HanziJS is a Chinese character and NLP module for Chinese language processing for Node.js. It is primarily written to help provide a framework for Chinese language learners to explore Chinese.
At present features include:
- Character decomposition into components
- Dictionary definition lookup using CC-CEDICT
- Phonetic Regularity Computation
- Example Word Calculations
- Chinese sentence segmentation
Among others. See below all the functions available and how to use them.
npm install hanzi//Require
var hanzi = require("hanzi");
//Initiate
hanzi.start();A function that takes a Chinese character and returns an object with decomposition data. Type of decomposition is optional.
Type of decomposition levels:
- 1 - "Once" (only decomposes character once),
- 2 - "Radical" (decomposes character into its lowest radical components),
- 3 - "Graphical" (decomposes into lowest forms, will be mostly strokes and small indivisable units)
- 4 - "Primitive" (decomposes into meaningful building blocks — 基础部件 — stopping at recognisable components instead of strokes; keeps fused/contained/surrounding shapes whole, preserves independent single strokes, and surfaces a coherent component that has no Unicode glyph as
No glyph available)
var decomposition = hanzi.decompose('爱');
console.log(decomposition);
{ character: '爱',
components1: [ 'No glyph available', '友' ],
components2: [ '爫', '冖', '𠂇', '又' ],
components3: [ '爫', '冖', '𠂇', '㇇', '㇏' ],
components4: [ '爫', '冖', '𠂇', '又' ] }
//Example of forced level decomposition
var decomposition = hanzi.decompose('爱', 2);
console.log(decomposition);
{ character: '爱', components: [ '爫', '冖', '𠂇', '又' ] }The "Primitive" level (4) is designed to stop at meaningful components rather than
over-splitting into strokes the way "Radical" (2) can. For example hanzi.decompose('成', 2)
returns [ '一', '丿', '㇆', '戈' ], while hanzi.decompose('成', 4) returns [ '万', '戈' ].
It stops at a recognisable radical rather than breaking it into strokes — e.g.
hanzi.decompose('当', 4) returns [ '⺌', '彐' ] (not [ '⺌', '㇕', '二' ]) and
hanzi.decompose('所', 4) returns [ '戶', '斤' ]. When part of a glyph is a
coherent shape with no Unicode codepoint, the whole recognisable glyph is kept
rather than shredding it into loose strokes — so hanzi.decompose('售', 4)
returns [ '隹', '口' ] and hanzi.decompose('钱', 4) returns [ '钅', '戋' ].
A function that takes a string of characters and returns one object for all characters.
var decomposition = hanzi.decomposeMany('爱橄黃');
console.log(decomposition);
{ '爱':
{ character: '爱',
components1: [ 'No glyph available', '友' ],
components2: [ '爫', '冖', '𠂇', '又' ],
components3: [ '爫', '冖', '𠂇', '㇇', '㇏' ],
components4: [ '爫', '冖', '𠂇', '又' ] },
'橄':
{ character: '橄',
components1: [ '木', '敢' ],
components2: [ '木', 'No glyph available', '耳', '⺙' ],
components3: [ '一', '丨', '㇒', '㇒', '匚', '一', '一', '丨', '丨', '一', '一', '丿', '一', '㇒', '㇒' ],
components4: [ '木', '匚', '耳', '⺙' ] },
'黃':
{ character: '黃',
components1: [ '廿', 'No glyph available' ],
components2: [ '黃' ],
components3: [ '卄', '一', '一', '一', '一', '丨', '凵', '㇒', '㇒' ],
components4: [ '廿', '一', '由', '八' ] } }Returns a nested decomposition tree for a character (rather than the flat
array hanzi.decompose gives). type is either:
'primitive'(default) — recurses down until it reaches primitives (基础部件), the same stopping points asdecompose(character, 4). Keeps a fused/contained frame whole.'graphical'— continues all the way down to strokes, the same leaves asdecompose(character, 3).
