@chaffjs/lang-ja 0.13.0 → 0.15.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/analyser-pieces.d.ts +6 -0
- package/dist/analyser-pieces.d.ts.map +1 -0
- package/dist/analyser-pieces.js +37 -0
- package/dist/analyser-pieces.js.map +1 -0
- package/dist/citation.d.ts +3 -1
- package/dist/citation.d.ts.map +1 -1
- package/dist/citation.js +8 -3
- package/dist/citation.js.map +1 -1
- package/dist/counter-tsu.d.ts +5 -0
- package/dist/counter-tsu.d.ts.map +1 -1
- package/dist/counter-tsu.js.map +1 -1
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +12 -8
- package/dist/index.js.map +1 -1
- package/dist/lexicons.d.ts.map +1 -1
- package/dist/lexicons.js +5 -6
- package/dist/lexicons.js.map +1 -1
- package/dist/particle-after.d.ts +7 -0
- package/dist/particle-after.d.ts.map +1 -0
- package/dist/particle-after.js +9 -0
- package/dist/particle-after.js.map +1 -0
- package/dist/passive-reading.d.ts +29 -0
- package/dist/passive-reading.d.ts.map +1 -0
- package/dist/passive-reading.js +143 -0
- package/dist/passive-reading.js.map +1 -0
- package/dist/pos.d.ts +4 -0
- package/dist/pos.d.ts.map +1 -1
- package/dist/pos.js +75 -19
- package/dist/pos.js.map +1 -1
- package/dist/reduplication.d.ts +33 -1
- package/dist/reduplication.d.ts.map +1 -1
- package/dist/reduplication.js +112 -1
- package/dist/reduplication.js.map +1 -1
- package/dist/sentence-split.js +1 -1
- package/dist/sentence-split.js.map +1 -1
- package/dist/structure.d.ts.map +1 -1
- package/dist/structure.js +28 -13
- package/dist/structure.js.map +1 -1
- package/dist/tokens-within.d.ts +7 -0
- package/dist/tokens-within.d.ts.map +1 -0
- package/dist/tokens-within.js +29 -0
- package/dist/tokens-within.js.map +1 -0
- package/dist/well-formed.d.ts +3 -0
- package/dist/well-formed.d.ts.map +1 -0
- package/dist/well-formed.js +3 -0
- package/dist/well-formed.js.map +1 -0
- package/lexicons/calendar-unit.yaml +8 -0
- package/lexicons/calendar-year-unit.yaml +7 -0
- package/lexicons/common-acronym.yaml +4 -0
- package/lexicons/date-time-unit.yaml +12 -0
- package/lexicons/distributive-noun.yaml +2 -1
- package/lexicons/doubled-word.yaml +6 -0
- package/lexicons/emphasis-word.yaml +1 -0
- package/lexicons/excessive-hedging.yaml +9 -5
- package/lexicons/hedge-frame.yaml +14 -0
- package/lexicons/hedge-scope.yaml +7 -0
- package/lexicons/honorific-formula.yaml +10 -0
- package/lexicons/honorific.yaml +15 -0
- package/lexicons/http-method.yaml +14 -0
- package/lexicons/intransitive-verb.yaml +18 -0
- package/lexicons/iteration-kanji.yaml +18 -0
- package/lexicons/name-joiner.yaml +5 -0
- package/lexicons/naming-verb.yaml +9 -0
