fix(handler): add support for zip64 values in Central Directory Headers

Author: nyuware

blue.txt

@@ -0,0 +1 @@
+blue

red.txt

@@ -0,0 +1 @@
+red

tests/integration/archive/zip/zip64/__input__/colors.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a59995571929d2047445ef9ead25ebe1560a6b52a8204ed342c72f8eb624bec9
+size 457

tests/integration/archive/zip/zip64/__output__/colors.zip_extract/blue.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a0bee6616b5e5eae6799cb4525a884a82e7161614f11122bbdf4383b2ac05998
+size 5

tests/integration/archive/zip/zip64/__output__/colors.zip_extract/red.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ace33171ce0acb6891e3cc311d75a97aa429d77c05cba600d49ed9652ed49de
+size 4

unblob/handlers/archive/zip.py

@@ -111,6 +111,14 @@ def is_zip64_eocd(end_of_central_directory: Instance):
             or end_of_central_directory.offset_of_cd == 0xFFFFFFFF
         )
 
+    @staticmethod
+    def is_zip64_cd(central_directory: Instance):
+        # see https://pkware.cachefly.net/webdocs/casestudies/APPNOTE.TXT section 4.3.9.2
+        return (
+            central_directory.file_size == 0xFFFFFFFF
+            or central_directory.compress_size == 0xFFFFFFFF
+        )
+
     def _parse_zip64(self, file: File, start_offset: int, offset: int) -> Instance:
         file.seek(start_offset, io.SEEK_SET)
         for eocd_locator_offset in iterate_patterns(
@@ -147,7 +155,21 @@ def calculate_chunk(self, file: File, start_offset: int) -> Optional[ValidChunk]
             file.seek(offset, io.SEEK_SET)
             end_of_central_directory = self.parse_header(file)
 
-            if self.is_zip64_eocd(end_of_central_directory):
+            # some values in the CD can be FFFF, indicating its a zip64
+            # if the offset of the CD is 0xFFFFFFFF, its definitely one
+            # otherwise we check every other header indicating zip64
+            if end_of_central_directory.offset_of_cd == 0xFFFFFFFF:
+                file.seek(offset, io.SEEK_SET)
+                end_of_central_directory = self._parse_zip64(file, start_offset, offset)
+                break
+
+            file.seek(end_of_central_directory.offset_of_cd, io.SEEK_SET)
+            central_directory = self.cparser_le.cd_file_header_t(file)
+
+            if self.is_zip64_eocd(end_of_central_directory) or self.is_zip64_cd(
+                central_directory
+            ):
+                file.seek(offset, io.SEEK_SET)
                 end_of_central_directory = self._parse_zip64(file, start_offset, offset)
                 break