Awesome

zhconv-rs 中文简繁及地區詞轉換

zhconv-rs converts Chinese text among traditional/simplified scripts or regional variants (e.g. zh-TW <-> zh-CN <-> zh-HK <-> zh-Hans <-> zh-Hant), backed by rulesets from MediaWiki/Wikipedia and OpenCC.

It leverages the Aho-Corasick algorithm for linear time complexity with respect to the length of input text and conversion rules (O(n+m)), processing dozens of MiBs text per second.

🔗 Web app (Wasm): https://zhconv.pages.dev (w/ OpenCC dicts)

⚙️ Cli: cargo install zhconv-cli or check releases.

🦀 Rust crate: cargo add zhconv (check docs for examples)

🐍 Python package w/ wheels: pip install zhconv-rs or pip install zhconv-rs-opencc (w/ OpenCC dicts)

🧩 API demo: https://zhconv.bamboo.workers.dev

# > pip install zhconv_rs
# Convert with builtin rulesets:
from zhconv_rs import zhconv
assert zhconv("天干物燥 小心火烛", "zh-tw") == "天乾物燥 小心火燭"
assert zhconv("霧失樓臺，月迷津渡", "zh-hans") == "雾失楼台，月迷津渡"
assert zhconv("《-{zh-hans:三个火枪手;zh-hant:三劍客;zh-tw:三劍客}-》是亞歷山大·仲馬的作品。", "zh-cn", mediawiki=True) == "《三个火枪手》是亚历山大·仲马的作品。"
assert zhconv("-{H|zh-cn:雾都孤儿;zh-tw:孤雛淚;zh-hk:苦海孤雛;zh-sg:雾都孤儿;zh-mo:苦海孤雛;}-《雾都孤儿》是查尔斯·狄更斯的作品。", "zh-tw", True) == "《孤雛淚》是查爾斯·狄更斯的作品。"

# Convert with custom rules:
from zhconv_rs import make_converter
assert make_converter(None, [("天", "地"), ("水", "火")])("甘肅天水") == "甘肅地火"

import io
convert = make_converter("zh-hans", io.StringIO("䖏 处\n罨畫 掩画")) # or path to rule file
assert convert("秀州西去湖州近 幾䖏樓臺罨畫間") == "秀州西去湖州近 几处楼台掩画间"

JS (Webpack): npm install zhconv or yarn add zhconv (Wasm, instructions)

JS in browser: https://cdn.jsdelivr.net/npm/zhconv-web@latest (Wasm)

<script type="module">
    // Use ES module import syntax to import functionality from the module
    // that we have compiled.
    //
    // Note that the `default` import is an initialization function which
    // will "boot" the module and make it ready to use. Currently browsers
    // don't support natively imported WebAssembly as an ES module, but
    // eventually the manual initialization won't be required!
    import init, { zhconv } from 'https://cdn.jsdelivr.net/npm/zhconv-web@latest/zhconv.js'; // specify a version tag if in prod

    async function run() {
        await init();

        alert(zhconv(prompt("Text to convert to zh-hans:"), "zh-hans"));
    }

    run();
</script>

Supported variants

Note: zh-TW and zh-HK are based on zh-Hant. zh-CN are based on zh-Hans. Currently, zh-MO shares the same rulesets with zh-HK unless additional rules are manually configured; zh-MY and zh-SG shares the same rulesets with zh-CN unless additional rules are manually configured.

Performance

cargo bench on AMD EPYC 7B13 (GitPod) by v0.3:

load/zh2Hant            time:   [4.6368 ms 4.6862 ms 4.7595 ms]
load/zh2Hans            time:   [2.2670 ms 2.2891 ms 2.3138 ms]
load/zh2TW              time:   [4.7115 ms 4.7543 ms 4.8001 ms]
load/zh2HK              time:   [5.4438 ms 5.5474 ms 5.6573 ms]
load/zh2MO              time:   [4.9503 ms 4.9673 ms 4.9850 ms]
load/zh2CN              time:   [3.0809 ms 3.1046 ms 3.1323 ms]
load/zh2SG              time:   [3.0543 ms 3.0637 ms 3.0737 ms]
load/zh2MY              time:   [3.0514 ms 3.0640 ms 3.0787 ms]
zh2CN wikitext basic    time:   [385.95 µs 388.53 µs 391.39 µs]
zh2TW wikitext basic    time:   [393.70 µs 395.16 µs 396.89 µs]
zh2TW wikitext extended time:   [1.5105 ms 1.5186 ms 1.5271 ms]
zh2CN 天乾物燥          time:   [46.970 ns 47.312 ns 47.721 ns]
zh2TW data54k           time:   [200.72 µs 201.54 µs 202.41 µs]
zh2CN data54k           time:   [231.55 µs 232.86 µs 234.30 µs]
zh2Hant data689k        time:   [2.0330 ms 2.0513 ms 2.0745 ms]
zh2TW data689k          time:   [1.9710 ms 1.9790 ms 1.9881 ms]
zh2Hant data3185k       time:   [15.199 ms 15.260 ms 15.332 ms]
zh2TW data3185k         time:   [15.346 ms 15.464 ms 15.629 ms]
zh2TW data55m           time:   [329.54 ms 330.53 ms 331.58 ms]
is_hans data55k         time:   [404.73 µs 407.11 µs 409.59 µs]
infer_variant data55k   time:   [1.0468 ms 1.0515 ms 1.0570 ms]
is_hans data3185k       time:   [22.442 ms 22.589 ms 22.757 ms]
infer_variant data3185k time:   [60.205 ms 60.412 ms 60.627 ms]

