1: <?php
  2: 
  3: /**
  4:  * This file is part of the Nette Framework (https://nette.org)
  5:  * Copyright (c) 2004 David Grudl (http://davidgrudl.com)
  6:  */
  7: 
  8: namespace Nette\Utils;
  9: 
 10: use Nette;
 11: 
 12: 
 13: /**
 14:  * Simple parser & generator for Nette Object Notation.
 15:  *
 16:  * @author     David Grudl
 17:  */
 18: class Neon extends Nette\Object
 19: {
 20:     const BLOCK = 1;
 21: 
 22:     /** @var array */
 23:     private static $patterns = array(
 24:         '
 25:             \'[^\'\n]*\' |
 26:             "(?: \\\\. | [^"\\\\\n] )*"
 27:         ', // string
 28:         '
 29:             (?: [^#"\',:=[\]{}()\x00-\x20!`-] | [:-][^"\',\]})\s] )
 30:             (?:
 31:                 [^,:=\]})(\x00-\x20]+ |
 32:                 :(?! [\s,\]})] | $ ) |
 33:                 [\ \t]+ [^#,:=\]})(\x00-\x20]
 34:             )*
 35:         ', // literal / boolean / integer / float
 36:         '
 37:             [,:=[\]{}()-]
 38:         ', // symbol
 39:         '?:\#.*', // comment
 40:         '\n[\t\ ]*', // new line + indent
 41:         '?:[\t\ ]+', // whitespace
 42:     );
 43: 
 44:     /** @var Tokenizer */
 45:     private static $tokenizer;
 46: 
 47:     private static $brackets = array(
 48:         '[' => ']',
 49:         '{' => '}',
 50:         '(' => ')',
 51:     );
 52: 
 53:     /** @var int */
 54:     private $n = 0;
 55: 
 56:     /** @var bool */
 57:     private $indentTabs;
 58: 
 59: 
 60:     /**
 61:      * Returns the NEON representation of a value.
 62:      * @param  mixed
 63:      * @param  int
 64:      * @return string
 65:      */
 66:     public static function encode($var, $options = NULL)
 67:     {
 68:         if ($var instanceof \DateTime) {
 69:             return $var->format('Y-m-d H:i:s O');
 70: 
 71:         } elseif ($var instanceof NeonEntity) {
 72:             return self::encode($var->value) . '(' . substr(self::encode($var->attributes), 1, -1) . ')';
 73:         }
 74: 
 75:         if (is_object($var)) {
 76:             $obj = $var; $var = array();
 77:             foreach ($obj as $k => $v) {
 78:                 $var[$k] = $v;
 79:             }
 80:         }
 81: 
 82:         if (is_array($var)) {
 83:             $isList = Validators::isList($var);
 84:             $s = '';
 85:             if ($options & self::BLOCK) {
 86:                 if (count($var) === 0) {
 87:                     return "[]";
 88:                 }
 89:                 foreach ($var as $k => $v) {
 90:                     $v = self::encode($v, self::BLOCK);
 91:                     $s .= ($isList ? '-' : self::encode($k) . ':')
 92:                         . (Strings::contains($v, "\n") ? "\n\t" . str_replace("\n", "\n\t", $v) : ' ' . $v)
 93:                         . "\n";
 94:                     continue;
 95:                 }
 96:                 return $s;
 97: 
 98:             } else {
 99:                 foreach ($var as $k => $v) {
100:                     $s .= ($isList ? '' : self::encode($k) . ': ') . self::encode($v) . ', ';
101:                 }
102:                 return ($isList ? '[' : '{') . substr($s, 0, -2) . ($isList ? ']' : '}');
103:             }
104: 
105:         } elseif (is_string($var) && !is_numeric($var)
106:             && !preg_match('~[\x00-\x1F]|^\d{4}|^(true|false|yes|no|on|off|null)\z~i', $var)
107:             && preg_match('~^' . self::$patterns[1] . '\z~x', $var) // 1 = literals
108:         ) {
109:             return $var;
110: 
111:         } elseif (is_float($var)) {
112:             $var = json_encode($var);
113:             return Strings::contains($var, '.') ? $var : $var . '.0';
114: 
115:         } else {
116:             return json_encode($var);
117:         }
118:     }
119: 
120: 
121:     /**
122:      * Decodes a NEON string.
