WIP: sort globs

Author: MHendricks

hab/distro_finders/df_zip.py

@@ -95,7 +95,7 @@ def distro_path_info(self):
             cached: Will always be `False`. The path is not stored in a .habcache
                 file so this data is not cached across processes.
         """
-        for path in self.root.glob(self.glob_str):
+        for path in sorted(self.root.glob(self.glob_str)):
             member_path = path / self.hab_filename
             if self.safe:
                 # Opening archives on cloud based systems is slow, this allows us

hab/utils.py

@@ -298,7 +298,7 @@ def glob_path(path):
     While a `pathlib.Path` can be created with a glob string you won't be able to
     resolve the glob into files. This function breaks the path down to its top level
     item and calls `Path.glob` on the remaining path. So `Path("/mnt/*/.hab.json")`
-    gets converted into `Path("/mnt").glob("*/.hab.json")`.
+    gets converted into `sorted(Path("/mnt").glob("*/.hab.json"))`.
 
     The input path will be converted to a `pathlib.Path` object for this operation.
 
@@ -307,7 +307,7 @@ def glob_path(path):
     # Strip the path into its parts removing the root of absolute paths.
     parts = path.parts[1:]
     # From the root run a glob search on all parts of the glob string
-    return Path(path.parts[0]).glob(str(Path(*parts)))
+    return sorted(Path(path.parts[0]).glob(str(Path(*parts))))
 
 
 class HabJsonEncoder(_json.JSONEncoder):

semgrep-json.yaml

@@ -6,6 +6,6 @@ rules:
           - pattern: json.dumps(..., sort_keys=True, ...)
           - pattern: json.dump(..., sort_keys=True, ...)
     message: Do not use sort_keys in json, hab relies on preserving the original
-      defined order.
+      defined order. You will need to sort glob calls for consistency.
     languages: [python]
     severity: ERROR

tests/test_env_var_expansion.py

@@ -38,7 +38,7 @@ def __post_init__(self) -> None:
     @classmethod
     def from_glob(cls, path, glob_str="*.json"):
         ret = []
-        for filename in path.glob(glob_str):
+        for filename in sorted(path.glob(glob_str)):
             ret.append(cls(filename))
         return ret