Raphaël G. Git Repositories - packbundle/blob - Util/SluggerUtil.php

   1 <?php declare(strict_types=1);
   2
   3 /*
   4  * This file is part of the Rapsys PackBundle package.
   5  *
   6  * (c) Raphaël Gertz <symfony@rapsys.eu>
   7  *
   8  * For the full copyright and license information, please view the LICENSE
   9  * file that was distributed with this source code.
  10  */
  11
  12 namespace Rapsys\PackBundle\Util;
  13
  14 /**
  15  * Manages string conversions
  16  */
  17 class SluggerUtil {
  18         /**
  19          * The alpha array
  20          */
  21         protected array $alpha;
  22
  23         /**
  24          * The rev array
  25          */
  26         protected array $rev;
  27
  28         /**
  29          * The alpha array key number
  30          */
  31         protected int $count;
  32
  33         /**
  34          * The offset reduced from secret
  35          */
  36         protected int $offset;
  37
  38         /**
  39          * Construct slugger util
  40          *
  41          * TODO: use a recipe to generate in .env.local an env variable RAPSYSPACK_SECRET="ayl[...]z9w"
  42          *
  43          * @todo Add a command to generate alpha array or generate it on first run with cache storage ?
  44          * @todo Use Cache like in calendar controller through FilesystemAdapter
  45          *
  46          * @param string $secret The secret string
  47          */
  48         public function __construct(protected string $secret) {
  49                 /**
  50                  * Pseudo-random alphabet
  51                  * @xxx use array flip and keys to workaround php "smart" that cast range('0', '9') as int instead of string
  52                  * @xxx The key count mismatch, count(alpha)>count(rev), resulted in a data corruption due to duplicate numeric values
  53                  * @todosee required range by json_encode result and short input (0->255 ???)
  54                  */
  55                 $this->alpha = array_keys(array_flip(array_merge(
  56                         range('^', '[', -1),
  57                         range('V', 'Z'),
  58                         range('9', '7', -1),
  59                         range('L', 'O'),
  60                         range('f', 'a', -1),
  61                         range('_', '`'),
  62                         range('3', '0', -1),
  63                         range('E', 'H'),
  64                         range('v', 'r', -1),
  65                         range('+', '/'),
  66                         range('K', 'I', -1),
  67                         range('g', 'j'),
  68                         range('=', ':', -1),
  69                         range('>', '@'),
  70                         range('m', 'k', -1),
  71                         range('4', '6'),
  72                         range('*', '%', -1),
  73                         range('n', 'q'),
  74                         range('U', 'P', -1),
  75                         range(' ', '$'),
  76                         range('D', 'A', -1),
  77                         range('w', 'z'),
  78                         range('~', '!', -1)
  79                 )));
  80
  81                 //Init rev array
  82                 $this->count = count($rev = $this->rev = array_flip($this->alpha));
  83
  84                 //Init split
  85                 $split = str_split($this->secret);
  86
  87                 //Set offset
  88                 $this->offset = array_reduce($split, function ($res, $a) use ($rev) { return $res += $rev[$a]; }, count($split)) % $this->count;
  89         }
  90
  91         /**
  92          * Flatten recursively an array
  93          *
  94          * @param array|string $data The data tree
  95          * @param string|null $current The current prefix
  96          * @param string $sep The key separator
  97          * @param string $prefix The key prefix
  98          * @param string $suffix The key suffix
  99          * @return array The flattened data
 100          */
 101         public function flatten($data, ?string $current = null, string $sep = '.', string $prefix = '', string $suffix = ''): array {
 102                 //Init result
 103                 $ret = [];
 104
 105                 //Look for data array
 106                 if (is_array($data)) {
 107                         //Iteare on each pair
 108                         foreach($data as $k => $v) {
 109                                 //Merge flattened value in return array
 110                                 $ret += $this->flatten($v, empty($current) ? $k : $current.$sep.$k, $sep, $prefix, $suffix);
 111                         }
 112                 //Look flat data
 113                 } else {
 114                         //Store data in flattened key
 115                         $ret[$prefix.$current.$suffix] = $data;
 116                 }
 117
 118                 //Return result
 119                 return $ret;
 120         }
 121
 122         /**
 123          * Crypt and base64uri encode string
 124          *
 125          * @param array|string $data The data string
 126          * @return string The hashed data
 127          */
 128         public function hash(array|string $data): string {
 129                 //With array
 130                 if (is_array($data)) {
 131                         //Json encode array
 132                         $data = json_encode($data);
 133                 }
 134
 135                 //Return hashed data
 136                 //XXX: we use hash_hmac with md5 hash
