Merge pull request #39 from mike42/bugfix/38-lzw

Fix LZW deferred clear code
mike42 · Feb 26, 2019 · 2ee1b09 · 2ee1b09
2 parents 0ea8f50 + 4e88acc
commit 2ee1b09
Show file tree

Hide file tree

Showing 6 changed files with 9 additions and 39 deletions.
diff --git a/src/Mike42/GfxPhp/Util/LzwCompression.php b/src/Mike42/GfxPhp/Util/LzwCompression.php
@@ -32,14 +32,13 @@ public static function compress(string $inp, int $minCodeSize)
                 $outp -> add($dict -> get($word), $bits);
                 $dict -> add($ch);
                 if ($dict -> getSize() > (2 ** $bits)) {
+                    // expand bit size
                     $bits++;
-                    //echo "compress Expanded to $bits\n";
                 }
                 $word = $inp[$i];
             }
             // Separately to parsing the input stream, periodically reset to keep under 12 bits
             if ($dict -> getSize() >= AbstractLzwDictionary::MAX_SIZE) {
-                //echo "truncating dict\n";
                 if ($word !== "") {
                     $outp -> add($dict -> get($word), $bits);
                 }
@@ -75,20 +74,19 @@ public static function decompress(string $inp, int $minCodeSize)
         $prevcodeStr = $dict -> get($prevcode);
         $outp .= $prevcodeStr;
         while (($code = $buffer -> read($bits)) !== false) {
-            //echo "DECODING " . $code . "\n";
+            // Decode
             if ($code > $dict -> getSize()) {
                 throw new \Exception("Bad LZW, code too large");
             }
 
             if ($code === $dict -> getClearCode()) {
-                //echo "got RESET\n";
+                // got a clear code
                 $bits = $minCodeSize + 1;
                 $prevcodeStr = "";
                 $dict -> clear();
                 // Manually do an iteration with no dictionary modification.
                 // This involves reading ahead 1 character, same as what we do above the
                 // 'while' loop. There is probably some way to combine these
-                // way to combine them.
                 $prevcode = $buffer -> read($bits);
                 while ($prevcode == $dict -> getClearCode()) {
                     // In case of consecutive clear codes
@@ -105,7 +103,7 @@ public static function decompress(string $inp, int $minCodeSize)
                 $outp .= $prevcodeStr;
                 continue;
             } else if ($code === $dict -> getEodCode()) {
-                //$outp .= $prevcodeStr;
+                // End the stream
                 break;
             }
 
@@ -116,19 +114,20 @@ public static function decompress(string $inp, int $minCodeSize)
             }
             $outp .= $codeStr;
             $ch = $codeStr[0];
-            $dict -> add($prevcodeStr . $ch);
+            if ($dict -> getSize() < 4096) { // Don't modify dictionary after a certain point
+                $dict -> add($prevcodeStr . $ch);
+            }
             // assign current code to prev code
             $prevcode = $code;
             $prevcodeStr = $codeStr;
             if ($dict -> getSize() >= (2 ** $bits)) {
                 if (($bits + 1) > 12) {
                     // Don't exceed 12 bits. If table gets max'd out at
                     // 4096, the following RESET or END code remains 12-bit encoded.
-                    //echo "not expanding";
                     continue;
                 } else {
+                    // Expand
                     $bits++;
-                    //echo "decompress: expanded to $bits\n";
                 }
             }
         }

diff --git a/src/Mike42/GfxPhp/Util/LzwDecodeBuffer.php b/src/Mike42/GfxPhp/Util/LzwDecodeBuffer.php
@@ -24,33 +24,26 @@ public function __construct(string $contents)
      */
     public function read(int $readBits)
     {
-        //echo "\n read($readBits)\n";
-        //echo "  pointer is " . $this -> ptr . "\n";
         $num = 0;
         $firstBit = $this -> ptr - $readBits + 1;
         $lastBit = $this -> ptr;
         if ($firstBit < 0) {
             return false;
         }
-        //echo "  firstBit = $firstBit, lastBit = $lastBit\n";
         $val = 0;
         for ($i = $firstBit; $i <= $lastBit; $i++) {
             $bit = $this -> readBit($i);
             $val = ($val << 1) | $bit;
         }
         $this -> ptr -= $readBits;
-        //echo "\n  returning $val\n";
         return $val;
     }
 
