| 1 |
<?php |
| 2 |
|
| 3 |
/** |
| 4 |
* This file is part of the PacketeryNette Framework (https://nette.org) |
| 5 |
* Copyright (c) 2004 David Grudl (https://davidgrudl.com) |
| 6 |
*/ |
| 7 |
|
| 8 |
declare(strict_types=1); |
| 9 |
|
| 10 |
namespace PacketeryNette\Neon; |
| 11 |
|
| 12 |
|
| 13 |
/** |
| 14 |
* Parser for PacketeryNette Object Notation. |
| 15 |
* @internal |
| 16 |
*/ |
| 17 |
final class Decoder |
| 18 |
{ |
| 19 |
public const PATTERNS = [ |
| 20 |
// strings |
| 21 |
' |
| 22 |
\'\'\'\n (?:(?: [^\n] | \n(?![\t\ ]*+\'\'\') )*+ \n)?[\t\ ]*+\'\'\' | |
| 23 |
"""\n (?:(?: [^\n] | \n(?![\t\ ]*+""") )*+ \n)?[\t\ ]*+""" | |
| 24 |
\' (?: \'\' | [^\'\n] )*+ \' | |
| 25 |
" (?: \\\\. | [^"\\\\\n] )*+ " |
| 26 |
', |
| 27 |
|
| 28 |
// literal / boolean / integer / float |
| 29 |
' |
| 30 |
(?: [^#"\',:=[\]{}()\n\t\ `-] | (?<!["\']) [:-] [^"\',=[\]{}()\n\t\ ] ) |
| 31 |
(?: |
| 32 |
[^,:=\]})(\n\t\ ]++ | |
| 33 |
:(?! [\n\t\ ,\]})] | $ ) | |
| 34 |
[\ \t]++ [^#,:=\]})(\n\t\ ] |
| 35 |
)*+ |
| 36 |
', |
| 37 |
|
| 38 |
// punctuation |
| 39 |
'[,:=[\]{}()-]', |
| 40 |
|
| 41 |
// comment |
| 42 |
'?:\#.*+', |
| 43 |
|
| 44 |
// new line + indent |
| 45 |
'\n[\t\ ]*+', |
| 46 |
|
| 47 |
// whitespace |
| 48 |
'?:[\t\ ]++', |
| 49 |
]; |
| 50 |
|
| 51 |
private const PATTERN_DATETIME = '#\d\d\d\d-\d\d?-\d\d?(?:(?:[Tt]| ++)\d\d?:\d\d:\d\d(?:\.\d*+)? *+(?:Z|[-+]\d\d?(?::?\d\d)?)?)?$#DA'; |
| 52 |
|
| 53 |
private const PATTERN_HEX = '#0x[0-9a-fA-F]++$#DA'; |
| 54 |
|
| 55 |
private const PATTERN_OCTAL = '#0o[0-7]++$#DA'; |
| 56 |
|
| 57 |
private const PATTERN_BINARY = '#0b[0-1]++$#DA'; |
| 58 |
|
| 59 |
private const SIMPLE_TYPES = [ |
| 60 |
'true' => 'TRUE', 'True' => 'TRUE', 'TRUE' => 'TRUE', 'yes' => 'TRUE', 'Yes' => 'TRUE', 'YES' => 'TRUE', 'on' => 'TRUE', 'On' => 'TRUE', 'ON' => 'TRUE', |
| 61 |
'false' => 'FALSE', 'False' => 'FALSE', 'FALSE' => 'FALSE', 'no' => 'FALSE', 'No' => 'FALSE', 'NO' => 'FALSE', 'off' => 'FALSE', 'Off' => 'FALSE', 'OFF' => 'FALSE', |
| 62 |
'null' => 'NULL', 'Null' => 'NULL', 'NULL' => 'NULL', |
| 63 |
]; |
| 64 |
|
| 65 |
private const DEPRECATED_TYPES = ['on' => 1, 'On' => 1, 'ON' => 1, 'off' => 1, 'Off' => 1, 'OFF' => 1]; |
| 66 |
|
| 67 |
private const ESCAPE_SEQUENCES = [ |
| 68 |
't' => "\t", 'n' => "\n", 'r' => "\r", 'f' => "\x0C", 'b' => "\x08", '"' => '"', '\\' => '\\', '/' => '/', '_' => "\u{A0}", |
| 69 |
]; |
| 70 |
|
| 71 |
private const BRACKETS = [ |
| 72 |
'[' => ']', |
| 73 |
'{' => '}', |
| 74 |
'(' => ')', |
| 75 |
]; |
| 76 |
|
| 77 |
/** @var string */ |
| 78 |
private $input; |
| 79 |
|
| 80 |
/** @var array */ |
| 81 |
private $tokens; |
| 82 |
|
| 83 |
/** @var int */ |
| 84 |
private $pos; |
| 85 |
|
| 86 |
|
| 87 |
/** |
| 88 |
* Decodes a NEON string. |