- package/lexicons/quantity-noun.yaml +42 -0
- package/lexicons/self-prefix.yaml +2 -1
- package/lexicons/spontaneous-verb.yaml +18 -0
- package/lexicons/stative-passive-verb.yaml +44 -0
- package/package.json +1 -1
- package/src/analyser-pieces.ts +38 -0
- package/src/citation.ts +11 -3
- package/src/counter-tsu.ts +5 -0
- package/src/index.ts +10 -11
- package/src/lexicons.ts +9 -8
- package/src/particle-after.ts +11 -0
- package/src/passive-reading.ts +195 -0
- package/src/pos.ts +84 -19
- package/src/reduplication.ts +175 -2
- package/src/sentence-split.ts +1 -1
- package/src/structure.ts +29 -14
- package/src/tokens-within.ts +27 -0
- package/src/well-formed.ts +2 -0
package/src/reduplication.ts
CHANGED
|
@@ -26,7 +26,180 @@ const isParticleAfter = (next: Token | undefined, second: Token, particles: Read
|
|
|
26
26
|
|
|
27
27
|
const echoed = (token: Token): Token => ({ ...token, features: { ...token.features, Echo: "Rdp" } });
|
|
28
28
|
|
|
29
|
-
|
|
29
|
+
const isDistributive = (tokens: readonly Token[], index: number, vocabulary: Distributive): boolean => {
|
|
30
|
+
const token = tokens[index];
|
|
31
|
+
return token !== undefined && isSameNoun(tokens[index - 1], token, vocabulary.nouns) && isParticleAfter(tokens[index + 1], token, vocabulary.particles);
|
|
32
|
+
};
|
|
33
|
+
|
|
34
|
+
/** 文字列を解析器が一語の副詞と読むか。解析器を持つ index.ts が渡す。 */
|
|
35
|
+
export type ReadsAsAdverb = (text: string) => boolean;
|
|
36
|
+
|
|
37
|
+
/** 一字の漢字に 々 を付けて畳語になるか(家 → 家々)。解析器の辞書と語彙表を持つ index.ts が渡す。 */
|
|
38
|
+
export type TakesIterationMark = (kanji: string) => boolean;
|
|
39
|
+
|
|
40
|
+
/**
|
|
41
|
+
* 解析器の辞書が 々 付きの一語として持つ漢字(家々・国々)と、辞書に無い畳語の漢字(朝々・神々)の語彙表。
|
|
42
|
+
* readsAsOneWord は解析器を持つ index.ts が渡す。
|
|
43
|
+
*/
|
|
44
|
+
export const iterationMarkReading = (lexicons: Readonly<Record<string, Lexicon>>, readsAsOneWord: (text: string) => boolean): TakesIterationMark => {
|
|
45
|
+
const listed = patternsOf(lexicons["iteration-kanji"]);
|
|
46
|
+
return (kanji) => listed.has(kanji) || readsAsOneWord(`${kanji}々`);
|
|
47
|
+
};
|
|
48
|
+
|
|
49
|
+
const CONTENT_WORD = new Set(["NOUN", "VERB", "ADJ"]);
|
|
50
|
+
|
|
51
|
+
/** 助動詞が付く語。 */
|
|
52
|
+
const PREDICATE = new Set(["VERB", "ADJ", "AUX"]);
|
|
53
|
+
|
|
54
|
+
/**
|
|
55
|
+
* 助動詞の重なりは、前に付く述語が無ければ解析器が仮名の擬音を切った片割れ(たんたらたら → たん・たら・たら)。
|
|
56
|
+
* 述語に付いた重なり(行ったたので・着いたらたら)は書き損じ。
|
|
57
|
+
*/
|
|
58
|
+
const isStrandedAuxiliary = (before: Token | undefined, first: Token): boolean => first.pos === "AUX" && (before === undefined || !PREDICATE.has(before.pos));
|
|
59
|
+
