| @@ -32,9 +32,9 @@ | ||
| 32 | 32 | protected $_padFinalGroup; |
| 33 | 33 | protected $_padCharacter; |
| 34 | 34 | protected $_caseSensitive; |
| 35 | 35 | protected $_charmap; |
| 36 | - | |
| 36 | + | |
| 37 | 37 | /** |
| 38 | 38 | * Constructor |
| 39 | 39 | * |
| 40 | 40 | * @param integer $bitsPerCharacter Bits to use for each encoded character |
| @@ -55,59 +55,59 @@ | ||
| 55 | 55 | // Ensure validity of $chars |
| 56 | 56 | if ( ! is_string( $chars ) || ( $charLength = strlen( $chars ) ) < 2 ) { |
| 57 | 57 | throw new InvalidArgumentException( '$chars must be a string of at least two characters' ); |
| 58 | 58 | } |
| 59 | - | |
| 59 | + | |
| 60 | 60 | // Ensure validity of $padCharacter |
| 61 | 61 | if ( $padFinalGroup ) { |
| 62 | 62 | if ( ! is_string( $padCharacter ) || ! isset( $padCharacter[0] ) ) { |
| 63 | 63 | throw new InvalidArgumentException( '$padCharacter must be a string of one character' ); |
| 64 | 64 | } |
| 65 | - | |
| 65 | + | |
| 66 | 66 | if ( $caseSensitive ) { |
| 67 | 67 | $padCharFound = strpos( $chars, $padCharacter[0] ); |
| 68 | 68 | } else { |
| 69 | 69 | $padCharFound = stripos( $chars, $padCharacter[0] ); |
| 70 | 70 | } |
| 71 | - | |
| 71 | + | |
| 72 | 72 | if ( $padCharFound !== false ) { |
| 73 | 73 | throw new InvalidArgumentException( '$padCharacter can not be a member of $chars' ); |
| 74 | 74 | } |
| 75 | 75 | } |
| 76 | - | |
| 76 | + | |
| 77 | 77 | // Ensure validity of $bitsPerCharacter |
| 78 | 78 | if ( ! is_int( $bitsPerCharacter ) ) { |
| 79 | 79 | throw new InvalidArgumentException( '$bitsPerCharacter must be an integer' ); |
| 80 | 80 | } |
| 81 | - | |
| 81 | + | |
| 82 | 82 | if ( $bitsPerCharacter < 1 ) { |
| 83 | 83 | // $bitsPerCharacter must be at least 1 |
| 84 | 84 | throw new InvalidArgumentException( '$bitsPerCharacter can not be less than 1' ); |
| 85 | - | |
| 85 | + | |
| 86 | 86 | } elseif ( $charLength < 1 << $bitsPerCharacter ) { |
| 87 | 87 | // Character length of $chars is too small for $bitsPerCharacter |
| 88 | 88 | // Find greatest acceptable value of $bitsPerCharacter |
| 89 | 89 | $bitsPerCharacter = 1; |
| 90 | 90 | $radix = 2; |
| 91 | - | |
| 91 | + | |
| 92 | 92 | while ( $charLength >= ( $radix <<= 1 ) && $bitsPerCharacter < 8 ) { |
| 93 | 93 | $bitsPerCharacter ++; |
| 94 | 94 | } |
| 95 | - | |
| 95 | + | |
| 96 | 96 | $radix >>= 1; |
| 97 | 97 | throw new InvalidArgumentException( |
| 98 | 98 | '$bitsPerCharacter can not be more than ' . $bitsPerCharacter |
| 99 | 99 | . ' given $chars length of ' . $charLength |
| 100 | 100 | . ' (max radix ' . $radix . ')' ); |
| 101 | - | |
| 101 | + | |
| 102 | 102 | } elseif ( $bitsPerCharacter > 8 ) { |
| 103 | 103 | // $bitsPerCharacter must not be greater than 8 |
| 104 | 104 | throw new InvalidArgumentException( '$bitsPerCharacter can not be greater than 8' ); |
| 105 | - | |
| 105 | + | |
| 106 | 106 | } else { |
| 107 | 107 | $radix = 1 << $bitsPerCharacter; |
| 108 | 108 | } |
| 109 | - | |
| 109 | + | |
| 110 | 110 | $this->_chars = $chars; |
| 111 | 111 | $this->_bitsPerCharacter = $bitsPerCharacter; |