| 89 |
* @return mixed |
| 90 |
*/ |
| 91 |
public function decode(string $input) |
| 92 |
{ |
| 93 |
if (substr($input, 0, 3) === "\u{FEFF}") { // BOM |
| 94 |
$input = substr($input, 3); |
| 95 |
} |
| 96 |
$this->input = "\n" . str_replace("\r", '', $input); // \n forces indent detection |
| 97 |
|
| 98 |
$pattern = '~(' . implode(')|(', self::PATTERNS) . ')~Amixu'; |
| 99 |
$this->tokens = preg_split($pattern, $this->input, -1, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE | PREG_SPLIT_DELIM_CAPTURE); |
| 100 |
if ($this->tokens === false) { |
| 101 |
throw new Exception('Invalid UTF-8 sequence.'); |
| 102 |
} |
| 103 |
|
| 104 |
$last = end($this->tokens); |
| 105 |
if ($this->tokens && !preg_match($pattern, $last[0])) { |
| 106 |
$this->pos = count($this->tokens) - 1; |
| 107 |
$this->error(); |
| 108 |
} |
| 109 |
|
| 110 |
$this->pos = 0; |
| 111 |
$res = $this->parse(null); |
| 112 |
|
| 113 |
while (isset($this->tokens[$this->pos])) { |
| 114 |
if ($this->tokens[$this->pos][0][0] === "\n") { |
| 115 |
$this->pos++; |
| 116 |
} else { |
| 117 |
$this->error(); |
| 118 |
} |
| 119 |
} |
| 120 |
return $res; |
| 121 |
} |
| 122 |
|
| 123 |
|
| 124 |
/** |
| 125 |
* @param string|bool|null $indent indentation (for block-parser) |
| 126 |
* @return mixed |
| 127 |
*/ |
| 128 |
private function parse($indent, array $result = null, $key = null, bool $hasKey = false) |
| 129 |
{ |
| 130 |
$inlineParser = $indent === false; |
| 131 |
$value = null; |
| 132 |
$hasValue = false; |
| 133 |
$tokens = $this->tokens; |
| 134 |
$n = &$this->pos; |
| 135 |
$count = count($tokens); |
| 136 |
$mainResult = &$result; |
| 137 |
|
| 138 |
for (; $n < $count; $n++) { |
| 139 |
$t = $tokens[$n][0]; |
| 140 |
|
| 141 |
if ($t === ',') { // ArrayEntry separator |
| 142 |
if ((!$hasKey && !$hasValue) || !$inlineParser) { |
| 143 |
$this->error(); |
| 144 |
} |
| 145 |
$this->addValue($result, $hasKey ? $key : null, $hasValue ? $value : null); |
| 146 |
$hasKey = $hasValue = false; |
| 147 |
|
| 148 |
} elseif ($t === ':' || $t === '=') { // KeyValuePair separator |
| 149 |
if ($hasValue && (is_array($value) || is_object($value))) { |
| 150 |
$this->error('Unacceptable key'); |
| 151 |
|
| 152 |
} elseif ($hasKey && $key === null && $hasValue && !$inlineParser) { |
| 153 |
$n++; |
| 154 |
$result[] = $this->parse($indent . ' ', [], $value, true); |
| 155 |
$newIndent = isset($tokens[$n], $tokens[$n + 1]) // not last |
| 156 |
? (string) substr($tokens[$n][0], 1) |
| 157 |
: ''; |
| 158 |
if (strlen($newIndent) > strlen($indent)) { |
| 159 |
$n++; |
| 160 |
$this->error('Bad indentation'); |
| 161 |
} elseif (strlen($newIndent) < strlen($indent)) { |
| 162 |
return $mainResult; // block parser exit point |
| 163 |
} |