|
|
60
|
+
/**
|
|
61
|
+
* 重ね言葉の副詞(がんがん)は、前の語に引かれると同じ名詞二つに切られる(労働者ががんがん → が・がん・がん)。
|
|
62
|
+
* 二語を続けて読み直して一語の副詞なら、重ね言葉。機能語の重なりは、付く先の無い助動詞のほかは読み直さない。
|
|
63
|
+
* 書き損じの「行ったたので」の「たた」も一語の副詞に読める。
|
|
64
|
+
*/
|
|
65
|
+
const isSplitAdverb = (tokens: readonly Token[], index: number, readsAsAdverb: ReadsAsAdverb): boolean => {
|
|
66
|
+
const [before, first, second] = [tokens[index - 2], tokens[index - 1], tokens[index]];
|
|
67
|
+
return (
|
|
68
|
+
first !== undefined &&
|
|
69
|
+
second !== undefined &&
|
|
70
|
+
first.surface === second.surface &&
|
|
71
|
+
first.pos === second.pos &&
|
|
72
|
+
(CONTENT_WORD.has(second.pos) || isStrandedAuxiliary(before, first)) &&
|
|
73
|
+
first.span.end === second.span.start &&
|
|
74
|
+
readsAsAdverb(`${first.surface}${second.surface}`)
|
|
75
|
+
);
|
|
76
|
+
};
|
|
77
|
+
|
|
78
|
+
const ONE_KANJI = /^\p{Script=Han}$/u;
|
|
79
|
+
|
|
80
|
+
const isKanjiNoun = (token: Token | undefined): token is Token =>
|
|
81
|
+
token !== undefined && token.pos === "NOUN" && ONE_KANJI.test(token.surface) && token.features?.["NumType"] !== "Card";
|
|
82
|
+
|
|
83
|
+
/**
|
|
84
|
+
* 一字の漢字の名詞を 々 を使わずに重ねた畳語(家家・朝朝)。々 を付けて畳語になる漢字だけ。
|
|
85
|
+
* 略称を重ねた「法法」は 法々 と書けないので、書き損じのまま。
|
|
86
|
+
*/
|
|
87
|
+
const isIteratedKanji = (tokens: readonly Token[], index: number, takesIterationMark: TakesIterationMark): boolean => {
|
|
88
|
+
const [first, second] = [tokens[index - 1], tokens[index]];
|
|
89
|
+
return (
|
|
90
|
+
isKanjiNoun(first) && isKanjiNoun(second) && first.surface === second.surface && first.span.end === second.span.start && takesIterationMark(second.surface)
|
|
91
|
+
);
|
|
92
|
+
};
|
|
93
|
+
|
|
94
|
+
export const markReduplication = (
|
|
95
|
+
tokens: readonly Token[],
|
|
96
|
+
vocabulary: Distributive,
|
|
97
|
+
readsAsAdverb: ReadsAsAdverb,
|
|
98
|
+
takesIterationMark: TakesIterationMark,
|
|
99
|
+
): Token[] =>
|
|
30
100
|
tokens.map((token, index) =>
|
|
31
|
-
|
|
101
|
+
isDistributive(tokens, index, vocabulary) || isSplitAdverb(tokens, index, readsAsAdverb) || isIteratedKanji(tokens, index, takesIterationMark)
|
|
102
|
+
? echoed(token)
|
|
103
|
+
: token,
|
|
32
104
|
);
|
|
105
|
+
|
|
106
|
+
/** 解析器が読んだ一語。pos は IPADIC の品詞、detail はその細分類、form は活用形、conjugation は活用型(活用しない語はどちらも *)。 */
|
|
107
|
+
export type Inflection = {
|
|
108
|
+
readonly surface: string;
|
|
109
|
+
readonly pos: string;
|
|
110
|
+
readonly detail: string;
|
|
111
|
+
readonly form: string;
|
|
112
|
+
readonly conjugation: string;
|
|
113
|
+
readonly start: number;
|
|
114
|
+
};
|
|
115
|
+
|
|
116
|
+