| 112 | 112 | $this->_radix = $radix; |
| 113 | 113 | $this->_rightPadFinalBits = $rightPadFinalBits; |
| @@ -114,9 +114,9 @@ | ||
| 114 | 114 | $this->_padFinalGroup = $padFinalGroup; |
| 115 | 115 | $this->_padCharacter = $padCharacter[0]; |
| 116 | 116 | $this->_caseSensitive = $caseSensitive; |
| 117 | 117 | } |
| 118 | - | |
| 118 | + | |
| 119 | 119 | /** |
| 120 | 120 | * Encode a string |
| 121 | 121 | * |
| 122 | 122 | * @param string $rawString Binary data to encode |
| @@ -126,26 +126,26 @@ | ||
| 126 | 126 | public function encode( $rawString ) { |
| 127 | 127 | // Unpack string into an array of bytes |
| 128 | 128 | $bytes = unpack( 'C*', $rawString ); |
| 129 | 129 | $byteCount = count( $bytes ); |
| 130 | - | |
| 130 | + | |
| 131 | 131 | $encodedString = ''; |
| 132 | 132 | $byte = array_shift( $bytes ); |
| 133 | 133 | $bitsRead = 0; |
| 134 | 134 | $oldBits = 0; |
| 135 | - | |
| 135 | + | |
| 136 | 136 | $chars = $this->_chars; |
| 137 | 137 | $bitsPerCharacter = $this->_bitsPerCharacter; |
| 138 | 138 | $rightPadFinalBits = $this->_rightPadFinalBits; |
| 139 | 139 | $padFinalGroup = $this->_padFinalGroup; |
| 140 | 140 | $padCharacter = $this->_padCharacter; |
| 141 | - | |
| 141 | + | |
| 142 | 142 | $charsPerByte = 8 / $bitsPerCharacter; |
| 143 | 143 | $encodedLength = $byteCount * $charsPerByte; |
| 144 | - | |
| 144 | + | |
| 145 | 145 | // Generate encoded output; each loop produces one encoded character |
| 146 | 146 | for ( $c = 0; $c < $encodedLength; $c ++ ) { |
| 147 | - | |
| 147 | + | |
| 148 | 148 | // Get the bits needed for this encoded character |
| 149 | 149 | if ( $bitsRead + $bitsPerCharacter > 8 ) { |
| 150 | 150 | // Not enough bits remain in this byte for the current character |
| 151 | 151 | // Save the remaining bits before getting the next byte |
| @@ -151,9 +151,9 @@ | ||
| 151 | 151 | // Save the remaining bits before getting the next byte |
| 152 | 152 | $oldBitCount = 8 - $bitsRead; |
| 153 | 153 | $oldBits = $byte ^ ( $byte >> $oldBitCount << $oldBitCount ); |
| 154 | 154 | $newBitCount = $bitsPerCharacter - $oldBitCount; |
| 155 | - | |
| 155 | + | |
| 156 | 156 | if ( ! $bytes ) { |
| 157 | 157 | // Last bits; match final character and exit loop |
| 158 | 158 | if ( $rightPadFinalBits ) { |
| 159 | 159 | $oldBits <<= $newBitCount; |
| @@ -158,9 +158,9 @@ | ||
| 158 | 158 | if ( $rightPadFinalBits ) { |
| 159 | 159 | $oldBits <<= $newBitCount; |
| 160 | 160 | } |
| 161 | 161 | $encodedString .= $chars[ $oldBits ]; |
| 162 | - | |
| 162 | + | |
| 163 | 163 | if ( $padFinalGroup ) { |
| 164 | 164 | // Array of the lowest common multiples of $bitsPerCharacter and 8, divided by 8 |
| 165 | 165 | $lcmMap = array( 1 => 1, 2 => 1, 3 => 3, 4 => 1, 5 => 5, 6 => 3, 7 => 7, 8 => 1 ); |
| 166 | 166 | $bytesPerGroup = $lcmMap[ $bitsPerCharacter ]; |
| @@ -166,37 +166,37 @@ | ||
| 166 | 166 | $bytesPerGroup = $lcmMap[ $bitsPerCharacter ]; |
| 167 | 167 | $pads = $bytesPerGroup * $charsPerByte - ceil( ( strlen( $rawString ) % $bytesPerGroup ) * $charsPerByte ); |