Each node is { character, children }; a leaf has an empty children array. A
coherent component that has no Unicode glyph is surfaced as
{ character: 'No glyph available', id: <number>, children: [] }, where id is
its identifier in the underlying data. The flat level-3 and level-4
decompositions are exactly the in-order leaves of the 'graphical' and
'primitive' trees, so the two views never disagree.
// primitive tree — stops at recognisable building blocks
console.log(hanzi.getDecompositionTree('哀', 'primitive'));
{ character: '哀',
children:
[ { character: '衣', children: [] },
{ character: '口', children: [] } ] }
// a glyph-less stroke-blob is one leaf carrying its data id
console.log(hanzi.getDecompositionTree('师', 'primitive'));
{ character: '师',
children:
[ { character: 'No glyph available', id: '37207', children: [] },
{ character: '帀', children: [] } ] }
// graphical tree — continues down to strokes
console.log(hanzi.getDecompositionTree('大', 'graphical'));
{ character: '大',
children:
[ { character: '人',
children:
[ { character: '㇒', children: [] },
{ character: '㇒', children: [] } ] },
{ character: '一', children: [] } ] }Check if a component/character exists in the data. Returns boolean value.
console.log(hanzi.ifComponentExists('乂'));
true
console.log(hanzi.ifComponentExists('$'));
falseCheck if a character/component is a primitive building block — i.e. whether primitive decomposition (level 4) stops at it rather than breaking it into smaller components. Fused/contained shapes (大, 木, 哀), stroke-cluster units (八, 二) and single strokes count as primitives; compounds do not. Returns a boolean.
console.log(hanzi.isPrimitive('木'));
true
console.log(hanzi.isPrimitive('成')); // decomposes to 万, 戈
falseReturns a dictionary entry object. Script type is optional.
Script type parameters:
- 's' - Simplified
- 't' - Traditional
console.log(hanzi.definitionLookup('雪'));
[ { traditional: '雪',
simplified: '雪',
pinyin: 'Xue3',
definition: 'surname Xue' },
{ traditional: '雪',
simplified: '雪',
pinyin: 'xue3',
definition: 'snow/snowfall/CL:場|场[chang2]/to have the appearance of snow/to wipe away, off or out/to clean' } ]Searches the dictionary based on input. Search type changes what data it returns. Defaults to
Search type paramaters:
- 'only' - this parameter returns only entries with the characters specfied. This is a means to find all compounds words with the characters specified.
- null - returns all occurences of the character.
console.log(hanzi.dictionarySearch('雪'));
[ [ { traditional: '下雪',
simplified: '下雪',
pinyin: 'xia4 xue3',
definition: 'to snow' } ],
[ { traditional: '似雪',
simplified: '似雪',
pinyin: 'si4 xue3',
definition: 'snowy' } ],
[ { traditional: '冰天雪地',
simplified: '冰天雪地',
pinyin: 'bing1 tian1 xue3 di4',
definition: 'a world of ice and snow' } ],
[ { traditional: '冰雪',
simplified: '冰雪',
pinyin: 'bing1 xue3',
definition: 'ice and snow' } ],
[ { traditional: '冰雪皇后',
simplified: '冰雪皇后',
pinyin: 'bing1 xue3 huang2 hou4',
definition: 'Dairy Queen (brand)' } ],
[ { traditional: '冰雪聰明',
simplified: '冰雪聪明',
pinyin: 'bing1 xue3 cong1 ming5',
definition: 'exceptionally intelligent (idiom)' } ],
[ { traditional: '各人自掃門前雪，莫管他家瓦上霜',
simplified: '各人自扫门前雪，莫管他家瓦上霜',
pinyin: 'ge4 ren2 zi4 sao3 men2 qian2 xue3 , mo4 guan3 ta1 jia1 wa3 shang4 shuang1',
