Fiz long encoding/decoding

flix-tech · Jan 23, 2025 · 2de9b2f · 2de9b2f
1 parent fcf815e
commit 2de9b2f
Show file tree

Hide file tree

Showing 2 changed files with 119 additions and 23 deletions.
diff --git a/lib/avro/datum.php b/lib/avro/datum.php
@@ -43,6 +43,79 @@ public function __construct($expected_schema, $datum)
   }
 }
 
+/**
+ * Zigzag implementation to encode longs
+ * https://en.wikipedia.org/wiki/Variable-length_quantity#Zigzag_encoding
+ *
+ * @package Avro
+ */
+class Zigzag {
+
+  /**
+   * Implementation of unsigned shift right as PHP does not have the `>>>` operator
+   *
+   * @param  int  $n
+   * @param  int  $x
+   *
+   * @return int
+   */
+  public static function unsigned_right_shift(int $n, int $x): int
+  {
+    $byte_bits = 8;
+    $platform_bits = PHP_INT_SIZE * $byte_bits;
+
+    // $sign_mask will be a row of 1s (0xFFFFFFFF) if $n is
+    // negative, and a row of 0s (0x00000000) if $n is positive
+    $sign_mask = ($n >> ($platform_bits - 1));
+
+    // $filler_removal_mask will have the $x left most bits set
+    // to 1 and the other bits set to 0 if $n is negative.
+    // if $n is positive, sign mask will have all bit set to 0
+    $filler_removal_mask = $sign_mask << $platform_bits - $x;
+
+    // right shift $n with $x and apply the mask to flip the
+    // left most bits to 0 if $n is negative and do nothing otherwise
+    return ($n >> $x) ^ $filler_removal_mask;
+  }
+
+  /**
+   * @param int|string $n
+   * @return string long $n encoded as bytes
+   * @internal This relies on 64-bit PHP.
+   */
+  public static function encode_long($n): string
+  {
+    $n = (int) $n;
+    $n = ($n << 1) ^ ($n >> 63);
+    $str = '';
+    if (($n & ~0x7F) != 0) {
+      $str .= chr(($n | 0x80) & 0xFF);
+      $n = self::unsigned_right_shift($n, 7);
+
+      while ($n > 0x7F) {
+        $str .= chr(($n | 0x80) & 0xFF);
+        $n = self::unsigned_right_shift($n, 7);
+      }
+    }
+
+    $str .= chr($n);
+    return $str;
+  }
+
+  public static function decode_long(array $bytes): int {
+    $b = array_shift($bytes);
+    $n = $b & 0x7f;
+    $shift = 7;
+    while (0 != ($b & 0x80))
+    {
+      $b = array_shift($bytes);
+      $n |= (($b & 0x7f) << $shift);
+      $shift += 7;
+    }
+    return self::unsigned_right_shift($n,  1) ^ -($n & 1);
+  }
+}
+
 /**
  * Exceptions arising from incompatibility between
  * reader and writer schemas.
@@ -304,18 +377,9 @@ static function double_to_long_bits($double)
    * @return string long $n encoded as bytes
    * @internal This relies on 64-bit PHP.
    */
-  static public function encode_long($n)
+  public static function encode_long($n): string
   {
-    $n = (int) $n;
-    $n = ($n << 1) ^ ($n >> 63);
-    $str = '';
-    while (0 != ($n & ~0x7F))
-    {
-      $str .= chr(($n & 0x7F) | 0x80);
-      $n >>= 7;
-    }
-    $str .= chr($n);
-    return $str;
+    return Zigzag::encode_long($n);
   }
 
   /**
@@ -931,16 +995,7 @@ class AvroIOBinaryDecoder
    */
   public static function decode_long_from_array($bytes)
   {
-    $b = array_shift($bytes);
-    $n = $b & 0x7f;
-    $shift = 7;
-    while (0 != ($b & 0x80))
-    {
-      $b = array_shift($bytes);
-      $n |= (($b & 0x7f) << $shift);
-      $shift += 7;
-    }
-    return (($n >> 1) ^ -($n & 1));
+    return Zigzag::decode_long($bytes);
   }
 
   /**

diff --git a/test/DatumIOTest.php b/test/DatumIOTest.php
@@ -51,6 +51,37 @@ function test_datum_round_trip($schema_json, $datum, $binary)
     $this->assertEquals($datum, $read_datum);
   }
 
+  /**
+   * @dataProvider zigzag_unsigned_right_shift_provider
+   */
+  function test_zigzag_unsigned_right_shift(int $expected, int $n, int $x) {
+    $this->assertEquals($expected, Zigzag::unsigned_right_shift($n, $x));
+  }
+
+  public static function zigzag_unsigned_right_shift_provider(): array {
+    return [
+      [4611686018427387902, -8, 2],
+      [2, 8, 2],
+      [144115188075855871, -2, 7],
+      [1125899906842623, 144115188075855871, 7],
+      [8796093022207, 1125899906842623, 7],
+      [68719476735, 8796093022207, 7],
+      [536870911, 68719476735, 7],
+      [4194303, 536870911, 7],
+      [32767, 4194303, 7],
+      [255, 32767, 7],
+      [1, 255, 7],
+      [144115188059078656, -2147483648, 7],
+      [1125899906711552, 144115188059078656, 7],
+      [8796093021184, 1125899906711552, 7],
+      [68719476728, 8796093021184, 7],
+      [536870911, 68719476728, 7],
+      [4194303, 536870911, 7],
+      [32767, 4194303, 7],
+      [255, 32767, 7],
+    ];
+  }
+
   /**
    * @return array
    */
@@ -67,11 +98,21 @@ function data_provider()
                  array('"int"', 1, "\002"),
                  array('"int"', 2147483647, "\xFE\xFF\xFF\xFF\x0F"),
 
-                 // array('"long"', (int) -9223372036854775808, "\001"),
+                 array('"long"', (int) -9223372036854775808, "\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x01"),
+                 array('"long"', -(1<<62), "\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x7F"),
+                 array('"long"', -4294967295, "\xFD\xFF\xFF\xFF\x1F"),
+                 array('"long"', -10, "\x13"),
+                 array('"long"', -3, "\005"),
+                 array('"long"', -2, "\003"),
                  array('"long"', -1, "\001"),
                  array('"long"',  0, "\000"),
                  array('"long"',  1, "\002"),
-                 // array('"long"', 9223372036854775807, "\002")
+                 array('"long"',  2, "\004"),
+                 array('"long"',  3, "\006"),
+                 array('"long"', 10, "\x14"),
+                 array('"long"', 4294967295, "\xFE\xFF\xFF\xFF\x1F"),
+                 array('"long"', 1<<62, "\x80\x80\x80\x80\x80\x80\x80\x80\x80\x01"),
+                 array('"long"', 9223372036854775807, "\xFE\xFF\xFF\xFF\xFF\xFF\xFF\xFF\xFF\x01"),
 
                  array('"float"', (float) -10.0, "\000\000 \301"),
                  array('"float"', (float)  -1.0, "\000\000\200\277"),