| 1 |
<?php |
| 2 |
|
| 3 |
if ( ! defined( 'ABSPATH' ) ) { |
| 4 |
exit; |
| 5 |
} |
| 6 |
|
| 7 |
class M_Chart_Parse { |
| 8 |
const LABELS_NONE = 'none'; |
| 9 |
const LABELS_FIRST_ROW = 'first_row'; |
| 10 |
const LABELS_FIRST_COLUMN = 'first_column'; |
| 11 |
const LABELS_BOTH = 'both'; |
| 12 |
const PARSE_ROWS = 'rows'; |
| 13 |
const PARSE_COLUMNS = 'columns'; |
| 14 |
|
| 15 |
public array $data = []; |
| 16 |
public array $value_labels = []; |
| 17 |
public string $value_labels_position = ''; |
| 18 |
public array $set_data = []; |
| 19 |
public array $raw_data = []; |
| 20 |
public string $parse_in = ''; |
| 21 |
|
| 22 |
private ?NumberFormatter $formatter = null; |
| 23 |
|
| 24 |
/** |
| 25 |
* Parses a chart's data array for labels, data sets, and raw values |
| 26 |
* |
| 27 |
* @param array $data the raw two-dimensional data array from the chart post |
| 28 |
* @param string $parse_in whether to parse by rows or columns (PARSE_ROWS or PARSE_COLUMNS) |
| 29 |
* |
| 30 |
* @return static $this |
| 31 |
*/ |
| 32 |
public function parse_data( array $data, string $parse_in ): static { |
| 33 |
$this->data = $data; |
| 34 |
$this->parse_in = $parse_in; |
| 35 |
$this->value_labels_position = $this->get_value_labels_position(); |
| 36 |
$this->parse_value_labels(); |
| 37 |
$this->parse_set_data(); |
| 38 |
|
| 39 |
return $this; |
| 40 |
} |
| 41 |
|
| 42 |
/** |
| 43 |
* Populates $this->value_labels by reading label values from the data array based on the detected labels position |
| 44 |
*/ |
| 45 |
private function parse_value_labels(): void { |
| 46 |
$this->value_labels = []; |
| 47 |
|
| 48 |
switch ( $this->value_labels_position ) { |
| 49 |
case self::LABELS_NONE: |
| 50 |
return; |
| 51 |
|
| 52 |
case self::LABELS_FIRST_COLUMN: |
| 53 |
foreach ( (array) $this->data as $columns ) { |
| 54 |
if ( '' != trim( (string) $columns[0] ) ) { |
| 55 |
$this->value_labels[] = $this->clean_labels( $columns[0] ); |
| 56 |
} |
| 57 |
} |
| 58 |
break; |
| 59 |
|
| 60 |
case self::LABELS_FIRST_ROW: |
| 61 |
foreach ( (array) $this->data[0] as $column ) { |
| 62 |
if ( '' != trim( (string) $column ) ) { |
| 63 |
$this->value_labels[] = $this->clean_labels( $column ); |
| 64 |
} |
| 65 |
} |
| 66 |
break; |
| 67 |
|
| 68 |
case self::LABELS_BOTH: |
| 69 |
foreach ( (array) $this->data as $columns ) { |
| 70 |
if ( '' != trim( (string) $columns[0] ) ) { |
| 71 |
$this->value_labels[ self::LABELS_FIRST_COLUMN ][] = $this->clean_labels( $columns[0] ); |
| 72 |
} |
| 73 |
} |
| 74 |
|
| 75 |
foreach ( (array) $this->data[0] as $column ) { |
| 76 |
if ( '' != trim( (string) $column ) ) { |
| 77 |
$this->value_labels[ self::LABELS_FIRST_ROW ][] = $this->clean_labels( $column ); |
| 78 |
} |
| 79 |
} |
| 80 |
break; |
| 81 |
} |
| 82 |
|
| 83 |
$this->value_labels = apply_filters( 'm_chart_value_labels', $this->value_labels, $this->value_labels_position, $this->data ); |
| 84 |
} |
| 85 |
|
| 86 |
/** |
| 87 |
* Helper function returns a string describing where the value labels are (first_row, first_column, both) |