| 164 |
$hasKey = $hasValue = false; |
| 165 |
|
| 166 |
} elseif ($hasKey || !$hasValue) { |
| 167 |
$this->error(); |
| 168 |
|
| 169 |
} else { |
| 170 |
$key = (string) $value; |
| 171 |
$hasKey = true; |
| 172 |
$hasValue = false; |
| 173 |
$result = &$mainResult; |
| 174 |
} |
| 175 |
|
| 176 |
} elseif ($t === '-') { // BlockArray bullet |
| 177 |
if ($hasKey || $hasValue || $inlineParser) { |
| 178 |
$this->error(); |
| 179 |
} |
| 180 |
$key = null; |
| 181 |
$hasKey = true; |
| 182 |
|
| 183 |
} elseif (isset(self::BRACKETS[$t])) { // Opening bracket [ ( { |
| 184 |
if ($hasValue) { |
| 185 |
if ($t !== '(') { |
| 186 |
$this->error(); |
| 187 |
} |
| 188 |
$n++; |
| 189 |
if ($value instanceof Entity && $value->value === Neon::CHAIN) { |
| 190 |
end($value->attributes)->attributes = $this->parse(false, []); |
| 191 |
} else { |
| 192 |
$value = new Entity($value, $this->parse(false, [])); |
| 193 |
} |
| 194 |
} else { |
| 195 |
$n++; |
| 196 |
$value = $this->parse(false, []); |
| 197 |
} |
| 198 |
$hasValue = true; |
| 199 |
if ( // unexpected type of bracket or block-parser |
| 200 |
!isset($tokens[$n]) |
| 201 |
|| $tokens[$n][0] !== self::BRACKETS[$t] |
| 202 |
) { |
| 203 |
$this->error(); |
| 204 |
} |
| 205 |
|
| 206 |
} elseif ($t === ']' || $t === '}' || $t === ')') { // Closing bracket ] ) } |
| 207 |
if (!$inlineParser) { |
| 208 |
$this->error(); |
| 209 |
} |
| 210 |
break; |
| 211 |
|
| 212 |
} elseif ($t[0] === "\n") { // Indent |
| 213 |
if ($inlineParser) { |
| 214 |
if ($hasKey || $hasValue) { |
| 215 |
$this->addValue($result, $hasKey ? $key : null, $hasValue ? $value : null); |
| 216 |
$hasKey = $hasValue = false; |
| 217 |
} |
| 218 |
|
| 219 |
} else { |
| 220 |
while (isset($tokens[$n + 1]) && $tokens[$n + 1][0][0] === "\n") { |
| 221 |
$n++; // skip to last indent |
| 222 |
} |
| 223 |
if (!isset($tokens[$n + 1])) { |
| 224 |
break; |
| 225 |
} |
| 226 |
|
| 227 |
$newIndent = (string) substr($tokens[$n][0], 1); |
| 228 |
if ($indent === null) { // first iteration |
| 229 |
$indent = $newIndent; |
| 230 |
} |
| 231 |
$minlen = min(strlen($newIndent), strlen($indent)); |
| 232 |
if ($minlen && (string) substr($newIndent, 0, $minlen) !== (string) substr($indent, 0, $minlen)) { |
| 233 |
$n++; |
| 234 |
$this->error('Invalid combination of tabs and spaces'); |
| 235 |
} |
| 236 |
|
| 237 |
if (strlen($newIndent) > strlen($indent)) { // open new block-array or hash |
| 238 |
if ($hasValue || !$hasKey) { |
| 239 |
$n++; |
| 240 |
$this->error('Bad indentation'); |
| 241 |
} |
| 242 |
$this->addValue($result, $key, $this->parse($newIndent)); |
| 243 |
$newIndent = isset($tokens[$n], $tokens[$n + 1]) // not last |
| 244 |
? (string) substr($tokens[$n][0], 1) |
| 245 |
: ''; |
| 246 |
