Support rejecting insecure hashes when updating

Signed-off-by: Michał Górny <mgorny@gentoo.org>

4 files changed, 59 insertions, 4 deletions

diff --git a/gemato/exceptions.py b/gemato/exceptions.py
index db754ac..ad9002a 100644
--- a/gemato/exceptions.py
+++ b/gemato/exceptions.py
@@ -1,6 +1,6 @@
 # gemato: exceptions
 # vim:fileencoding=utf-8
-# (c) 2017-2020 Michał Górny
+# (c) 2017-2022 Michał Górny
 # Licensed under the terms of 2-clause BSD license
 
 class GematoException(Exception):
@@ -279,3 +279,17 @@ class ManifestNoSupportedHashes(GematoException):
     def __str__(self):
         return (f"No hashes provided for file {self.entry.path!r} are "
                 f"supported (out of {' '.join(self.entry.checksums)})")
+
+
+class ManifestInsecureHashes(GematoException):
+    """
+    An exception raised when some of the requested hashes are insecure.
+    """
+
+    def __init__(self, hashes):
+        super().__init__()
+        self.hashes = hashes
+
+    def __str__(self):
+        return ("Some of the requested hashes are considered insecure: "
+                f"{' '.join(self.hashes)}")
diff --git a/gemato/manifest.py b/gemato/manifest.py
index 2ce4811..d62fa14 100644
--- a/gemato/manifest.py
+++ b/gemato/manifest.py
@@ -564,3 +564,11 @@ def is_hash_supported(h):
     Return True if the passed hash (in Manifest naming) is supported.
     """
     return h in MANIFEST_HASH_MAPPING
+
+
+def is_hash_secure(h):
+    """
+    Return True if the passed hash (in Manifest naming) is considered
+    cryptographically secure.
+    """
+    return h not in ("MD5", "SHA1")
diff --git a/gemato/verify.py b/gemato/verify.py
index b54c946..44a639f 100644
--- a/gemato/verify.py
+++ b/gemato/verify.py
@@ -13,9 +13,14 @@ from gemato.exceptions import (
     ManifestCrossDevice,
     ManifestInvalidPath,
     ManifestNoSupportedHashes,
+    ManifestInsecureHashes,
     )
 from gemato.hash import hash_file
-from gemato.manifest import manifest_hashes_to_hashlib, is_hash_supported
+from gemato.manifest import (
+    manifest_hashes_to_hashlib,
+    is_hash_supported,
+    is_hash_secure,
+    )
 
 
 def get_file_metadata(path, hashes):
@@ -220,7 +225,7 @@ def verify_path(path, e, expected_dev=None, last_mtime=None):
 
 
 def update_entry_for_path(path, e, hashes=None, expected_dev=None,
-                          last_mtime=None):
+                          last_mtime=None, require_secure_hashes=False):
     """
     Update the data in entry @e to match the current state of file
     at path @path. Uses hashes listed in @hashes (using Manifest names),
@@ -241,12 +246,18 @@ def update_entry_for_path(path, e, hashes=None, expected_dev=None,
     If @last_mtime is not None, it specifies the timestamp corresponding
     to the previous file update. If the file is not newer than that,
     the checksum calculation is skipped.
+
+    If @require_secure_hashes is True, only secure hashes can be used.
     """
 
     assert e.tag not in ('IGNORE', 'TIMESTAMP')
 
     if hashes is None:
         hashes = list(e.checksums)
+    if require_secure_hashes:
+        insecure = list(filter(lambda x: not is_hash_secure(x), hashes))
+        if insecure:
+            raise ManifestInsecureHashes(insecure)
 
     with contextlib.closing(get_file_metadata(path, hashes)) as g:
         # 1. verify whether the file existed in the first place
diff --git a/tests/test_verify.py b/tests/test_verify.py
index 1f3170b..7ee4545 100644
--- a/tests/test_verify.py
+++ b/tests/test_verify.py
@@ -1,8 +1,9 @@
 # gemato: Verification tests
 # vim:fileencoding=utf-8
-# (c) 2017-2020 Michał Górny
+# (c) 2017-2022 Michał Górny
 # Licensed under the terms of 2-clause BSD license
 
+import contextlib
 import itertools
 import os
 import os.path
@@ -14,6 +15,7 @@ import pytest
 from gemato.exceptions import (
     ManifestInvalidPath,
     ManifestCrossDevice,
+    ManifestInsecureHashes,
     )
 from gemato.hash import hash_path
 from gemato.manifest import new_manifest_entry
@@ -444,6 +446,26 @@ def test_unreadable_file(test_tree, function, args):
 
 
 @pytest.mark.parametrize(
+    "entry_hash,hashes_arg,insecure",
+    [("MD5", None, True),
+     ("SHA1", None, True),
+     ("SHA512", None, False),
+     ("MD5", "SHA1 SHA512", True),
+     ("MD5", "SHA512", False),
+     ("SHA512", "MD5 SHA512", True),
+     ])
+def test_insecure_hashes(test_tree, entry_hash, hashes_arg, insecure):
+    ctx = (pytest.raises(ManifestInsecureHashes) if insecure
+           else contextlib.nullcontext())
+    with ctx:
+        update_entry_for_path(
+            test_tree / "empty-file",
+            new_manifest_entry("DATA", "empty-file", 0, {entry_hash: ""}),
+            hashes=hashes_arg.split() if hashes_arg else None,
+            require_secure_hashes=True)
+
+
+@pytest.mark.parametrize(
     'a_cls,a_name,a_args,b_cls,b_name,b_args,expected,diff',
     [('DATA', 'test', [0, {'MD5': 'd41d8cd98f00b204e9800998ecf8427e'}],
       'DATA', 'test', [0, {'MD5': 'd41d8cd98f00b204e9800998ecf8427e'}],