| 88 |
* |
| 89 |
* @return string the position of the labels in the given data set |
| 90 |
*/ |
| 91 |
private function get_value_labels_position(): string { |
| 92 |
if ( ! isset( $this->data[0][0] ) && ! isset( $this->data[1][0] ) ) { |
| 93 |
return self::LABELS_NONE; |
| 94 |
} |
| 95 |
|
| 96 |
if ( '' == $this->data[0][0] ) { |
| 97 |
return self::LABELS_BOTH; |
| 98 |
} |
| 99 |
|
| 100 |
// Structural pre-check |
| 101 |
// A sheet with 2 effective columns (rows mode) or 2 effective rows (columns mode) is the simple single-series shape and is unambiguous |
| 102 |
// Works regardless of whether the labels look numeric (years, ordinals, etc) |
| 103 |
if ( self::PARSE_ROWS === $this->parse_in && 2 === $this->effective_max_columns() ) { |
| 104 |
return self::LABELS_FIRST_COLUMN; |
| 105 |
} |
| 106 |
|
| 107 |
if ( self::PARSE_COLUMNS === $this->parse_in && 2 === $this->effective_row_count() ) { |
| 108 |
return self::LABELS_FIRST_ROW; |
| 109 |
} |
| 110 |
|
| 111 |
// Existing content-based heuristic for 3+ effective columns/rows |
| 112 |
if ( ! is_numeric( trim( (string) $this->data[0][0] ) ) ) { |
| 113 |
// If the first row has multiple non-numeric headers and the data rows start with numeric values the entire first row is column labels (e.g. scatter format) |
| 114 |
if ( |
| 115 |
isset( $this->data[0][1] ) && ! is_numeric( trim( (string) $this->data[0][1] ) ) |
| 116 |
&& isset( $this->data[1][0] ) && is_numeric( trim( (string) $this->data[1][0] ) ) |
| 117 |
) { |
| 118 |
return self::LABELS_FIRST_ROW; |
| 119 |
} |
| 120 |
|
| 121 |
return self::LABELS_FIRST_COLUMN; |
| 122 |
} |
| 123 |
|
| 124 |
return self::LABELS_FIRST_ROW; |
| 125 |
} |
| 126 |
|
| 127 |
/** |
| 128 |
* Max effective column count across rows |
| 129 |
* Rightmost non-empty cell index + 1, taken as a max over all rows |
| 130 |
* Trailing empty cells (typical of Jspreadsheet's minDimensions padding) don't count |
| 131 |
* |
| 132 |
* @return int |
| 133 |
*/ |
| 134 |
private function effective_max_columns(): int { |
| 135 |
if ( ! is_array( $this->data ) ) { |
| 136 |
return 0; |
| 137 |
} |
| 138 |
|
| 139 |
$max = 0; |
| 140 |
|
| 141 |
foreach ( $this->data as $row ) { |
| 142 |
if ( ! is_array( $row ) ) { |
| 143 |
continue; |
| 144 |
} |
| 145 |
|
| 146 |
for ( $i = count( $row ) - 1; $i >= 0; $i-- ) { |
| 147 |
if ( '' !== trim( (string) ( $row[ $i ] ?? '' ) ) ) { |
| 148 |
if ( $i + 1 > $max ) { |
| 149 |
$max = $i + 1; |
| 150 |
} |
| 151 |
|
| 152 |
break; |
| 153 |
} |
| 154 |
} |
| 155 |
} |
| 156 |
|
| 157 |
return $max; |
| 158 |
} |
| 159 |
|
| 160 |
/** |
| 161 |
* Count of rows that contain at least one non-empty cell |
| 162 |
* Trailing empty rows (typical of Jspreadsheet's minDimensions padding) don't count |
| 163 |
* |
| 164 |
* @return int |
| 165 |
*/ |
| 166 |
private function effective_row_count(): int { |
| 167 |
if ( ! is_array( $this->data ) ) { |
| 168 |
return 0; |
| 169 |
} |
| 170 |
|
| 171 |
$last = -1; |
