Add tests for composite rule annotation script

Author: Kaushik-Kumar-CEG

etc/scripts/dataset_pipeline/test_annotate_composites.py

@@ -0,0 +1,96 @@
+# tests for annotate_composites.py
+import sys
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).parent))
+from annotate_composites import strip_version_suffix, get_candidate_names, find_in_text
+
+
+class TestStripVersionSuffix:
+
+    def test_strips_trailing_version(self):
+        assert strip_version_suffix('GNU General Public License 3.0') == 'GNU General Public License'
+
+    def test_strips_version_with_v_prefix(self):
+        assert strip_version_suffix('Mozilla Public License v2.0') == 'Mozilla Public License'
+
+    def test_strips_or_later(self):
+        result = strip_version_suffix('GNU Lesser General Public License 2.1 or later')
+        assert result == 'GNU Lesser General Public License'
+
+    def test_returns_none_when_result_too_short(self):
+        assert strip_version_suffix('EPL 1.0') is None
+
+    def test_returns_none_when_no_version(self):
+        assert strip_version_suffix('Boost Software License') is None
+
+
+class FakeLicense:
+    """minimal mock for License objects"""
+    def __init__(self, key, name=None, short_name=None, spdx_license_key=None):
+        self.key = key
+        self.name = name
+        self.short_name = short_name
+        self.spdx_license_key = spdx_license_key
+
+
+class TestGetCandidateNames:
+
+    def test_collects_all_fields(self):
+        lic = FakeLicense(
+            key='apache-2.0',
+            name='Apache License 2.0',
+            short_name='Apache 2.0',
+            spdx_license_key='Apache-2.0',
+        )
+        names = get_candidate_names(lic)
+        assert 'Apache License 2.0' in names
+        assert 'Apache 2.0' in names
+        assert 'Apache-2.0' in names
+        assert 'apache-2.0' in names
+
+    def test_includes_extra_names_for_known_keys(self):
+        lic = FakeLicense(key='gpl-2.0', name='GNU General Public License 2.0')
+        names = get_candidate_names(lic)
+        assert 'GPLv2' in names
+        assert 'GPL 2.0' in names
+
+    def test_includes_version_stripped_base(self):
+        lic = FakeLicense(key='agpl-3.0', name='GNU Affero General Public License 3.0')
+        names = get_candidate_names(lic)
+        assert 'GNU Affero General Public License' in names
+
+    def test_sorted_longest_first(self):
+        lic = FakeLicense(key='mit', name='MIT License', short_name='MIT')
+        names = get_candidate_names(lic)
+        for i in range(len(names) - 1):
+            assert len(names[i]) >= len(names[i + 1])
+
+    def test_no_duplicates(self):
+        lic = FakeLicense(key='isc', name='ISC License', short_name='ISC License',
+                          spdx_license_key='ISC')
+        names = get_candidate_names(lic)
+        assert len(names) == len(set(names))
+
+
+class TestFindInText:
+
+    def test_basic_match(self):
+        text = 'Licensed under the Apache License, Version 2.0'
+        assert find_in_text(text, ['Apache License']) == 'Apache License'
+
+    def test_case_insensitive(self):
+        text = 'distributed under the gnu general public license'
+        result = find_in_text(text, ['GNU General Public License'])
+        assert result == 'gnu general public license'
+
+    def test_not_found(self):
+        text = 'this software uses the BSD license'
+        assert find_in_text(text, ['Apache License', 'MIT License']) is None
+
+    def test_skips_very_short_candidates(self):
+        assert find_in_text('released under XY terms', ['XY']) is None
+
+    def test_longer_candidate_matched_first(self):
+        text = 'Released under the MIT License'
+        assert find_in_text(text, ['MIT License', 'MIT']) == 'MIT License'