     public function readBit(int $i)
     {
-        //echo "  readBit($i)\n";
         $byte = intdiv($i, 8);
-        //echo "   byte is $byte\n";
         $bit = $i % 8;
         $rightIgnore = 7 - $bit;
-        //echo "   bit is is $bit ($rightIgnore bits to the right to remove)\n";
         return (($this -> contents[$byte]) >> $rightIgnore) & 1;
     }
 }
diff --git a/src/Mike42/GfxPhp/Util/LzwDecodeDictionary.php b/src/Mike42/GfxPhp/Util/LzwDecodeDictionary.php
@@ -10,15 +10,11 @@ public function clear()
     {
         $count = 2 << ($this -> minCodeSize - 1);
         $this -> decodeDict = range(chr(0), chr($count - 1));
-        //$this -> encodeDict = array_flip($this -> decodeDict);
         $this -> clearCode = $count;
         $count++;
         $this -> eodCode = $count;
         $count++;
         $this -> size = $count;
-
-        //print_r($this -> decodeDict);
-        //print_r($this -> encodeDict);
     }
 
     public function get(int $code)
@@ -39,9 +35,6 @@ public function add(string $entry)
         if ($this -> size == AbstractLzwDictionary::MAX_SIZE) {
             throw new \Exception("LZW code table overflow");
         }
-        //$hexVal = bin2hex($entry);
-        //$l = strlen($entry);
-        //echo "LzwDecodeDictionary ADDING 0x$hexVal (len=$l) to dict @ position " . $this -> size ."\n";
         $this -> decodeDict[$this -> size] = $entry;
         $this -> size++;
     }

diff --git a/src/Mike42/GfxPhp/Util/LzwEncodeBuffer.php b/src/Mike42/GfxPhp/Util/LzwEncodeBuffer.php
@@ -6,8 +6,6 @@
  */
 class LzwEncodeBuffer
 {
-
-
     public function __construct()
     {
         $this -> textBuffer = "";
@@ -17,42 +15,33 @@ public function __construct()
 
     public function add(int $code, int $bits)
     {
-        //echo "ADD $code to output ($bits bits)\n";
         $mask = [];
         $byte = 0;
         for ($i = 0; $i <= 8; $i++) {
             $mask[] = $byte;
             $byte = ($byte << 1) | 1;
         }
-        //print_r($mask);
         while ($bits > 0) {
-            //echo "  $code, $bits bits remaining\n";
             // Determine number of bits to write to this byte
             $capacity = (8 - $this -> bitPos);
             $writeBits = min($capacity, $bits);
             // Extract bits to use, shift and OR them on
             $maskedByte = ($code & $mask[$writeBits]);
-            //echo "  masked byte for $writeBits bits is $maskedByte\n";
             $shiftDir = ($this -> bitPos);
-            //echo "  need to shift by $shiftDir to place\n";
             $theBits = $maskedByte << $shiftDir;
             $this -> bitBuffer = $this -> bitBuffer | $theBits;
             // Truncate lower bits of code and repeat
             $code = $code >> $writeBits;
             $bits = $bits - $writeBits;
             $this -> bitPos += $writeBits;
-            //echo "  curstor is at position " . $this -> bitPos . " \n";
             if ($this -> bitPos >= 8) {
                 $this -> bitPos = 0;
                 $this -> textBuffer .= chr($this -> bitBuffer);
                 $this -> bitBuffer = 0;
             }
-            //echo "  string is: " . bin2hex($this -> textBuffer) . "\n";
-            //echo "  buffer is: " . $this -> bitBuffer . ", position " . $this -> bitPos . "\n";
         }
     }
-
-
+
     public function asString()
     {
         if ($this -> bitPos !== 0) {

diff --git a/src/Mike42/GfxPhp/Util/LzwEncodeDictionary.php b/src/Mike42/GfxPhp/Util/LzwEncodeDictionary.php
@@ -34,9 +34,6 @@ public function add(string $entry)
         if ($this -> size == self::MAX_SIZE) {
             throw new \Exception("LZW code table overflow");
         }
-        //$hexVal = bin2hex($entry);
-        //$l = strlen($entry);
-        //echo "LzwEncodeDictionary ADDING 0x$hexVal (len=$l) to dict @ position " . $this -> size ."\n";
         $this -> encodeDict[$entry] = $this -> size;
         $this -> size++;
     }

diff --git a/test/integration/GifsuiteTest.php b/test/integration/GifsuiteTest.php
@@ -28,7 +28,6 @@ function test_4095_codes_clear() {
     }
 
     function test_4095_codes() {
-        $this -> markTestSkipped("Known bug: LZW overflow");
         $img = $this -> loadImage("4095-codes.gif");
         $this -> assertEquals(100, $img -> getWidth());
         $this -> assertEquals(100, $img -> getHeight());