From 29e33aa07c1fe77d9df04a1c257668c6958bbb1b Mon Sep 17 00:00:00 2001 From: mittei <181688194+mittei@users.noreply.github.com> Date: Fri, 21 Mar 2025 13:14:52 +0000 Subject: [PATCH] Fix jp search (#8) This PR merges the use of lunr search instead of the default mdbook search to enable support of JP characters. --- book.toml | 4 + js/NOTES.md | 34 +++++ js/load-lunr-plugin.js | 69 +++++++++ js/lunr.jp.js | 120 +++++++++++++++ js/lunr.multi.js | 65 ++++++++ js/lunr.stemmer.support.js | 297 +++++++++++++++++++++++++++++++++++++ js/tinyseg.js | 178 ++++++++++++++++++++++ 7 files changed, 767 insertions(+) create mode 100644 js/NOTES.md create mode 100644 js/load-lunr-plugin.js create mode 100644 js/lunr.jp.js create mode 100644 js/lunr.multi.js create mode 100644 js/lunr.stemmer.support.js create mode 100644 js/tinyseg.js diff --git a/book.toml b/book.toml index a0665d7..e5adc19 100644 --- a/book.toml +++ b/book.toml @@ -6,8 +6,12 @@ title = "Yokubi" [output.html] default-theme = "Navy" +additional-js = ["js/lunr.stemmer.support.js", "js/tinyseg.js", "js/lunr.jp.js", "js/lunr.multi.js", "js/load-lunr-plugin.js"] additional-css = ["style/custom.css"] no-section-label = true [output.html.fold] enable = true + +[output.html.search] +use-boolean-and = true # multiple search terms combined with AND opposed to OR diff --git a/js/NOTES.md b/js/NOTES.md new file mode 100644 index 0000000..17b0197 --- /dev/null +++ b/js/NOTES.md @@ -0,0 +1,34 @@ +## elasticlunr.js (included in mdbook output by default) + +- implements indexing and searching for English text +- index + 1. tokenizer: splits the text into tokens (i.e. words) + 1. pipeline: For each token: + - trimmer: removes prefix/suffix non-word characters `token.replace(/^\W+/, '').replace(/\W+$/, '')` (which removes any Japanese characters) + - stopWordFilter: remove common and non-meaningful words like "a", "we" + - stemmer: removes inflection e.g. both "applies" and "apply" become "appli" + 1. builds a prefix tree of the tokens +- search + - applies the same process to the search query, and search the page that contains all/any of the resulting tokens (depending on the `use-boolean-and` option in book.toml) + +## Plugins from https://github.com/weixsong/lunr-languages + +### lunr.jp.js + +- provides stopWordFilter, stemmer for Japanese +- implements a custom tokenizer (which is unusual) + +### tinyseg.js + +- used by lunr.jp.js +- segments Japanese text + +### lunr.multi.js + +- add support for multiple languages +- combines pipeline functions from all specified languages, as well as constructs a new trimmer based on each lang's `wordCharacters` (which lunr.jp.js doesn't have, so we added it) + +### lunr.stemmer.support.js + +- provides `trimmerSupport` and `stemmerSupport` functions +- `trimmerSupport` is used by lunr.multi.js diff --git a/js/load-lunr-plugin.js b/js/load-lunr-plugin.js new file mode 100644 index 0000000..eda2d52 --- /dev/null +++ b/js/load-lunr-plugin.js @@ -0,0 +1,69 @@ +// Modified version of this code: +// https://github.com/rust-lang/mdBook/issues/1081#issuecomment-2030246067 + +// Required for `elasticlunr.multiLanguage()` to work. +// This is something lunr.jp should implement +// but it doesn't. +// copied from: https://github.com/MihaiValentin/lunr-languages/blob/master/lunr.ja.js +elasticlunr.jp.wordCharacters = "一二三四五六七八九十百千万億兆一-龠々〆ヵヶぁ-んァ-ヴーア-ン゙a-zA-Za-zA-Z0-90-9"; + +var re_ENchar = new RegExp("[" + elasticlunr.wordCharacters + "]") + +// Patch load method, which is used by searcher.js. +// This swaps the prebuilt index with a new index +// that supports JP keyword search. +// Since this happens on client side, it may increase load time. +elasticlunr.Index.load = function (index) { + // `index` contains the index built by mdbook during build. + // It also contains all raw page texts so we'll use them + // to rebuild a new index. + + return elasticlunr(function () { + this.use(elasticlunr.multiLanguage('en', 'jp')); + + var en_tokenizer = elasticlunr.tokenizer; + + // first tokenize with the English tokenizer, and then + // only apply the Japanese tokenizer on tokens that + // don't look like English. + // + // Why should we not use just the JP tokenizer? + // Because a query like "they're" would be + // tokenized into "they" and "re" and + // EN stopWordFilter removes "they" and + // you're left with "re" which would match + // any words that start with re. + elasticlunr.tokenizer = function (obj) { + var result = []; + var en_tokens = en_tokenizer(obj); + for (var i = 