fix(handler): improve tar handler to support sparse archives.

A custom TarInfo that keeps the actual entry size rather than the
original file size is used when calculating the end offset.

unblob/handlers/archive/_safe_tarfile.py

@@ -1,6 +1,9 @@
 import os
 import tarfile
 from pathlib import Path
+from tarfile import (
+    TarInfo,
+)
 
 from structlog import get_logger
 
@@ -13,6 +16,14 @@
 MAX_PATH_LEN = 255
 
 
+class SafeTarInfo(TarInfo):
+    def _proc_member(self, tarfile) -> "SafeTarInfo":
+        orig_size = self.size
+        self = super()._proc_member(tarfile)
+        self.size = orig_size
+        return self
+
+
 class SafeTarFile:
     def __init__(self, inpath: Path):
         self.inpath = inpath

unblob/handlers/archive/tar.py

@@ -16,7 +16,7 @@
     StructHandler,
     ValidChunk,
 )
-from ._safe_tarfile import SafeTarFile
+from ._safe_tarfile import SafeTarFile, SafeTarInfo
 
 logger = get_logger()
 
@@ -43,7 +43,7 @@ def _get_tar_end_offset(file: File, offset=0):
 
 def _get_end_of_last_tar_entry(file) -> int:
     try:
-        tf = tarfile.TarFile(mode="r", fileobj=file)
+        tf = tarfile.TarFile(mode="r", fileobj=file, tarinfo=SafeTarInfo)
     except tarfile.TarError:
         return -1