/** 文を終えずに後ろへ続く形。連用形(連用テ接続・連用タ接続を含む)と命令形。 */
|
|
117
|
+
const CONTINUING_FORM = /^(?:連用|命令)/u;
|
|
118
|
+
|
|
119
|
+
/** 一文字の語(し・い)は重ねて強める形にならない。「確認ししました」は書き損じ。 */
|
|
120
|
+
const MIN_ECHO_LENGTH = 2;
|
|
121
|
+
|
|
122
|
+
/** IPADIC で細分類が「自立」なのは動詞と形容詞だけで、活用形を持つ。 */
|
|
123
|
+
const continuesAsWord = (word: Inflection | undefined): word is Inflection =>
|
|
124
|
+
word !== undefined && word.detail === "自立" && word.surface.length >= MIN_ECHO_LENGTH && CONTINUING_FORM.test(word.form);
|
|
125
|
+
|
|
126
|
+
const GODAN = /^五段/u;
|
|
127
|
+
|
|
128
|
+
/**
|
|
129
|
+
* 語尾として続く語。助動詞が続く重なり(できできます)は語幹の書き損じ。接続助詞も同じだが、五段動詞の連用形(売り・行き)は
|
|
130
|
+
* 語幹に音を足した形なので、「て」を続けた重ね言葉(売り売りて・行き行きて)になる。一段動詞の連用形(でき)は語幹そのもの。
|
|
131
|
+
*/
|
|
132
|
+
const isEnding = (next: Inflection | undefined, echo: Inflection): boolean =>
|
|
133
|
+
next !== undefined && (next.pos === "助動詞" || (next.detail === "接続助詞" && !GODAN.test(echo.conjugation)));
|
|
134
|
+
|
|
135
|
+
const touches = (first: Inflection, second: Inflection): boolean => first.start + first.surface.length === second.start;
|
|
136
|
+
|
|
137
|
+
/**
|
|
138
|
+
* 自立の動詞・形容詞を、連用形か命令形のまま重ねた形(泣き泣き・売り売り・長く長く・待て待て)は重ね言葉。二つ目に Echo=Rdp。
|
|
139
|
+
* 解析器は一つ目を別の活用に読むことがある(待て待て の一つ目は「待てる」の連用形)ので、形は二つとも続く形であればよい。
|
|
140
|
+
* 終止形の重なり(行く行く)、非自立の語(くださいください)、語尾が続く重なり(できできます)は書き損じのまま。
|
|
141
|
+
*/
|
|
142
|
+
export const isInflectedEcho = (words: readonly Inflection[], index: number): boolean => {
|
|
143
|
+
const [first, second] = [words[index - 1], words[index]];
|
|
144
|
+
return continuesAsWord(first) && continuesAsWord(second) && first.surface === second.surface && touches(first, second) && !isEnding(words[index + 1], second);
|
|
145
|
+
};
|
|
146
|
+
|
|
147
|
+
/** 片仮名の語はたいてい外来語(ユーザー・データ)で、重ねれば書き損じ。 */
|
|
148
|
+
const HIRAGANA = /^[\p{Script=Hiragana}ー]+$/u;
|
|
149
|
+
|
|
150
|
+
/** 解析器が擬音・擬態語の切れ端に当てる品詞。助詞・助動詞の重なり(のの・ですです)は書き損じ。 */
|
|
151
|
+
const FREE_WORD = new Set(["名詞", "動詞", "形容詞"]);
|
|
152
|
+
|
|
153
|
+
/** 接尾語が付く語。 */
|
|
154
|
+
const HOST = new Set(["名詞", "動詞", "形容詞", "助動詞"]);
|
|
155
|
+
|
|
156
|
+
/** 拍に数えない小さい仮名(ちょ・しゃ)。 */
|
|
157
|
+
const SMALL_KANA = /[ぁぃぅぇぉゃゅょゎ]/gu;
|
|
158
|
+
|
|
159
|
+
/** 擬音・擬態語の根は二拍(すう・きし・ちょん)。動詞に読まれた重なりは二拍までが擬音で、長い動詞の重なり(できるできる)は書き損じ。 */
|
|
160
|
+
const MIMETIC_ROOT_MORAE = 2;
|
|
161
|
+
|
|
162
|
+
const moraeOf = (text: string): number => text.length - (text.match(SMALL_KANA)?.length ?? 0);
|
|
163
|
+
|
|
164
|
+
const fitsMimeticRoot = (word: Inflection): boolean => word.pos !== "動詞" || moraeOf(word.surface) <= MIMETIC_ROOT_MORAE;
|
|
165
|
+
|
|
166
|
+