| 172 |
|
| 173 |
foreach ( $this->data as $i => $row ) { |
| 174 |
if ( ! is_array( $row ) ) { |
| 175 |
continue; |
| 176 |
} |
| 177 |
|
| 178 |
foreach ( $row as $cell ) { |
| 179 |
if ( '' !== trim( (string) ( $cell ?? '' ) ) ) { |
| 180 |
$last = $i; |
| 181 |
|
| 182 |
break; |
| 183 |
} |
| 184 |
} |
| 185 |
} |
| 186 |
|
| 187 |
return $last + 1; |
| 188 |
} |
| 189 |
|
| 190 |
/** |
| 191 |
* Helper function cleans data point values |
| 192 |
* |
| 193 |
* @param mixed $data_point a data point that may need to be cleaned or typed as an int |
| 194 |
* |
| 195 |
* @return float|string a float of the cleaned data point or string if the cleaned value was not numeric |
| 196 |
*/ |
| 197 |
public function clean_data_point( mixed $data_point ): float|string { |
| 198 |
$data_point = trim( (string) $data_point ); |
| 199 |
|
| 200 |
if ( preg_match( '/-?\d[\d,]*(?:\.\d+)?/', $data_point, $matches ) ) { |
| 201 |
return floatval( str_replace( ',', '', $matches[0] ) ); |
| 202 |
} |
| 203 |
|
| 204 |
return $data_point; |
| 205 |
} |
| 206 |
|
| 207 |
/** |
| 208 |
* Helper function parses a data point into an M_Chart_Parsed_Data_Point value object for localized display |
| 209 |
* Splits the cell string into prefix, numeric value, and suffix |
| 210 |
* This means the number can be reformatted for any locale while preserving surrounding context |
| 211 |
* |
| 212 |
* @param mixed $data_point a raw cell value |
| 213 |
* |
| 214 |
* @return M_Chart_Parsed_Data_Point |
| 215 |
*/ |
| 216 |
public function parse_data_point( mixed $data_point ): M_Chart_Parsed_Data_Point { |
| 217 |
$data_point = trim( (string) $data_point ); |
| 218 |
|
| 219 |
if ( preg_match( '/(-?\d[\d,]*(?:\.\d+)?)/', $data_point, $matches, PREG_OFFSET_CAPTURE ) ) { |
| 220 |
$number = $matches[1][0]; |
| 221 |
$offset = $matches[1][1]; |
| 222 |
|
| 223 |
return M_Chart_Parsed_Data_Point::numeric( |
| 224 |
floatval( str_replace( ',', '', $number ) ), |
| 225 |
substr( $data_point, 0, $offset ), |
| 226 |
substr( $data_point, $offset + strlen( $number ) ) |
| 227 |
); |
| 228 |
} |
| 229 |
|
| 230 |
return M_Chart_Parsed_Data_Point::text( $data_point ); |
| 231 |
} |
| 232 |
|
| 233 |
/** |
| 234 |
* Helper function cleans out label values |
| 235 |
* |
| 236 |
* @param mixed $label a label string |
| 237 |
* |
| 238 |
* @return string the label string cleaned of any problem content |
| 239 |
*/ |
| 240 |
public function clean_labels( mixed $label ): string { |
| 241 |
$label = trim( html_entity_decode( (string) $label, ENT_QUOTES ) ); |
| 242 |
|
| 243 |
// PHP's strip_tags() is case-insensitive, recursive, handles unclosed/malformed tags, and is XSS-safe |
| 244 |
return strip_tags( $label ); |
| 245 |
} |
| 246 |
|
| 247 |
/** |
| 248 |
* Populates $this->set_data and $this->raw_data by delegating to the appropriate collector based on the parse direction and labels position, then normalizing both arrays |
| 249 |
*/ |
| 250 |
private function parse_set_data(): void { |
| 251 |
