | |
| 165 | - return $candidate; | |
| 166 | - } | |
| 167 | - } | |
| 168 | - | |
| 169 | - return ''; | |
| 170 | - } | |
| 171 | - | |
| 172 | - /** | |
| 173 | - * Determine whether a decoded JSON key is metadata and should not be shown as answer text. | |
| 174 | - * | |
| 175 | - * @param string $key JSON key. | |
| 176 | - * | |
| 177 | - * @return bool | |
| 178 | - */ | |
| 179 | - private function is_metadata_text_key( string $key ): bool { | |
| 180 | - | |
| 181 | - $normalized = strtolower( trim( $key ) ); | |
| 182 | - if ( $normalized === '' ) { | |
| 183 | - return false; | |
| 184 | - } | |
| 185 | - | |
| 186 | - return in_array( | |
| 187 | - $normalized, | |
| 188 | - array( | |
| 189 | - 'language', | |
| 190 | - 'locale', | |
| 191 | - 'intent', | |
| 192 | - 'action', | |
| 193 | - 'type', | |
| 194 | - 'status', | |
| 195 | - 'model', | |
| 196 | - 'provider', | |
| 197 | - 'metadata', | |
| 198 | - ), | |
| 199 | - true | |
| 200 | - ); | |
| 201 | - } | |
| 202 | - | |
| 203 | - /** | |
| 204 | - * Check whether a text candidate looks like metadata output. | |
| 205 | - * | |
| 206 | - * @param string $value Text candidate. | |
| 207 | - * | |
| 208 | - * @return bool | |
| 209 | - */ | |
| 210 | - private function looks_like_metadata_value( string $value ): bool { | |
| 211 | - | |
| 212 | - $trimmed = trim( $value ); | |
| 213 | - if ( $trimmed === '' ) { | |
| 214 | - return true; | |
| 215 | - } | |
| 216 | - | |
| 217 | - if ( preg_match( '/^(language|locale|intent|action|type|status)\s*[:=]\s*[\w\-]+$/ui', $trimmed ) ) { | |
| 218 | - return true; | |
| 219 | - } | |
| 220 | - | |
| 221 | - return preg_match( '/^[a-z]{2}[-_][a-z]{2}$/i', $trimmed ) === 1; | |
| 222 | - } | |
| 223 | - | |
| 224 | - /** | |
| 225 | - * Determine whether candidate text is substantial enough for user-facing response. | |
| 226 | - * | |
| 227 | - * @param string $value Text candidate. | |
| 228 | - * | |
| 229 | - * @return bool | |
| 230 | - */ | |
| 231 | - private function is_substantial_text( string $value ): bool { | |
| 232 | - | |
| 233 | - $trimmed = trim( $value ); | |
| 234 | - if ( $trimmed === '' ) { | |
| 235 | - return false; | |
| 236 | - } | |
| 237 | - | |
| 238 | - if ( preg_match( '/[.!?。！？]/u', $trimmed ) ) { | |
| 239 | - return true; | |
| 240 | - } | |
| 241 | - | |
| 242 | - $length = mb_strlen( $trimmed, 'UTF-8' ); | |
| 243 | - if ( $length >= 20 ) { | |
| 244 | - return true; | |
| 245 | - } | |
| 246 | - | |
| 247 | - return $length >= 14 && preg_match( '/\s/u', $trimmed ) === 1; | |
| 248 | - } | |
| 249 | -} | |
| 1 | +<?php | |
| 2 | + | |
| 3 | +namespace LearnPress\AI\Assistant; | |
| 4 | + | |
| 5 | +/** | |
| 6 | + * ResponseNormalizer — decodes JSON model output and extracts human-readable text. | |
| 7 | + * | |
| 8 | + * Handles the common pattern where OpenAI returns JSON-wrapped text such as | |
| 9 | + * {"message":"..."} and extracts the best text candidate for frontend rendering. | |
| 10 | + * | |
| 11 | + * @package LearnPress\AI\Assistant | |
| 12 | + * @since 4.3.5 | |
| 13 | + */ | |
| 14 | +class ResponseNormalizer { | |
| 15 | + | |
| 16 | + /** | |
| 17 | + * Decode JSON content safely without throwing exceptions. | |
| 18 | + * | |
| 19 | + * Accepts either pure JSON or content containing a JSON object substring. | |
| 20 | + * Returns empty array when content is plain text or invalid JSON. | |
| 21 | + * | |
| 22 | + * @param string $content Raw model content. | |
| 23 | + * | |
| 24 | + * @return array | |
| 25 | + */ | |
| 26 | + public function decode_json( string $content ): array { | |
| 27 | + $content = trim( $content ); | |
| 28 | + if ( '' === $content ) { | |
