chaffjs 0.11.0 → 0.13.0
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/date-stamp.d.ts +14 -0
- package/dist/date-stamp.d.ts.map +1 -0
- package/dist/date-stamp.js +40 -0
- package/dist/date-stamp.js.map +1 -0
- package/dist/declared-lists.d.ts +4 -0
- package/dist/declared-lists.d.ts.map +1 -0
- package/dist/declared-lists.js +3 -0
- package/dist/declared-lists.js.map +1 -0
- package/dist/detectors/acronym-context.d.ts +25 -0
- package/dist/detectors/acronym-context.d.ts.map +1 -0
- package/dist/detectors/acronym-context.js +28 -0
- package/dist/detectors/acronym-context.js.map +1 -0
- package/dist/detectors/acronym-expansion.d.ts +9 -0
- package/dist/detectors/acronym-expansion.d.ts.map +1 -0
- package/dist/detectors/acronym-expansion.js +94 -0
- package/dist/detectors/acronym-expansion.js.map +1 -0
- package/dist/detectors/char-shape.d.ts.map +1 -1
- package/dist/detectors/char-shape.js +20 -34
- package/dist/detectors/char-shape.js.map +1 -1
- package/dist/detectors/doubled-word.d.ts +27 -0
- package/dist/detectors/doubled-word.d.ts.map +1 -0
- package/dist/detectors/doubled-word.js +94 -0
- package/dist/detectors/doubled-word.js.map +1 -0
- package/dist/detectors/heading-echo.d.ts.map +1 -1
- package/dist/detectors/heading-echo.js +22 -12
- package/dist/detectors/heading-echo.js.map +1 -1
- package/dist/detectors/index.d.ts.map +1 -1
- package/dist/detectors/index.js +2 -0
- package/dist/detectors/index.js.map +1 -1
- package/dist/detectors/lead-in.d.ts +8 -0
- package/dist/detectors/lead-in.d.ts.map +1 -0
- package/dist/detectors/lead-in.js +13 -0
- package/dist/detectors/lead-in.js.map +1 -0
- package/dist/detectors/lexicon-match.d.ts +6 -0
- package/dist/detectors/lexicon-match.d.ts.map +1 -1
- package/dist/detectors/lexicon-match.js +8 -0
- package/dist/detectors/lexicon-match.js.map +1 -1
- package/dist/detectors/lexicon.d.ts +5 -0
- package/dist/detectors/lexicon.d.ts.map +1 -1
- package/dist/detectors/lexicon.js +21 -4
- package/dist/detectors/lexicon.js.map +1 -1
- package/dist/detectors/orthography.d.ts.map +1 -1
- package/dist/detectors/orthography.js +17 -1
- package/dist/detectors/orthography.js.map +1 -1
- package/dist/detectors/place-run.d.ts +8 -0
- package/dist/detectors/place-run.d.ts.map +1 -0
- package/dist/detectors/place-run.js +54 -0
- package/dist/detectors/place-run.js.map +1 -0
- package/dist/detectors/register.d.ts +32 -0
- package/dist/detectors/register.d.ts.map +1 -0
- package/dist/detectors/register.js +62 -0
- package/dist/detectors/register.js.map +1 -0
- package/dist/detectors/sentence-ending.d.ts.map +1 -1
- package/dist/detectors/sentence-ending.js +17 -28
- package/dist/detectors/sentence-ending.js.map +1 -1
- package/dist/detectors/signals.d.ts.map +1 -1
- package/dist/detectors/signals.js +20 -115
- package/dist/detectors/signals.js.map +1 -1
- package/dist/detectors/structure.d.ts.map +1 -1
- package/dist/detectors/structure.js +14 -1
- package/dist/detectors/structure.js.map +1 -1
- package/dist/detectors/superlative-scope.d.ts +16 -0
- package/dist/detectors/superlative-scope.d.ts.map +1 -0
- package/dist/detectors/superlative-scope.js +40 -0
- package/dist/detectors/superlative-scope.js.map +1 -0
- package/dist/document.d.ts.map +1 -1
- package/dist/document.js +17 -11
- package/dist/document.js.map +1 -1
- package/dist/eval.d.ts.map +1 -1
- package/dist/eval.js +3 -2
- package/dist/eval.js.map +1 -1
- package/dist/heading-title.d.ts +6 -0
- package/dist/heading-title.d.ts.map +1 -0
- package/dist/heading-title.js +8 -0
- package/dist/heading-title.js.map +1 -0
- package/dist/in-page-nav.d.ts +19 -0
- package/dist/in-page-nav.d.ts.map +1 -0
- package/dist/in-page-nav.js +44 -0
- package/dist/in-page-nav.js.map +1 -0
- package/dist/number-name.d.ts +9 -0
- package/dist/number-name.d.ts.map +1 -0
- package/dist/number-name.js +57 -0
- package/dist/number-name.js.map +1 -0
- package/dist/orthography.d.ts +7 -2
- package/dist/orthography.d.ts.map +1 -1
- package/dist/orthography.js +79 -17
- package/dist/orthography.js.map +1 -1
- package/dist/plugin.d.ts +4 -0
- package/dist/plugin.d.ts.map +1 -1
- package/dist/rule-load.d.ts.map +1 -1
- package/dist/rule-load.js +1 -0
- package/dist/rule-load.js.map +1 -1
- package/dist/run.d.ts.map +1 -1
- package/dist/run.js +5 -4
- package/dist/run.js.map +1 -1
- package/package.json +3 -3
- package/rules/doubled-word.yaml +27 -0
- package/rules/latin-spacing.yaml +4 -2
- package/rules/max-kanji-continuous.yaml +2 -0
- package/rules/undefined-acronym.yaml +2 -0
- package/rules/unqualified-superlative.yaml +1 -0
- package/src/date-stamp.ts +65 -0
- package/src/declared-lists.ts +5 -0
- package/src/detectors/acronym-context.ts +50 -0
- package/src/detectors/acronym-expansion.ts +109 -0
- package/src/detectors/char-shape.ts +12 -30
- package/src/detectors/doubled-word.ts +128 -0
- package/src/detectors/heading-echo.ts +13 -3
- package/src/detectors/index.ts +2 -0
- package/src/detectors/lead-in.ts +14 -0
- package/src/detectors/lexicon-match.ts +10 -0
- package/src/detectors/lexicon.ts +32 -6
- package/src/detectors/orthography.ts +20 -2
- package/src/detectors/place-run.ts +63 -0
- package/src/detectors/register.ts +83 -0
- package/src/detectors/sentence-ending.ts +24 -28
- package/src/detectors/signals.ts +24 -120
- package/src/detectors/structure.ts +21 -2
- package/src/detectors/superlative-scope.ts +66 -0
- package/src/document.ts +19 -11
- package/src/eval.ts +2 -1
- package/src/heading-title.ts +8 -0
- package/src/in-page-nav.ts +67 -0
- package/src/number-name.ts +68 -0
- package/src/orthography.ts +77 -15
- package/src/plugin.ts +4 -0
- package/src/rule-load.ts +1 -0
- package/src/run.ts +5 -4
package/dist/detectors/index.js
CHANGED
|
@@ -5,6 +5,7 @@ import { repeatedHead } from "./repeated-head.js";
|
|
|
5
5
|
import { sentenceRhythm } from "./sentence-rhythm.js";
|
|
6
6
|
import { phraseMatch } from "./phrase-match.js";
|
|
7
7
|
import { agentlessPassive } from "./agentless-passive.js";
|
|
8
|
+
import { doubledWord } from "./doubled-word.js";
|
|
8
9
|
import { sentenceEnding } from "./sentence-ending.js";
|
|
9
10
|
import { doubledParticle, nounEnding } from "./token-shape.js";
|
|
10
11