 137                 //XXX: crypt was dropped because it provided identical signature for string starting with same pattern
 138                 return str_replace(['+','/'], ['-','_'], base64_encode(hash_hmac('md5', $data, $this->secret, true)));
 139         }
 140
 141         /**
 142          * Serialize then short
 143          *
 144          * @param array $data The data array
 145          * @return string The serialized and shorted data
 146          */
 147         public function serialize(array $data): string {
 148                 //Return shorted serialized data
 149                 //XXX: dropped serialize use to prevent short function from dropping utf-8 characters
 150                 return $this->short(json_encode($data));
 151         }
 152
 153         /**
 154          * Short
 155          *
 156          * @param string $data The data string
 157          * @return string The shorted data
 158          */
 159         public function short(string $data): string {
 160                 //Return string
 161                 $ret = '';
 162
 163                 //With data
 164                 if (!empty($data)) {
 165                         //Iterate on each character
 166                         foreach(str_split($data) as $k => $c) {
 167                                 if (isset($this->rev[$c]) && isset($this->alpha[($this->rev[$c]+$this->offset)%$this->count])) {
 168                                         //XXX: Remap char to an other one
 169                                         $ret .= chr(($this->rev[$c] - $this->offset + $this->count) % $this->count);
 170                                 } else {
 171                                         throw new \RuntimeException(sprintf('Unable to retrieve character: %c', $c));
 172                                 }
 173                         }
 174                 }
 175
 176                 //Send result
 177                 return str_replace(['+','/'], ['-','_'], base64_encode($ret));
 178         }
 179
 180         /**
 181          * Convert string to safe slug
 182          *
 183          * @param string $data The data string
 184          * @return ?string The slugged data
 185          */
 186         function slug(?string $data): ?string {
 187                 //With null
 188                 if ($data === null) {
 189                         //Return null
 190                         return $data;
 191                 }
 192
 193                 //Use Transliterator if available
 194                 if (class_exists('Transliterator')) {
 195                         //Convert from any to latin, then to ascii and lowercase
 196                         $trans = \Transliterator::create('Any-Latin; Latin-ASCII; Lower()');
 197                         //Replace every non alphanumeric character by dash then trim dash
 198                         return trim(preg_replace('/[^a-zA-Z0-9]+/', '-', $trans->transliterate($data)), '-');
 199                 }
 200
 201                 //Convert from utf-8 to ascii, replace quotes with space, remove non alphanumericseparator, replace separator with dash and trim dash
 202                 return trim(preg_replace('/[\/_|+ -]+/', '-', strtolower(preg_replace('/[^a-zA-Z0-9\/_|+ -]/', '', str_replace(['\'', '"'], ' ', iconv('UTF-8', 'ASCII//TRANSLIT', $data))))), '-');
 203         }
 204
 205         /**
 206          * Convert string to latin
 207          *
 208          * @param string $data The data string
 209          * @return ?string The slugged data
 210          */
 211         function latin(?string $data): ?string {
 212                 //With null
 213                 if ($data === null) {
 214                         //Return null
 215                         return $data;
 216                 }
 217
 218                 //Use Transliterator if available
 219                 if (class_exists('Transliterator')) {
 220                         //Convert from any to latin, then to ascii and lowercase
 221                         $trans = \Transliterator::create('Any-Latin; Latin-ASCII');
 222                         //Replace every non alphanumeric character by dash then trim dash
 223                         return trim($trans->transliterate($data));
 224                 }
 225
 226                 //Convert from utf-8 to ascii
 227                 return trim(iconv('UTF-8', 'ASCII//TRANSLIT', $data));
 228         }
 229
 230         /**
 231          * Unshort then unserialize
 232          *
 233          * @param string $data The data string
 234          * @return array The unshorted and unserialized data
 235          */
 236         public function unserialize(string $data): array {
 237                 //Return unshorted unserialized string
 238                 return json_decode($this->unshort($data), true);
 239         }
 240
 241         /**
 242          * Unshort
 243          *
 244          * @param string $data The data string
 245          * @return string The unshorted data
 246          */
 247         public function unshort(string $data): string {
 248                 //Return string
 249                 $ret = '';
 250
 251                 //Iterate on each character
 252                 foreach(str_split(base64_decode(str_replace(['-','_'], ['+','/'], $data))) as $c) {
 253                         //XXX: Reverse map char to an other one
 254                         $ret .= $this->alpha[(ord($c) + $this->offset) % $this->count];
 255                 }
 256
 257                 //Send result
 258                 return $ret;
 259         }
 260 }