""" FreeExile Milestone 2 Empirical Challenger Test Suite: Adversarial Verification of Static Linter (tools/lint/check_i18n_hygiene.py). Verifies Negative Baselines and Edge Cases: - Test 1: Injected hardcoded Vietnamese string in UI file -> code 1 under --strict (RULE-1-HARDCODED-VI). - Test 2: Injected missing key in 1 of 9 languages in catalog / mock catalog -> code 1 (RULE-2-DICT-PARITY). - Test 3: Injected dangling data-i18n key in HTML -> code 1 (RULE-3-DANGLING-KEY). - Test 4: Current production codebase -> code 0 (100% clean). - Test 5: Allowlisted seed structures and fallback arguments do NOT trigger false positives. - Test 6: Single-line and multi-line comments do NOT trigger false positives. - Test 7: JSON audit report generation format and data integrity. - Test 8-11: Deep empirical edge-case boundary challenges (path resolution, multiline, braces). """ from __future__ import annotations import json import subprocess import sys import unittest from pathlib import Path PROJECT_ROOT = Path(__file__).resolve().parent.parent.parent LINTER_SCRIPT = PROJECT_ROOT / "tools" / "lint" / "check_i18n_hygiene.py" CHAT_UI_JS = PROJECT_ROOT / "client" / "webapp" / "js" / "ui" / "chat_ui.js" CHAT_CATALOG_JS = PROJECT_ROOT / "client" / "webapp" / "js" / "data" / "chat_i18n_catalog.js" INDEX_HTML = PROJECT_ROOT / "client" / "webapp" / "index.html" def run_linter(*extra_args: str) -> subprocess.CompletedProcess[str]: """Executes check_i18n_hygiene.py via subprocess with utf-8 encoding.""" cmd = [sys.executable, str(LINTER_SCRIPT)] + list(extra_args) return subprocess.run( cmd, capture_output=True, text=True, encoding="utf-8", cwd=str(PROJECT_ROOT), timeout=15.0, ) class TestI18nHygieneLinterAdversarial(unittest.TestCase): """Empirical negative and positive baseline verification of check_i18n_hygiene.py.""" def test_01_negative_baseline_hardcoded_vietnamese_ui_fails_strict(self) -> None: """Test 1: Injected hardcoded Vietnamese string in UI file MUST exit with code 1.""" temp_file = PROJECT_ROOT / "client" / "webapp" / "js" / "ui" / "temp_adversarial_vi.js" try: temp_file.write_text( 'export function renderAlert() { return "Thông báo khẩn cấp"; }\n', encoding="utf-8", ) res = run_linter("--strict", "--target-files", str(temp_file)) self.assertEqual(res.returncode, 1, f"Expected code 1, got {res.returncode}") self.assertIn("RULE-1-HARDCODED-VI", res.stdout) self.assertIn("Hardcoded Vietnamese string detected", res.stdout) self.assertIn("Thông báo khẩn cấp", res.stdout) finally: if temp_file.exists(): temp_file.unlink() def test_02_negative_baseline_missing_key_in_catalog_fails_strict(self) -> None: """Test 2: Injected missing key in 1 of 9 languages MUST trigger parity failure and code 1.""" original_content = CHAT_CATALOG_JS.read_text(encoding="utf-8") try: # Delete one key from English dictionary in chat catalog target_entry = '"channel_world": "World",' self.assertIn(target_entry, original_content, "Target entry missing from catalog") modified = original_content.replace(target_entry, "", 1) CHAT_CATALOG_JS.write_text(modified, encoding="utf-8") res = run_linter("--strict") self.assertEqual(res.returncode, 1, f"Expected code 1, got {res.returncode}") self.assertIn("RULE-2-DICT-PARITY", res.stdout) self.assertIn("missing 1 keys", res.stdout) self.assertIn("channel_world", res.stdout) finally: CHAT_CATALOG_JS.write_text(original_content, encoding="utf-8") def test_03_negative_baseline_dangling_data_i18n_key_in_html_fails_strict(self) -> None: """Test 3: Injected dangling data-i18n key in HTML MUST detect missing key and exit code 1.""" original_html = INDEX_HTML.read_text(encoding="utf-8") dangling_tag = '\n
\n' try: INDEX_HTML.write_text(original_html + dangling_tag, encoding="utf-8") res = run_linter("--strict") self.assertEqual(res.returncode, 1, f"Expected code 1, got {res.returncode}") self.assertIn("RULE-3-DANGLING-KEY", res.stdout) self.assertIn("adversarial_dangling_ghost_key_999", res.stdout) finally: INDEX_HTML.write_text(original_html, encoding="utf-8") def test_04_positive_baseline_production_codebase_passes_strict(self) -> None: """Test 4: Current production codebase MUST pass cleanly with code 0.""" res = run_linter("--strict") self.assertEqual(res.returncode, 0, f"Production codebase failed:\n{res.stdout}\n{res.stderr}") self.assertIn("SUCCESS: 100% i18n hygiene compliance", res.stdout) self.assertIn("[*] Rule 1 (Zero Hardcoded VI Strings) Violations: 0", res.stdout) self.assertIn("[*] Rule 2 (9-Language Parity) Violations : 0", res.stdout) self.assertIn("[*] Rule 3 (Missing / Dangling Keys) Violations : 0", res.stdout) def test_05_allowlisted_seed_structures_and_fallback_params_pass(self) -> None: """Rule 1 MUST allow fallback string arguments and seed dictionaries.""" temp_file = PROJECT_ROOT / "client" / "webapp" / "js" / "ui" / "temp_allowlist_check.js" code = """ export const CHANNELS = { world: { name: "Thế Giới" } }; export const RARITY_INFO = { 1: { name: "Phàm Phẩm" } }; const text = t('my_key', null, 'Dữ liệu dự phòng'); """ try: temp_file.write_text(code, encoding="utf-8") res = run_linter("--strict", "--target-files", str(temp_file)) self.assertEqual(res.returncode, 0, f"False positive on allowlisted code:\n{res.stdout}") self.assertIn("Violations: 0", res.stdout) finally: if temp_file.exists(): temp_file.unlink() def test_06_comments_with_vietnamese_do_not_trigger_rule_1(self) -> None: """Rule 1 MUST ignore single-line and multi-line comments containing Vietnamese.""" temp_file = PROJECT_ROOT / "client" / "webapp" / "js" / "ui" / "temp_comment_check.js" code = """ // Chú thích tiếng Việt một dòng: Thế Giới /* * Chú thích tiếng Việt nhiều dòng: * Bang Hội, Đội Ngũ, Mật Thư */ const key = "chat_btn_send"; """ try: temp_file.write_text(code, encoding="utf-8") res = run_linter("--strict", "--target-files", str(temp_file)) self.assertEqual(res.returncode, 0, f"False positive on comments:\n{res.stdout}") finally: if temp_file.exists(): temp_file.unlink() def test_07_mock_catalog_parity_checker_isolated(self) -> None: """Rule 2 isolated verification on a mock 9-language catalog with extra key in 'vi'.""" sys.path.insert(0, str(PROJECT_ROOT)) from tools.lint.check_i18n_hygiene import check_rule2_dictionary_parity mock_catalog = PROJECT_ROOT / "tests" / "temp_mock_catalog.js" mock_content = """ export const MOCK_CATALOG = { vi: { "key_a": "A", "key_b": "B" }, en: { "key_a": "A" }, zh: { "key_a": "A", "key_b": "B" }, ja: { "key_a": "A", "key_b": "B" }, ko: { "key_a": "A", "key_b": "B" }, th: { "key_a": "A", "key_b": "B" }, de: { "key_a": "A", "key_b": "B" }, ru: { "key_a": "A", "key_b": "B" }, es: { "key_a": "A", "key_b": "B" } }; """ try: mock_catalog.write_text(mock_content, encoding="utf-8") violations = check_rule2_dictionary_parity(mock_catalog, "MOCK_CATALOG") self.assertEqual(len(violations), 1) self.assertIn("Locale 'en' missing 1 keys", violations[0].message) self.assertIn("key_b", violations[0].message) finally: if mock_catalog.exists(): mock_catalog.unlink() def test_08_json_report_export_schema(self) -> None: """Linter with --json-out MUST write valid JSON matching report schema.""" out_json = PROJECT_ROOT / "tests" / "temp_i18n_report.json" try: res = run_linter("--strict", "--json-out", str(out_json)) self.assertEqual(res.returncode, 0) self.assertTrue(out_json.exists(), "JSON report file was not created") data = json.loads(out_json.read_text(encoding="utf-8")) self.assertIn("total_files_scanned", data) self.assertIn("total_errors", data) self.assertIn("rule1", data) self.assertIn("rule2", data) self.assertIn("rule3", data) self.assertEqual(data["total_errors"], 0) finally: if out_json.exists(): out_json.unlink() def test_09_multiline_template_string_blindspot_investigation(self) -> None: """Adversarial stress: documents that multiline template literals bypass line regex.""" temp_file = PROJECT_ROOT / "client" / "webapp" / "js" / "ui" / "temp_multiline.js" # String broken across lines lacks opening/closing quotes on inner line multiline_code = "const unlocalized = `\nTiếng Việt nhiều dòng\n`;\n" try: temp_file.write_text(multiline_code, encoding="utf-8") res = run_linter("--strict", "--target-files", str(temp_file)) # Demonstrates current implementation blind spot: returns 0 because line scan misses it bypassed = (res.returncode == 0) self.assertTrue(bypassed, "Empirical proof of multiline template literal evasion") finally: if temp_file.exists(): temp_file.unlink() def test_10_catalog_brace_in_string_parser_limit_investigation(self) -> None: """Adversarial stress: documents that naive brace matching truncates on '}' in strings.""" sys.path.insert(0, str(PROJECT_ROOT)) from tools.lint.check_i18n_hygiene import parse_catalog_keys mock_catalog = PROJECT_ROOT / "tests" / "temp_brace_catalog.js" mock_content = """ export const MOCK = { vi: { "key1": "Value with closing } brace", "key2": "Second key", "key3": "Third key" } }; """ try: mock_catalog.write_text(mock_content, encoding="utf-8") keys = parse_catalog_keys(mock_catalog) vi_keys = keys.get("vi", set()) # Truncated because '}' in string decrements brace counter prematurely self.assertIn("key1", vi_keys) self.assertNotIn("key2", vi_keys, "Empirical proof of brace truncation limitation") finally: if mock_catalog.exists(): mock_catalog.unlink() if __name__ == "__main__": unittest.main()