|
import { kanjiRun, middleDot } from "./char-shape.js";
|
|
@@ -27,6 +28,7 @@ export const DETECTORS = {
|
|
|
27
28
|
"sentence-ending": sentenceEnding,
|
|
28
29
|
"noun-ending": nounEnding,
|
|
29
30
|
"doubled-particle": doubledParticle,
|
|
31
|
+
"doubled-word": doubledWord,
|
|
30
32
|
"kanji-run": kanjiRun,
|
|
31
33
|
"middle-dot": middleDot,
|
|
32
34
|
"adverb-density": adverbDensity,
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/detectors/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAClD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC/D,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AACtD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,SAAS,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AACpG,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,cAAc,EAAE,WAAW,EAAE,iBAAiB,EAAE,MAAM,gBAAgB,CAAC;AACpH,OAAO,EAAE,gBAAgB,EAAE,WAAW,EAAE,YAAY,EAAE,eAAe,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAC;AAC9G,OAAO,EAAE,MAAM,EAAE,cAAc,EAAE,cAAc,EAAE,cAAc,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,MAAM,cAAc,CAAC;AACnI,OAAO,EAAE,cAAc,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,WAAW,CAAC;AAChF,OAAO,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAC;AAC/D,OAAO,EAAE,iBAAiB,EAAE,SAAS,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAE1I,kDAAkD;AAClD,MAAM,CAAC,MAAM,SAAS,GAAuC;IAC3D,mBAAmB,EAAE,eAAe;IACpC,iBAAiB,EAAE,cAAc;IACjC,cAAc,EAAE,WAAW;IAC3B,eAAe,EAAE,YAAY;IAC7B,iBAAiB,EAAE,cAAc;IACjC,cAAc,EAAE,WAAW;IAC3B,mBAAmB,EAAE,gBAAgB;IACrC,iBAAiB,EAAE,cAAc;IACjC,aAAa,EAAE,UAAU;IACzB,kBAAkB,EAAE,eAAe;IACnC,WAAW,EAAE,QAAQ;IACrB,YAAY,EAAE,SAAS;IACvB,gBAAgB,EAAE,aAAa;IAC/B,SAAS,EAAE,SAAS;IACpB,iBAAiB,EAAE,cAAc;IACjC,gBAAgB,EAAE,YAAY;IAC9B,cAAc,EAAE,WAAW;IAC3B,kBAAkB,EAAE,eAAe;IACnC,oBAAoB,EAAE,iBAAiB;IACvC,oBAAoB,EAAE,iBAAiB;IACvC,eAAe,EAAE,WAAW;IAC5B,iBAAiB,EAAE,cAAc;IACjC,eAAe,EAAE,YAAY;IAC7B,cAAc,EAAE,WAAW;IAC3B,kBAAkB,EAAE,eAAe;IACnC,mBAAmB,EAAE,gBAAgB;IACrC,mBAAmB,EAAE,gBAAgB;IACrC,iBAAiB,EAAE,cAAc;IACjC,iBAAiB,EAAE,cAAc;IACjC,yBAAyB,EAAE,sBAAsB;IACjD,sBAAsB,EAAE,mBAAmB;IAC3C,SAAS,EAAE,MAAM;IACjB,iBAAiB,EAAE,cAAc;IACjC,iBAAiB,EAAE,cAAc;IACjC,mBAAmB,EAAE,gBAAgB;IACrC,qBAAqB,EAAE,iBAAiB;IACxC,gBAAgB,EAAE,aAAa;IAC/B,eAAe,EAAE,YAAY;IAC7B,oBAAoB,EAAE,iBAAiB;IACvC,uBAAuB,EAAE,mBAAmB;IAC5C,YAAY,EAAE,SAAS;IACvB,gBAAgB,EAAE,aAAa;IAC/B,eAAe,EAAE,YAAY;IAC7B,sBAAsB,EAAE,mBAAmB;CAC5C,CAAC"}
|
|
1
|
+
{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/detectors/index.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAC;AACzD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAClD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAC1D,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,cAAc,EAAE,MAAM,sBAAsB,CAAC;AACtD,OAAO,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC/D,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,iBAAiB,CAAC;AACtD,OAAO,EAAE,aAAa,EAAE,cAAc,EAAE,SAAS,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,eAAe,CAAC;AACpG,OAAO,EAAE,eAAe,EAAE,iBAAiB,EAAE,cAAc,EAAE,WAAW,EAAE,iBAAiB,EAAE,MAAM,gBAAgB,CAAC;AACpH,OAAO,EAAE,gBAAgB,EAAE,WAAW,EAAE,YAAY,EAAE,eAAe,EAAE,gBAAgB,EAAE,MAAM,cAAc,CAAC;AAC9G,OAAO,EAAE,MAAM,EAAE,cAAc,EAAE,cAAc,EAAE,cAAc,EAAE,mBAAmB,EAAE,sBAAsB,EAAE,MAAM,cAAc,CAAC;AACnI,OAAO,EAAE,cAAc,EAAE,iBAAiB,EAAE,gBAAgB,EAAE,MAAM,WAAW,CAAC;AAChF,OAAO,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,kBAAkB,CAAC;AAC/D,OAAO,EAAE,iBAAiB,EAAE,SAAS,EAAE,mBAAmB,EAAE,mBAAmB,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAE1I,kDAAkD;AAClD,MAAM,CAAC,MAAM,SAAS,GAAuC;IAC3D,mBAAmB,EAAE,eAAe;IACpC,iBAAiB,EAAE,cAAc;IACjC,cAAc,EAAE,WAAW;IAC3B,eAAe,EAAE,YAAY;IAC7B,iBAAiB,EAAE,cAAc;IACjC,cAAc,EAAE,WAAW;IAC3B,mBAAmB,EAAE,gBAAgB;IACrC,iBAAiB,EAAE,cAAc;IACjC,aAAa,EAAE,UAAU;IACzB,kBAAkB,EAAE,eAAe;IACnC,cAAc,EAAE,WAAW;IAC3B,WAAW,EAAE,QAAQ;IACrB,YAAY,EAAE,SAAS;IACvB,gBAAgB,EAAE,aAAa;IAC/B,SAAS,EAAE,SAAS;IACpB,iBAAiB,EAAE,cAAc;IACjC,gBAAgB,EAAE,YAAY;IAC9B,cAAc,EAAE,WAAW;IAC3B,kBAAkB,EAAE,eAAe;IACnC,oBAAoB,EAAE,iBAAiB;IACvC,oBAAoB,EAAE,iBAAiB;IACvC,eAAe,EAAE,WAAW;IAC5B,iBAAiB,EAAE,cAAc;IACjC,eAAe,EAAE,YAAY;IAC7B,cAAc,EAAE,WAAW;IAC3B,kBAAkB,EAAE,eAAe;IACnC,mBAAmB,EAAE,gBAAgB;IACrC,mBAAmB,EAAE,gBAAgB;IACrC,iBAAiB,EAAE,cAAc;IACjC,iBAAiB,EAAE,cAAc;IACjC,yBAAyB,EAAE,sBAAsB;IACjD,sBAAsB,EAAE,mBAAmB;IAC3C,SAAS,EAAE,MAAM;IACjB,iBAAiB,EAAE,cAAc;IACjC,iBAAiB,EAAE,cAAc;IACjC,mBAAmB,EAAE,gBAAgB;IACrC,qBAAqB,EAAE,iBAAiB;IACxC,gBAAgB,EAAE,aAAa;IAC/B,eAAe,EAAE,YAAY;IAC7B,oBAAoB,EAAE,iBAAiB;IACvC,uBAAuB,EAAE,mBAAmB;IAC5C,YAAY,EAAE,SAAS;IACvB,gBAAgB,EAAE,aAAa;IAC/B,eAAe,EAAE,YAAY;IAC7B,sBAAsB,EAAE,mBAAmB;CAC5C,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
/**
|
|
2
|
+
* 後ろの箇条書きや表へ読者を渡す文(「次のとおりとする。」「The following expenses require receipts:」)。
|
|
3
|
+
* 見出しの語を繰り返していても、中身は後ろにある。言い回しは言語パッケージの語彙表 lead-in が持つ。
|
|
4
|
+
*/
|
|
5
|
+
export declare const isLeadIn: (sentence: string, phrases: readonly string[]) => boolean;
|
|
6
|
+
/** 渡す先(following: 同じ節の、その文より後ろ)が空なら、渡す文ではなくただの繰り返し。 */
|
|
7
|
+
export declare const handsOver: (sentence: string, following: string, phrases: readonly string[]) => boolean;
|
|
8
|
+
//# sourceMappingURL=lead-in.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"lead-in.d.ts","sourceRoot":"","sources":["../../src/detectors/lead-in.ts"],"names":[],"mappings":"AAGA;;;GAGG;AACH,eAAO,MAAM,QAAQ,GAAI,UAAU,MAAM,EAAE,SAAS,SAAS,MAAM,EAAE,KAAG,OAGvE,CAAC;AAEF,uDAAuD;AACvD,eAAO,MAAM,SAAS,GAAI,UAAU,MAAM,EAAE,WAAW,MAAM,EAAE,SAAS,SAAS,MAAM,EAAE,KAAG,OAAiE,CAAC"}
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
/** 文末のコロン。後ろに強調の印(**・_)や空白が残っていても見る。 */
|
|
2
|
+
const ENDS_WITH_COLON = /[::][\s*_]*$/u;
|
|
3
|
+
/**
|
|
4
|
+
* 後ろの箇条書きや表へ読者を渡す文(「次のとおりとする。」「The following expenses require receipts:」)。
|
|
5
|
+
* 見出しの語を繰り返していても、中身は後ろにある。言い回しは言語パッケージの語彙表 lead-in が持つ。
|
|
6
|
+
*/
|
|
7
|
+
export const isLeadIn = (sentence, phrases) => {
|
|
8
|
+
const text = sentence.trim().toLowerCase();
|
|
9
|
+
return ENDS_WITH_COLON.test(text) || phrases.some((phrase) => phrase !== "" && text.includes(phrase.toLowerCase()));
|
|
10
|
+
};
|
|
11
|
+
/** 渡す先(following: 同じ節の、その文より後ろ)が空なら、渡す文ではなくただの繰り返し。 */
|
|
12
|
+
export const handsOver = (sentence, following, phrases) => following.trim() !== "" && isLeadIn(sentence, phrases);
