base64 hash

Author: cmcmarrow

salt/utils/gitfs.py

@@ -3,6 +3,7 @@
 """
 
 
+import base64
 import contextlib
 import copy
 import errno
@@ -16,6 +17,7 @@
 import shlex
 import shutil
 import stat
+import string
 import subprocess
 import time
 import weakref
@@ -457,7 +459,7 @@ def __init__(
 
         hash_type = getattr(hashlib, self.opts.get("hash_type", "md5"))
         # Generate full id. The full id is made from these parts name-id-env-_root.
-        # Full id stops collections in the gitfs cache.
+        # Full id helps decrease the chances of collections in the gitfs cache.
         self._full_id = "-".join(
             [
                 getattr(self, "name", ""),
@@ -468,7 +470,22 @@ def __init__(
         )
         # We loaded this data from yaml configuration files, so, its safe
         # to use UTF-8
-        self.cachedir_basename = f"{getattr(self, 'name', '')}-{hash_type(self._full_id.encode('utf-8')).hexdigest()}"
+        base64_hash = str(
+            base64.b64encode(hash_type(self._full_id.encode("utf-8")).digest()),
+            encoding="ascii",  # base64 only outputs ascii
+        ).replace(
+            "/", "_"
+        )  # replace "/" with "_" to not cause trouble with file system
+
+        # limit name length to 19, so we don't eat up all the path length for windows
+        # this is due to pygit2 limitations
+        # replace any unknown char with "_" to not cause trouble with file system
+        name_chars = string.ascii_letters + string.digits + "-"
+        cache_name = "".join(
+            c if c in name_chars else "_" for c in getattr(self, "name", "")[:19]
+        )
+
+        self.cachedir_basename = f"{cache_name}-{base64_hash}"
         self.cachedir = salt.utils.path.join(cache_root, self.cachedir_basename)
         self.linkdir = salt.utils.path.join(cache_root, "links", self.cachedir_basename)
 

tests/pytests/unit/utils/test_gitfs.py

@@ -254,7 +254,9 @@ def test_full_id_pygit2(_prepare_provider):
 )
 def test_get_cachedir_basename_pygit2(_prepare_provider):
     basename = _prepare_provider.get_cachedir_basename()
-    assert len(basename) > 1
+    # Note: if you are changing the length of basename
+    # keep in mind that pygit2 will error out on large file paths on Windows
+    assert len(basename) == 45
     assert basename[0] == "-"
-    # check that a valid hex is given
-    assert all(c in string.hexdigits for c in basename[1:])
+    # check that a valid base64 is given '/' -> '_'
+    assert all(c in string.ascii_letters + string.digits + "+_=" for c in basename[1:])

tests/unit/utils/test_gitfs.py

@@ -124,13 +124,13 @@ def test_full_id_with_name(self):
     def test_get_cachedir_basename(self):
         self.assertEqual(
             self.main_class.remotes[0].get_cachedir_basename(),
-            "-b4dcbd51b08742ec23eaf96d192d29b417ec137ea7ca0c0de2515cfaf6e26860",
+            "-tNy9UbCHQuwj6vltGS0ptBfsE36nygwN4lFc+vbiaGA=",
         )
 
     def test_get_cachedir_base_with_name(self):
         self.assertEqual(
             self.main_class.remotes[1].get_cachedir_basename(),
-            "repo2-4218c2f8e303c6ea24cc541d8748e523d5b443c3050170a43a1a00be253b56aa",
+            "repo2-QhjC+OMDxuokzFQdh0jlI9W0Q8MFAXCkOhoAviU7Vqo=",
         )
 
     def test_git_provider_mp_lock(self):