| 1 | import copy |
| 2 | from contextlib import contextmanager |
| 3 | import json |
| 4 | from pathlib import Path |
| 5 | import subprocess |
| 6 | import tempfile |
| 7 | import unittest |
| 8 | from unittest.mock import patch |
| 9 | import native_stress |
| 10 | from native_stress import edit_history, native_history, verify_capture |
| 11 | from native_collaboration import reachable_page_text |
| 12 | from document_model import DEFAULT_CONTEXT, EXPORTER |
| 13 | |
| 14 | |
| 15 | class NativeHistoryTests(unittest.TestCase): |
| 16 | def test_conflict_oracle_excludes_unreferenced_history_and_spaces(self): |
| 17 | root = Path(__file__).resolve().parent.parent |
| 18 | with tempfile.TemporaryDirectory() as folder: |
| 19 | output = Path(folder) / 'model' |
| 20 | subprocess.run([EXPORTER, |
| 21 | root / 'corpus/collaboration/round-01/offline/notebook/synthetic.one', output], check=True) |
| 22 | model = json.loads((output / 'document.json').read_text()) |
| 23 | retained = reachable_page_text(model) |
| 24 | self.assertEqual(list(retained.values()), [['Rust edited during the native outage.'], |
| 25 | ['Native edited while disconnected.']]) |
| 26 | main, conflict = retained |
| 27 | current = model['spaces'][main] |
| 28 | revision = current['revisions'][current['contexts'][DEFAULT_CONTEXT]] |
| 29 | current['revisions']['unreferenced'] = copy.deepcopy(revision) |
| 30 | for node in current['revisions']['unreferenced']['nodes'].values(): |
| 31 | if node['kind']['type'] == 'RichText': node['kind']['text'] = 'History is not recovery' |
| 32 | self.assertEqual(reachable_page_text(model), retained) |
| 33 | revision['nodes'][revision['roots']['1']]['spaces'] = [] |
| 34 | self.assertIn(conflict, model['spaces']) |
| 35 | self.assertEqual(reachable_page_text(model), {main: retained[main]}) |
| 36 | page, = revision['nodes'][revision['roots']['1']]['content'] |
| 37 | revision['nodes'][page]['children'] *= 2 |
| 38 | self.assertEqual(reachable_page_text(model), {main: retained[main] * 2}) |
| 39 | |
| 40 | def setUp(self): |
| 41 | self.logs = {} |
| 42 | text = 'Concurrent edits:' |
| 43 | for i in range(2): |
| 44 | token = f' [w{i}:0]' |
| 45 | at = len(text.encode('utf-16-le')) // 2 |
| 46 | self.logs[f'w{i}'] = [{'event': 'ready'}, |
| 47 | {'event': 'intent', 'attempt': 1, 'operation': 0, 'token': token, |
| 48 | 'replacement': token, 'before': text, 'range': [at, at], 'source_transaction': 1}, |
| 49 | {'event': 'commit', 'attempt': 1, 'operation': 0, 'token': token, |
| 50 | 'source_transaction': 1, 'started_us': i * 20, 'finished_us': i * 20 + 10}, |
| 51 | {'event': 'done'}] |
| 52 | text += token |
| 53 | self.logs['r0'] = [{'event': 'ready'}, {'event': 'read', 'text': text, |
| 54 | 'started_us': 40, 'finished_us': 50}, {'event': 'done'}] |
| 55 | |
| 56 | def test_incomplete_capture_requires_explicit_retention_mode(self): |
| 57 | self.logs['w0'].pop() |
| 58 | with self.assertRaisesRegex(AssertionError, 'Incomplete client log'): |
| 59 | edit_history(self.logs, 2) |
| 60 | self.assertEqual(len(edit_history(self.logs, 2, partial=True)[0]), 2) |
| 61 | self.logs['w1'][2]['operation'] = 1 |
| 62 | with self.assertRaisesRegex(AssertionError, 'acknowledgements'): |
| 63 | edit_history(self.logs, 2, partial=True) |
| 64 | self.logs['w1'][2]['operation'] = 0 |
| 65 | self.logs['w1'].pop(2) |
| 66 | with self.assertRaisesRegex(AssertionError, 'reader observed'): |
| 67 | edit_history(self.logs, 2, partial=True) |
| 68 | |
| 69 | def test_counter_renumbering_preserves_one_content_history(self): |
