1 files changed, 26 insertions, 24 deletions

diff --git a/dedup/debpkg.py b/dedup/debpkg.py
index de00e60..0d1b7da 100644
--- a/dedup/debpkg.py
+++ b/dedup/debpkg.py
@@ -1,26 +1,29 @@
 import tarfile
+import typing
 
 import arpy
 from debian import deb822
 
 from dedup.compression import decompress
-from dedup.hashing import hash_file
+from dedup.hashing import HashlibLike, hash_file
 
 class MultiHash:
-    def __init__(self, *hashes):
+    def __init__(self, *hashes: HashlibLike):
         self.hashes = hashes
 
-    def update(self, data):
+    def update(self, data: bytes) -> None:
         for hasher in self.hashes:
             hasher.update(data)
 
-def get_tar_hashes(tar, hash_functions):
+
+def get_tar_hashes(
+    tar: tarfile.TarFile,
+    hash_functions: typing.Sequence[typing.Callable[[], HashlibLike]],
+) -> typing.Iterator[typing.Tuple[str, int, typing.Dict[str, str]]]:
     """Given a TarFile read all regular files and compute all of the given hash
     functions on each file.
-    @type tar: tarfile.TarFile
     @param hash_functions: a sequence of parameter-less functions each creating a
             new hashlib-like object
-    @rtype: gen((str, int, {str: str}}
     @returns: an iterable of (filename, filesize, hashes) tuples where
             hashes is a dict mapping hash function names to hash values
     """
@@ -29,7 +32,9 @@ def get_tar_hashes(tar, hash_functions):
         if not elem.isreg(): # excludes hard links as well
             continue
         hasher = MultiHash(*[func() for func in hash_functions])
-        hash_file(hasher, tar.extractfile(elem))
+        extracted = tar.extractfile(elem)
+        assert extracted is not None
+        hash_file(hasher, extracted)
         hashes = {}
         for hashobj in hasher.hashes:
             hashvalue = hashobj.hexdigest()
@@ -37,17 +42,18 @@ def get_tar_hashes(tar, hash_functions):
                 hashes[hashobj.name] = hashvalue
         yield (elem.name, elem.size, hashes)
 
-def opentar(filelike):
+
+def opentar(filelike: typing.BinaryIO) -> tarfile.TarFile:
     return tarfile.open(fileobj=filelike, mode="r|", encoding="utf8",
                         errors="surrogateescape")
 
 class DebExtractor:
     "Base class for extracting desired features from a Debian package."
 
-    def __init__(self):
+    def __init__(self) -> None:
         self.arstate = "start"
 
-    def process(self, filelike):
+    def process(self, filelike: typing.BinaryIO) -> None:
         """Process a Debian package.
         @param filelike: is a file-like object containing the contents of the
                          Debian packge and can be read once without seeks.
@@ -89,22 +95,20 @@ class DebExtractor:
         else:
             assert self.arstate == "data"
 
-    def handle_ar_end(self):
+    def handle_ar_end(self) -> None:
         "Handle the end of the ar archive of the Debian package."
         if self.arstate != "data":
             raise ValueError("data.tar not found")
 
-    def handle_debversion(self, version):
+    def handle_debversion(self, version: bytes) -> None:
         """Handle the debian-binary member of the Debian package.
-        @type version: bytes
         @param version: The full contents of the ar member.
         """
 
-    def handle_control_tar(self, tarfileobj):
+    def handle_control_tar(self, tarfileobj: tarfile.TarFile) -> None:
         """Handle the control.tar member of the Debian package.
         If you replace this method, none of handle_control_member,
         handle_control_info or handle_control_end are called.
-        @type tarfileobj: tarfile.TarFile
         @param tarfile: is opened for streaming reads
         """
         controlseen = False
@@ -113,7 +117,9 @@ class DebExtractor:
                 name = elem.name
                 if name.startswith("./"):
                     name = name[2:]
-                content = tarfileobj.extractfile(elem).read()
+                extracted = tarfileobj.extractfile(elem)
+                assert extracted is not None
+                content = extracted.read()
                 self.handle_control_member(name, content)
                 if name == "control":
                     self.handle_control_info(deb822.Packages(content))
@@ -125,24 +131,20 @@ class DebExtractor:
             raise ValueError("control missing from control.tar")
         self.handle_control_end()
 
-    def handle_control_member(self, name, content):
+    def handle_control_member(self, name: str, content: bytes) -> None:
         """Handle a file member of the control.tar member of the Debian package.
-        @type name: str
         @param name: is the plain member name
-        @type content: bytes
         """
 
-    def handle_control_info(self, info):
+    def handle_control_info(self, info: deb822.Packages) -> None:
         """Handle the control member of the control.tar member of the Debian
         package.
-        @type info: deb822.Packages
         """
 
-    def handle_control_end(self):
+    def handle_control_end(self) -> None:
         "Handle the end of the control.tar member of the Debian package."
 
-    def handle_data_tar(self, tarfileobj):
+    def handle_data_tar(self, tarfileobj: tarfile.TarFile) -> None:
         """Handle the data.tar member of the Debian package.
-        @type tarfileobj: tarfile.TarFile
         @param tarfile: is opened for streaming reads
         """