|
|
13
|
+
//# sourceMappingURL=lead-in.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"lead-in.js","sourceRoot":"","sources":["../../src/detectors/lead-in.ts"],"names":[],"mappings":"AAAA,wCAAwC;AACxC,MAAM,eAAe,GAAG,eAAe,CAAC;AAExC;;;GAGG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,CAAC,QAAgB,EAAE,OAA0B,EAAW,EAAE;IAChF,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC;IAC3C,OAAO,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,OAAO,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,KAAK,EAAE,IAAI,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC;AACtH,CAAC,CAAC;AAEF,uDAAuD;AACvD,MAAM,CAAC,MAAM,SAAS,GAAG,CAAC,QAAgB,EAAE,SAAiB,EAAE,OAA0B,EAAW,EAAE,CAAC,SAAS,CAAC,IAAI,EAAE,KAAK,EAAE,IAAI,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC"}
|
|
@@ -4,6 +4,12 @@ import type { LexiconEntry, Sentence } from "../plugin.ts";
|
|
|
4
4
|
* 語の途中には当たらない(「また」は「またいで」に当たらない)。品詞が無ければ、空白をまとめた文字列で照らす。
|
|
5
5
|
*/
|
|
6
6
|
export declare const entryIn: (sentence: Sentence, entry: LexiconEntry) => boolean;
|
|
7
|
+
export type TokenRange = {
|
|
8
|
+
readonly start: number;
|
|
9
|
+
readonly end: number;
|
|
10
|
+
};
|
|
11
|
+
/** 語彙の語が文のどの語の並びに当たったか(始まりと、終わった次の位置)。品詞が無ければ位置を言えないので空。 */
|
|
12
|
+
export declare const entryRanges: (sentence: Sentence, entry: LexiconEntry) => TokenRange[];
|
|
7
13
|
/** 語彙の語で文が始まるか。文頭の記号は飛ばす。品詞が無ければ文字列の前方一致。 */
|
|
8
14
|
export declare const entryOpens: (sentence: Sentence, entry: LexiconEntry) => boolean;
|
|
9
15
|
//# sourceMappingURL=lexicon-match.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"lexicon-match.d.ts","sourceRoot":"","sources":["../../src/detectors/lexicon-match.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,QAAQ,EAAS,MAAM,cAAc,CAAC;AAuBlE;;;GAGG;AACH,eAAO,MAAM,OAAO,GAAI,UAAU,QAAQ,EAAE,OAAO,YAAY,KAAG,OAKjE,CAAC;AAIF,6CAA6C;AAC7C,eAAO,MAAM,UAAU,GAAI,UAAU,QAAQ,EAAE,OAAO,YAAY,KAAG,OAMpE,CAAC"}
|
|
1
|
+
{"version":3,"file":"lexicon-match.d.ts","sourceRoot":"","sources":["../../src/detectors/lexicon-match.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,YAAY,EAAE,QAAQ,EAAS,MAAM,cAAc,CAAC;AAuBlE;;;GAGG;AACH,eAAO,MAAM,OAAO,GAAI,UAAU,QAAQ,EAAE,OAAO,YAAY,KAAG,OAKjE,CAAC;AAEF,MAAM,MAAM,UAAU,GAAG;IAAE,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAA;CAAE,CAAC;AAE1E,4DAA4D;AAC5D,eAAO,MAAM,WAAW,GAAI,UAAU,QAAQ,EAAE,OAAO,YAAY,KAAG,UAAU,EAK/E,CAAC;AAIF,6CAA6C;AAC7C,eAAO,MAAM,UAAU,GAAI,UAAU,QAAQ,EAAE,OAAO,YAAY,KAAG,OAMpE,CAAC"}
|
|
@@ -27,6 +27,14 @@ export const entryIn = (sentence, entry) => {
|
|
|
27
27
|
return proseText(sentence).toLowerCase().includes(entry.pattern.toLowerCase());
|
|
28
28
|
return tokens.some((_token, start) => runsAt(tokens, words, start));
|
|
29
29
|
};
|
|
30
|
+
/** 語彙の語が文のどの語の並びに当たったか(始まりと、終わった次の位置)。品詞が無ければ位置を言えないので空。 */
|
|
31
|
+
export const entryRanges = (sentence, entry) => {
|
|
32
|
+
const tokens = sentence.tokens;
|
|
33
|
+
const words = entry.tokens;
|
|
34
|
+
if (tokens === undefined || words === undefined || words.length === 0)
|
|
35
|
+
return [];
|
|
36
|
+
return tokens.flatMap((_token, start) => (runsAt(tokens, words, start) ? [{ start, end: start + words.length }] : []));
|
|
37
|
+
};
|
|
30
38
|
const LEADING_MARK = new Set(["PUNCT", "SYM"]);
|
|
31
39
|
/** 語彙の語で文が始まるか。文頭の記号は飛ばす。品詞が無ければ文字列の前方一致。 */
|
|
32
40
|
export const entryOpens = (sentence, entry) => {
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"lexicon-match.js","sourceRoot":"","sources":["../../src/detectors/lexicon-match.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAE1C,wCAAwC;AACxC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAE5C;;;GAGG;AACH,MAAM,QAAQ,GAAG,CAAC,OAAc,EAAE,KAAY,EAAW,EAAE;IACzD,MAAM,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;IAC5C,IAAI,OAAO,CAAC,OAAO,CAAC,WAAW,EAAE,KAAK,OAAO;QAAE,OAAO,IAAI,CAAC;IAC3D,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC1E,OAAO,QAAQ,IAAI,KAAK,CAAC,KAAK,EAAE,WAAW,EAAE,KAAK,OAAO,IAAI,OAAO,CAAC,KAAK,EAAE,WAAW,EAAE,KAAK,OAAO,CAAC;AACxG,CAAC,CAAC;AAEF,MAAM,MAAM,GAAG,CAAC,MAAwB,EAAE,KAAuB,EAAE,KAAa,EAAW,EAAE,CAC3F,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,EAAE;IAC3B,MAAM,OAAO,GAAG,MAAM,CAAC,KAAK,GAAG,MAAM,CAAC,CAAC;IACvC,OAAO,OAAO,KAAK,SAAS,IAAI,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;AAC1D,CAAC,CAAC,CAAC;AAEL;;;GAGG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE;IAC1E,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC/B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC;IAC3B,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IACtJ,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AACtE,CAAC,CAAC;AAEF,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC;AAE/C,6CAA6C;AAC7C,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE;IAC7E,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC/B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC;IAC3B,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IACxJ,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;IACxE,OAAO,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC;AACtC,CAAC,CAAC"}
|
|
1
|
+
{"version":3,"file":"lexicon-match.js","sourceRoot":"","sources":["../../src/detectors/lexicon-match.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,SAAS,EAAE,MAAM,eAAe,CAAC;AAE1C,wCAAwC;AACxC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,CAAC,CAAC,CAAC;AAE5C;;;GAGG;AACH,MAAM,QAAQ,GAAG,CAAC,OAAc,EAAE,KAAY,EAAW,EAAE;IACzD,MAAM,OAAO,GAAG,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC;IAC5C,IAAI,OAAO,CAAC,OAAO,CAAC,WAAW,EAAE,KAAK,OAAO;QAAE,OAAO,IAAI,CAAC;IAC3D,MAAM,QAAQ,GAAG,UAAU,CAAC,GAAG,CAAC,OAAO,CAAC,GAAG,CAAC,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;IAC1E,OAAO,QAAQ,IAAI,KAAK,CAAC,KAAK,EAAE,WAAW,EAAE,KAAK,OAAO,IAAI,OAAO,CAAC,KAAK,EAAE,WAAW,EAAE,KAAK,OAAO,CAAC;AACxG,CAAC,CAAC;AAEF,MAAM,MAAM,GAAG,CAAC,MAAwB,EAAE,KAAuB,EAAE,KAAa,EAAW,EAAE,CAC3F,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,EAAE;IAC3B,MAAM,OAAO,GAAG,MAAM,CAAC,KAAK,GAAG,MAAM,CAAC,CAAC;IACvC,OAAO,OAAO,KAAK,SAAS,IAAI,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;AAC1D,CAAC,CAAC,CAAC;AAEL;;;GAGG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE;IAC1E,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC/B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC;IAC3B,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IACtJ,OAAO,MAAM,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AACtE,CAAC,CAAC;AAIF,4DAA4D;AAC5D,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAgB,EAAE;IACnF,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC/B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC;IAC3B,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IACjF,OAAO,MAAM,CAAC,OAAO,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,GAAG,EAAE,KAAK,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;AACzH,CAAC,CAAC;AAEF,MAAM,YAAY,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC,CAAC;AAE/C,6CAA6C;AAC7C,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAW,EAAE;IAC7E,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,CAAC;IAC/B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC;IAC3B,IAAI,MAAM,KAAK,SAAS,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAC,QAAQ,CAAC,CAAC,WAAW,EAAE,CAAC,UAAU,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC;IACxJ,MAAM,KAAK,GAAG,MAAM,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,YAAY,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;IACxE,OAAO,MAAM,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC;AACtC,CAAC,CAAC"}
|
|
@@ -1,6 +1,11 @@
|
|
|
1
1
|
import type { Detector } from "../plugin.ts";
|
|
2
2
|
export declare const hedgingDensity: Detector;
|
|
3
3
|
export declare const cushionDensity: Detector;
|
|
4
|
+
/**
|
|
5
|
+
* 限定のない最上級。「最も速い」だけでは、何と比べて最もなのかが無い。
|
|
6
|
+
* 同じ文に比較対象・条件・範囲(日本で最も、the best in the world)があれば、それは主張であって誇張ではない。
|
|
7
|
+
* 比較と範囲を言う語は言語パッケージの語彙表が持つ。
|
|
8
|
+
*/
|
|
4
9
|
export declare const unqualifiedSuperlative: Detector;
|
|
5
10
|
/**
|
|
6
11
|
* 段落が同じ接続詞で始まり続ける。1 つなら流れを作るが、続くとどの段落も
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"lexicon.d.ts","sourceRoot":"","sources":["../../src/detectors/lexicon.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,QAAQ,
|
|
1
|
+
{"version":3,"file":"lexicon.d.ts","sourceRoot":"","sources":["../../src/detectors/lexicon.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,QAAQ,EAA2D,MAAM,cAAc,CAAC;AAoCtG,eAAO,MAAM,cAAc,UAAmC,CAAC;AAC/D,eAAO,MAAM,cAAc,UAAwC,CAAC;AA4BpE;;;;GAIG;AACH,eAAO,MAAM,sBAAsB,EAAE,QAWpC,CAAC;AAQF;;;GAGG;AACH,eAAO,MAAM,mBAAmB,EAAE,QA0BjC,CAAC;AAEF;;;GAGG;AACH,eAAO,MAAM,MAAM,EAAE,QAyBpB,CAAC;AAiBF,eAAO,MAAM,cAAc,EAAE,QAwB5B,CAAC"}
|
|
@@ -1,6 +1,7 @@
|
|
|
1
1
|
import { joinWords } from "./word-list.js";
|
|
2
2
|
import { wordsOf } from "./structure.js";
|
|
3
|
-
import { entryIn, entryOpens } from "./lexicon-match.js";
|
|
3
|
+
import { entryIn, entryOpens, entryRanges } from "./lexicon-match.js";
|
|
4
|
+
import { scoped, scopeMarkersOf } from "./superlative-scope.js";
|
|
4
5
|
const PER = 1000;
|
|
5
6
|
/** 密度を見る rule は短い文書を測らない。単位は言語で違うので床も分ける。 */
|
|
6
7
|
const FLOOR = { word: 200, char: 500 };
|
|
@@ -27,13 +28,29 @@ const densityRule = (rule) => (doc, options) => {
|
|
|
27
28
|
};
|
|
28
29
|
export const hedgingDensity = densityRule("excessive-hedging");
|
|
29
30
|
export const cushionDensity = densityRule("cushion-phrase-density");
|
|
31
|
+
/** 数字は語ではないので言語を問わない。数があれば測った結果を言っている。品詞の数(NUM)は "the best one" の one まで含むので使わない。 */
|
|
32
|
+
const DIGIT = /\d/u;
|
|
33
|
+
const qualifiersOf = (doc) => ({
|
|
34
|
+
comparison: doc.lexicons["comparison-marker"] ?? [],
|
|
35
|
+
scope: scopeMarkersOf(doc.lexicons["superlative-scope"] ?? []),
|
|
36
|
+
});
|
|
37
|
+
/** どの出現も範囲を持つときだけ。1 つでも範囲の無い出現があれば、その文には限定の無い最上級がある。 */
|
|
38
|
+
const everyScoped = (sentence, entry, scope) => {
|
|
39
|
+
const ranges = entryRanges(sentence, entry);
|
|
40
|
+
return ranges.length > 0 && ranges.every((range) => scoped(sentence.tokens ?? [], range, scope));
|
|
41
|
+
};
|
|
42
|
+
const qualified = (sentence, entry, qualifiers) => DIGIT.test(sentence.text) || qualifiers.comparison.some((marker) => entryIn(sentence, marker)) || everyScoped(sentence, entry, qualifiers.scope);
|
|
43
|
+
const bareHits = (doc, lexicon) => {
|
|
44
|
+
const qualifiers = qualifiersOf(doc);
|
|
45
|
+
return doc.sentences.flatMap((sentence) => lexicon.filter((entry) => entryIn(sentence, entry) && !qualified(sentence, entry, qualifiers)).map((entry) => ({ sentence, matched: entry.pattern })));
|
|
46
|
+
};
|
|
30
47
|
/**
|
|
31
48
|
* 限定のない最上級。「最も速い」だけでは、何と比べて最もなのかが無い。
|
|
32
|
-
*
|
|
49
|
+
* 同じ文に比較対象・条件・範囲(日本で最も、the best in the world)があれば、それは主張であって誇張ではない。
|
|
50
|
+
* 比較と範囲を言う語は言語パッケージの語彙表が持つ。
|
|
33
51
|
*/
|
|
34
|
-
const QUALIFIER = /\d|より|に比べ|のうち|among|than|compared|based on|according/iu;
|
|
35
52
|
export const unqualifiedSuperlative = (doc, options) => {
|
|
36
|
-
const bare =
|
|
53
|
+
const bare = bareHits(doc, options.lexicon ?? []);
|
|
37
54
|
if (bare.length < options.limit)
|
|
38
55
|
return [];
|
|
39
56
|
return bare.map((hit) => ({
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"lexicon.js","sourceRoot":"","sources":["../../src/detectors/lexicon.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAC3C,OAAO,EAAE,OAAO,EAAE,MAAM,gBAAgB,CAAC;AAEzC,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,oBAAoB,CAAC;
|
|
1
|
+
{"version":3,"file":"lexicon.js","sourceRoot":"","sources":["../../src/detectors/lexicon.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,gBAAgB,CAAC;AAC3C,OAAO,EAAE,OAAO,EAAE,MAAM,gBAAgB,CAAC;AAEzC,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AACtE,OAAO,EAAE,MAAM,EAAE,cAAc,EAAqB,MAAM,wBAAwB,CAAC;AAEnF,MAAM,GAAG,GAAG,IAAI,CAAC;AAEjB,6CAA6C;AAC7C,MAAM,KAAK,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;AAIvC,MAAM,OAAO,GAAG,CAAC,GAAkB,EAAE,OAAgB,EAAS,EAAE,CAC9D,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CAAC,CAAC;AAElJ;;;GAGG;AACH,MAAM,WAAW,GACf,CAAC,IAAY,EAAY,EAAE,CAC3B,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAC1B,MAAM,IAAI,GAAG,OAAO,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC;IACjD,MAAM,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,CAAC;IAC5B,MAAM,IAAI,GAAG,MAAM,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC,GAAG,GAAG,CAAC,CAAC;IACzE,MAAM,KAAK,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;IACtB,IAAI,MAAM,GAAG,KAAK,CAAC,GAAG,CAAC,UAAU,CAAC,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,IAAI,OAAO,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IAC9F,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QACxB,IAAI;QACJ,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC/B,MAAM,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,MAAM,EAAE,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;KAC3H,CAAC,CAAC,CAAC;AACN,CAAC,CAAC;AAEJ,MAAM,CAAC,MAAM,cAAc,GAAG,WAAW,CAAC,mBAAmB,CAAC,CAAC;AAC/D,MAAM,CAAC,MAAM,cAAc,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAC;AAEpE,qFAAqF;AACrF,MAAM,KAAK,GAAG,KAAK,CAAC;AAIpB,MAAM,YAAY,GAAG,CAAC,GAAkB,EAAc,EAAE,CAAC,CAAC;IACxD,UAAU,EAAE,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,IAAI,EAAE;IACnD,KAAK,EAAE,cAAc,CAAC,GAAG,CAAC,QAAQ,CAAC,mBAAmB,CAAC,IAAI,EAAE,CAAC;CAC/D,CAAC,CAAC;AAEH,wDAAwD;AACxD,MAAM,WAAW,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAE,KAAmB,EAAW,EAAE;IAC5F,MAAM,MAAM,GAAG,WAAW,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC;IAC5C,OAAO,MAAM,CAAC,MAAM,GAAG,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AACnG,CAAC,CAAC;AAEF,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,KAAmB,EAAE,UAAsB,EAAW,EAAE,CAC7F,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,UAAU,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC,IAAI,WAAW,CAAC,QAAQ,EAAE,KAAK,EAAE,UAAU,CAAC,KAAK,CAAC,CAAC;AAEnJ,MAAM,QAAQ,GAAG,CAAC,GAAkB,EAAE,OAAgB,EAAS,EAAE;IAC/D,MAAM,UAAU,GAAG,YAAY,CAAC,GAAG,CAAC,CAAC;IACrC,OAAO,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CACxC,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,QAAQ,EAAE,KAAK,EAAE,UAAU,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,OAAO,EAAE,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC,CACtJ,CAAC;AACJ,CAAC,CAAC;AAEF;;;;GAIG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAC1E,MAAM,IAAI,GAAG,QAAQ,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC;IAClD,IAAI,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IAC3C,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QACxB,IAAI,EAAE,yBAAyB;QAC/B,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC/B,MAAM,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,KAAK,EAAE,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;KAC5G,CAAC,CAAC,CAAC;AACN,CAAC,CAAC;AAEF,MAAM,QAAQ,GAAG,CAAC,GAAkB,EAAE,OAAgB,EAA0B,EAAE,CAChF,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,SAAS,EAAE,EAAE;IAC/B,MAAM,IAAI,GAAG,SAAS,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;IACpC,OAAO,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,UAAU,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC;AACpG,CAAC,CAAC,CAAC;AAEL;;;GAGG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IACvE,MAAM,OAAO,GAAG,QAAQ,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC;IACrD,MAAM,IAAI,GAAG,OAAO,CAAC,MAAM,CACzB,CAAC,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,EAAE;QACrB,IAAI,MAAM,KAAK,SAAS;YAAE,OAAO,EAAE,IAAI,EAAE,CAAC,GAAG,GAAG,CAAC,IAAI,EAAE,GAAG,CAAC,OAAO,CAAC,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC;QACnF,OAAO,EAAE,IAAI,EAAE,GAAG,CAAC,IAAI,EAAE,OAAO,EAAE,CAAC,GAAG,GAAG,CAAC,OAAO,EAAE,KAAK,CAAC,EAAE,CAAC;IAC9D,CAAC,EACD,EAAE,IAAI,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,EAAE,CAC1B,CAAC;IACF,OAAO,CAAC,GAAG,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,OAAO,CAAC;SAChC,MAAM,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK,CAAC;SAC3C,OAAO,CAAC,CAAC,GAAG,EAAE,EAAE;QACf,MAAM,EAAE,GAAG,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC;QACvC,OAAO,EAAE,KAAK,SAAS;YACrB,CAAC,CAAC,EAAE;YACJ,CAAC,CAAC;gBACE;oBACE,IAAI,EAAE,sBAAsB;oBAC5B,QAAQ,EAAE,SAAkB;oBAC5B,IAAI,EAAE,CAAC;oBACP,MAAM,EAAE,CAAC;oBACT,KAAK,EAAE,EAAE,CAAC,SAAS,CAAC,CAAC,CAAC,EAAE,IAAI,CAAC,IAAI,EAAE,IAAI,EAAE;oBACzC,MAAM,EAAE,EAAE,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,EAAE;iBAC3E;aACF,CAAC;IACR,CAAC,CAAC,CAAC;AACP,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,CAAC,MAAM,MAAM,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAC1D,MAAM,OAAO,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,CAAC;IACpH,MAAM,KAAK,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC;IACpG,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,KAAK,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC,CAAC;IAC1F,MAAM,KAAK,GAAG,GAAG,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,OAAO,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC;IAChG,IAAI,KAAK,KAAK,SAAS,IAAI,KAAK,IAAI,OAAO,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IAC7D,OAAO;QACL;YACE,IAAI,EAAE,SAAS;YACf,QAAQ,EAAE,MAAM;YAChB,IAAI,EAAE,CAAC;YACP,MAAM,EAAE,CAAC;YACT,KAAK,EAAE,KAAK,CAAC,IAAI,CAAC,IAAI,EAAE;YACxB,MAAM,EAAE;gBACN,IAAI,EAAE,SAAS,CACb,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,EACjD,GAAG,CAAC,QAAQ,CACb;gBACD,KAAK,EAAE,KAAK,CAAC,MAAM;gBACnB,OAAO,EAAE,KAAK;gBACd,KAAK,EAAE,OAAO,CAAC,KAAK;gBACpB,MAAM,EAAE,KAAK,CAAC,IAAI,CAAC,KAAK;aACzB;SACF;KACF,CAAC;AACJ,CAAC,CAAC;AAEF;;;;;;;;GAQG;AACH;;;GAGG;AACH,MAAM,YAAY,GAAG,CAAC,IAAY,EAAE,IAAY,EAAW,EAAE,CAAC,IAAI,MAAM,CAAC,MAAM,IAAI,CAAC,OAAO,CAAC,sBAAsB,EAAE,MAAM,CAAC,KAAK,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AAElJ,MAAM,CAAC,MAAM,cAAc,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAClE,MAAM,KAAK,GAAG,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,OAAO,EAAE,IAAI,EAAE,KAAK,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC,CAAC;IACrJ,MAAM,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IACpF,MAAM,IAAI,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC;IAClF,MAAM,OAAO,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,EAAE,IAAI,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC;IACpF,oCAAoC;IACpC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,OAAO,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IACzD,MAAM,eAAe,GAAG,IAAI,CAAC,MAAM,IAAI,OAAO,CAAC,MAAM,CAAC;IACtD,MAAM,GAAG,GAAG,eAAe,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC;IAC7C,IAAI,GAAG,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IAC1C,MAAM,MAAM,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC;IAC3F,MAAM,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE;QAC9C,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC;QACzC,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,YAAY,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,CAAC;QAChE,OAAO,OAAO,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC,CAAC;IAC9D,CAAC,CAAC,CAAC;IACH,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QACxB,IAAI,EAAE,yBAAyB;QAC/B,QAAQ,EAAE,MAAM;QAChB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC/B,MAAM,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,EAAE,KAAK,EAAE,GAAG,CAAC,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;KAC3G,CAAC,CAAC,CAAC;AACN,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"orthography.d.ts","sourceRoot":"","sources":["../../src/detectors/orthography.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAqB,MAAM,cAAc,CAAC;