if ( self::PARSE_ROWS == $this->parse_in && self::LABELS_FIRST_COLUMN == $this->value_labels_position ) { |
| 252 |
[ $set_data_array, $raw_data_array ] = $this->collect_rows_first_column(); |
| 253 |
} elseif ( self::PARSE_ROWS == $this->parse_in && self::LABELS_BOTH == $this->value_labels_position ) { |
| 254 |
[ $set_data_array, $raw_data_array ] = $this->collect_rows_both(); |
| 255 |
} elseif ( self::PARSE_COLUMNS == $this->parse_in && self::LABELS_BOTH == $this->value_labels_position ) { |
| 256 |
[ $set_data_array, $raw_data_array ] = $this->collect_columns_both(); |
| 257 |
} else { |
| 258 |
[ $set_data_array, $raw_data_array ] = $this->collect_default(); |
| 259 |
} |
| 260 |
|
| 261 |
$set_data_array = $this->normalize_data_array( $set_data_array ); |
| 262 |
$raw_data_array = $this->normalize_data_array( $raw_data_array ); |
| 263 |
|
| 264 |
$this->set_data = apply_filters( 'm_chart_set_data', $set_data_array, $this->data, $this->parse_in ); |
| 265 |
$this->raw_data = apply_filters( 'm_chart_raw_data', $raw_data_array, $this->data, $this->parse_in ); |
| 266 |
} |
| 267 |
|
| 268 |
/** |
| 269 |
* Collects data when parsing rows with labels in the first column |
| 270 |
* |
| 271 |
* @return array {0: array, 1: array} Two-element array of [set_data, raw_data] |
| 272 |
*/ |
| 273 |
private function collect_rows_first_column(): array { |
| 274 |
$set_data_array = []; |
| 275 |
$raw_data_array = []; |
| 276 |
|
| 277 |
foreach ( $this->data as $row ) { |
| 278 |
if ( ! is_array( $row ) ) { |
| 279 |
continue; |
| 280 |
} |
| 281 |
|
| 282 |
foreach ( $row as $key => $column ) { |
| 283 |
if ( '' == $column || 0 == $key ) { |
| 284 |
continue; |
| 285 |
} |
| 286 |
|
| 287 |
$set_data_array[] = $this->clean_data_point( $column ); |
| 288 |
$raw_data_array[] = $this->parse_data_point( $column ); |
| 289 |
} |
| 290 |
} |
| 291 |
|
| 292 |
return [ $set_data_array, $raw_data_array ]; |
| 293 |
} |
| 294 |
|
| 295 |
/** |
| 296 |
* Collects data when parsing rows with labels in both the first row and first column |
| 297 |
* |
| 298 |
* @return array {0: array, 1: array} Two-element array of [set_data, raw_data] |
| 299 |
*/ |
| 300 |
private function collect_rows_both(): array { |
| 301 |
$set_data_array = []; |
| 302 |
$raw_data_array = []; |
| 303 |
$limit = count( $this->data ); |
| 304 |
$this_sets = []; |
| 305 |
$this_raw = []; |
| 306 |
|
| 307 |
for ( $i = 1; $i < $limit; $i++ ) { |
| 308 |
if ( ! is_array( $this->data[ $i ] ) ) { |
| 309 |
continue; |
| 310 |
} |
| 311 |
|
| 312 |
foreach ( $this->data[ $i ] as $c_key => $column ) { |
| 313 |
if ( 0 != $c_key ) { |
| 314 |
$data_point = $this->clean_data_point( $column ); |
| 315 |
$key = $i - 1; |
| 316 |
|
| 317 |
if ( ! isset( $this_sets[ $key ]['is_null'] ) ) { |
| 318 |
$this_sets[ $key ]['is_null'] = true; |
| 319 |
} |
| 320 |
|
| 321 |
if ( is_numeric( $data_point ) ) { |
| 322 |
$this_sets[ $key ]['is_null'] = false; |
| 323 |
} |
| 324 |
|
| 325 |
$this_sets[ $key ]['data'][] = $data_point; |
| 326 |