| 29 | + return array(); | |
| 30 | + } | |
| 31 | + | |
| 32 | + $decoded = json_decode( $content, true ); | |
| 33 | + if ( is_array( $decoded ) && JSON_ERROR_NONE === json_last_error() ) { | |
| 34 | + return $decoded; | |
| 35 | + } | |
| 36 | + | |
| 37 | + $first_brace = strpos( $content, '{' ); | |
| 38 | + $last_brace = strrpos( $content, '}' ); | |
| 39 | + if ( false === $first_brace || false === $last_brace || $last_brace <= $first_brace ) { | |
| 40 | + return array(); | |
| 41 | + } | |
| 42 | + | |
| 43 | + $json_slice = substr( $content, $first_brace, $last_brace - $first_brace + 1 ); | |
| 44 | + if ( ! is_string( $json_slice ) || '' === $json_slice ) { | |
| 45 | + return array(); | |
| 46 | + } | |
| 47 | + | |
| 48 | + $decoded = json_decode( $json_slice, true ); | |
| 49 | + | |
| 50 | + return ( is_array( $decoded ) && JSON_ERROR_NONE === json_last_error() ) ? $decoded : array(); | |
| 51 | + } | |
| 52 | + | |
| 53 | + /** | |
| 54 | + * Convert model JSON output into readable assistant text. | |
| 55 | + * | |
| 56 | + * @param string $content Raw model content. | |
| 57 | + * | |
| 58 | + * @return string | |
| 59 | + */ | |
| 60 | + public function normalize( string $content ): string { | |
| 61 | + $content = trim( $content ); | |
| 62 | + if ( '' === $content ) { | |
| 63 | + return ''; | |
| 64 | + } | |
| 65 | + | |
| 66 | + $decoded = $this->decode_json( $content ); | |
| 67 | + if ( empty( $decoded ) ) { | |
| 68 | + if ( $this->looks_like_metadata_value( $content ) ) { | |
| 69 | + return __( 'I could not generate a complete answer. Please try rephrasing your request.', 'learnpress' ); | |
| 70 | + } | |
| 71 | + | |
| 72 | + return $content; | |
| 73 | + } | |
| 74 | + | |
| 75 | + $extracted = $this->extract_text_from_array( $decoded ); | |
| 76 | + | |
| 77 | + if ( '' !== $extracted ) { | |
| 78 | + return $extracted; | |
| 79 | + } | |
| 80 | + | |
| 81 | + return $this->looks_like_metadata_value( $content ) | |
| 82 | + ? __( 'I could not generate a complete answer. Please try rephrasing your request.', 'learnpress' ) | |
| 83 | + : $content; | |
| 84 | + } | |
| 85 | + | |
| 86 | + /** | |
| 87 | + * Build normalized response from assistant content. | |
| 88 | + * | |
| 89 | + * @param string $content The raw assistant text content. | |
| 90 | + * | |
| 91 | + * @return array{type: string, message: string, quiz: array|null} | |
| 92 | + */ | |
| 93 | + public function build_response( string $content ): array { | |
| 94 | + return array( | |
| 95 | + 'type' => 'text', | |
| 96 | + 'message' => $this->normalize( $content ), | |
| 97 | + 'quiz' => null, | |
| 98 | + ); | |
| 99 | + } | |
| 100 | + | |
| 101 | + /** | |
| 102 | + * Extract first meaningful text value from a decoded JSON object. | |
| 103 | + * | |
| 104 | + * @param array $data Decoded JSON. | |
| 105 | + * | |
| 106 | + * @return string | |
| 107 | + */ | |
| 108 | + private function extract_text_from_array( array $data ): string { | |
| 109 | + $preferred_keys = array( | |
| 110 | + 'message', | |
| 111 | + 'answer', | |
| 112 | + 'content', | |
| 113 | + 'summary', | |
| 114 | + 'explanation', | |
| 115 | + 'text', | |
| 116 | + 'response', | |
| 117 | + 'result', | |
| 118 | + 'key_points', | |
| 119 | + 'main_points', | |
| 120 | + 'bullet_points', | |
| 121 | + ); | |
| 122 | + | |
| 123 | + foreach ( $preferred_keys as $key ) { | |
| 124 | + if ( empty( $data[ $key ] ) ) { | |
| 125 | + continue; | |
| 126 | + } | |
| 127 | + | |
| 128 | + $value = $data[ $key ]; | |
| 129 | + if ( is_string( $value ) && '' !== trim( $value ) ) { | |
| 130 | + return trim( $value ); | |
| 131 | + } | |
| 132 | + | |