0; i < en_tokens.length; i++) { + if (en_tokens[i].match(re_ENchar)) { + result.push(en_tokens[i]); + } else { + var jp_tokens = elasticlunr.jp.tokenizer(en_tokens[i]); + result.push.apply(result, jp_tokens); + } + } + return result; + }; + + // At least the `seperator` property is required for + // EN tokenizer to work correctly + for (var prop in en_tokenizer) { + elasticlunr.tokenizer[prop] = en_tokenizer[prop] + } + + // fields to index + this.addField('title'); + this.addField('body'); + this.addField('breadcrumbs'); + + // Identify documents field + this.setRef('id'); + + // Build index of all pages + for (let key in index.documentStore.docs) { + this.addDoc(index.documentStore.docs[key]); + } + }); +}; diff --git a/js/lunr.jp.js b/js/lunr.jp.js new file mode 100644 index 0000000..cebbd28 --- /dev/null +++ b/js/lunr.jp.js @@ -0,0 +1,120 @@ +/* Copied from https://github.com/weixsong/lunr-languages */ + +/*! + * Lunr languages, `Japanese` language + * https://github.com/MihaiValentin/lunr-languages + * + * Copyright 2014, Chad Liu + * http://www.mozilla.org/MPL/ + */ +/*! + * based on + * Snowball JavaScript Library v0.3 + * http://code.google.com/p/urim/ + * http://snowball.tartarus.org/ + * + * Copyright 2010, Oleg Mazko + * http://www.mozilla.org/MPL/ + */ + +/** + * export the module via AMD, CommonJS or as a browser global + * Export code from https://github.com/umdjs/umd/blob/master/returnExports.js + */ +; +(function(root, factory) { + if (typeof define === 'function' && define.amd) { + // AMD. Register as an anonymous module. + define(factory) + } else if (typeof exports === 'object') { + /** + * Node. Does not work with strict CommonJS, but + * only CommonJS-like environments that support module.exports, + * like Node. + */ + module.exports = factory() + } else { + // Browser globals (root is window) + factory()(root.lunr); + } +}(this, function() { + /** + * Just return a value to define the module export. + * This example returns an object, but the module + * can return a function as the exported value. + */ + return function(lunr) { + /* throw error if lunr is not yet included */ + if ('undefined' === typeof lunr) { + throw new Error('Lunr is not present. Please include / require Lunr before this script.'); + } + + /* throw error if lunr stemmer support is not yet included */ + if ('undefined' === typeof lunr.stemmerSupport) { + throw new Error('Lunr stemmer support is not present. Please include / require Lunr stemmer support before this script.'); + } + + /* register specific locale function */ + lunr.jp = function() { + this.pipeline.reset(); + this.pipeline.add( + lunr.jp.stopWordFilter, + lunr.jp.stemmer + ); + // change the tokenizer for japanese one + lunr.tokenizer = lunr.jp.tokenizer; + }; + var segmenter = new TinySegmenter(); // インスタンス生成 + + lunr.jp.tokenizer = function (obj) { + if (!arguments.length || obj == null || obj == undefined) return [] + if (Array.isArray(obj)) return obj.map(function (t) { return t.toLowerCase() }) + + var str = obj.toString().replace(/^\s+/, '') + + for (var i = str.length - 1; i >= 0; i--) { + if (/\S/.test(str.charAt(i))) { + str = str.substring(0, i + 1) + break + } + } + + + var segs = segmenter.segment(str); // 単語の配列が返る + return segs.filter(function (token) { + return !!token + }) + .map(function (token) { + return token + }) + } + + /* lunr stemmer function */ + lunr.jp.stemmer = (function() { + + /* TODO japanese stemmer */ + return function(word) { + return word; + } + })(); + + lunr.Pipeline.registerFunction(lunr.jp.stemmer, 'stemmer-jp'); + + /* stop word filter function */ + lunr.jp.stopWordFilter = function(token) { + if (lunr.jp.stopWordFilter.stopWords.indexOf(token) === -1) { + return token; + } + }; + + lunr.jp.stopWordFilter.stopWords = new lunr.SortedSet(); + lunr.jp.stopWordFilter.stopWords.length = 45; + + // The space at the beginning is crucial: It marks the empty string + // as a stop word. lunr.js crashes during search when documents + // processed by the pipeline still contain the empty string. + // stopword for japanese is from http://www.ranks.nl/stopwords/japanese + lunr.jp.stopWordFilter.stopWords.elements = ' これ それ あれ この その あの ここ そこ あそこ こちら どこ だれ なに なん 何 私 貴方 貴方方 我々 私達 あの人 あのかた 彼女 彼 です あります おります います は が の に を で え から まで より も どの と し それで しかし'.split(' '); + lunr.Pipeline.registerFunction(lunr.jp.stopWordFilter, 'stopWordFilter-jp'); + }; +})) \ No newline at end of file diff --git a/js/lunr.multi.js b/js/lunr.multi.js new file mode 100644 index 0000000..6bdb14d --- /dev/null +++ b/js/lunr.multi.js @@ -0,0 +1,65 @@ +/* Copied from https://github.com/weixsong/lunr-languages */ + +/** + * export the module via AMD, CommonJS or as a browser global + * Export code from https://github.com/umdjs/umd/blob/master/returnExports.js + */ +;(function (root, factory) { + if (typeof define === 'function' && define.amd) { + // AMD. Register as an anonymous module. + define(factory) + } else if (typeof exports === 'object') { + /** + * Node. Does not work with strict CommonJS, but + * only CommonJS-like environments that support module.exports, + * like Node. + */ + module.exports = factory() + } else { + // Browser globals (root is window) + factory()(root.lunr); + } +}(this, function () { + /** + * Just return a value to define the module export. + * This example returns an object, but the module + * can return a function as the exported value. + */ + return function(lunr) { + /* Set up the pipeline for indexing content in multiple languages. The + corresponding lunr.{lang} files must be loaded before calling this + function; English ('en') is built in. + + Returns: a lunr plugin for use in your indexer. + + Known drawback: every word will be stemmed with stemmers for every + language. This could mean that sometimes words that have the same + stemming root will not be stemmed as such. + */ + lunr.multiLanguage = function(/* lang1, lang2, ... */) { + var languages = Array.prototype.slice.call(arguments); + var nameSuffix = languages.join('-'); + var wordCharacters = ""; + var pipeline = []; + for (var i = 0; i < languages.length; ++i) { + if (languages[i] == 'en') { + wordCharacters += '\\w'; + pipeline.unshift(lunr.stopWordFilter); + pipeline.push(lunr.stemmer); + } else { + wordCharacters += lunr[languages[i]].wordCharacters; + pipeline.unshift(lunr[languages[i]].stopWordFilter); + pipeline.push(lunr[languages[i]].stemmer); + } + }; + var multiTrimmer = lunr.trimmerSupport.generateTrimmer(wordCharacters); + lunr.Pipeline.registerFunction(multiTrimmer, 'lunr-multi-trimmer-' + nameSuffix); + pipeline.unshift(multiTrimmer); + + return function() { + this.pipeline.reset(); + this.pipeline.add.apply(this.pipeline, pipeline); + }; + } + } +})); diff --git a/js/lunr.stemmer.support.js b/js/lunr.stemmer.support.js new file mode 100644 index 0000000..a520963 --- /dev/null +++ b/js/lunr.stemmer.support.js @@ -0,0 +1,297 @@ +/* Copied from https://github.com/weixsong/lunr-languages */ + +/*! + * Snowball JavaScript Library v0.3 + * http://code.google.com/p/urim/ + * http://snowball.tartarus.org/ + * + * Copyright 2010, Oleg Mazko + * http://www.mozilla.org/MPL/ + */ + +/** + * export the module via AMD, CommonJS or as a browser global + * Export code from https://github.com/umdjs/umd/blob/master/returnExports.js + */ +;(function (root, factory) { + if (typeof define === 'function' && define.amd) { + // AMD. Register as an anonymous module. + define(factory) + } else if (typeof exports === 'object') { + /** + * Node. Does not work with strict CommonJS, but + * only CommonJS-like environments that support module.exports, + * like Node. + */ + module.exports = factory() + } else { + // Browser globals (root is window) + factory()(root.lunr); + } +}(this, function () { + /** + * Just return a value to define the module export. + * This example returns an object, but the module + * can return a function as the exported value. + */ + return function(lunr) { + /* provides utilities for the included stemmers */ + lunr.stemmerSupport = { + Among: function(s, substring_i, result, method) { + this.toCharArray = function(s) { + var sLength = s.length, charArr = new Array(sLength); + for (var i = 0; i < sLength; i++) + charArr[i] = s.charCodeAt(i); + return charArr; + }; + + if ((!s && s != "") || (!substring_i && (substring_i != 0)) || !result) + throw ("Bad Among initialisation: s:" + s + ", substring_i: " + + substring_i + ", result: " + result); + this.s_size = s.length; + this.s = this.toCharArray(s); + this.substring_i = substring_i; + this.result = result; + this.method = method; + }, + SnowballProgram: function() { + var current; + return { + bra : 0, + ket : 0, + limit : 0, + cursor : 0, + limit_backward : 0, + setCurrent : function(word) { + current = word; + this.cursor = 0; + this.limit = word.length; + this.limit_backward = 0; + this.bra = this.cursor; + this.ket = this.limit; + }, + getCurrent : function() { + var result = current; + current = null; + return result; + }, + in_grouping : function(s, min, max) { + if (this.cursor < this.limit) { + var ch = current.charCodeAt(this.cursor); + if (ch <= max && ch >= min) { + ch -= min; + if (s[ch >> 3] & (0X1 << (ch & 0X7))) { + this.cursor++; + return true; + } + } + } + return false; + }, + in_grouping_b : function(s, min, max) { + if (this.cursor > this.limit_backward) { + var ch = current.charCodeAt(this.cursor - 1); + if (ch <= max && ch >= min) { + ch -= min; + if (s[ch >> 3] & (0X1 << (ch & 0X7))) { + this.cursor--; + return true; + } + } + } + return false; + }, + out_grouping : function(s, min, max) { + if (this.cursor < this.limit) { + var ch = current.charCodeAt(this.cursor); + if (ch > max || ch < min) { + this.cursor++; + return true; + } + ch -= min; + if (!(s[ch >> 3] & (0X1 << (ch & 0X7)))) { + this.cursor++; + return true; + } + } + return false; + }, + out_grouping_b : function(s, min, max) { + if (this.cursor > this.limit_backward) { + var ch = current.charCodeAt(this.cursor - 1); + if (ch > max || ch < min) { + this.cursor--; + return true; + } + ch -= min; + if (!(s[ch >> 3] & (0X1 << (ch & 0X7)))) { + this.cursor--; + return true; + } + } + return false; + }, + eq_s : function(s_size, s) { + if (this.limit - this.cursor < s_size) + return false; + for (var i = 0; i < s_size; i++) + if (current.charCodeAt(this.cursor + i) != s.charCodeAt(i)) + return false; + this.cursor += s_size; + return true; + }, + eq_s_b : function(s_size, s) { + if (this.cursor - this.limit_backward < s_size) + return false; + for (var i = 0; i < s_size; i++) + if (current.charCodeAt(this.cursor - s_size + i) != s + .charCodeAt(i)) + return false; + this.cursor -= s_size; + return true; + }, + find_among : function(v, v_size) { + var i = 0, j = v_size, c = this.cursor, l = this.limit, common_i = 0, common_j = 0, first_key_inspected = false; + while (true) { + var k = i + ((j - i) >> 1), diff = 0, common = common_i < common_j + ? common_i + : common_j, w = v[k]; + for (var i2 = common; i2 < w.s_size; i2++) { + if (c + common == l) { + diff = -1; + break; + } + diff = current.charCodeAt(c + common) - w.s[i2]; + if (diff) + break; + common++; + } + if (diff < 0) { + j = k; + common_j = common; + } else { + i = k; + common_i = common; + } + if (j - i <= 1) { + if (i > 0 || j == i || first_key_inspected) + break; + first_key_inspected = true; + } + } + while (true) { + var w = v[i]; + if (common_i >= w.s_size) { + this.cursor = c + w.s_size; + if (!w.method) + return w.result; + var res = w.method(); + this.cursor = c + w.s_size; + if (res) + return w.result; + } + i = w.substring_i; + if (i < 0) + return 0; + } + }, + find_among_b : function(v, v_size) { + var i = 0, j = v_size, c = this.cursor, lb = this.limit_backward, common_i = 0, common_j = 0, first_key_inspected = false; + while (true) { + var k = i + ((j - i) >> 1), diff = 0, common = common_i < common_j + ? common_i + : common_j, w = v[k]; + for (var i2 = w.s_size - 1 - common; i2 >= 0; i2--) { + if (c - common == lb) { + diff = -1; + break; + } + diff = current.charCodeAt(c - 1 - common) - w.s[i2]; + if (diff) + break; + common++; + } + if (diff < 0) { + j = k; + common_j = common; + } else { + i = k; + common_i = common; + } + if (j - i <= 1) { + if (i > 0 || j == i || first_key_inspected) + break; + first_key_inspected = true; + } + } + while (true) { + var w = v[i]; + if (common_i >= w.s_size) { + this.cursor = c - w.s_size; + if (!w.method) + return w.result; + var res = w.method(); + this.cursor = c - w.s_size; + if (res) + return w.result; + } + i = w.substring_i; + if (i < 0) + return 0; + } + }, + replace_s : function(c_bra, c_ket, s) { + var adjustment = s.length - (c_ket - c_bra), left = current + .substring(0, c_bra), right = current.substring(c_ket); + current = left + s + right; + this.limit += adjustment; + if (this.cursor >= c_ket) + this.cursor += adjustment; + else if (this.cursor > c_bra) + this.cursor = c_bra; + return adjustment; + }, + slice_check : function() { + if (this.bra < 0 || this.bra > this.ket || this.ket > this.limit + || this.limit > current.length) + throw ("faulty slice operation"); + }, + slice_from : function(s) { + this.slice_check(); + this.replace_s(this.bra, this.ket, s); + }, + slice_del : function() { + this.slice_from(""); + }, + insert : function(c_bra, c_ket, s) { + var adjustment = this.replace_s(c_bra, c_ket, s); + if (c_bra <= this.bra) + this.bra += adjustment; + if (c_bra <= this.ket) + this.ket += adjustment; + }, + slice_to : function() { + this.slice_check(); + return current.substring(this.bra, this.ket); + }, + eq_v_b : function(s) { + return this.eq_s_b(s.length, s); + } + }; + } + }; + + lunr.trimmerSupport = { + generateTrimmer: function(wordCharacters) { + var startRegex = new RegExp("^[^" + wordCharacters + "]+") + var endRegex = new RegExp("[^" + wordCharacters + "]+$") + + return function(token) { + return token + .replace(startRegex, '') + .replace(endRegex, ''); + }; + } + } + } +})); diff --git a/js/tinyseg.js b/js/tinyseg.js new file mode 100644 index 0000000..b514415 --- /dev/null +++ b/js/tinyseg.js @@ -0,0 +1,178 @@ +/* Copied from https://github.com/weixsong/lunr-languages */ + +// TinySegmenter 0.1 -- Super compact Japanese tokenizer in Javascript +// (c) 2008 Taku Kudo +// TinySegmenter is freely distributable under the terms of a new BSD licence. +// For details, see http://chasen.org/~taku/software/TinySegmenter/LICENCE.txt + +function TinySegmenter() { + var patterns = { + "[一二三四五六七八九十百千万億兆]":"M", + "[一-龠々〆ヵヶ]":"H", + "[ぁ-ん]":"I", + "[ァ-ヴーア-ン゙ー]":"K", + "[a-zA-Za-zA-Z]":"A", + "[0-90-9]":"N" + } + this.chartype_ = []; + for (var i in patterns) { + var regexp = new RegExp; + regexp.compile(i) + this.chartype_.push([regexp, patterns[i]]); + } + + this.BIAS__ = -332 + this.BC1__ = {"HH":6,"II":2461,"KH":406,"OH":-1378}; + this.BC2__ = {"AA":-3267,"AI":2744,"AN":-878,"HH":-4070,"HM":-1711,"HN":4012,"HO":3761,"IA":1327,"IH":-1184,"II":-1332,"IK":1721,"IO":5492,"KI":3831,"KK":-8741,"MH":-3132,"MK":3334,"OO":-2920}; + this.BC3__ = {"HH":996,"HI":626,"HK":-721,"HN":-1307,"HO":-836,"IH":-301,"KK":2762,"MK":1079,"MM":4034,"OA":-1652,"OH":266}; + this.BP1__ = {"BB":295,"OB":304,"OO":-125,"UB":352}; + this.BP2__ = {"BO":60,"OO":-1762}; + this.BQ1__ = {"BHH":1150,"BHM":1521,"BII":-1158,"BIM":886,"BMH":1208,"BNH":449,"BOH":-91,"BOO":-2597,"OHI":451,"OIH":-296,"OKA":1851,"OKH":-1020,"OKK":904,"OOO":2965}; + this.BQ2__ = {"BHH":118,"BHI":-1159,"BHM":466,"BIH":-919,"BKK":-1720,"BKO":864,"OHH":-1139,"OHM":-181,"OIH":153,"UHI":-1146}; + this.BQ3__ = {"BHH":-792,"BHI":2664,"BII":-299,"BKI":419,"BMH":937,"BMM":8335,"BNN":998,"BOH":775,"OHH":2174,"OHM":439,"OII":280,"OKH":1798,"OKI":-793,"OKO":-2242,"OMH":-2402,"OOO":11699}; + this.BQ4__ = {"BHH":-3895,"BIH":3761,"BII":-4654,"BIK":1348,"BKK":-1806,"BMI":-3385,"BOO":-12396,"OAH":926,"OHH":266,"OHK":-2036,"ONN":-973}; + this.BW1__ = {",と":660,",同":727,"B1あ":1404,"B1同":542,"、と":660,"、同":727,"」と":1682,"あっ":1505,"いう":1743,"いっ":-2055,"いる":672,"うし":-4817,"うん":665,"から":3472,"がら":600,"こう":-790,"こと":2083,"こん":-1262,"さら":-4143,"さん":4573,"した":2641,"して":1104,"すで":-3399,"そこ":1977,"それ":-871,"たち":1122,"ため":601,"った":3463,"つい":-802,"てい":805,"てき":1249,"でき":1127,"です":3445,"では":844,"とい":-4915,"とみ":1922,"どこ":3887,"ない":5713,"なっ":3015,"など":7379,"なん":-1113,"にし":2468,"には":1498,"にも":1671,"に対":-912,"の一":-501,"の中":741,"ませ":2448,"まで":1711,"まま":2600,"まる":-2155,"やむ":-1947,"よっ":-2565,"れた":2369,"れで":-913,"をし":1860,"を見":731,"亡く":-1886,"京都":2558,"取り":-2784,"大き":-2604,"大阪":1497,"平方":-2314,"引き":-1336,"日本":-195,"本当":-2423,"毎日":-2113,"目指":-724,"B1あ":1404,"B1同":542,"」と":1682}; + this.BW2__ = {"..":-11822,"11":-669,"――":-5730,"−−":-13175,"いう":-1609,"うか":2490,"かし":-1350,"かも":-602,"から":-7194,"かれ":4612,"がい":853,"がら":-3198,"きた":1941,"くな":-1597,"こと":-8392,"この":-4193,"させ":4533,"され":13168,"さん":-3977,"しい":-1819,"しか":-545,"した":5078,"して":972,"しな":939,"その":-3744,"たい":-1253,"たた":-662,"ただ":-3857,"たち":-786,"たと":1224,"たは":-939,"った":4589,"って":1647,"っと":-2094,"てい":6144,"てき":3640,"てく":2551,"ては":-3110,"ても":-3065,"でい":2666,"でき":-1528,"でし":-3828,"です":-4761,"でも":-4203,"とい":1890,"とこ":-1746,"とと":-2279,"との":720,"とみ":5168,"とも":-3941,"ない":-2488,"なが":-1313,"など":-6509,"なの":2614,"なん":3099,"にお":-1615,"にし":2748,"にな":2454,"によ":-7236,"に対":-14943,"に従":-4688,"に関":-11388,"のか":2093,"ので":-7059,"のに":-6041,"のの":-6125,"はい":1073,"はが":-1033,"はず":-2532,"ばれ":1813,"まし":-1316,"まで":-6621,"まれ":5409,"めて":-3153,"もい":2230,"もの":-10713,"らか":-944,"らし":-1611,"らに":-1897,"りし":651,"りま":1620,"れた":4270,"れて":849,"れば":4114,"ろう":6067,"われ":7901,"を通":-11877,"んだ":728,"んな":-4115,"一人":602,"一方":-1375,"一日":970,"一部":-1051,"上が":-4479,"会社":-1116,"出て":2163,"分の":-7758,"同党":970,"同日":-913,"大阪":-2471,"委員":-1250,"少な":-1050,"年度":-8669,"年間":-1626,"府県":-2363,"手権":-1982,"新聞":-4066,"日新":-722,"日本":-7068,"日米":3372,"曜日":-601,"朝鮮":-2355,"本人":-2697,"東京":-1543,"然と":-1384,"社会":-1276,"立て":-990,"第に":-1612,"米国":-4268,"11":-669}; + this.BW3__ = {"あた":-2194,"あり":719,"ある":3846,"い.":-1185,"い。":-1185,"いい":5308,"いえ":2079,"いく":3029,"いた":2056,"いっ":1883,"いる":5600,"いわ":1527,"うち":1117,"うと":4798,"えと":1454,"か.":2857,"か。":2857,"かけ":-743,"かっ":-4098,"かに":-669,"から":6520,"かり":-2670,"が,":1816,"が、":1816,"がき":-4855,"がけ":-1127,"がっ":-913,"がら":-4977,"がり":-2064,"きた":1645,"けど":1374,"こと":7397,"この":1542,"ころ":-2757,"さい":-714,"さを":976,"し,":1557,"し、":1557,"しい":-3714,"した":3562,"して":1449,"しな":2608,"しま":1200,"す.":-1310,"す。":-1310,"する":6521,"ず,":3426,"ず、":3426,"ずに":841,"そう":428,"た.":8875,"た。":8875,"たい":-594,"たの":812,"たり":-1183,"たる":-853,"だ.":4098,"だ。":4098,"だっ":1004,"った":-4748,"って":300,"てい":6240,"てお":855,"ても":302,"です":1437,"でに":-1482,"では":2295,"とう":-1387,"とし":2266,"との":541,"とも":-3543,"どう":4664,"ない":1796,"なく":-903,"など":2135,"に,":-1021,"に、":-1021,"にし":1771,"にな":1906,"には":2644,"の,":-724,"の、":-724,"の子":-1000,"は,":1337,"は、":1337,"べき":2181,"まし":1113,"ます":6943,"まっ":-1549,"まで":6154,"まれ":-793,"らし":1479,"られ":6820,"るる":3818,"れ,":854,"れ、":854,"れた":1850,"れて":1375,"れば":-3246,"れる":1091,"われ":-605,"んだ":606,"んで":798,"カ月":990,"会議":860,"入り":1232,"大会":2217,"始め":1681,"市":965,"新聞":-5055,"日,":974,"日、":974,"社会":2024,"カ月":990}; + this.TC1__ = {"AAA":1093,"HHH":1029,"HHM":580,"HII":998,"HOH":-390,"HOM":-331,"IHI":1169,"IOH":-142,"IOI":-1015,"IOM":467,"MMH":187,"OOI":-1832}; + this.TC2__ = {"HHO":2088,"HII":-1023,"HMM":-1154,"IHI":-1965,"KKH":703,"OII":-2649}; + this.TC3__ = {"AAA":-294,"HHH":346,"HHI":-341,"HII":-1088,"HIK":731,"HOH":-1486,"IHH":128,"IHI":-3041,"IHO":-1935,"IIH":-825,"IIM":-1035,"IOI":-542,"KHH":-1216,"KKA":491,"KKH":-1217,"KOK":-1009,"MHH":-2694,"MHM":-457,"MHO":123,"MMH":-471,"NNH":-1689,"NNO":662,"OHO":-3393}; + this.TC4__ = {"HHH":-203,"HHI":1344,"HHK":365,"HHM":-122,"HHN":182,"HHO":669,"HIH":804,"HII":679,"HOH":446,"IHH":695,"IHO":-2324,"IIH":321,"III":1497,"IIO":656,"IOO":54,"KAK":4845,"KKA":3386,"KKK":3065,"MHH":-405,"MHI":201,"MMH":-241,"MMM":661,"MOM":841}; + this.TQ1__ = {"BHHH":-227,"BHHI":316,"BHIH":-132,"BIHH":60,"BIII":1595,"BNHH":-744,"BOHH":225,"BOOO":-908,"OAKK":482,"OHHH":281,"OHIH":249,"OIHI":200,"OIIH":-68}; + this.TQ2__ = {"BIHH":-1401,"BIII":-1033,"BKAK":-543,"BOOO":-5591}; + this.TQ3__ = {"BHHH":478,"BHHM":-1073,"BHIH":222,"BHII":-504,"BIIH":-116,"BIII":-105,"BMHI":-863,"BMHM":-464,"BOMH":620,"OHHH":346,"OHHI":1729,"OHII":997,"OHMH":481,"OIHH":623,"OIIH":1344,"OKAK":2792,"OKHH":587,"OKKA":679,"OOHH":110,"OOII":-685}; + this.TQ4__ = {"BHHH":-721,"BHHM":-3604,"BHII":-966,"BIIH":-607,"BIII":-2181,"OAAA":-2763,"OAKK":180,"OHHH":-294,"OHHI":2446,"OHHO":480,"OHIH":-1573,"OIHH":1935,"OIHI":-493,"OIIH":626,"OIII":-4007,"OKAK":-8156}; + this.TW1__ = {"につい":-4681,"東京都":2026}; + this.TW2__ = {"ある程":-2049,"いった":-1256,"ころが":-2434,"しょう":3873,"その後":-4430,"だって":-1049,"ていた":1833,"として":-4657,"ともに":-4517,"もので":1882,"一気に":-792,"初めて":-1512,"同時に":-8097,"大きな":-1255,"対して":-2721,"社会党":-3216}; + this.TW3__ = {"いただ":-1734,"してい":1314,"として":-4314,"につい":-5483,"にとっ":-5989,"に当た":-6247,"ので,":-727,"ので、":-727,"のもの":-600,"れから":-3752,"十二月":-2287}; + this.TW4__ = {"いう.":8576,"いう。":8576,"からな":-2348,"してい":2958,"たが,":1516,"たが、":1516,"ている":1538,"という":1349,"ました":5543,"ません":1097,"ようと":-4258,"よると":5865}; + this.UC1__ = {"A":484,"K":93,"M":645,"O":-505}; + this.UC2__ = {"A":819,"H":1059,"I":409,"M":3987,"N":5775,"O":646}; + this.UC3__ = {"A":-1370,"I":2311}; + this.UC4__ = {"A":-2643,"H":1809,"I":-1032,"K":-3450,"M":3565,"N":3876,"O":6646}; + this.UC5__ = {"H":313,"I":-1238,"K":-799,"M":539,"O":-831}; + this.UC6__ = {"H":-506,"I":-253,"K":87,"M":247,"O":-387}; + this.UP1__ = {"O":-214}; + this.UP2__ = {"B":69,"O":935}; + this.UP3__ = {"B":189}; + this.UQ1__ = {"BH":21,"BI":-12,"BK":-99,"BN":142,"BO":-56,"OH":-95,"OI":477,"OK":410,"OO":-2422}; + this.UQ2__ = {"BH":216,"BI":113,"OK":1759}; + this.UQ3__ = {"BA":-479,"BH":42,"BI":1913,"BK":-7198,"BM":3160,"BN":6427,"BO":14761,"OI":-827,"ON":-3212}; + this.UW1__ = {",":156,"、":156,"「":-463,"あ":-941,"う":-127,"が":-553,"き":121,"こ":505,"で":-201,"と":-547,"ど":-123,"に":-789,"の":-185,"は":-847,"も":-466,"や":-470,"よ":182,"ら":-292,"り":208,"れ":169,"を":-446,"ん":-137,"・":-135,"主":-402,"京":-268,"区":-912,"午":871,"国":-460,"大":561,"委":729,"市":-411,"日":-141,"理":361,"生":-408,"県":-386,"都":-718,"「":-463,"・":-135}; + this.UW2__ = {",":-829,"、":-829,"〇":892,"「":-645,"」":3145,"あ":-538,"い":505,"う":134,"お":-502,"か":1454,"が":-856,"く":-412,"こ":1141,"さ":878,"ざ":540,"し":1529,"す":-675,"せ":300,"そ":-1011,"た":188,"だ":1837,"つ":-949,"て":-291,"で":-268,"と":-981,"ど":1273,"な":1063,"に":-1764,"の":130,"は":-409,"ひ":-1273,"べ":1261,"ま":600,"も":-1263,"や":-402,"よ":1639,"り":-579,"る":-694,"れ":571,"を":-2516,"ん":2095,"ア":-587,"カ":306,"キ":568,"ッ":831,"三":-758,"不":-2150,"世":-302,"中":-968,"主":-861,"事":492,"人":-123,"会":978,"保":362,"入":548,"初":-3025,"副":-1566,"北":-3414,"区":-422,"大":-1769,"天":-865,"太":-483,"子":-1519,"学":760,"実":1023,"小":-2009,"市":-813,"年":-1060,"強":1067,"手":-1519,"揺":-1033,"政":1522,"文":-1355,"新":-1682,"日":-1815,"明":-1462,"最":-630,"朝":-1843,"本":-1650,"東":-931,"果":-665,"次":-2378,"民":-180,"気":-1740,"理":752,"発":529,"目":-1584,"相":-242,"県":-1165,"立":-763,"第":810,"米":509,"自":-1353,"行":838,"西":-744,"見":-3874,"調":1010,"議":1198,"込":3041,"開":1758,"間":-1257,"「":-645,"」":3145,"ッ":831,"ア":-587,"カ":306,"キ":568}; + this.UW3__ = {",":4889,"1":-800,"−":-1723,"、":4889,"々":-2311,"〇":5827,"」":2670,"〓":-3573,"あ":-2696,"い":1006,"う":2342,"え":1983,"お":-4864,"か":-1163,"が":3271,"く":1004,"け":388,"げ":401,"こ":-3552,"ご":-3116,"さ":-1058,"し":-395,"す":584,"せ":3685,"そ":-5228,"た":842,"ち":-521,"っ":-1444,"つ":-1081,"て":6167,"で":2318,"と":1691,"ど":-899,"な":-2788,"に":2745,"の":4056,"は":4555,"ひ":-2171,"ふ":-1798,"へ":1199,"ほ":-5516,"ま":-4384,"み":-120,"め":1205,"も":2323,"や":-788,"よ":-202,"ら":727,"り":649,"る":5905,"れ":2773,"わ":-1207,"を":6620,"ん":-518,"ア":551,"グ":1319,"ス":874,"ッ":-1350,"ト":521,"ム":1109,"ル":1591,"ロ":2201,"ン":278,"・":-3794,"一":-1619,"下":-1759,"世":-2087,"両":3815,"中":653,"主":-758,"予":-1193,"二":974,"人":2742,"今":792,"他":1889,"以":-1368,"低":811,"何":4265,"作":-361,"保":-2439,"元":4858,"党":3593,"全":1574,"公":-3030,"六":755,"共":-1880,"円":5807,"再":3095,"分":457,"初":2475,"別":1129,"前":2286,"副":4437,"力":365,"動":-949,"務":-1872,"化":1327,"北":-1038,"区":4646,"千":-2309,"午":-783,"協":-1006,"口":483,"右":1233,"各":3588,"合":-241,"同":3906,"和":-837,"員":4513,"国":642,"型":1389,"場":1219,"外":-241,"妻":2016,"学":-1356,"安":-423,"実":-1008,"家":1078,"小":-513,"少":-3102,"州":1155,"市":3197,"平":-1804,"年":2416,"広":-1030,"府":1605,"度":1452,"建":-2352,"当":-3885,"得":1905,"思":-1291,"性":1822,"戸":-488,"指":-3973,"政":-2013,"教":-1479,"数":3222,"文":-1489,"新":1764,"日":2099,"旧":5792,"昨":-661,"時":-1248,"曜":-951,"最":-937,"月":4125,"期":360,"李":3094,"村":364,"東":-805,"核":5156,"森":2438,"業":484,"氏":2613,"民":-1694,"決":-1073,"法":1868,"海":-495,"無":979,"物":461,"特":-3850,"生":-273,"用":914,"町":1215,"的":7313,"直":-1835,"省":792,"県":6293,"知":-1528,"私":4231,"税":401,"立":-960,"第":1201,"米":7767,"系":3066,"約":3663,"級":1384,"統":-4229,"総":1163,"線":1255,"者":6457,"能":725,"自":-2869,"英":785,"見":1044,"調":-562,"財":-733,"費":1777,"車":1835,"軍":1375,"込":-1504,"通":-1136,"選":-681,"郎":1026,"郡":4404,"部":1200,"金":2163,"長":421,"開":-1432,"間":1302,"関":-1282,"雨":2009,"電":-1045,"非":2066,"駅":1620,"1":-800,"」":2670,"・":-3794,"ッ":-1350,"ア":551,"グ":1319,"ス":874,"ト":521,"ム":1109,"ル":1591,"ロ":2201,"ン":278}; + this.UW4__ = {",":3930,".":3508,"―":-4841,"、":3930,"。":3508,"〇":4999,"「":1895,"」":3798,"〓":-5156,"あ":4752,"い":-3435,"う":-640,"え":-2514,"お":2405,"か":530,"が":6006,"き":-4482,"ぎ":-3821,"く":-3788,"け":-4376,"げ":-4734,"こ":2255,"ご":1979,"さ":2864,"し":-843,"じ":-2506,"す":-731,"ず":1251,"せ":181,"そ":4091,"た":5034,"だ":5408,"ち":-3654,"っ":-5882,"つ":-1659,"て":3994,"で":7410,"と":4547,"な":5433,"に":6499,"ぬ":1853,"ね":1413,"の":7396,"は":8578,"ば":1940,"ひ":4249,"び":-4134,"ふ":1345,"へ":6665,"べ":-744,"ほ":1464,"ま":1051,"み":-2082,"む":-882,"め":-5046,"も":4169,"ゃ":-2666,"や":2795,"ょ":-1544,"よ":3351,"ら":-2922,"り":-9726,"る":-14896,"れ":-2613,"ろ":-4570,"わ":-1783,"を":13150,"ん":-2352,"カ":2145,"コ":1789,"セ":1287,"ッ":-724,"ト":-403,"メ":-1635,"ラ":-881,"リ":-541,"ル":-856,"ン":-3637,"・":-4371,"ー":-11870,"一":-2069,"中":2210,"予":782,"事":-190,"井":-1768,"人":1036,"以":544,"会":950,"体":-1286,"作":530,"側":4292,"先":601,"党":-2006,"共":-1212,"内":584,"円":788,"初":1347,"前":1623,"副":3879,"力":-302,"動":-740,"務":-2715,"化":776,"区":4517,"協":1013,"参":1555,"合":-1834,"和":-681,"員":-910,"器":-851,"回":1500,"国":-619,"園":-1200,"地":866,"場":-1410,"塁":-2094,"士":-1413,"多":1067,"大":571,"子":-4802,"学":-1397,"定":-1057,"寺":-809,"小":1910,"屋":-1328,"山":-1500,"島":-2056,"川":-2667,"市":2771,"年":374,"庁":-4556,"後":456,"性":553,"感":916,"所":-1566,"支":856,"改":787,"政":2182,"教":704,"文":522,"方":-856,"日":1798,"時":1829,"最":845,"月":-9066,"木":-485,"来":-442,"校":-360,"業":-1043,"氏":5388,"民":-2716,"気":-910,"沢":-939,"済":-543,"物":-735,"率":672,"球":-1267,"生":-1286,"産":-1101,"田":-2900,"町":1826,"的":2586,"目":922,"省":-3485,"県":2997,"空":-867,"立":-2112,"第":788,"米":2937,"系":786,"約":2171,"経":1146,"統":-1169,"総":940,"線":-994,"署":749,"者":2145,"能":-730,"般":-852,"行":-792,"規":792,"警":-1184,"議":-244,"谷":-1000,"賞":730,"車":-1481,"軍":1158,"輪":-1433,"込":-3370,"近":929,"道":-1291,"選":2596,"郎":-4866,"都":1192,"野":-1100,"銀":-2213,"長":357,"間":-2344,"院":-2297,"際":-2604,"電":-878,"領":-1659,"題":-792,"館":-1984,"首":1749,"高":2120,"「":1895,"」":3798,"・":-4371,"ッ":-724,"ー":-11870,"カ":2145,"コ":1789,"セ":1287,"ト":-403,"メ":-1635,"ラ":-881,"リ":-541,"ル":-856,"ン":-3637}; + this.UW5__ = {",":465,".":-299,"1":-514,"E2":-32768,"]":-2762,"、":465,"。":-299,"「":363,"あ":1655,"い":331,"う":-503,"え":1199,"お":527,"か":647,"が":-421,"き":1624,"ぎ":1971,"く":312,"げ":-983,"さ":-1537,"し":-1371,"す":-852,"だ":-1186,"ち":1093,"っ":52,"つ":921,"て":-18,"で":-850,"と":-127,"ど":1682,"な":-787,"に":-1224,"の":-635,"は":-578,"べ":1001,"み":502,"め":865,"ゃ":3350,"ょ":854,"り":-208,"る":429,"れ":504,"わ":419,"を":-1264,"ん":327,"イ":241,"ル":451,"ン":-343,"中":-871,"京":722,"会":-1153,"党":-654,"務":3519,"区":-901,"告":848,"員":2104,"大":-1296,"学":-548,"定":1785,"嵐":-1304,"市":-2991,"席":921,"年":1763,"思":872,"所":-814,"挙":1618,"新":-1682,"日":218,"月":-4353,"査":932,"格":1356,"機":-1508,"氏":-1347,"田":240,"町":-3912,"的":-3149,"相":1319,"省":-1052,"県":-4003,"研":-997,"社":-278,"空":-813,"統":1955,"者":-2233,"表":663,"語":-1073,"議":1219,"選":-1018,"郎":-368,"長":786,"間":1191,"題":2368,"館":-689,"1":-514,"E2":-32768,"「":363,"イ":241,"ル":451,"ン":-343}; + this.UW6__ = {",":227,".":808,"1":-270,"E1":306,"、":227,"。":808,"あ":-307,"う":189,"か":241,"が":-73,"く":-121,"こ":-200,"じ":1782,"す":383,"た":-428,"っ":573,"て":-1014,"で":101,"と":-105,"な":-253,"に":-149,"の":-417,"は":-236,"も":-206,"り":187,"る":-135,"を":195,"ル":-673,"ン":-496,"一":-277,"中":201,"件":-800,"会":624,"前":302,"区":1792,"員":-1212,"委":798,"学":-960,"市":887,"広":-695,"後":535,"業":-697,"相":753,"社":-507,"福":974,"空":-822,"者":1811,"連":463,"郎":1082,"1":-270,"E1":306,"ル":-673,"ン":-496}; + + return this; +} + +TinySegmenter.prototype.ctype_ = function(str) { + for (var i in this.chartype_) { + if (str.match(this.chartype_[i][0])) { + return this.chartype_[i][1]; + } + } + return "O"; +} + +TinySegmenter.prototype.ts_ = function(v) { + if (v) { return v; } + return 0; +} + +TinySegmenter.prototype.segment = function(input) { + if (input == null || input == undefined || input == "") { + return []; + } + var result = []; + var seg = ["B3","B2","B1"]; + var ctype = ["O","O","O"]; + var o = input.split(""); + for (i = 0; i < o.length; ++i) { + seg.push(o[i]); + ctype.push(this.ctype_(o[i])) + } + seg.push("E1"); + seg.push("E2"); + seg.push("E3"); + ctype.push("O"); + ctype.push("O"); + ctype.push("O"); + var word = seg[3]; + var p1 = "U"; + var p2 = "U"; + var p3 = "U"; + for (var i = 4; i < seg.length - 3; ++i) { + var score = this.BIAS__; + var w1 = seg[i-3]; + var w2 = seg[i-2]; + var w3 = seg[i-1]; + var w4 = seg[i]; + var w5 = seg[i+1]; + var w6 = seg[i+2]; + var c1 = ctype[i-3]; + var c2 = ctype[i-2]; + var c3 = ctype[i-1]; + var c4 = ctype[i]; + var c5 = ctype[i+1]; + var c6 = ctype[i+2]; + score += this.ts_(this.UP1__[p1]); + score += this.ts_(this.UP2__[p2]); + score += this.ts_(this.UP3__[p3]); + score += this.ts_(this.BP1__[p1 + p2]); + score += this.ts_(this.BP2__[p2 + p3]); + score += this.ts_(this.UW1__[w1]); + score += this.ts_(this.UW2__[w2]); + score += this.ts_(this.UW3__[w3]); + score += this.ts_(this.UW4__[w4]); + score += this.ts_(this.UW5__[w5]); + score += this.ts_(this.UW6__[w6]); + score += this.ts_(this.BW1__[w2 + w3]); + score += this.ts_(this.BW2__[w3 + w4]); + score += this.ts_(this.BW3__[w4 + w5]); + score += this.ts_(this.TW1__[w1 + w2 + w3]); + score += this.ts_(this.TW2__[w2 + w3 + w4]); + score += this.ts_(this.TW3__[w3 + w4 + w5]); + score += this.ts_(this.TW4__[w4 + w5 + w6]); + score += this.ts_(this.UC1__[c1]); + score += this.ts_(this.UC2__[c2]); + score += this.ts_(this.UC3__[c3]); + score += this.ts_(this.UC4__[c4]); + score += this.ts_(this.UC5__[c5]); + score += this.ts_(this.UC6__[c6]); + score += this.ts_(this.BC1__[c2 + c3]); + score += this.ts_(this.BC2__[c3 + c4]); + score += this.ts_(this.BC3__[c4 + c5]); + score += this.ts_(this.TC1__[c1 + c2 + c3]); + score += this.ts_(this.TC2__[c2 + c3 + c4]); + score += this.ts_(this.TC3__[c3 + c4 + c5]); + score += this.ts_(this.TC4__[c4 + c5 + c6]); +// score += this.ts_(this.TC5__[c4 + c5 + c6]); + score += this.ts_(this.UQ1__[p1 + c1]); + score += this.ts_(this.UQ2__[p2 + c2]); + score += this.ts_(this.UQ3__[p3 + c3]); + score += this.ts_(this.BQ1__[p2 + c2 + c3]); + score += this.ts_(this.BQ2__[p2 + c3 + c4]); + score += this.ts_(this.BQ3__[p3 + c2 + c3]); + score += this.ts_(this.BQ4__[p3 + c3 + c4]); + score += this.ts_(this.TQ1__[p2 + c1 + c2 + c3]); + score += this.ts_(this.TQ2__[p2 + c2 + c3 + c4]); + score += this.ts_(this.TQ3__[p3 + c1 + c2 + c3]); + score += this.ts_(this.TQ4__[p3 + c2 + c3 + c4]); + var p = "O"; + if (score > 0) { + result.push(word); + word = ""; + p = "B"; + } + p1 = p2; + p2 = p3; + p3 = p; + word += seg[i]; + } + result.push(word); + + return result; +}