$this_raw[ $key ][] = $this->parse_data_point( $column ); |
| 327 |
} |
| 328 |
} |
| 329 |
} |
| 330 |
|
| 331 |
foreach ( $this_sets as $key => $set ) { |
| 332 |
if ( false == $set['is_null'] ) { |
| 333 |
$set_data_array[ $key ] = $set['data']; |
| 334 |
$raw_data_array[ $key ] = $this_raw[ $key ]; |
| 335 |
} |
| 336 |
} |
| 337 |
|
| 338 |
return [ $set_data_array, $raw_data_array ]; |
| 339 |
} |
| 340 |
|
| 341 |
/** |
| 342 |
* Collects data when parsing columns with labels in both the first row and first column |
| 343 |
* |
| 344 |
* @return array {0: array, 1: array} Two-element array of [set_data, raw_data] |
| 345 |
*/ |
| 346 |
private function collect_columns_both(): array { |
| 347 |
$set_data_array = []; |
| 348 |
$raw_data_array = []; |
| 349 |
$limit = count( $this->data ); |
| 350 |
$this_sets = []; |
| 351 |
$this_raw = []; |
| 352 |
|
| 353 |
for ( $i = 1; $i < $limit; $i++ ) { |
| 354 |
if ( ! is_array( $this->data[ $i ] ) ) { |
| 355 |
continue; |
| 356 |
} |
| 357 |
|
| 358 |
foreach ( $this->data[ $i ] as $key => $column ) { |
| 359 |
if ( 0 == $key ) { |
| 360 |
continue; |
| 361 |
} |
| 362 |
|
| 363 |
$data_point = $this->clean_data_point( $column ); |
| 364 |
$a_key = $key - 1; |
| 365 |
|
| 366 |
if ( ! isset( $this_sets[ $a_key ]['is_null'] ) ) { |
| 367 |
$this_sets[ $a_key ]['is_null'] = true; |
| 368 |
} |
| 369 |
|
| 370 |
if ( is_numeric( $data_point ) ) { |
| 371 |
$this_sets[ $a_key ]['is_null'] = false; |
| 372 |
} |
| 373 |
|
| 374 |
$this_sets[ $a_key ]['data'][] = $data_point; |
| 375 |
$this_raw[ $a_key ][] = $this->parse_data_point( $column ); |
| 376 |
} |
| 377 |
} |
| 378 |
|
| 379 |
foreach ( $this_sets as $key => $set ) { |
| 380 |
if ( false == $set['is_null'] ) { |
| 381 |
$set_data_array[ $key ] = $set['data']; |
| 382 |
$raw_data_array[ $key ] = $this_raw[ $key ]; |
| 383 |
} |
| 384 |
} |
| 385 |
|
| 386 |
return [ $set_data_array, $raw_data_array ]; |
| 387 |
} |
| 388 |
|
| 389 |
/** |
| 390 |
* Collects data for the default case (first-row labels only, or no labels) |
| 391 |
* |
| 392 |
* @return array {0: array, 1: array} Two-element array of [set_data, raw_data] |
| 393 |
*/ |
| 394 |
private function collect_default(): array { |
| 395 |
$set_data_array = []; |
| 396 |
$raw_data_array = []; |
| 397 |
|
| 398 |
if ( ! isset( $this->data[1] ) ) { |
| 399 |
return [ $set_data_array, $raw_data_array ]; |
| 400 |
} |
| 401 |
|
| 402 |
foreach ( $this->data as $key => $columns ) { |
| 403 |
if ( ! is_array( $columns ) ) { |
| 404 |
continue; |
| 405 |
} |
| 406 |
|
| 407 |
foreach ( $columns as $column ) { |
| 408 |
if ( '' == $column || 0 == $key ) { |
| 409 |
continue; |
| 410 |
} |
| 411 |
|
| 412 |
$set_data_array[] = $this->clean_data_point( $column ); |
| 413 |
$raw_data_array[] = $this->parse_data_point( $column ); |
| 414 |
} |
| 415 |
} |
| 416 |
|
| 417 |
return [ $set_data_array, $raw_data_array ]; |
| 418 |
} |
| 419 |
|
| 420 |
/** |
| 421 |
* Helper function normalizes the data array so that the number of data values matches the number of value labels |