definition: 'sweep the snow from your own door step, don\'t worry about the frost on your neighbor\'s roof (idiom)' } ],
[ { traditional: '哈巴雪山',
simplified: '哈巴雪山',
pinyin: 'Ha1 ba1 xue3 shan1',
definition: 'Mt Haba (Nakhi: golden flower), in Lijiang 麗江|丽江, northwest Yunnan' } ],
[ { traditional: '單板滑雪',
simplified: '单板滑雪',
pinyin: 'dan1 ban3 hua2 xue3',
definition: 'to snowboard' } ],
[ { traditional: '報仇雪恥',
simplified: '报仇雪耻',
pinyin: 'bao4 chou2 xue3 chi3',
definition: 'to take revenge and erase humiliation (idiom)' } ],
[....] //Truncated for display purposes
console.log(hanzi.dictionarySearch('心的小孩真', 'only'));
[ [ { traditional: '孩',
simplified: '孩',
pinyin: 'hai2',
definition: 'child' } ],
[ { traditional: '小',
simplified: '小',
pinyin: 'xiao3',
definition: 'small/tiny/few/young' } ],
[ { traditional: '小孩',
simplified: '小孩',
pinyin: 'xiao3 hai2',
definition: 'child/CL:個|个[ge4]' } ],
[ { traditional: '小小',
simplified: '小小',
pinyin: 'xiao3 xiao3',
definition: 'very small/very few/very minor' } ],
[ { traditional: '小心',
simplified: '小心',
pinyin: 'xiao3 xin1',
definition: 'to be careful/to take care' } ],
[ { traditional: '小的',
simplified: '小的',
pinyin: 'xiao3 de5',
definition: 'I (when talking to a superior)' } ],
[ { traditional: '心',
simplified: '心',
pinyin: 'xin1',
definition: 'heart/mind/intention/centre/core/CL:顆|颗[ke1],個|个[ge4]' } ],
[ { traditional: '的',
simplified: '的',
pinyin: 'de5',
definition: 'of/~\'s (possessive particle)/(used after an attribute)/(used to form a nominal expression)/(used at the end of a declarative sentence for emphasis)' },
{ traditional: '的',
simplified: '的',
pinyin: 'di2',
definition: 'really and truly' },
{ traditional: '的',
simplified: '的',
pinyin: 'di4',
definition: 'aim/clear' } ],
[ { traditional: '真',
simplified: '真',
pinyin: 'zhen1',
definition: 'really/truly/indeed/real/true/genuine' } ],
[ { traditional: '真心',
simplified: '真心',
pinyin: 'zhen1 xin1',
definition: 'sincere/heartfelt/CL:片[pian4]' } ],
[ { traditional: '真真',
simplified: '真真',
pinyin: 'zhen1 zhen1',
definition: 'really/in fact/genuinely/scrupulously' } ] ]This function does a dictionarySearch(), then compares that to the Leiden University corpus for vocabulary frequency, then sorts the dictionary entries into three categories in an array: [high frequency, medium frequency and low frequency].
The frequency categories are determined relative to the frequency distribution of the dictionarySearch data compared to the corpus.
console.log(hanzi.getExamples('橄'));
[ [ { traditional: '橄欖',
simplified: '橄榄',
pinyin: 'gan3 lan3',
definition: 'Chinese olive/olive' },
{ traditional: '橄欖油',
simplified: '橄榄油',
pinyin: 'gan3 lan3 you2',
definition: 'olive oil' } ],
[ { traditional: '橄欖球',
simplified: '橄榄球',
pinyin: 'gan3 lan3 qiu2',
definition: 'football played with oval-shaped ball (rugby, American football, Australian rules etc)' } ],
[ { traditional: '橄欖枝',
simplified: '橄榄枝',
pinyin: 'gan3 lan3 zhi1',
definition: 'olive branch/symbol of peace' },
{ traditional: '橄欖樹',
simplified: '橄榄树',
pinyin: 'gan3 lan3 shu4',
definition: 'olive tree' },
{ traditional: '橄欖石',
simplified: '橄榄石',
pinyin: 'gan3 lan3 shi2',
definition: 'olivine (rock-forming mineral magnesium-iron silicate (Mg,Fe)2SiO4)/peridot' } ] ]Returns an array of characters that are segmented based on a longest match lookup.