if (strlen($newIndent) > strlen($indent)) { |
| 247 |
$n++; |
| 248 |
$this->error('Bad indentation'); |
| 249 |
} |
| 250 |
$hasKey = false; |
| 251 |
|
| 252 |
} else { |
| 253 |
if ($hasValue && !$hasKey) { // block items must have "key"; null key means list item |
| 254 |
break; |
| 255 |
|
| 256 |
} elseif ($hasKey) { |
| 257 |
$this->addValue($result, $key, $hasValue ? $value : null); |
| 258 |
if ( |
| 259 |
$key !== null |
| 260 |
&& !$hasValue |
| 261 |
&& $newIndent === $indent |
| 262 |
&& isset($tokens[$n + 1]) |
| 263 |
&& $tokens[$n + 1][0] === '-' |
| 264 |
) { |
| 265 |
$result = &$result[$key]; |
| 266 |
} |
| 267 |
$hasKey = $hasValue = false; |
| 268 |
} |
| 269 |
} |
| 270 |
|
| 271 |
if (strlen($newIndent) < strlen($indent)) { // close block |
| 272 |
return $mainResult; // block parser exit point |
| 273 |
} |
| 274 |
} |
| 275 |
|
| 276 |
} else { // Value |
| 277 |
$isKey = ($tmp = $tokens[$n + 1][0] ?? null) && ($tmp === ':' || $tmp === '='); |
| 278 |
|
| 279 |
if ($t[0] === '"' || $t[0] === "'") { |
| 280 |
if (preg_match('#^...\n++([\t ]*+)#', $t, $m)) { |
| 281 |
$converted = substr($t, 3, -3); |
| 282 |
$converted = str_replace("\n" . $m[1], "\n", $converted); |
| 283 |
$converted = preg_replace('#^\n|\n[\t ]*+$#D', '', $converted); |
| 284 |
} else { |
| 285 |
$converted = substr($t, 1, -1); |
| 286 |
if ($t[0] === "'") { |
| 287 |
$converted = str_replace("''", "'", $converted); |
| 288 |
} |
| 289 |
} |
| 290 |
if ($t[0] === '"') { |
| 291 |
$converted = preg_replace_callback('#\\\\(?:ud[89ab][0-9a-f]{2}\\\\ud[c-f][0-9a-f]{2}|u[0-9a-f]{4}|x[0-9a-f]{2}|.)#i', [$this, 'cbString'], $converted); |
| 292 |
} |
| 293 |
} elseif (!$isKey && isset(self::SIMPLE_TYPES[$t])) { |
| 294 |
$converted = constant(self::SIMPLE_TYPES[$t]); |
| 295 |
if (isset(self::DEPRECATED_TYPES[$t])) { |
| 296 |
trigger_error("Neon: keyword '$t' is deprecated, use true/yes or false/no.", E_USER_DEPRECATED); |
| 297 |
} |
| 298 |
} elseif (is_numeric($t)) { |
| 299 |
$converted = $t * 1; |
| 300 |
} elseif (preg_match(self::PATTERN_HEX, $t)) { |
| 301 |
$converted = hexdec($t); |
| 302 |
} elseif (preg_match(self::PATTERN_OCTAL, $t)) { |
| 303 |
$converted = octdec($t); |
| 304 |
} elseif (preg_match(self::PATTERN_BINARY, $t)) { |
| 305 |
$converted = bindec($t); |
| 306 |
} elseif (!$isKey && preg_match(self::PATTERN_DATETIME, $t)) { |
| 307 |
$converted = new \DateTimeImmutable($t); |
| 308 |
} else { // literal |
| 309 |
$converted = $t; |
| 310 |
} |
| 311 |
if ($hasValue) { |
| 312 |
if ($value instanceof Entity) { // Entity chaining |
| 313 |
if ($value->value !== Neon::CHAIN) { |
| 314 |
$value = new Entity(Neon::CHAIN, [$value]); |
| 315 |
} |
| 316 |
$value->attributes[] = new Entity($converted); |
| 317 |