| 422 |
* |
| 423 |
* @param array $data_array an already parsed array of data |
| 424 |
* |
| 425 |
* @return array a normalized array of parsed data |
| 426 |
*/ |
| 427 |
private function normalize_data_array( array $data_array ): array { |
| 428 |
if ( self::PARSE_ROWS == $this->parse_in && self::LABELS_BOTH == $this->value_labels_position ) { |
| 429 |
$first_row_labels = $this->value_labels[ self::LABELS_FIRST_ROW ] ?? []; |
| 430 |
$label_count = is_array( $first_row_labels ) ? count( $first_row_labels ) - 1 : 0; |
| 431 |
|
| 432 |
foreach ( $data_array as $key => $data ) { |
| 433 |
foreach ( $data as $t_key => $value ) { |
| 434 |
if ( $t_key > $label_count ) { |
| 435 |
unset( $data_array[ $key ][ $t_key ] ); |
| 436 |
} |
| 437 |
} |
| 438 |
} |
| 439 |
} elseif ( self::PARSE_COLUMNS == $this->parse_in && self::LABELS_BOTH == $this->value_labels_position ) { |
| 440 |
$first_col_labels = $this->value_labels[ self::LABELS_FIRST_COLUMN ] ?? []; |
| 441 |
$label_count = is_array( $first_col_labels ) ? count( $first_col_labels ) - 1 : 0; |
| 442 |
|
| 443 |
foreach ( $data_array as $key => $data ) { |
| 444 |
foreach ( $data as $t_key => $value ) { |
| 445 |
if ( $t_key > $label_count ) { |
| 446 |
unset( $data_array[ $key ][ $t_key ] ); |
| 447 |
} |
| 448 |
} |
| 449 |
} |
| 450 |
} |
| 451 |
|
| 452 |
return $data_array; |
| 453 |
} |
| 454 |
|
| 455 |
/** |
| 456 |
* Formats an M_Chart_Parsed_Data_Point for table display |
| 457 |
* Numeric cells are formatted with the locale-aware NumberFormatter |
| 458 |
* Non-numeric cells are returned as plain text |
| 459 |
* |
| 460 |
* @param ?M_Chart_Parsed_Data_Point $raw the parsed data point to format, or null for an empty cell |
| 461 |
* |
| 462 |
* @return string the formatted cell value |
| 463 |
*/ |
| 464 |
public function format_raw( ?M_Chart_Parsed_Data_Point $raw ): string { |
| 465 |
if ( null === $raw ) { |
| 466 |
return ''; |
| 467 |
} |
| 468 |
|
| 469 |
// If the value is a number return the formatted and prefixed/suffixed version of it |
| 470 |
if ( $raw->is_numeric() ) { |
| 471 |
$formatter = $this->get_formatter(); |
| 472 |
$number = $formatter ? $formatter->format( $raw->value ) : (string) $raw->value; |
| 473 |
|
| 474 |
return $raw->prefix . $number . $raw->suffix; |
| 475 |
} |
| 476 |
|
| 477 |
return $raw->text; |
| 478 |
} |
| 479 |
|
| 480 |
/** |
| 481 |
* Returns a locale-aware NumberFormatter, creating and caching it on first use |
| 482 |
* |
| 483 |
* @return ?NumberFormatter a NumberFormatter instance, or null if the intl extension is unavailable |
| 484 |
*/ |
| 485 |
private function get_formatter(): ?NumberFormatter { |
| 486 |
if ( null === $this->formatter ) { |
| 487 |
$locale = m_chart()->get_settings( 'locale' ); |
| 488 |
$this->formatter = class_exists( 'NumberFormatter' ) ? new NumberFormatter( $locale, NumberFormatter::DECIMAL ) : null; |
| 489 |
} |
| 490 |
|
| 491 |
return $this->formatter; |
| 492 |
} |
| 493 |
} |
| 494 |
|