console.log(hanzi.segment("我們都是陌生人。"));
[ '我們', '都', '是', '陌生人', '。' ]Returns all possible pinyin data for a character.
console.log(hanzi.getPinyin('的'));
[ 'de5', 'di2', 'di4' ]Returns frequency data for a character based on the Junda corpus. The data is in simplified characters, but I made the function script agnostic. So both traditional and simplified will return the same data.
console.log(hanzi.getCharacterFrequency('热'));
{ number: '530',
character: '热',
count: '31190',
percentage: '76.4970999352',
pinyin: 're4',
meaning: 'heat/to heat up/fervent/hot (of weather)/warm up' }Gets a character based on its position the frequency list. This only goes up to 9933 based on the Junda Frequency list.
console.log(hanzi.getCharacterInFrequencyListByPosition(111));
{ number: '111',
character: '机',
count: '339823',
percentage: '43.7756134862',
pinyin: 'ji1',
meaning: 'machine/opportunity/secret' }Returns an array of characters with the given component. If a component has bound forms, such as 手 and 扌, they're considered the same and returns all the characters with the component.
NB: This feature is new. Data might not be hundred percent correct and consistent.
console.log(hanzi.getCharactersWithComponent('囗'));
[ '国','因','西','回','口','四','团','图','围','困',
'固','园','圆','圈','囚','圃','囤','囿','囡','囫',
'圜','囵','囹','圄','囝','圉','圊','釦', /* ...and more */ ]The sample above is truncated for brevity. The full result is longer, because a character is matched whenever 囗 appears at its once- or radical-decomposition level — including characters where 囗 only surfaces after deeper decomposition (e.g. 回 = 囗 within 囗, so 徊, 茴 and 蛔 are matched too).
This function takes a decomposition object created by hanzi.decompose() or a character, then returns an object that displays all possible combinations of phonetic regularity relationship of the character to all its components.
The components it scores are drawn from the whole primitive decomposition tree — every node in the tree, both internal grouping nodes and leaves (deduplicated). This lets it catch a phonetic component that sits at an intermediate node (e.g. 相 inside 想, or 果 inside 裹) rather than only the outermost split.
Phonetic Regularity Scale:
- 0 = No regularity
- 1 = Exact Match (with tone)
- 2 = Syllable Match (without tone)
- 3 = Similar in Initial (alliterates)
- 4 = Similar in Final (rhymes)
The object returned is organized by the possible pronunciations of the character. A component may appear more than once when it has multiple readings (each reading is scored separately). It is up to the developer to use this data or not.
console.log(hanzi.determinePhoneticRegularity('洋'));
{ yang2:
{ character: '洋',
component: [ '氵', '羊', '羊' ],
phoneticpinyin: [ 'shui3', 'Yang2', 'yang2' ],
regularity: [ 0, 1, 1 ] } }
// 果 (guo3) is an internal tree node of 裹, and is still found (exact match):
console.log(hanzi.determinePhoneticRegularity('裹'));
{ guo3:
{ character: '裹',
component: [ '衣', '衣', '果', '田', '田', '木', '木' ],
phoneticpinyin: [ 'yi1', 'yi4', 'guo3', 'Tian2', 'tian2', 'Mu4', 'mu4' ],
regularity: [ 0, 0, 1, 0, 0, 0, 0 ] } }Returns a short, usually one-word, meaning of a radical.
console.log(hanzi.getRadicalMeaning('氵'));
waterHanziJS is used in the following projects:
HanziJS uses data from various sources:
Other data files are either generated by HanziJS or are not in use at present in the software.
The MIT License below thus applies to HanziJS as software, not as data. If you want to use HanziJS commercially please do, but please consider the copyright and license of the data. In short, DO NOT SELL THE DATA! For example, do not sell Leiden's Word Frequency list by making users purchase the list. This goes against the terms.
The MIT License (MIT)
Copyright (c) 2012-2026 Jeremiah Daneil de la Rouviere
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.