|
|
1
|
+
{"version":3,"file":"orthography.d.ts","sourceRoot":"","sources":["../../src/detectors/orthography.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAqB,MAAM,cAAc,CAAC;AAIhE,6DAA6D;AAC7D,eAAO,MAAM,aAAa,EAAE,QAc3B,CAAC;AAoBF;;;GAGG;AACH,eAAO,MAAM,YAAY,EAAE,QA6B1B,CAAC"}
|
|
@@ -1,4 +1,5 @@
|
|
|
1
1
|
import { latinBoundaries, minorityStyle, occurrencesOutside } from "../orthography.js";
|
|
2
|
+
import { digitRunAround, isNumberName } from "../number-name.js";
|
|
2
3
|
/** チームが chaff.yaml の prefer に書いた「使わない書き方」。書いていなければ何も言わない。 */
|
|
3
4
|
export const preferredTerm = (doc, options) => {
|
|
4
5
|
const pairs = (options.lexicon ?? []).flatMap((entry) => (entry.instead_of === undefined ? [] : [{ avoid: entry.pattern, use: entry.instead_of }]));
|
|
@@ -16,12 +17,27 @@ export const preferredTerm = (doc, options) => {
|
|
|
16
17
|
};
|
|
17
18
|
const KIND_NAME = { letter: "英字", "before-digit": "後ろの数字", "after-digit": "前の数字" };
|
|
18
19
|
const KINDS = ["letter", "before-digit", "after-digit"];
|
|
20
|
+
/** 境目の隣の数字の、並びの中の 1 字の位置。数字の後ろなら左の字、数字の前なら(空白を越えた)右の字。 */
|
|
21
|
+
const digitBeside = (boundary) => {
|
|
22
|
+
if (boundary.kind === "after-digit")
|
|
23
|
+
return boundary.offset - 1;
|
|
24
|
+
return boundary.spaced ? boundary.offset + 1 : boundary.offset;
|
|
25
|
+
};
|
|
26
|
+
/** 番号・識別子として書かれた数の境目は、空け方の好みではないので数えない(number-name.ts)。 */
|
|
27
|
+
const isCounted = (sentence, boundary) => {
|
|
28
|
+
if (boundary.kind === "letter")
|
|
29
|
+
return true;
|
|
30
|
+
const run = digitRunAround(sentence.text, digitBeside(boundary));
|
|
31
|
+
return run === undefined || !isNumberName(sentence.text, run, sentence.tokens, sentence.span.start);
|
|
32
|
+
};
|
|
19
33
|
/**
|
|
20
34
|
* 日本語と英字・数字のあいだを、空けるか詰めるか。文書の中で混ざっていたら、少ないほうを指摘する。
|
|
21
35
|
* どちらが正しいかは決めない。決めるのはチームで、chaff はそろっているかだけを見る。
|
|
22
36
|
*/
|
|
23
37
|
export const latinSpacing = (doc, options) => {
|
|
24
|
-
const located = doc.sentences.flatMap((sentence) => latinBoundaries(sentence.text
|
|
38
|
+
const located = doc.sentences.flatMap((sentence) => latinBoundaries(sentence.text, doc.source.slice(sentence.span.start, sentence.span.end))
|
|
39
|
+
.filter((boundary) => isCounted(sentence, boundary))
|
|
40
|
+
.map((boundary) => ({ sentence, ...boundary })));
|
|
25
41
|
return KINDS.flatMap((kind) => {
|
|
26
42
|
const ofKind = located.filter((entry) => entry.kind === kind);
|
|
27
43
|
const minority = minorityStyle(ofKind);
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"orthography.js","sourceRoot":"","sources":["../../src/detectors/orthography.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,kBAAkB,
|
|
1
|
+
{"version":3,"file":"orthography.js","sourceRoot":"","sources":["../../src/detectors/orthography.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,kBAAkB,EAAmC,MAAM,mBAAmB,CAAC;AACxH,OAAO,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAEjE,6DAA6D;AAC7D,MAAM,CAAC,MAAM,aAAa,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IACjE,MAAM,KAAK,GAAG,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,UAAU,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,CAAC,OAAO,EAAE,GAAG,EAAE,KAAK,CAAC,UAAU,EAAE,CAAC,CAAC,CAAC,CAAC;IACpJ,MAAM,IAAI,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CAC9C,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,kBAAkB,CAAC,QAAQ,CAAC,IAAI,EAAE,IAAI,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,EAAE,EAAE,IAAI,EAAE,CAAC,CAAC,CAAC,CACvH,CAAC;IACF,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,IAAI,IAAI,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK;QAAE,OAAO,EAAE,CAAC;IAChE,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;QACxB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC/B,MAAM,EAAE,EAAE,OAAO,EAAE,GAAG,CAAC,IAAI,CAAC,KAAK,EAAE,SAAS,EAAE,GAAG,CAAC,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,IAAI,CAAC,MAAM,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC,EAAE,EAAE;KACjJ,CAAC,CAAC,CAAC;AACN,CAAC,CAAC;AAEF,MAAM,SAAS,GAA0C,EAAE,MAAM,EAAE,IAAI,EAAE,cAAc,EAAE,OAAO,EAAE,aAAa,EAAE,MAAM,EAAE,CAAC;AAC1H,MAAM,KAAK,GAA2B,CAAC,QAAQ,EAAE,cAAc,EAAE,aAAa,CAAC,CAAC;AAIhF,0DAA0D;AAC1D,MAAM,WAAW,GAAG,CAAC,QAAkB,EAAU,EAAE;IACjD,IAAI,QAAQ,CAAC,IAAI,KAAK,aAAa;QAAE,OAAO,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;IAChE,OAAO,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC;AACjE,CAAC,CAAC;AAEF,2DAA2D;AAC3D,MAAM,SAAS,GAAG,CAAC,QAAkB,EAAE,QAAkB,EAAW,EAAE;IACpE,IAAI,QAAQ,CAAC,IAAI,KAAK,QAAQ;QAAE,OAAO,IAAI,CAAC;IAC5C,MAAM,GAAG,GAAG,cAAc,CAAC,QAAQ,CAAC,IAAI,EAAE,WAAW,CAAC,QAAQ,CAAC,CAAC,CAAC;IACjE,OAAO,GAAG,KAAK,SAAS,IAAI,CAAC,YAAY,CAAC,QAAQ,CAAC,IAAI,EAAE,GAAG,EAAE,QAAQ,CAAC,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC;AACtG,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,CAAC,MAAM,YAAY,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAChE,MAAM,OAAO,GAAc,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CAC5D,eAAe,CAAC,QAAQ,CAAC,IAAI,EAAE,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;SACrF,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,SAAS,CAAC,QAAQ,EAAE,QAAQ,CAAC,CAAC;SACnD,GAAG,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,EAAE,QAAQ,EAAE,GAAG,QAAQ,EAAE,CAAC,CAAC,CAClD,CAAC;IACF,OAAO,KAAK,CAAC,OAAO,CAAC,CAAC,IAAI,EAAE,EAAE;QAC5B,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,IAAI,CAAC,CAAC;QAC9D,MAAM,QAAQ,GAAG,aAAa,CAAC,MAAM,CAAC,CAAC;QACvC,IAAI,QAAQ,KAAK,SAAS;YAAE,OAAO,EAAE,CAAC;QACtC,MAAM,GAAG,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,KAAK,QAAQ,CAAC,CAAC;QAChE,IAAI,GAAG,CAAC,MAAM,GAAG,OAAO,CAAC,KAAK;YAAE,OAAO,EAAE,CAAC;QAC1C,OAAO,GAAG,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YACzB,IAAI,EAAE,EAAE;YACR,QAAQ,EAAE,SAAS;YACnB,IAAI,EAAE,CAAC;YACP,MAAM,EAAE,CAAC;YACT,KAAK,EAAE,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;YACjC,MAAM,EAAE;gBACN,IAAI,EAAE,SAAS,CAAC,IAAI,CAAC;gBACrB,KAAK,EAAE,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,QAAQ;gBACrC,KAAK,EAAE,QAAQ,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK;gBAC/B,KAAK,EAAE,GAAG,CAAC,MAAM;gBACjB,EAAE,EAAE,MAAM,CAAC,MAAM;gBACjB,KAAK,EAAE,OAAO,CAAC,KAAK;gBACpB,MAAM,EAAE,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,MAAM;aACjD;SACF,CAAC,CAAC,CAAC;IACN,CAAC,CAAC,CAAC;AACL,CAAC,CAAC"}
