Extend license check to file header check

arthurdejong · arthurdejong · commit b8ee83071e63 · 2023-08-06T15:53:40.000+02:00
This also checks that the file name referenced in the file header is
correct.
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -60,7 +60,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        tox_job: [docs, flake8, license]
+        tox_job: [docs, flake8, headers]
     steps:
       - uses: actions/checkout@v3
       - name: Set up Python
diff --git a/scripts/check_headers.py b/scripts/check_headers.py
@@ -1,6 +1,6 @@
 #!/usr/bin/env python3
 
-# check_license_headers - check that all source files have licensing information
+# check_headers.py - check that all source files have licensing information
 #
 # Copyright (C) 2023 Arthur de Jong
 #
@@ -22,11 +22,19 @@
 """This script checks that all source files have licensing information."""
 
 import glob
+import os
 import re
 import sys
 import textwrap
 
 
+# Regext to match the first line
+identification_re = re.compile(
+    r'^(#!/usr/bin/env python3?\s+|/\*\s+)?'
+    r'(# coding: utf-8\s+)?'
+    r'(\s?#\s+)?'
+    r'(?P<filename>[^ :]+) - ', re.MULTILINE)
+
 # Regex to match standard license blurb
 license_re = re.compile(textwrap.dedent(r'''
     [# ]*This library is free software; you can redistribute it and/or
@@ -46,13 +54,12 @@
     ''').strip(), re.MULTILINE)
 
 
-def file_has_correct_license(filename):
-    """Check that the file contains a valid license header."""
+def get_file_header(filename):
+    """Read the file header from the file."""
     with open(filename, 'rt') as f:
         # Only read the first 2048 bytes to avoid loading too much and the
         # license information should be in the first part anyway.
-        contents = f.read(2048)
-        return bool(license_re.search(contents))
+        return f.read(2048)
 
 
 if __name__ == '__main__':
@@ -66,8 +73,17 @@ def file_has_correct_license(filename):
         glob.glob('online_check/check.js', recursive=True)
     )
 
-    incorrect_files = [f for f in files_to_check if not file_has_correct_license(f)]
-    if incorrect_files:
-        print('Files with incorrect license information:')
-        print('\n'.join(incorrect_files))
+    # Look for files with incorrect first lines or license
+    fail = False
+    for filename in sorted(files_to_check):
+        contents = get_file_header(filename)
+        m = identification_re.match(contents)
+        if not bool(m) or m.group('filename') not in (filename, os.path.basename(filename)):
+            print('%s: Incorrect file identification' % filename)
+            fail = True
+        if not bool(license_re.search(contents)):
+            print('%s: Incorrect license text' % filename)
+            fail = True
+
+    if fail:
         sys.exit(1)
diff --git a/tox.ini b/tox.ini
@@ -1,5 +1,5 @@
 [tox]
-envlist = py{27,35,36,37,38,39,310,311,py,py3},flake8,docs,license
+envlist = py{27,35,36,37,38,39,310,311,py,py3},flake8,docs,headers
 skip_missing_interpreters = true
 
 [testenv]
@@ -35,7 +35,7 @@ use_develop = true
 deps = Sphinx
 commands = sphinx-build -N -b html docs {envtmpdir}/sphinx -W
 
-[testenv:license]
+[testenv:headers]
 skip_install = true
 deps =
-commands = python scripts/check_license_headers.py
+commands = python scripts/check_headers.py
