| | |
| | /** |
| | - * Fix: Extracts size from the Git header without inflating the entire blob. |
| | + * Fix: Uses incremental inflation to safely read headers for both loose and packed objects. |
| | + * Resolves the true uncompressed size for Delta objects in packfiles. |
| | */ |
| | public function getObjectSize(string $sha): int { |
| | $loose = "{$this->objPath}/" . substr($sha, 0, 2) . "/" . substr($sha, 2); |
| | if (file_exists($loose)) { |
| | $f = @fopen($loose, 'rb'); |
| | if (!$f) return 0; |
| | - $data = fread($f, 128); |
| | + |
| | + // Use inflate_init for safe partial decompression of loose headers |
| | + $ctx = inflate_init(ZLIB_ENCODING_DEFLATE); |
| | + $data = ''; |
| | + while (!feof($f)) { |
| | + $chunk = fread($f, 128); // Read small chunks for the header |
| | + $data .= inflate_add($ctx, $chunk, ZLIB_NO_FLUSH); |
| | + if (strpos($data, "\0") !== false) break; // Stop once we have the header |
| | + } |
| | fclose($f); |
| | - $inflated = @gzuncompress($data); |
| | - if (!$inflated) return 0; |
| | - $header = explode("\0", $inflated, 2)[0]; |
| | + |
| | + $header = explode("\0", $data, 2)[0]; |
| | $parts = explode(' ', $header); |
| | return isset($parts[1]) ? (int)$parts[1] : 0; |
| | } |
| | return $this->getPackedObjectSize($sha); |
| | } |
| | |
| | /** |
| | - * Reads the Variable Length Quantity (VLQ) size from the packfile header. |
| | + * Reads size from packfile. Handles OBJ_OFS_DELTA and OBJ_REF_DELTA to return full file size. |
| | */ |
| | private function getPackedObjectSize(string $sha): int { |
| | $info = $this->getPackOffset($sha); |
| | if (!$info) return 0; |
| | |
| | $pf = @fopen($info['file'], 'rb'); |
| | if (!$pf) return 0; |
| | |
| | fseek($pf, $info['offset']); |
| | + |
| | + // Read Pack Object Header |
| | $byte = ord(fread($pf, 1)); |
| | + $type = ($byte >> 4) & 7; |
| | $size = $byte & 15; |
| | $shift = 4; |
| | |
| | while ($byte & 128) { |
| | $byte = ord(fread($pf, 1)); |
| | $size |= (($byte & 127) << $shift); |
| | $shift += 7; |
| | + } |
| | + |
| | + // If it's a Delta, we must read the delta header to get the target (actual) size. |
| | + // Type 6 = OBJ_OFS_DELTA, Type 7 = OBJ_REF_DELTA |
| | + if ($type === 6 || $type === 7) { |
| | + if ($type === 6) { |
| | + // OFS_DELTA: Skip the variable-length offset |
| | + $byte = ord(fread($pf, 1)); |
| | + while ($byte & 128) { $byte = ord(fread($pf, 1)); } |
| | + } else { |
| | + // REF_DELTA: Skip the 20-byte base SHA |
| | + fread($pf, 20); |
| | + } |
| | + |
| | + // Inflate the start of the delta stream to get the target size |
| | + $ctx = inflate_init(ZLIB_ENCODING_DEFLATE); |
| | + $buffer = ''; |
| | + $found = false; |
| | + |
| | + // We only need the first two VLQ integers from the stream |
| | + while (!$found && !feof($pf)) { |
| | + $chunk = fread($pf, 512); |
| | + $buffer .= inflate_add($ctx, $chunk, ZLIB_NO_FLUSH); |
| | + |
| | + // Check if we have enough bytes to decode two VLQs |
| | + // (Just a heuristic check, the decoding loop below handles the actual logic) |
| | + if (strlen($buffer) > 20) $found = true; |
| | + } |
| | + |
| | + // Decode Delta Header: [Source Size VLQ] [Target Size VLQ] |
| | + $pos = 0; |
| | + |
| | + // Skip Source Size |
| | + $byte = ord($buffer[$pos++]); |
| | + while ($byte & 128) { $byte = ord($buffer[$pos++]); } |
| | + |
| | + // Read Target Size (The full file size) |
| | + $byte = ord($buffer[$pos++]); |
| | + $size = $byte & 127; |
| | + $shift = 7; |
| | + while ($byte & 128) { |
| | + $byte = ord($buffer[$pos++]); |
| | + $size |= (($byte & 127) << $shift); |
| | + $shift += 7; |
| | + } |
| | } |
| | |
 |
| | } |
| | |
| | - /** |
| | - * Refactored fromPack to handle larger objects and accurate offsets. |
| | - */ |
| | private function fromPack(string $sha): ?string { |
| | $info = $this->getPackOffset($sha); |
|
| | while ($byte & 128) { $byte = ord(fread($pf, 1)); } |
| | |
| | - // Read a large enough chunk for decompression (up to 32MB for blobs) |
| | $data = @gzuncompress(fread($pf, 33554432)); |
| | fclose($pf); |
| | return $data ?: null; |
| | } |
| | |
| | - /** |
| | - * Helper to find an object's location in .idx files. |
| | - */ |
| | private function getPackOffset(string $sha): ?array { |
| | $packs = glob("{$this->objPath}/pack/*.idx"); |