| 133 | + if ( is_array( $value ) ) { | |
| 134 | + $nested = $this->extract_text_from_array( $value ); | |
| 135 | + if ( '' !== $nested ) { | |
| 136 | + return $nested; | |
| 137 | + } | |
| 138 | + } | |
| 139 | + } | |
| 140 | + | |
| 141 | + foreach ( $data as $key => $value ) { | |
| 142 | + if ( $this->is_metadata_text_key( (string) $key ) ) { | |
| 143 | + continue; | |
| 144 | + } | |
| 145 | + | |
| 146 | + if ( is_array( $value ) ) { | |
| 147 | + $nested = $this->extract_text_from_array( $value ); | |
| 148 | + if ( '' !== $nested ) { | |
| 149 | + return $nested; | |
| 150 | + } | |
| 151 | + } | |
| 152 | + } | |
| 153 | + | |
| 154 | + foreach ( $data as $key => $value ) { | |
| 155 | + if ( $this->is_metadata_text_key( (string) $key ) || ! is_string( $value ) ) { | |
| 156 | + continue; | |
| 157 | + } | |
| 158 | + | |
| 159 | + $candidate = trim( $value ); | |
| 160 | + if ( '' === $candidate || $this->looks_like_metadata_value( $candidate ) ) { | |
| 161 | + continue; | |
| 162 | + } | |
| 163 | + | |
| 164 | + if ( $this->is_substantial_text( $candidate ) ) { | |
| 165 | + return $candidate; | |
| 166 | + } | |
| 167 | + } | |
| 168 | + | |
| 169 | + return ''; | |
| 170 | + } | |
| 171 | + | |
| 172 | + /** | |
| 173 | + * Determine whether a decoded JSON key is metadata and should not be shown as answer text. | |
| 174 | + * | |
| 175 | + * @param string $key JSON key. | |
| 176 | + * | |
| 177 | + * @return bool | |
| 178 | + */ | |
| 179 | + private function is_metadata_text_key( string $key ): bool { | |
| 180 | + | |
| 181 | + $normalized = strtolower( trim( $key ) ); | |
| 182 | + if ( $normalized === '' ) { | |
| 183 | + return false; | |
| 184 | + } | |
| 185 | + | |
| 186 | + return in_array( | |
| 187 | + $normalized, | |
| 188 | + array( | |
| 189 | + 'language', | |
| 190 | + 'locale', | |
| 191 | + 'intent', | |
| 192 | + 'action', | |
| 193 | + 'type', | |
| 194 | + 'status', | |
| 195 | + 'model', | |
| 196 | + 'provider', | |
| 197 | + 'metadata', | |
| 198 | + ), | |
| 199 | + true | |
| 200 | + ); | |
| 201 | + } | |
| 202 | + | |
| 203 | + /** | |
| 204 | + * Check whether a text candidate looks like metadata output. | |
| 205 | + * | |
| 206 | + * @param string $value Text candidate. | |
| 207 | + * | |
| 208 | + * @return bool | |
| 209 | + */ | |
| 210 | + private function looks_like_metadata_value( string $value ): bool { | |
| 211 | + | |
| 212 | + $trimmed = trim( $value ); | |
| 213 | + if ( $trimmed === '' ) { | |
| 214 | + return true; | |
| 215 | + } | |
| 216 | + | |
| 217 | + if ( preg_match( '/^(language|locale|intent|action|type|status)\s*[:=]\s*[\w\-]+$/ui', $trimmed ) ) { | |
| 218 | + return true; | |
| 219 | + } | |
| 220 | + | |
| 221 | + return preg_match( '/^[a-z]{2}[-_][a-z]{2}$/i', $trimmed ) === 1; | |
| 222 | + } | |
| 223 | + | |
| 224 | + /** | |
| 225 | + * Determine whether candidate text is substantial enough for user-facing response. | |
| 226 | + * | |
| 227 | + * @param string $value Text candidate. | |
| 228 | + * | |
| 229 | + * @return bool | |
| 230 | + */ | |
| 231 | + private function is_substantial_text( string $value ): bool { | |
| 232 | + | |
| 233 | + $trimmed = trim( $value ); | |
| 234 | + if ( $trimmed === '' ) { | |
| 235 | + return false; | |
| 236 | + } | |
| 237 | + | |
| 238 | + if ( preg_match( '/[.!?。！？]/u', $trimmed ) ) { | |
| 239 | + return true; | |
| 240 | + } | |
| 241 | + | |
| 242 | + $length = mb_strlen( $trimmed, 'UTF-8' ); | |
| 243 | + if ( $length >= 20 ) { | |
| 244 | + return true; | |
| 245 | + } | |
| 246 | + | |
| 247 | + return $length >= 14 && preg_match( '/\s/u', $trimmed ) === 1; | |
| 248 | + } | |
| 249 | +} | |