load/zh2Hant            time:   [22.074 ms 22.338 ms 22.624 ms]
load/zh2Hans            time:   [2.7913 ms 2.8126 ms 2.8355 ms]
load/zh2TW              time:   [23.068 ms 23.286 ms 23.520 ms]
load/zh2HK              time:   [23.358 ms 23.630 ms 23.929 ms]
load/zh2MO              time:   [23.363 ms 23.627 ms 23.913 ms]
load/zh2CN              time:   [3.6778 ms 3.7222 ms 3.7722 ms]
load/zh2SG              time:   [3.6522 ms 3.6848 ms 3.7202 ms]
load/zh2MY              time:   [3.6642 ms 3.7079 ms 3.7545 ms]
zh2CN wikitext basic    time:   [396.17 µs 402.51 µs 409.36 µs]
zh2TW wikitext basic    time:   [442.16 µs 447.53 µs 453.27 µs]
zh2TW wikitext extended time:   [1.5795 ms 1.6007 ms 1.6233 ms]
zh2CN 天乾物燥          time:   [47.884 ns 48.878 ns 49.953 ns]
zh2TW data54k           time:   [255.25 µs 259.01 µs 262.92 µs]
zh2CN data54k           time:   [233.74 µs 236.99 µs 240.67 µs]
zh2Hant data689k        time:   [3.9696 ms 4.0005 ms 4.0327 ms]
zh2TW data689k          time:   [3.4593 ms 3.4896 ms 3.5203 ms]
zh2Hant data3185k       time:   [27.710 ms 27.955 ms 28.206 ms]
zh2TW data3185k         time:   [30.298 ms 30.858 ms 31.428 ms]
zh2TW data55m           time:   [500.95 ms 515.80 ms 531.34 ms]
is_hans data55k         time:   [461.22 µs 470.99 µs 481.20 µs]
infer_variant data55k   time:   [1.1669 ms 1.1759 ms 1.1852 ms]
is_hans data3185k       time:   [26.609 ms 26.964 ms 27.385 ms]
infer_variant data3185k time:   [74.878 ms 76.262 ms 77.818 ms]

zhconv-rs does not maintain any conversion rulesets/dicts. Instead, it relies on two upstream sources: MediaWiki and OpenCC. These rulesets are merged and compiled into an automaton at compile-time for optimal performance, which means rulesets cannot be dynamically selected at runtime. However, it is possible to load custom rulesets manually.

By default, only MediaWiki rulesets are used. For a Rust project, to enable additional OpenCC rulesets, activate the opencc feature: zhconv = { version = "...", features = [ "opencc" ] }. For a Python project, there are two standalone packages zhconv-rs (w/ MediaWiki rulesets only) and zhconv-rs-opencc (w/ additional OpenCC rulesets) to be installed as needed. For the API on Workers, check worker.yml for instructions on configuring OpenCC rulesets. The web app is always shiped with additional OpenCC rulesets for now.-->

Note: Enabling OpenCC rulesets increases the build size by several MiBs and noticeably impacts performance, even though it still outperforms other implementations.

Limitations

Accuracy

A rule-based converter cannot capture every possible linguistic nuance, resulting in limited accuracy. Besides, the converter employs a leftmost-longest matching strategy, prioritizing to the earliest and longest matches in the text. For instance, if a ruleset includes both 干 -> 幹 and 天干物燥 -> 天乾物燥, the converter would prioritize 天乾物燥 because 天干物燥 gets matched earlier compared to 干 at a later position. This approach generally produces accurate results but may occasionally lead to incorrect conversions.

Wikitext support

While the implementation supports most MediaWiki conversion rules, it is not fully compliant with the original MediaWiki implementation.

For wikitext inputs containing global conversion rules (e.g., -{H|zh-hans:鹿|zh-hant:马}- in MediaWiki syntax), the implementation's time complexity may degrade to O(n*m) in the worst case, where n is the input text length and m is the maximum length of source words in the ruleset. This is equivalent to a brute-force approach.

Credits

Rulesets/Dictionaries: MediaWiki and OpenCC.

References:

• https://github.com/gumblex/zhconv : Python implementation of zhConver{ter,sion}.php.
• https://github.com/BYVoid/OpenCC/ : Widely adopted Chinese converter.
• https://zh.wikipedia.org/wiki/Wikipedia:字詞轉換處理
• https://zh.wikipedia.org/wiki/Help:高级字词转换语法
• https://github.com/wikimedia/mediawiki/blob/master/includes/language/LanguageConverter.php