chore(阶段0): 验证 jieba 替代方案

@node-rs/jieba@2.0.1 在 Bun 1.3.11 下通过验证:NAPI 绑定正常加载,
cut 结果符合预期,1000 次切词 2ms。API 与预期有出入——没有
insertWord/addWord,改用 loadDict 加载自定义词典缓冲区达到同等效果。
结论写入设计文档 7.3 节。

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
2026-08-06 19:48:27 -06:00
parent b4cbfd85b0
commit ec563ef64b
2 changed files with 38 additions and 2 deletions

View File

@@ -0,0 +1,20 @@
#!/usr/bin/env bun
// 验证 @node-rs/jieba 能否替代 Python jieba
// 对照 flowchart/views/admin.py:65,191 的用法add_word + cut
import { Jieba } from "@node-rs/jieba"
import { dict } from "@node-rs/jieba/dict"
const jieba = Jieba.withDict(dict)
const text = "输入两个整数并输出它们的和"
console.log("默认切词:", jieba.cut(text).join(" / "))
// 对应 jieba.add_word(_w, freq=9999)
// 注:@node-rs/jieba@2.0.1 没有 insertWord/addWord 方法(.d.ts 未导出),
// 改用 loadDict 加载一份自定义词条缓冲区,格式同 Python jieba 用户词典:"词 词频"
jieba.loadDict(Buffer.from("两个整数 9999\n"))
console.log("加词后 :", jieba.cut(text).join(" / "))
const t0 = performance.now()
for (let i = 0; i < 1000; i++) jieba.cut(text)
console.log("1000 次切词耗时:", (performance.now() - t0).toFixed(0), "ms")