123:      * @param  string
124:      * @return mixed
125:      */
126:     public static function decode($input)
127:     {
128:         if (!is_string($input)) {
129:             throw new Nette\InvalidArgumentException("Argument must be a string, " . gettype($input) . " given.");
130:         }
131:         if (!self::$tokenizer) {
132:             self::$tokenizer = new Tokenizer(self::$patterns, 'mix');
133:         }
134: 
135:         if (substr($input, 0, 3) === "\xEF\xBB\xBF") { // BOM
136:             $input = substr($input, 3);
137:         }
138:         $input = str_replace("\r", '', $input);
139:         self::$tokenizer->tokenize($input);
140: 
141:         $parser = new static;
142:         $res = $parser->parse(0);
143: 
144:         while (isset(self::$tokenizer->tokens[$parser->n])) {
145:             if (self::$tokenizer->tokens[$parser->n][0] === "\n") {
146:                 $parser->n++;
147:             } else {
148:                 $parser->error();
149:             }
150:         }
151:         return $res;
152:     }
153: 
154: 
155:     /**
156:      * @param  int  indentation (for block-parser)
157:      * @param  mixed
158:      * @return array
159:      */
160:     private function parse($indent = NULL, $result = NULL)
161:     {
162:         $inlineParser = $indent === NULL;
163:         $value = $key = NULL;
164:         $hasValue = $hasKey = FALSE;
165:         $tokens = self::$tokenizer->tokens;
166:         $n = & $this->n;
167:         $count = count($tokens);
168: 
169:         for (; $n < $count; $n++) {
170:             $t = $tokens[$n];
171: 
172:             if ($t === ',') { // ArrayEntry separator
173:                 if ((!$hasKey && !$hasValue) || !$inlineParser) {
174:                     $this->error();
175:                 }
176:                 $this->addValue($result, $hasKey, $key, $hasValue ? $value : NULL);
177:                 $hasKey = $hasValue = FALSE;
178: 
179:             } elseif ($t === ':' || $t === '=') { // KeyValuePair separator
180:                 if ($hasKey || !$hasValue) {
181:                     $this->error();
182:                 }
183:                 if (is_array($value) || is_object($value)) {
184:                     $this->error('Unacceptable key');
185:                 }
186:                 $key = (string) $value;
187:                 $hasKey = TRUE;
188:                 $hasValue = FALSE;
189: 
190:             } elseif ($t === '-') { // BlockArray bullet
191:                 if ($hasKey || $hasValue || $inlineParser) {
192:                     $this->error();
193:                 }
194:                 $key = NULL;
195:                 $hasKey = TRUE;
196: 
197:             } elseif (isset(self::$brackets[$t])) { // Opening bracket [ ( {
198:                 if ($hasValue) {
199:                     if ($t !== '(') {
200:                         $this->error();
201:                     }
202:                     $n++;
203:                     $entity = new NeonEntity;
204:                     $entity->value = $value;
205:                     $entity->attributes = $this->parse(NULL, array());
206:                     $value = $entity;
207:                 } else {
208:                     $n++;
209:                     $value = $this->parse(NULL, array());
210:                 }
211:                 $hasValue = TRUE;
212:                 if (!isset($tokens[$n]) || $tokens[$n] !== self::$brackets[$t]) { // unexpected type of bracket or block-parser
213:                     $this->error();
214:                 }
215: 
216:             } elseif ($t === ']' || $t === '}' || $t === ')') { // Closing bracket ] ) }
217:                 if (!$inlineParser) {
218:                     $this->error();
219:                 }
220:                 break;
221: 
222:             } elseif ($t[0] === "\n") { // Indent
223:                 if ($inlineParser) {
224:                     if ($hasKey || $hasValue) {
225:                         $this->addValue($result, $hasKey, $key, $hasValue ? $value : NULL);
226:                         $hasKey = $hasValue = FALSE;
227:                     }
228: 
229:                 } else {
230:                     while (isset($tokens[$n+1]) && $tokens[$n+1][0] === "\n") $n++; // skip to last indent
231:                     if (!isset($tokens[$n+1])) {
232:                         break;
233:                     }
234: 
235:                     $newIndent = strlen($tokens[$n]) - 1;
236:                     if ($indent === NULL) { // first iteration
237:                         $indent = $newIndent;
238:                     }
239:                     if ($newIndent) {
240:                         if ($this->indentTabs === NULL) {
241:                             $this->indentTabs = $tokens[$n][1] === "\t";
242:                         }
243:                         if (strpos($tokens[$n], $this->indentTabs ? ' ' : "\t")) {
244:                             $n++;
245:                             $this->error('Either tabs or spaces may be used as indenting chars, but not both.');
246:                         }
247:                     }
248: 
249:                     if ($newIndent > $indent) { // open new block-array or hash
250:                         if ($hasValue || !$hasKey) {