const isKanaWord = (word: Inflection | undefined): word is Inflection =>
|
|
167
|
+
word !== undefined &&
|
|
168
|
+
word.surface.length >= MIN_ECHO_LENGTH &&
|
|
169
|
+
HIRAGANA.test(word.surface) &&
|
|
170
|
+
FREE_WORD.has(word.pos) &&
|
|
171
|
+
word.detail !== "非自立" &&
|
|
172
|
+
fitsMimeticRoot(word);
|
|
173
|
+
|
|
174
|
+
/** 前の語に付いた接尾語(田中さん・見られ)。読点の後ろの「しだい」は付く先が無く、解析器が擬態語を切った片割れ。 */
|
|
175
|
+
const isAttachedSuffix = (before: Inflection | undefined, word: Inflection): boolean =>
|
|
176
|
+
word.detail === "接尾" && before !== undefined && HOST.has(before.pos) && touches(before, word);
|
|
177
|
+
|
|
178
|
+
/**
|
|
179
|
+
* 擬音・擬態語は副詞として立つ。後ろに「と」「に」が続くか、行が終わる(歌の行末。解析器は改行を「記号,空白」と読む)。
|
|
180
|
+
* 句読点の前は、名詞の書き損じ(まとめまとめ、)と見分けられない。
|
|
181
|
+
*/
|
|
182
|
+
const ADVERB_MARK = new Set(["と", "に"]);
|
|
183
|
+
|
|
184
|
+
const endsAsAdverb = (second: Inflection, next: Inflection | undefined): boolean =>
|
|
185
|
+
next === undefined || !touches(second, next) || next.detail === "空白" || (next.pos === "助詞" && ADVERB_MARK.has(next.surface));
|
|
186
|
+
|
|
187
|
+
/**
|
|
188
|
+
* 平仮名だけの語を丸ごと重ねた形(きしきし・ちょんちょん・すうすう・しだいしだい・あはれあはれ)は擬音・擬態語か感動詞。
|
|
189
|
+
* 解析器は辞書に無い擬音を、表層の合う名詞や動詞に切って読む(すう = 吸う)ので、品詞ではなく平仮名であることと、副詞の位置に立つことで見分ける。
|
|
190
|
+
* 前の語に付いた接尾語(田中さんさん)と、副詞の位置に立たない重なり(まとめまとめを・できるできるように)は書き損じのまま。
|
|
191
|
+
*/
|
|
192
|
+
export const isKanaEcho = (words: readonly Inflection[], index: number): boolean => {
|
|
193
|
+
const [before, first, second] = [words[index - 2], words[index - 1], words[index]];
|
|
194
|
+
return (
|
|
195
|
+
isKanaWord(first) &&
|
|
196
|
+
isKanaWord(second) &&
|
|
197
|
+
first.surface === second.surface &&
|
|
198
|
+
touches(first, second) &&
|
|
199
|
+
!isAttachedSuffix(before, first) &&
|
|
200
|
+
endsAsAdverb(second, words[index + 1])
|
|
201
|
+
);
|
|
202
|
+
};
|
|
203
|
+
|
|
204
|
+
/** 活用した語の重なりか、仮名の語の重なり。pos.ts が二つ目に Echo=Rdp を付ける。 */
|
|
205
|
+
export const isEchoAt = (words: readonly Inflection[], index: number): boolean => isInflectedEcho(words, index) || isKanaEcho(words, index);
|
package/src/sentence-split.ts
CHANGED
|
@@ -77,7 +77,7 @@ const indexesOf = (text: string, pattern: RegExp): number[] => [...text.matchAll
|
|
|
77
77
|
const stopEnds = (text: string): number[] => {
|
|
78
78
|
const japanese = [...text.matchAll(JAPANESE_STOPS)].filter((match) => isCJK(text[match.index - 1]));
|
|
79
79
|
const english = [...text.matchAll(ENGLISH_STOP)].filter((match) => !ABBREVIATIONS.has(match[0].toLowerCase()));
|
|
80
|
-
return [...japanese, ...english].map((match) => match.index + match[0].length).
