style: only test for files in the repo

Add a helper function to only consider files in the repo. This helps if
the local workspace has files coming from other sources, like the linux
kernel, or uncommitted changes.

Signed-off-by: Pablo Barbáchano <[email protected]>

Author: pb8o

tests/framework/utils.py

@@ -2,7 +2,6 @@
 # SPDX-License-Identifier: Apache-2.0
 """Generic utility functions that are used in the framework."""
 import functools
-import glob
 import json
 import logging
 import os
@@ -419,34 +418,6 @@ def search_output_from_cmd(cmd: str, find_regex: typing.Pattern) -> typing.Match
     )
 
 
-def get_files_from(
-    find_path: str, pattern: str, exclude_names: list = None, recursive: bool = True
-):
-    """
-    Return a list of files from a given path, recursively.
-
-    :param find_path: path where to look for files
-    :param pattern: what pattern to apply to file names
-    :param exclude_names: folder names to exclude
-    :param recursive: do a recursive search for the given pattern
-    :return: list of found files
-    """
-    found = []
-    # For each directory in the given path
-    for path_dir in os.scandir(find_path):
-        # Check if it should be skipped
-        if path_dir.name in exclude_names or os.path.isfile(path_dir):
-            continue
-        # Run glob inside the folder with the given pattern
-        found.extend(
-            glob.glob(f"{find_path}/{path_dir.name}/**/{pattern}", recursive=recursive)
-        )
-    # scandir will not look at the files matching the pattern in the
-    # current directory.
-    found.extend(glob.glob(f"{find_path}/./{pattern}"))
-    return found
-
-
 def get_free_mem_ssh(ssh_connection):
     """
     Get how much free memory in kB a guest sees, over ssh.

tests/framework/utils_repo.py

@@ -0,0 +1,25 @@
+# Copyright 2024 Amazon.com, Inc. or its affiliates. All Rights Reserved.
+# SPDX-License-Identifier: Apache-2.0
+
+"""Utilities to deal with the git repo."""
+
+import subprocess
+from fnmatch import fnmatch
+from pathlib import Path
+
+
+def git_repo_files(root: str = ".", glob: str = "*"):
+    """
+    Return a list of files in the git repo from a given path
+
+    :param root: path where to look for files, defaults to the current dir
+    :param glob: what pattern to apply to file names
+    :return: list of found files
+    """
+    files = subprocess.check_output(
+        ["git", "ls-files", root],
+        encoding="ascii",
+    ).splitlines()
+    for file in files:
+        if fnmatch(file, glob):
+            yield Path(file)

tests/integration_tests/style/test_licenses.py

@@ -4,7 +4,7 @@
 
 import datetime
 
-from framework import utils
+from framework import utils_repo
 
 AMAZON_COPYRIGHT_YEARS = range(2018, datetime.datetime.now().year + 1)
 AMAZON_COPYRIGHT = (
@@ -24,8 +24,6 @@
 ALIBABA_COPYRIGHT = "Copyright (C) 2019 Alibaba Cloud Computing. All rights reserved."
 ALIBABA_LICENSE = "SPDX-License-Identifier: Apache-2.0 OR BSD-3-Clause"
 
-EXCLUDE = ["build", ".kernel", ".git"]
-
 
 def _has_amazon_copyright(string):
     for year in AMAZON_COPYRIGHT_YEARS:
@@ -90,16 +88,10 @@ def test_for_valid_licenses():
     """
     Test that all *.py, *.rs and *.sh files contain a valid license.
     """
-    python_files = utils.get_files_from(
-        find_path="..", pattern="*.py", exclude_names=EXCLUDE
-    )
-    rust_files = utils.get_files_from(
-        find_path="..", pattern="*.rs", exclude_names=EXCLUDE
-    )
-    bash_files = utils.get_files_from(
-        find_path="..", pattern="*.sh", exclude_names=EXCLUDE
-    )
 
+    python_files = list(utils_repo.git_repo_files(root="..", glob="*.py"))
+    rust_files = list(utils_repo.git_repo_files(root="..", glob="*.rs"))
+    bash_files = list(utils_repo.git_repo_files(root="..", glob="*.sh"))
     all_files = rust_files + python_files + bash_files
     error_msg = []
     for file in all_files:

tests/integration_tests/style/test_markdown.py

@@ -2,17 +2,15 @@
 # SPDX-License-Identifier: Apache-2.0
 """Tests for markdown style checks."""
 
-from framework import utils
+from framework import utils, utils_repo
 
 
 def test_markdown_style():
     """
     Test that markdown files adhere to the style rules.
     """
     # Get all *.md files from the project
-    md_files = utils.get_files_from(
-        find_path="..", pattern="*.md", exclude_names=["build"]
-    )
+    md_files = list(utils_repo.git_repo_files(root="..", glob="*.md"))
 
     # Assert if somehow no markdown files were found.
     assert len(md_files) != 0

tests/integration_tests/style/test_repo.py

@@ -9,6 +9,8 @@
 
 import yaml
 
+from framework import utils_repo
+
 
 def test_repo_no_spaces_in_paths():
     """
@@ -30,10 +32,7 @@ def test_repo_validate_yaml():
     Ensure all YAML files are valid
     """
 
-    repo_root = Path("..")
-    for path in repo_root.rglob("*.y*ml"):
-        if str(path).startswith("../build/"):
-            continue
+    for path in utils_repo.git_repo_files(root="..", glob="*.y*ml"):
         yaml.safe_load(path.open(encoding="utf-8"))