| 70 | commits, text = edit_history(self.logs, 1) |
| 71 | self.assertEqual(len(commits), 2) |
| 72 | self.assertEqual(text, self.logs['r0'][1]['text']) |
| 73 | |
| 74 | def test_branches_unexplained_edits_and_incorrect_observations_fail(self): |
| 75 | for field, value in [('before', 'Concurrent edits:'), ('before', 'unknown'), |
| 76 | ('range', [0, 0]), ('replacement', 'wrong')]: |
| 77 | logs = copy.deepcopy(self.logs) |
| 78 | logs['w1'][1][field] = value |
| 79 | with self.assertRaises(AssertionError): edit_history(logs, 1) |
| 80 | for field, value in [('text', 'Concurrent edits:'), ('text', 'unknown'), ('finished_us', 1)]: |
| 81 | logs = copy.deepcopy(self.logs) |
| 82 | logs['r0'][1][field] = value |
| 83 | with self.assertRaises(AssertionError): edit_history(logs, 1) |
| 84 | |
| 85 | def test_content_branch_and_disconnected_history_are_detected(self): |
| 86 | for before, message in [('Concurrent edits:', 'branched'), ('Concurrent edits: unexplained', 'complete content history')]: |
| 87 | logs = copy.deepcopy(self.logs) |
| 88 | logs['w1'][1]['before'] = before |
| 89 | offset = len(before.encode('utf-16-le')) // 2 |
| 90 | logs['w1'][1]['range'] = [offset, offset] |
| 91 | with self.assertRaisesRegex(AssertionError, message): edit_history(logs, 1) |
| 92 | |
| 93 | def test_replacements_chain_by_content_after_native_counter_renumbering(self): |
| 94 | expected = 'Concurrent edits:' |
| 95 | for i in range(2): |
| 96 | events = self.logs[f'w{i}'] |
| 97 | intent = events[1] |
| 98 | intent['before'] = expected |
| 99 | intent['range'] = [len('Concurrent edits:'), len(expected.encode('utf-16-le')) // 2] |
| 100 | intent['replacement'] = ' café 🦀' + intent['token'] |
| 101 | expected = 'Concurrent edits:' + intent['replacement'] |
| 102 | self.logs['r0'][1]['text'] = expected |
| 103 | commits, text = edit_history(self.logs, 1, edit=True) |
| 104 | self.assertEqual((len(commits), text), (2, expected)) |
| 105 | self.logs['w1'][1]['range'][0] = len('Concurrent edits: café ') + 1 |
| 106 | with self.assertRaises(UnicodeDecodeError): edit_history(self.logs, 1, edit=True) |
| 107 | |
| 108 | def test_missing_acknowledgement_is_not_silently_dropped(self): |
| 109 | self.logs['w1'].pop(2) |
| 110 | with self.assertRaises(AssertionError): edit_history(self.logs, 1) |
| 111 | |
| 112 | def test_native_replacement_history_rejects_lost_edits_and_split_surrogates(self): |
| 113 | events = [{'operation': 0, 'token': ' [n0:0]', 'before': 'Native 0:', |
| 114 | 'range': [9, 9], 'replacement': ' café 🦀 [n0:0]'}, |
| 115 | {'operation': 1, 'token': ' [n0:1]', 'before': 'Native 0: café 🦀 [n0:0]', |
| 116 | 'range': [10, 17], 'replacement': ' café 🦀 [n0:1]'}] |
| 117 | self.assertEqual(native_history(events, 0, 2, True), 'Native 0: café 🦀 [n0:1] [n0:0]') |
| 118 | for field, value in [('before', 'Native 0:'), ('range', [15, 16]), |
| 119 | ('operation', 0), ('replacement', 'unrecorded')]: |
| 120 | changed = copy.deepcopy(events) |
| 121 | changed[1][field] = value |
| 122 | with self.assertRaises((AssertionError, UnicodeDecodeError)): |
| 123 | native_history(changed, 0, 2, True) |
| 124 | with self.assertRaises(AssertionError): native_history(events[:-1], 0, 2, True) |
| 125 | |
| 126 | def test_cold_capture_rejects_extra_missing_and_misformatted_content(self): |
| 127 | with tempfile.TemporaryDirectory() as folder: |
| 128 | root = Path(folder) |
| 129 | (root / 'rust').mkdir() |
| 130 | (root / 'n0').mkdir() |
| 131 | (root / 'run.json').write_text(json.dumps({'stress_operations': 1, 'edit': True, |
| 132 | 'rust_writers': 2, 'rust_readers': 1, 'stress_clients': 1})) |
| 133 | text = 'Concurrent edits:' |
| 134 | for i in range(2): |
| 135 | events = self.logs[f'w{i}'] |
| 136 | events[1].update(before=text, range=[17, len(text.encode('utf-16-le')) // 2], |
| 137 | replacement=' café 🦀' + events[1]['token']) |
| 138 | text = 'Concurrent edits:' + events[1]['replacement'] |
| 139 | self.logs['r0'][1]['text'] = text |
| 140 | for actor, events in self.logs.items(): |
| 141 | (root / 'rust' / f'{actor}.jsonl').write_text('\n'.join(map(json.dumps, events))) |
| 142 | event = {'operation': 0, 'token': ' [n0:0]', 'before': 'Native 0:', |
| 143 | 'range': [9, 9], 'replacement': ' café 🦀 [n0:0]', 'bold': True, 'italic': False} |
| 144 | (root / 'n0/stress-events.jsonl').write_text(json.dumps(event), encoding='utf-8-sig') |
| 145 | native = '<OE style="font-weight:bold"><T>Native 0: café 🦀 [n0:0]</T></OE>' |
| 146 | body = f'<OE><T>{text}</T></OE>{native}' |
| 147 | page = root / 'page-000.xml' |
| 148 | page.write_text(f'<Page xmlns="http://schemas.microsoft.com/office/onenote/2010/onenote"><Outline>{body}</Outline></Page>') |
| 149 | result = verify_capture(root, root) |
| 150 | self.assertEqual((result['rust_intents'], result['native_intents'], result['exact_paragraphs']), (2, 1, 2)) |
| 151 | self.assertGreater(result['native_intended_format_checks'], 0) |
| 152 | for changed in (body + native, body.replace(native, ''), body.replace('bold', 'normal')): |
| 153 | page.write_text(f'<Page xmlns="http://schemas.microsoft.com/office/onenote/2010/onenote"><Outline>{changed}</Outline></Page>') |
| 154 | with self.assertRaises(AssertionError): verify_capture(root, root) |
| 155 | |
| 156 | |
| 157 | class FailureArtifacts(unittest.TestCase): |
| 158 | def test_failed_clients_preserve_native_logs_without_masking_the_failure(self): |
| 159 | @contextmanager |
| 160 | def failed_clients(*args, **kwargs): |
| 161 | yield {} |
| 162 | raise RuntimeError('Rust client exited') |
| 163 | |
| 164 | with tempfile.TemporaryDirectory() as directory: |
| 165 | output = Path(directory) |
| 166 | (output / 'run.json').write_text(json.dumps({'maintenance': False})) |
| 167 | shared = output / 'shared' |
| 168 | shared.mkdir() |
| 169 | clients = [{'name': f'n{i}', 'folder': output / f'n{i}'} for i in range(2)] |
| 170 | for client in clients: client['folder'].mkdir() |
| 171 | |
| 172 | def capture(remote, local, name): |
| 173 | self.assertTrue(remote.endswith('\\outbox\\7\\events.jsonl')) |
| 174 | if name == 'n0': raise OSError('Native client unavailable') |
| 175 | local.write_text('{"operation":0}\n') |
| 176 | return {'error': None} |
| 177 | |
| 178 | with patch.object(native_stress, 'running_clients', failed_clients), \ |
| 179 | patch.object(native_stress.windows, 'do_health', return_value={'utc_us': 0}), \ |
| 180 | patch.object(native_stress.windows, 'do_cmd', return_value={'stdout': 'ready editing'}), \ |
| 181 | patch.object(native_stress.windows, 'do_get', side_effect=capture): |
| 182 | with self.assertRaisesRegex(RuntimeError, 'Rust client exited'): |
| 183 | native_stress.exercise(output, shared, clients, lambda *a, **kw: 7, |
| 184 | lambda *a: None, lambda *a: None, lambda *a: None, 40, 1, 4, 4) |
| 185 | self.assertEqual(json.loads((output / 'n0/stress-capture.json').read_text())['error'], 'Native client unavailable') |
| 186 | self.assertEqual((output / 'n1/stress-events.jsonl').read_text(), '{"operation":0}\n') |