|
|
80
|
+
return [...japanese, ...english].map((match) => match.index + match[0].length).toSorted((a, b) => a - b);
|
|
81
81
|
};
|
|
82
82
|
|
|
83
83
|
const skipSpaces = (text: string, from: number): number => {
|
package/src/structure.ts
CHANGED
|
@@ -4,6 +4,7 @@ import { citationVocabulary, citedDocument } from "./citation.ts";
|
|
|
4
4
|
import { loadLexicons } from "./lexicons.ts";
|
|
5
5
|
import { countedAfter, dates, quantities } from "./quantities.ts";
|
|
6
6
|
import { sectionReferences, sectionVocabulary } from "./section-reference.ts";
|
|
7
|
+
import { startsWithParticle } from "./particle-after.ts";
|
|
7
8
|
|
|
8
9
|
// 契約書・規程・法令の番号の書き方。core は番号の書き方を知らず、ここで読んだものを入れ子にする。
|
|
9
10
|
|
|
@@ -20,7 +21,7 @@ const ARTICLE_RANGE = new RegExp(
|
|
|
20
21
|
const PARAGRAPH = new RegExp(`^${SPACE}*第(?<n>${NUMBER})項(?<rest>(?:${SPACE}|(|\\().*|)$`, "u");
|
|
21
22
|
const ITEM = new RegExp(`^${SPACE}*第(?<n>${NUMBER})号(?<rest>(?:${SPACE}|(|\\().*|)$`, "u");
|
|
22
23
|
/**
|
|
23
|
-
*
|
|
24
|
+
* 法令の項と号は、番号だけを行頭に置き、全角空白で本文と区切る(「2」「一」)。条の中でだけ読み、見出しの行(章の番号)では読まない。
|
|
24
25
|
* 半角数字や漢数字に半角空白が続くだけなら本文として扱う。「3 人で」「一 人で」を項や号にしない。
|
|
25
26
|
*/
|
|
26
27
|
const BARE_PARAGRAPH = new RegExp(`^${SPACE}*(?<n>[0-9]{1,3})\\u3000+(?<rest>\\S.*)$`, "u");
|
|
@@ -43,10 +44,20 @@ const numberOf = (text: string | undefined): string | undefined => {
|
|
|
43
44
|
|
|
44
45
|
const insideArticle = (context: NumberingContext): boolean => context.open.some((open) => open.kind === "article");
|
|
45
46
|
|
|
46
|
-
/**
|
|
47
|
+
/** 条の行に本文があれば、番号の無い第 1 項がその行で始まっている。見出しだけの条(「第1条(目的)」「第1条 総則」)は違う。 */
|
|
48
|
+
const opensFirstParagraph = (open: NumberedLine | undefined): boolean => open?.kind === "article" && open.heading === "" && open.rest !== "";
|
|
49
|
+
|
|
50
|
+
/**
|
|
51
|
+
* 「(2)」は開いている「(1)」の兄弟。無ければ、開いているものより一段深い。
|
|
52
|
+
* ただし本文のある条の直下なら、番号の無い第 1 項の号。規則の「第3条 …次のとおりとする。」「(1) 学士」「2 …」の (1) は、第 2 項の兄弟ではない。
|
|
53
|
+
*/
|
|
47
54
|
const PAREN_LABEL = /^(\d+)$/u;
|
|
48
|
-
const parenDepth = (context: NumberingContext): number =>
|
|
49
|
-
|
|
55
|
+
const parenDepth = (context: NumberingContext): number => {
|
|
56
|
+
const sibling = context.open.findLast((open) => PAREN_LABEL.test(open.label));
|
|
57
|
+
if (sibling !== undefined) return sibling.depth;
|
|
58
|
+
const innermost = context.open.at(-1);
|
|
59
|
+
return opensFirstParagraph(innermost) ? ITEM_DEPTH : (innermost?.depth ?? 0) + 1;
|
|
60
|
+
};
|
|
50
61
|
|
|
51
62
|
/**
|
|
52
63
|
* 条の範囲の行は、最初の条の番地を持ち、並びの位置は範囲の最後まで進める。中の条は番地を持たない。
|
|
@@ -124,11 +135,11 @@ const ITEM_SHAPES: readonly ItemShape[] = [
|
|
|
124
135
|
* 「第十二条第二項」(12.2)がぶつからないようにする。
|
|
125
136
|
*/
|
|
126
137
|
const inFirstParagraph = (context: NumberingContext, depth: number): boolean =>
|
|
127
|
-
depth === ITEM_DEPTH &&
|
|
138
|
+
depth === ITEM_DEPTH && context.open.findLast((open) => open.depth < ITEM_DEPTH)?.kind === "article";
|
|
128
139
|
|
|
129