} else { |
| 318 |
$this->error(); |
| 319 |
} |
| 320 |
} else { |
| 321 |
$value = $converted; |
| 322 |
$hasValue = true; |
| 323 |
} |
| 324 |
} |
| 325 |
} |
| 326 |
|
| 327 |
if ($inlineParser) { |
| 328 |
if ($hasKey || $hasValue) { |
| 329 |
$this->addValue($result, $hasKey ? $key : null, $hasValue ? $value : null); |
| 330 |
} |
| 331 |
} else { |
| 332 |
if ($hasValue && !$hasKey) { // block items must have "key" |
| 333 |
if ($result === null) { |
| 334 |
$result = $value; // simple value parser |
| 335 |
} else { |
| 336 |
$this->error(); |
| 337 |
} |
| 338 |
} elseif ($hasKey) { |
| 339 |
$this->addValue($result, $key, $hasValue ? $value : null); |
| 340 |
} |
| 341 |
} |
| 342 |
return $mainResult; |
| 343 |
} |
| 344 |
|
| 345 |
|
| 346 |
private function addValue(&$result, $key, $value) |
| 347 |
{ |
| 348 |
if ($key === null) { |
| 349 |
$result[] = $value; |
| 350 |
} elseif ($result && array_key_exists($key, $result)) { |
| 351 |
$this->error("Duplicated key '$key'"); |
| 352 |
} else { |
| 353 |
$result[$key] = $value; |
| 354 |
} |
| 355 |
} |
| 356 |
|
| 357 |
|
| 358 |
private function cbString(array $m): string |
| 359 |
{ |
| 360 |
$sq = $m[0]; |
| 361 |
if (isset(self::ESCAPE_SEQUENCES[$sq[1]])) { |
| 362 |
return self::ESCAPE_SEQUENCES[$sq[1]]; |
| 363 |
} elseif ($sq[1] === 'u' && strlen($sq) >= 6) { |
| 364 |
$lead = hexdec(substr($sq, 2, 4)); |
| 365 |
$tail = hexdec(substr($sq, 8, 4)); |
| 366 |
$code = $tail ? (0x2400 + (($lead - 0xD800) << 10) + $tail) : $lead; |
| 367 |
if ($code >= 0xD800 && $code <= 0xDFFF) { |
| 368 |
$this->error("Invalid UTF-8 (lone surrogate) $sq"); |
| 369 |
} |
| 370 |
return function_exists('iconv') |
| 371 |
? iconv('UTF-32BE', 'UTF-8//IGNORE', pack('N', $code)) |
| 372 |
: mb_convert_encoding(pack('N', $code), 'UTF-8', 'UTF-32BE'); |
| 373 |
|
| 374 |
} elseif ($sq[1] === 'x' && strlen($sq) === 4) { |
| 375 |
trigger_error("Neon: '$sq' is deprecated, use '\\uXXXX' instead.", E_USER_DEPRECATED); |
| 376 |
return chr(hexdec(substr($sq, 2))); |
| 377 |
|
| 378 |
} else { |
| 379 |
$this->error("Invalid escaping sequence $sq"); |
| 380 |
} |
| 381 |
} |
| 382 |
|
| 383 |
|
| 384 |
private function error(string $message = "Unexpected '%s'") |
| 385 |
{ |
| 386 |
$last = $this->tokens[$this->pos] ?? null; |
| 387 |
$offset = $last ? $last[1] : strlen($this->input); |
| 388 |
$text = substr($this->input, 0, $offset); |
| 389 |
$line = substr_count($text, "\n"); |
| 390 |
$col = $offset - strrpos("\n" . $text, "\n") + 1; |
| 391 |
$token = $last |
| 392 |
? str_replace("\n", '<new line>', substr($last[0], 0, 40)) |
| 393 |
: 'end'; |
| 394 |
throw new Exception(str_replace('%s', $token, $message) . " on line $line, column $col."); |
| 395 |
} |
| 396 |
} |
| 397 |
|