| 168 | 168 | $encodedString .= str_repeat( $padCharacter, $pads ); |
| 169 | 169 | } |
| 170 | - | |
| 170 | + | |
| 171 | 171 | break; |
| 172 | 172 | } |
| 173 | - | |
| 173 | + | |
| 174 | 174 | // Get next byte |
| 175 | 175 | $byte = array_shift( $bytes ); |
| 176 | 176 | $bitsRead = 0; |
| 177 | - | |
| 177 | + | |
| 178 | 178 | } else { |
| 179 | 179 | $oldBitCount = 0; |
| 180 | 180 | $newBitCount = $bitsPerCharacter; |
| 181 | 181 | } |
| 182 | - | |
| 182 | + | |
| 183 | 183 | // Read only the needed bits from this byte |
| 184 | 184 | $bits = $byte >> 8 - ( $bitsRead + ( $newBitCount ) ); |
| 185 | 185 | $bits ^= $bits >> $newBitCount << $newBitCount; |
| 186 | 186 | $bitsRead += $newBitCount; |
| 187 | - | |
| 187 | + | |
| 188 | 188 | if ( $oldBitCount ) { |
| 189 | 189 | // Bits come from seperate bytes, add $oldBits to $bits |
| 190 | 190 | $bits = ( $oldBits << $newBitCount ) | $bits; |
| 191 | 191 | } |
| 192 | - | |
| 192 | + | |
| 193 | 193 | $encodedString .= $chars[ $bits ]; |
| 194 | 194 | } |
| 195 | - | |
| 195 | + | |
| 196 | 196 | return $encodedString; |
| 197 | 197 | } |
| 198 | - | |
| 198 | + | |
| 199 | 199 | /** |
| 200 | 200 | * Decode a string |
| 201 | 201 | * |
| 202 | 202 | * @param string $encodedString Data to decode |
| @@ -208,9 +208,9 @@ | ||
| 208 | 208 | if ( ! $encodedString || ! is_string( $encodedString ) ) { |
| 209 | 209 | // Empty string, nothing to decode |
| 210 | 210 | return ''; |
| 211 | 211 | } |
| 212 | - | |
| 212 | + | |
| 213 | 213 | $chars = $this->_chars; |
| 214 | 214 | $bitsPerCharacter = $this->_bitsPerCharacter; |
| 215 | 215 | $radix = $this->_radix; |
| 216 | 216 | $rightPadFinalBits = $this->_rightPadFinalBits; |
| @@ -216,26 +216,26 @@ | ||
| 216 | 216 | $rightPadFinalBits = $this->_rightPadFinalBits; |
| 217 | 217 | $padFinalGroup = $this->_padFinalGroup; |
| 218 | 218 | $padCharacter = $this->_padCharacter; |
| 219 | 219 | $caseSensitive = $this->_caseSensitive; |
| 220 | - | |
| 220 | + | |
| 221 | 221 | // Get index of encoded characters |
| 222 | 222 | if ( $this->_charmap ) { |
| 223 | 223 | $charmap = $this->_charmap; |
| 224 | - | |
| 224 | + | |
| 225 | 225 | } else { |
| 226 | 226 | $charmap = array(); |
| 227 | - | |
| 227 | + | |
| 228 | 228 | for ( $i = 0; $i < $radix; $i ++ ) { |
| 229 | 229 | $charmap[ $chars[ $i ] ] = $i; |
| 230 | 230 | } |
| 231 | - | |
| 231 | + | |
| 232 | 232 | $this->_charmap = $charmap; |
| 233 | 233 | } |
| 234 | - | |
| 234 | + | |
| 235 | 235 | // The last encoded character is $encodedString[$lastNotatedIndex] |
| 236 | 236 | $lastNotatedIndex = strlen( $encodedString ) - 1; |
| 237 | - | |
| 237 | + | |
| 238 | 238 | // Remove trailing padding characters |
| 239 | 239 | if ( $padFinalGroup ) { |
| 240 | 240 | while ( $encodedString[ $lastNotatedIndex ] === $padCharacter ) { |
| 241 | 241 | $encodedString = substr( $encodedString, 0, $lastNotatedIndex ); |
| @@ -241,53 +241,53 @@ | ||
| 241 | 241 | $encodedString = substr( $encodedString, 0, $lastNotatedIndex ); |
| 242 | 242 | $lastNotatedIndex --; |
| 243 | 243 | } |