140
|
const item = (line: string, context: NumberingContext): NumberedLine | undefined =>
|
|
130
141
|
ITEM_SHAPES.reduce<NumberedLine | undefined>((found, shape) => {
|
|
131
|
-
if (found !== undefined || (shape.needsArticle && !insideArticle(context))) return found;
|
|
142
|
+
if (found !== undefined || (shape.needsArticle && (context.isHeading || !insideArticle(context)))) return found;
|
|
132
143
|
const groups = shape.pattern.exec(line)?.groups;
|
|
133
144
|
const number = numberOf(groups?.["n"]);
|
|
134
145
|
if (groups === undefined || number === undefined) return undefined;
|
|
@@ -202,6 +213,9 @@ const definitions = (text: string): Mention[] =>
|
|
|
202
213
|
}),
|
|
203
214
|
);
|
|
204
215
|
|
|
216
|
+
/** 条を数える語。見出しに「1 目的」と番号を振っただけの文書には、この語で数えた条が無い。 */
|
|
217
|
+
const ARTICLE_UNIT = "条";
|
|
218
|
+
|
|
205
219
|
const REFERENCE = new RegExp(`第(?<a>${NUMBER})条(?:の(?<s>${NUMBER}))?(?:第(?<p>${NUMBER})項)?(?:第(?<i>${NUMBER})号)?`, "gu");
|
|
206
220
|
|
|
207
221
|
/**
|
|
@@ -254,12 +268,12 @@ const addressOfReference = (
|
|
|
254
268
|
const main = numberOf(groups["a"]);
|
|
255
269
|
if (main === undefined) return undefined;
|
|
256
270
|
const sub = numberOf(groups["s"]);
|
|
257
|
-
const
|
|
258
|
-
const
|
|
271
|
+
const articleNumber = sub === undefined ? main : `${main}-${sub}`;
|
|
272
|
+
const itemNumber = numberOf(groups["i"]);
|
|
259
273
|
// 「第二条第一号」は第 1 項の号。項を書かずに号を指すのは、項が一つしかない条。
|
|
260
|
-
const paragraph = numberOf(groups["p"]) ?? (
|
|
261
|
-
const target = [
|
|
262
|
-
return { target, fallback: paragraph === "1" &&
|
|
274
|
+
const paragraph = numberOf(groups["p"]) ?? (itemNumber === undefined ? undefined : "1");
|
|
275
|
+
const target = [articleNumber, paragraph, itemNumber].filter((part) => part !== undefined).join(".");
|
|
276
|
+
return { target, fallback: paragraph === "1" && itemNumber === undefined ? articleNumber : undefined };
|
|
263
277
|
};
|
|
264
278
|
|
|
265
279
|
/** 他の文書の名前を読む語と、章・節の番地でない言い方。語彙表から一度だけ作る。 */
|
|
@@ -278,7 +292,7 @@ const articlesOf = (text: string): Mention[] =>
|
|
|
278
292
|
const address = addressOfReference(match.groups ?? {});
|
|
279
293
|
if (address === undefined) return [];
|
|
280
294
|
const { target, fallback } = address;
|
|
281
|
-
const attrs = { target, label: match[0], ...(fallback === undefined ? {} : { fallback }) };
|
|
295
|
+
const attrs = { target, label: match[0], unitWord: ARTICLE_UNIT, ...(fallback === undefined ? {} : { fallback }) };
|
|
282
296
|
return [{ start: match.index, end: match.index + match[0].length, attrs }];
|
|
283
297
|
});
|
|
284
298
|
|
|
@@ -288,7 +302,7 @@ const references = (text: string): Mention[] => {
|
|
|
288
302
|
// 括弧書きの中の参照(「(同法第五十九条において準用する場合を含む。)」)は、外の並びを切らない。並びは括弧の深さごとに持つ。
|
|
289
303
|
const chains = new Map<number, { readonly end: number; readonly document: string | undefined }>();
|
|
290
304
|
return [...articlesOf(text), ...sectionsOf(text)]
|
|
291
|
-
.