251:                             $n++;
252:                             $this->error('Unexpected indentation.');
253:                         } else {
254:                             $this->addValue($result, $key !== NULL, $key, $this->parse($newIndent));
255:                         }
256:                         $newIndent = isset($tokens[$n]) ? strlen($tokens[$n]) - 1 : 0;
257:                         $hasKey = FALSE;
258: 
259:                     } else {
260:                         if ($hasValue && !$hasKey) { // block items must have "key"; NULL key means list item
261:                             break;
262: 
263:                         } elseif ($hasKey) {
264:                             $this->addValue($result, $key !== NULL, $key, $hasValue ? $value : NULL);
265:                             $hasKey = $hasValue = FALSE;
266:                         }
267:                     }
268: 
269:                     if ($newIndent < $indent) { // close block
270:                         return $result; // block parser exit point
271:                     }
272:                 }
273: 
274:             } else { // Value
275:                 if ($hasValue) {
276:                     $this->error();
277:                 }
278:                 static $consts = array(
279:                     'true' => TRUE, 'True' => TRUE, 'TRUE' => TRUE, 'yes' => TRUE, 'Yes' => TRUE, 'YES' => TRUE, 'on' => TRUE, 'On' => TRUE, 'ON' => TRUE,
280:                     'false' => FALSE, 'False' => FALSE, 'FALSE' => FALSE, 'no' => FALSE, 'No' => FALSE, 'NO' => FALSE, 'off' => FALSE, 'Off' => FALSE, 'OFF' => FALSE,
281:                 );
282:                 if ($t[0] === '"') {
283:                     $value = preg_replace_callback('#\\\\(?:u[0-9a-f]{4}|x[0-9a-f]{2}|.)#i', array($this, 'cbString'), substr($t, 1, -1));
284:                 } elseif ($t[0] === "'") {
285:                     $value = substr($t, 1, -1);
286:                 } elseif (isset($consts[$t])) {
287:                     $value = $consts[$t];
288:                 } elseif ($t === 'null' || $t === 'Null' || $t === 'NULL') {
289:                     $value = NULL;
290:                 } elseif (is_numeric($t)) {
291:                     $value = $t * 1;
292:                 } elseif (preg_match('#\d\d\d\d-\d\d?-\d\d?(?:(?:[Tt]| +)\d\d?:\d\d:\d\d(?:\.\d*)? *(?:Z|[-+]\d\d?(?::\d\d)?)?)?\z#A', $t)) {
293:                     $value = new Nette\DateTime($t);
294:                 } else { // literal
295:                     $value = $t;
296:                 }
297:                 $hasValue = TRUE;
298:             }
299:         }
300: 
301:         if ($inlineParser) {
302:             if ($hasKey || $hasValue) {
303:                 $this->addValue($result, $hasKey, $key, $hasValue ? $value : NULL);
304:             }
305:         } else {
306:             if ($hasValue && !$hasKey) { // block items must have "key"
307:                 if ($result === NULL) {
308:                     $result = $value; // simple value parser
309:                 } else {
310:                     $this->error();
311:                 }
312:             } elseif ($hasKey) {
313:                 $this->addValue($result, $key !== NULL, $key, $hasValue ? $value : NULL);
314:             }
315:         }
316:         return $result;
317:     }
318: 
319: 
320:     private function addValue(& $result, $hasKey, $key, $value)
321:     {
322:         if ($hasKey) {
323:             if ($result && array_key_exists($key, $result)) {
324:                 $this->error("Duplicated key '$key'");
325:             }
326:             $result[$key] = $value;
327:         } else {
328:             $result[] = $value;
329:         }
330:     }
331: 
332: 
333:     private function cbString($m)
334:     {
335:         static $mapping = array('t' => "\t", 'n' => "\n", 'r' => "\r", 'f' => "\x0C", 'b' => "\x08", '"' => '"', '\\' => '\\',  '/' => '/', '_' => "\xc2\xa0");
336:         $sq = $m[0];
337:         if (isset($mapping[$sq[1]])) {
338:             return $mapping[$sq[1]];
339:         } elseif ($sq[1] === 'u' && strlen($sq) === 6) {
340:             return Strings::chr(hexdec(substr($sq, 2)));
341:         } elseif ($sq[1] === 'x' && strlen($sq) === 4) {
342:             return chr(hexdec(substr($sq, 2)));
343:         } else {
344:             $this->error("Invalid escaping sequence $sq");
345:         }
346:     }
347: 
348: 
349:     private function error($message = "Unexpected '%s'")
350:     {
351:         list(, $line, $col) = self::$tokenizer->getOffset($this->n);
352:         $token = isset(self::$tokenizer->tokens[$this->n])
353:             ? str_replace("\n", '<new line>', Strings::truncate(self::$tokenizer->tokens[$this->n], 40))
354:             : 'end';
355:         throw new NeonException(str_replace('%s', $token, $message) . " on line $line, column $col.");
356:     }
357: 
358: }
359: 
360: 
361: /**
362:  * Representation of 'foo(bar=1)' literal
363:  */
364: class NeonEntity extends \stdClass
365: {
366:     public $value;
367:     public $attributes;
368: }
369: 
370: 
371: /**
372:  * The exception that indicates error of NEON decoding.
373:  */
374: class NeonException extends \Exception
375: {
376: }
377: