1  import { describe, it, expect } from 'vitest';
 2  import { groupTranscriptSegments, formatGroupedTranscript } from './transcript-group.js';
 3  describe('groupTranscriptSegments', () => {
 4      it('groups segments by sentence boundaries', () => {
 5          const segments = [
 6              { start: 0, text: 'Hello there.' },
 7              { start: 2, text: 'How are you doing today?' },
 8              { start: 5, text: 'I am' },
 9              { start: 6, text: 'doing well.' },
10          ];
11          const result = groupTranscriptSegments(segments);
12          expect(result).toHaveLength(3);
13          expect(result[0].text).toBe('Hello there.');
14          expect(result[1].text).toBe('How are you doing today?');
15          expect(result[2].text).toBe('I am doing well.');
16      });
17      it('flushes on large time gaps', () => {
18          const segments = [
19              { start: 0, text: 'First part' },
20              { start: 2, text: 'still first' },
21              { start: 25, text: 'second part after gap' },
22          ];
23          const result = groupTranscriptSegments(segments);
24          expect(result).toHaveLength(2);
25          expect(result[0].text).toBe('First part still first');
26          expect(result[1].text).toBe('second part after gap');
27      });
28      it('respects 30s max group span for unpunctuated text', () => {
29          // Simulate CJK captions without punctuation
30          const segments = Array.from({ length: 20 }, (_, i) => ({
31              start: i * 2,
32              text: `segment${i}`,
33          }));
34          const result = groupTranscriptSegments(segments);
35          // 20 segments * 2s = 40s total, should be split into at least 2 groups
36          expect(result.length).toBeGreaterThanOrEqual(2);
37          // No single group should span more than ~30s
38          for (const g of result) {
39              const words = g.text.split(' ');
40              // With 2s per segment and 30s max, each group should have at most ~16 segments
41              expect(words.length).toBeLessThanOrEqual(16);
42          }
43      });
44      it('detects speaker changes via >> markers', () => {
45          const segments = [
46              { start: 0, text: '>> How are you?' },
47              { start: 3, text: '>> I am fine.' },
48          ];
49          const result = groupTranscriptSegments(segments);
50          expect(result.some(g => g.speakerChange)).toBe(true);
51          expect(result.some(g => g.speaker !== undefined)).toBe(true);
52      });
53      it('recognizes CJK sentence-ending punctuation', () => {
54          const segments = [
55              { start: 0, text: '你好世界。' },
56              { start: 2, text: '这是测试' },
57              { start: 4, text: '内容。' },
58          ];
59          const result = groupTranscriptSegments(segments);
60          expect(result).toHaveLength(2);
61          expect(result[0].text).toBe('你好世界。');
62          expect(result[1].text).toBe('这是测试 内容。');
63      });
64      it('returns empty array for empty input', () => {
65          expect(groupTranscriptSegments([])).toEqual([]);
66      });
67  });
68  describe('formatGroupedTranscript', () => {
69      it('formats timestamps correctly', () => {
70          const segments = [
71              { start: 65, text: 'One minute five.', speakerChange: false },
72              { start: 3661, text: 'One hour one minute.', speakerChange: false },
73          ];
74          const { rows } = formatGroupedTranscript(segments);
75          expect(rows[0].timestamp).toBe('1:05');
76          expect(rows[1].timestamp).toBe('1:01:01');
77      });
78      it('inserts chapter headings at correct positions', () => {
79          const segments = [
80              { start: 0, text: 'Intro text.', speakerChange: false },
81              { start: 60, text: 'Chapter content.', speakerChange: false },
82          ];
83          const chapters = [{ title: 'Introduction', start: 0 }, { title: 'Main', start: 50 }];
84          const { rows } = formatGroupedTranscript(segments, chapters);
85          expect(rows[0].text).toBe('[Chapter] Introduction');
86          expect(rows[1].text).toBe('Intro text.');
87          expect(rows[2].text).toBe('[Chapter] Main');
88          expect(rows[3].text).toBe('Chapter content.');
89      });
90      it('labels speakers', () => {
91          const segments = [
92              { start: 0, text: 'Hello.', speakerChange: true, speaker: 0 },
93              { start: 5, text: 'Hi there.', speakerChange: true, speaker: 1 },
94          ];
95          const { rows } = formatGroupedTranscript(segments);
96          expect(rows[0].speaker).toBe('Speaker 1');
97          expect(rows[1].speaker).toBe('Speaker 2');
98      });
99  });