|
|
305
|
+
.toSorted((left, right) => left.start - right.start)
|
|
292
306
|
.map((mention) => {
|
|
293
307
|
const depth = depths[mention.start] ?? 0;
|
|
294
308
|
const previous = chains.get(depth);
|
|
@@ -346,7 +360,7 @@ const obligations = (text: string): Mention[] => {
|
|
|
346
360
|
kept.push({ start, end, attrs: { marker, type } });
|
|
347
361
|
});
|
|
348
362
|
});
|
|
349
|
-
return kept.
|
|
363
|
+
return kept.toSorted((left, right) => left.start - right.start);
|
|
350
364
|
};
|
|
351
365
|
|
|
352
366
|
/** 「二十二」「2」を数にする。相対の参照(前二項・前条第二項)を core が読むときに使う。 */
|
|
@@ -360,5 +374,6 @@ export const structure: StructurePatterns = {
|
|
|
360
374
|
quantities: quantitiesOutsideSections,
|
|
361
375
|
dates,
|
|
362
376
|
countedAfter: countedAfterNumber,
|
|
377
|
+
continuesSentence: startsWithParticle,
|
|
363
378
|
number,
|
|
364
379
|
};
|
|
@@ -0,0 +1,27 @@
|
|
|
1
|
+
import type { Span, Token } from "chaffjs/plugin";
|
|
2
|
+
|
|
3
|
+
/** start が from 以上になる最初の位置。tokens は start の昇順。 */
|
|
4
|
+
const firstFrom = (tokens: readonly Token[], from: number): number => {
|
|
5
|
+
let low = 0;
|
|
6
|
+
let high = tokens.length;
|
|
7
|
+
while (low < high) {
|
|
8
|
+
const middle = (low + high) >> 1;
|
|
9
|
+
if ((tokens[middle]?.span.start ?? Number.POSITIVE_INFINITY) < from) low = middle + 1;
|
|
10
|
+
else high = middle;
|
|
11
|
+
}
|
|
12
|
+
return low;
|
|
13
|
+
};
|
|
14
|
+
|
|
15
|
+
/**
|
|
16
|
+
* span に収まる token。tokens は start の昇順(解析器が本文の頭から出した順)。
|
|
17
|
+
* 文ごとに全 token をなめると、空行の無い何千行もの .txt で文の数 × token の数になる。
|
|
18
|
+
*/
|
|
19
|
+
export const tokensWithin = (tokens: readonly Token[], span: Span): Token[] => {
|
|
20
|
+
const found: Token[] = [];
|
|
21
|
+
for (let at = firstFrom(tokens, span.start); at < tokens.length; at += 1) {
|
|
22
|
+
const token = tokens[at];
|
|
23
|
+
if (token === undefined || token.span.start > span.end) break;
|
|
24
|
+
if (token.span.end <= span.end) found.push(token);
|
|
25
|
+
}
|
|
26
|
+
return found;
|
|
27
|
+
};
|