| 244 | 244 | } |
| 245 | - | |
| 245 | + | |
| 246 | 246 | $rawString = ''; |
| 247 | 247 | $byte = 0; |
| 248 | 248 | $bitsWritten = 0; |
| 249 | - | |
| 249 | + | |
| 250 | 250 | // Convert each encoded character to a series of unencoded bits |
| 251 | 251 | for ( $c = 0; $c <= $lastNotatedIndex; $c ++ ) { |
| 252 | - | |
| 252 | + | |
| 253 | 253 | if ( ! $caseSensitive && ! isset( $charmap[ $encodedString[ $c ] ] ) ) { |
| 254 | 254 | // Encoded character was not found; try other case |
| 255 | 255 | if ( isset( $charmap[ $cUpper = strtoupper( $encodedString[ $c ] ) ] ) ) { |
| 256 | 256 | $charmap[ $encodedString[ $c ] ] = $charmap[ $cUpper ]; |
| 257 | - | |
| 257 | + | |
| 258 | 258 | } elseif ( isset( $charmap[ $cLower = strtolower( $encodedString[ $c ] ) ] ) ) { |
| 259 | 259 | $charmap[ $encodedString[ $c ] ] = $charmap[ $cLower ]; |
| 260 | 260 | } |
| 261 | 261 | } |
| 262 | - | |
| 262 | + | |
| 263 | 263 | if ( isset( $charmap[ $encodedString[ $c ] ] ) ) { |
| 264 | 264 | $bitsNeeded = 8 - $bitsWritten; |
| 265 | 265 | $unusedBitCount = $bitsPerCharacter - $bitsNeeded; |
| 266 | - | |
| 266 | + | |
| 267 | 267 | // Get the new bits ready |
| 268 | 268 | if ( $bitsNeeded > $bitsPerCharacter ) { |
| 269 | 269 | // New bits aren't enough to complete a byte; shift them left into position |
| 270 | 270 | $newBits = $charmap[ $encodedString[ $c ] ] << $bitsNeeded - $bitsPerCharacter; |
| 271 | 271 | $bitsWritten += $bitsPerCharacter; |
| 272 | - | |
| 272 | + | |
| 273 | 273 | } elseif ( $c !== $lastNotatedIndex || $rightPadFinalBits ) { |
| 274 | 274 | // Zero or more too many bits to complete a byte; shift right |
| 275 | 275 | $newBits = $charmap[ $encodedString[ $c ] ] >> $unusedBitCount; |
| 276 | 276 | $bitsWritten = 8; //$bitsWritten += $bitsNeeded; |
| 277 | - | |
| 277 | + | |
| 278 | 278 | } else { |
| 279 | 279 | // Final bits don't need to be shifted |
| 280 | 280 | $newBits = $charmap[ $encodedString[ $c ] ]; |
| 281 | 281 | $bitsWritten = 8; |
| 282 | 282 | } |
| 283 | - | |
| 283 | + | |
| 284 | 284 | $byte |= $newBits; |
| 285 | - | |
| 285 | + | |
| 286 | 286 | if ( $bitsWritten === 8 || $c === $lastNotatedIndex ) { |
| 287 | 287 | // Byte is ready to be written |
| 288 | 288 | $rawString .= pack( 'C', $byte ); |
| 289 | - | |
| 289 | + | |
| 290 | 290 | if ( $c !== $lastNotatedIndex ) { |
| 291 | 291 | // Start the next byte |
| 292 | 292 | $bitsWritten = $unusedBitCount; |
| 293 | 293 | $byte = ( $charmap[ $encodedString[ $c ] ] ^ ( $newBits << $unusedBitCount ) ) << 8 - $bitsWritten; |
| @@ -292,14 +292,14 @@ | ||
| 292 | 292 | $bitsWritten = $unusedBitCount; |
| 293 | 293 | $byte = ( $charmap[ $encodedString[ $c ] ] ^ ( $newBits << $unusedBitCount ) ) << 8 - $bitsWritten; |
| 294 | 294 | } |
| 295 | 295 | } |
| 296 | - | |
| 296 | + | |
| 297 | 297 | } elseif ( $strict ) { |
| 298 | 298 | // Unable to decode character; abort |
| 299 | 299 | return null; |
| 300 | 300 | } |
| 301 | 301 | } |
| 302 | - | |
| 302 | + | |
| 303 | 303 | return $rawString; |
| 304 | 304 | } |
| 305 | 305 | } |