recreate project

src/mosaic.test.ts

@@ -0,0 +1,147 @@
+import { describe, expect, it } from 'bun:test'
+import {
+  generateMosaic,
+  selectKeywords,
+  type KeywordCandidate,
+} from './mosaic'
+
+const SOURCE =
+  '我一直以为记忆会忠实地保存一切，后来才发现，每一次回想都在重新编造过去。'
+
+function retain(
+  candidates: KeywordCandidate[],
+  words: string[],
+): KeywordCandidate[] {
+  return candidates.map((candidate) => ({
+    ...candidate,
+    selected: words.includes(candidate.text),
+  }))
+}
+
+describe('selectKeywords', () => {
+  it('优先提供中文实词候选并排除常见停用词', () => {
+    const candidates = selectKeywords(SOURCE, 14, 1)
+    const words = candidates.map((candidate) => candidate.text)
+
+    expect(words).toContain('记忆')
+    expect(words).toContain('保存')
+    expect(words).toContain('回想')
+    expect(words).toContain('编造')
+    expect(words).not.toContain('一直')
+    expect(words).not.toContain('后来')
+  })
+
+  it('在有候选词时至少自动保留一个', () => {
+    const candidates = selectKeywords('风穿过废弃车站', 5, 1)
+
+    expect(candidates.some((candidate) => candidate.selected)).toBe(true)
+  })
+
+  it('显影 0% 不保留词语，100% 保留全部候选词', () => {
+    const hidden = selectKeywords(SOURCE, 0, 1)
+    const revealed = selectKeywords(SOURCE, 100, 1)
+
+    expect(hidden.some((candidate) => candidate.selected)).toBe(false)
+    expect(revealed.every((candidate) => candidate.selected)).toBe(true)
+  })
+})
+
+describe('generateMosaic', () => {
+  it('保留指定词语并彻底替换原标点', () => {
+    const candidates = retain(selectKeywords(SOURCE, 14, 1), [
+      '记忆',
+      '保存',
+      '回想',
+      '编造',
+      '过去',
+    ])
+    const result = generateMosaic(SOURCE, candidates, {
+      volume: 100,
+      disorder: 58,
+      seed: 42,
+    })
+
+    expect(result.output).toContain('记忆')
+    expect(result.output).toContain('保存')
+    expect(result.output).toContain('回想')
+    expect(result.output).toContain('编造')
+    expect(result.output).toContain('过去')
+    expect(result.output).not.toMatch(/[，。！？；：“”‘’（）【】、]/u)
+  })
+
+  it('将总可见长度维持在设定体量附近', () => {
+    const candidates = selectKeywords(SOURCE, 14, 1)
+    const result = generateMosaic(SOURCE, candidates, {
+      volume: 100,
+      disorder: 58,
+      seed: 42,
+    })
+    const ratio = result.outputLength / result.inputLength
+
+    expect(ratio).toBeGreaterThanOrEqual(0.97)
+    expect(ratio).toBeLessThanOrEqual(1.03)
+  })
+
+  it('支持将体量压缩到 10% 或膨胀到 1000%', () => {
+    const compressed = generateMosaic(SOURCE, [], {
+      volume: 10,
+      disorder: 0,
+      seed: 42,
+    })
+    const expanded = generateMosaic(SOURCE, [], {
+      volume: 1000,
+      disorder: 0,
+      seed: 42,
+    })
+
+    expect(compressed.outputLength).toBe(Math.round(compressed.inputLength * 0.1))
+    expect(expanded.outputLength).toBe(expanded.inputLength * 10)
+  })
+
+  it('不生成可能跳出正常行框的 Unicode 组合符', () => {
+    const outputs = Array.from({ length: 20 }, (_, seed) =>
+      generateMosaic(SOURCE, [], {
+        volume: 1000,
+        disorder: 100,
+        seed,
+      }).output,
+    ).join('')
+
+    expect(outputs).not.toMatch(/\p{Mark}/u)
+  })
+
+  it('相同种子可复现，不同种子会重新侵蚀', () => {
+    const candidates = selectKeywords(SOURCE, 14, 1)
+    const first = generateMosaic(SOURCE, candidates, {
+      volume: 100,
+      disorder: 58,
+      seed: 42,
+    })
+    const repeated = generateMosaic(SOURCE, candidates, {
+      volume: 100,
+      disorder: 58,
+      seed: 42,
+    })
+    const rerolled = generateMosaic(SOURCE, candidates, {
+      volume: 100,
+      disorder: 58,
+      seed: 43,
+    })
+
+    expect(repeated.output).toBe(first.output)
+    expect(rerolled.output).not.toBe(first.output)
+  })
+
+  it('保留换行但替换每一行中的隐藏内容', () => {
+    const source = '记忆，在这里。\n过去，在那里。'
+    const candidates = retain(selectKeywords(source, 20, 1), ['记忆', '过去'])
+    const result = generateMosaic(source, candidates, {
+      volume: 100,
+      disorder: 30,
+      seed: 7,
+    })
+
+    expect(result.output.split('\n')).toHaveLength(2)
+    expect(result.output).not.toMatch(/[，。]/u)
+  })
+})