|
|
@@ -0,0 +1,8 @@
|
|
|
1
|
+
import type { Token } from "../plugin.ts";
|
|
2
|
+
/**
|
|
3
|
+
* 漢字の連なりの語(covering: tokens の添字)が住所か(東京都港区新橋二丁目)。地名で始まり、住所の語だけでできている。
|
|
4
|
+
* 住所は決まった形で、ひらがなを挟んで書けない。組織名(日本経済団体連合会)は地名の後ろに普通の語が続くので住所ではない。
|
|
5
|
+
* topUnits は、住所の先頭にしか来ない単位(都道府県)。言語の語彙表から渡す。
|
|
6
|
+
*/
|
|
7
|
+
export declare const isAddressRun: (tokens: readonly Token[], covering: readonly number[], topUnits: ReadonlySet<string>) => boolean;
|
|
8
|
+
//# sourceMappingURL=place-run.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"place-run.d.ts","sourceRoot":"","sources":["../../src/detectors/place-run.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,KAAK,EAAE,MAAM,cAAc,CAAC;AAoD1C;;;;GAIG;AACH,eAAO,MAAM,YAAY,GAAI,QAAQ,SAAS,KAAK,EAAE,EAAE,UAAU,SAAS,MAAM,EAAE,EAAE,UAAU,WAAW,CAAC,MAAM,CAAC,KAAG,OAKnH,CAAC"}
|
|
@@ -0,0 +1,54 @@
|
|
|
1
|
+
/** 地名(Geo)と、地名に付く単位(GeoUnit: 都・県・郡・市・区・町・村)。 */
|
|
2
|
+
const isGeoName = (token) => token?.features?.["NameType"] === "Geo";
|
|
3
|
+
const isGeoUnit = (token) => token?.features?.["NameType"] === "GeoUnit";
|
|
4
|
+
const isGeo = (token) => isGeoName(token) || isGeoUnit(token);
|
|
5
|
+
const isNumber = (token) => token?.features?.["NumType"] === "Card";
|
|
6
|
+
/**
|
|
7
|
+
* 辞書に無い町名は、形態素解析が固有名詞か 1 字ずつに割る(紀美野町は 紀+美野、南伊勢町は 南+伊勢、久米島町は固有名詞,一般)。
|
|
8
|
+
* 2 字以上の普通の語(政策・総務)は地名のかけらにしない。
|
|
9
|
+
*/
|
|
10
|
+
const isNamePiece = (token) => token !== undefined && (token.pos === "PROPN" || [...token.surface].length === 1);
|
|
11
|
+
/**
|
|
12
|
+
* index から地名のかけらだけを読み進めて、連なりの終わりまでに都道府県より下の単位(郡・市・町)で閉じるか。
|
|
13
|
+
* 閉じれば、そこまでが 1 つの地名。連なりの外(「に」「の」も 1 字)までは読まない。
|
|
14
|
+
*/
|
|
15
|
+
const closedByUnit = (run, index) => {
|
|
16
|
+
const token = run.tokens[index];
|
|
17
|
+
if (index > run.last || token === undefined)
|
|
18
|
+
return false;
|
|
19
|
+
if (isGeoUnit(token))
|
|
20
|
+
return !run.topUnits.has(token.surface);
|
|
21
|
+
return isNamePiece(token) && closedByUnit(run, index + 1);
|
|
22
|
+
};
|
|
23
|
+
/**
|
|
24
|
+
* 住所の語: 地名と地名の単位(NameType=Geo / GeoUnit)、数(NumType=Card)、数のすぐ後ろの助数詞(丁目、NounType=Class)。
|
|
25
|
+
* 辞書に無い町名(新橋)は、地名の後ろで数の前に来るときだけ住所の一部とする。
|
|
26
|
+
* 割られた町名(紀美野町)は、単位の後ろで、地名のかけらだけを挟んで次の単位で閉じるときだけ住所の一部とする。
|
|
27
|
+
*/
|
|
28
|
+
const isPlacePart = (run, index) => {
|
|
29
|
+
const tokens = run.tokens;
|
|
30
|
+
return (isGeo(tokens[index]) ||
|
|
31
|
+
isNumber(tokens[index]) ||
|
|
32
|
+
(isNumber(tokens[index - 1]) && tokens[index]?.features?.["NounType"] === "Class") ||
|
|
33
|
+
(isGeo(tokens[index - 1]) && isNumber(tokens[index + 1])) ||
|
|
34
|
+
(isGeo(tokens[index - 1]) && isNamePiece(tokens[index]) && closedByUnit(run, index + 1)));
|
|
35
|
+
};
|
|
36
|
+
/**
|
|
37
|
+
* 地名が単位を挟まずに続けば地名の並び(東京大阪名古屋福岡)。ただし 2 つだけで郡・市・町が閉じれば、割られた 1 つの地名
|
|
38
|
+
* (北海道虻田郡、富士河口湖町)。3 つ続けば、単位が付いても並び(京都奈良大阪神戸市)。
|
|
39
|
+
* 名古屋大阪市は北海道札幌市と品詞が同じ形なので、ここでは分けられない。
|
|
40
|
+
*/
|
|
41
|
+
const isListed = (run, index) => isGeoName(run.tokens[index]) && isGeoName(run.tokens[index + 1]) && (isGeoName(run.tokens[index + 2]) || !closedByUnit(run, index + 1));
|
|
42
|
+
/**
|
|
43
|
+
* 漢字の連なりの語(covering: tokens の添字)が住所か(東京都港区新橋二丁目)。地名で始まり、住所の語だけでできている。
|
|
44
|
+
* 住所は決まった形で、ひらがなを挟んで書けない。組織名(日本経済団体連合会)は地名の後ろに普通の語が続くので住所ではない。
|
|
45
|
+
* topUnits は、住所の先頭にしか来ない単位(都道府県)。言語の語彙表から渡す。
|
|
46
|
+
*/
|
|
47
|
+
export const isAddressRun = (tokens, covering, topUnits) => {
|
|
48
|
+
const [first, last] = [covering[0], covering.at(-1)];
|
|
49
|
+
if (first === undefined || last === undefined)
|
|
50
|
+
return false;
|
|
51
|
+
const run = { tokens, last, topUnits };
|
|
52
|
+
return isGeo(tokens[first]) && !covering.some((index) => isListed(run, index)) && covering.every((index) => isPlacePart(run, index));
|
|
53
|
+
};
|
|
54
|
+
//# sourceMappingURL=place-run.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"place-run.js","sourceRoot":"","sources":["../../src/detectors/place-run.ts"],"names":[],"mappings":"AAEA,gDAAgD;AAChD,MAAM,SAAS,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,UAAU,CAAC,KAAK,KAAK,CAAC;AACjG,MAAM,SAAS,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,UAAU,CAAC,KAAK,SAAS,CAAC;AACrG,MAAM,KAAK,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,SAAS,CAAC,KAAK,CAAC,IAAI,SAAS,CAAC,KAAK,CAAC,CAAC;AAC1F,MAAM,QAAQ,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,SAAS,CAAC,KAAK,MAAM,CAAC;AAEhG;;;GAGG;AACH,MAAM,WAAW,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,KAAK,SAAS,IAAI,CAAC,KAAK,CAAC,GAAG,KAAK,OAAO,IAAI,CAAC,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,MAAM,KAAK,CAAC,CAAC,CAAC;AAK7I;;;GAGG;AACH,MAAM,YAAY,GAAG,CAAC,GAAQ,EAAE,KAAa,EAAW,EAAE;IACxD,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;IAChC,IAAI,KAAK,GAAG,GAAG,CAAC,IAAI,IAAI,KAAK,KAAK,SAAS;QAAE,OAAO,KAAK,CAAC;IAC1D,IAAI,SAAS,CAAC,KAAK,CAAC;QAAE,OAAO,CAAC,GAAG,CAAC,QAAQ,CAAC,GAAG,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IAC9D,OAAO,WAAW,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,GAAG,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC;AAC5D,CAAC,CAAC;AAEF;;;;GAIG;AACH,MAAM,WAAW,GAAG,CAAC,GAAQ,EAAE,KAAa,EAAW,EAAE;IACvD,MAAM,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC;IAC1B,OAAO,CACL,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACpB,QAAQ,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QACvB,CAAC,QAAQ,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,KAAK,CAAC,EAAE,QAAQ,EAAE,CAAC,UAAU,CAAC,KAAK,OAAO,CAAC;QAClF,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,QAAQ,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC;QACzD,CAAC,KAAK,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,WAAW,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,YAAY,CAAC,GAAG,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC,CACzF,CAAC;AACJ,CAAC,CAAC;AAEF;;;;GAIG;AACH,MAAM,QAAQ,GAAG,CAAC,GAAQ,EAAE,KAAa,EAAW,EAAE,CACpD,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,YAAY,CAAC,GAAG,EAAE,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC;AAE1I;;;;GAIG;AACH,MAAM,CAAC,MAAM,YAAY,GAAG,CAAC,MAAwB,EAAE,QAA2B,EAAE,QAA6B,EAAW,EAAE;IAC5H,MAAM,CAAC,KAAK,EAAE,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IACrD,IAAI,KAAK,KAAK,SAAS,IAAI,IAAI,KAAK,SAAS;QAAE,OAAO,KAAK,CAAC;IAC5D,MAAM,GAAG,GAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,CAAC;IAC5C,OAAO,KAAK,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,QAAQ,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC,IAAI,QAAQ,CAAC,KAAK,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,WAAW,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC,CAAC;AACvI,CAAC,CAAC"}
|
|
@@ -0,0 +1,32 @@
|
|
|
1
|
+
import type { Sentence, Span, Token } from "../plugin.ts";
|
|
2
|
+
export type Register = "polite" | "plain";
|
|
3
|
+
/**
|
|
4
|
+
* 文末の語の調子。preceding は文末の語より前の語。
|
|
5
|
+
* 文末に述語(動詞・形容詞・助動詞)が無ければ、ですます調でもである調でもないので undefined。
|
|
6
|
+
* 「以下の通り。」「円錐形の麦わら帽子。」のような名詞で終わる文を数えると、そのままである調の少数派になっていた。
|
|
7
|
+
* 書いた形でも原形でも当てる。「ください」の原形は「くださる」で、原形だけを見ると丁寧な文末を見落とす。
|
|
8
|
+
*/
|
|
9
|
+
export declare const registerOf: (ending: readonly Token[], preceding: readonly Token[], polite: readonly string[]) => Register | undefined;
|
|
10
|
+
/**
|
|
11
|
+
* 「!」「?」で切れた文が、間を置かずに助詞で始まる次の文へ続くか。「作っていた!!という人」「導入しませんか?が断られた」は、
|
|
12
|
+
* 分割器が切っても一つの文で、切れ目の手前は書き手の文末ではない。「。」で切れた文は続きとは見ない。
|
|
13
|
+
*/
|
|
14
|
+
export declare const continuesInto: (sentence: Sentence, next: Sentence | undefined) => boolean;
|
|
15
|
+
/** 位置 offset を含む一番外側の箇条書きの始まり。本文なら undefined。入れ子の項目は外側の箇条書きと一緒に見る。 */
|
|
16
|
+
export declare const outermostList: (offset: number, lists: readonly Span[]) => number | undefined;
|
|
17
|
+
/** group: 文が入っている箇条書き(outermostList の値)。本文なら undefined。 */
|
|
18
|
+
export type Judged = {
|
|
19
|
+
readonly register: Register;
|
|
20
|
+
readonly group: number | undefined;
|
|
21
|
+
};
|
|
22
|
+
/** 混ざった文と、その文が属する本文または箇条書きの中の少数派の数。 */
|
|
23
|
+
export type Slip<T extends Judged> = {
|
|
24
|
+
readonly entry: T;
|
|
25
|
+
readonly count: number;
|
|
26
|
+
};
|
|
27
|
+
/**
|
|
28
|
+
* 調子が混ざった文を、judged の順で。本文は本文どうし、箇条書きは 1 つずつ、その中で揃っているかを見る。
|
|
29
|
+
* ですます調の本文に常体の箇条書きを置くのはよくある書き方で、箇条書きが丸ごと揃っていれば混在ではない。
|
|
30
|
+
*/
|
|
31
|
+
export declare const slipsOf: <T extends Judged>(judged: readonly T[], limit: number) => Slip<T>[];
|
|
32
|
+
//# sourceMappingURL=register.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"register.d.ts","sourceRoot":"","sources":["../../src/detectors/register.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,KAAK,EAAE,MAAM,cAAc,CAAC;AAE1D,MAAM,MAAM,QAAQ,GAAG,QAAQ,GAAG,OAAO,CAAC;AAqB1C;;;;;GAKG;AACH,eAAO,MAAM,UAAU,GAAI,QAAQ,SAAS,KAAK,EAAE,EAAE,WAAW,SAAS,KAAK,EAAE,EAAE,QAAQ,SAAS,MAAM,EAAE,KAAG,QAAQ,GAAG,SAKxH,CAAC;AAIF;;;GAGG;AACH,eAAO,MAAM,aAAa,GAAI,UAAU,QAAQ,EAAE,MAAM,QAAQ,GAAG,SAAS,KAAG,OACkD,CAAC;AAElI,sEAAsE;AACtE,eAAO,MAAM,aAAa,GAAI,QAAQ,MAAM,EAAE,OAAO,SAAS,IAAI,EAAE,KAAG,MAAM,GAAG,SAG6C,CAAC;AAE9H,2DAA2D;AAC3D,MAAM,MAAM,MAAM,GAAG;IAAE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,CAAC;IAAC,QAAQ,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,CAAA;CAAE,CAAC;AAEzF,uCAAuC;AACvC,MAAM,MAAM,IAAI,CAAC,CAAC,SAAS,MAAM,IAAI;IAAE,QAAQ,CAAC,KAAK,EAAE,CAAC,CAAC;IAAC,QAAQ,CAAC,KAAK,EAAE,MAAM,CAAA;CAAE,CAAC;AAoBnF;;;GAGG;AACH,eAAO,MAAM,OAAO,GAAI,CAAC,SAAS,MAAM,EAAE,QAAQ,SAAS,CAAC,EAAE,EAAE,OAAO,MAAM,KAAG,IAAI,CAAC,CAAC,CAAC,EAGtF,CAAC"}
|
|
@@ -0,0 +1,62 @@
|
|
|
1
|
+
const PREDICATE = new Set(["VERB", "ADJ", "AUX"]);
|
|
2
|
+
const isPredicate = (token) => PREDICATE.has(token.pos);
|
|
3
|
+
const isDependent = (token) => token?.features?.["NounType"] === "Dependent";
|
|
4
|
+
/** preceding の後ろの述語の連なり: 助動詞の連なりと、その前の語一つ。「おかけしましたこと」なら「し・まし・た」。 */
|
|
5
|
+
const chainAtEnd = (preceding) => {
|
|
6
|
+
const head = preceding.findLastIndex((token) => token.pos !== "AUX");
|
|
7
|
+
return preceding.slice(Math.max(head, 0));
|
|
8
|
+
};
|
|
9
|
+
/**
|
|
10
|
+
* 非自立名詞で終わる文末は、手前の述語の連なりまで見る。「予約できること。」「おかけしましたこと。」は手前の述語が調子を持つ。
|
|
11
|
+
* 要件や規程のである調は「〜こと。」で書く。「以下のとおり。」は手前も述語でないので、調子を持たないまま。
|
|
12
|
+
*/
|
|
13
|
+
const withPreceding = (ending, preceding) => isDependent(ending[0]) ? [...chainAtEnd(preceding), ...ending] : ending;
|
|
14
|
+
/**
|
|
15
|
+
* 文末の語の調子。preceding は文末の語より前の語。
|
|
16
|
+
* 文末に述語(動詞・形容詞・助動詞)が無ければ、ですます調でもである調でもないので undefined。
|
|
17
|
+
* 「以下の通り。」「円錐形の麦わら帽子。」のような名詞で終わる文を数えると、そのままである調の少数派になっていた。
|
|
18
|
+
* 書いた形でも原形でも当てる。「ください」の原形は「くださる」で、原形だけを見ると丁寧な文末を見落とす。
|
|
19
|
+
*/
|
|
20
|
+
export const registerOf = (ending, preceding, polite) => {
|
|
21
|
+
const judged = withPreceding(ending, preceding);
|
|
22
|
+
if (!judged.some(isPredicate))
|
|
23
|
+
return undefined;
|
|
24
|
+
const isPolite = judged.some((token) => polite.includes(token.surface) || (token.lemma !== undefined && polite.includes(token.lemma)));
|
|
25
|
+
return isPolite ? "polite" : "plain";
|
|
26
|
+
};
|
|
27
|
+
const EXCLAIMED = /[!?!?]$/u;
|
|
28
|
+
/**
|
|
29
|
+
* 「!」「?」で切れた文が、間を置かずに助詞で始まる次の文へ続くか。「作っていた!!という人」「導入しませんか?が断られた」は、
|
|
30
|
+
* 分割器が切っても一つの文で、切れ目の手前は書き手の文末ではない。「。」で切れた文は続きとは見ない。
|
|
31
|
+
*/
|
|
32
|
+
export const continuesInto = (sentence, next) => EXCLAIMED.test(sentence.text) && next !== undefined && next.span.start === sentence.span.end && next.tokens?.[0]?.pos === "ADP";
|
|
33
|
+
/** 位置 offset を含む一番外側の箇条書きの始まり。本文なら undefined。入れ子の項目は外側の箇条書きと一緒に見る。 */
|
|
34
|
+
export const outermostList = (offset, lists) => lists
|
|
35
|
+
.filter((list) => offset >= list.start && offset < list.end)
|
|
36
|
+
.reduce((outer, list) => (outer === undefined || list.start < outer ? list.start : outer), undefined);
|
|
37
|
+
const countIn = (judged, register) => judged.filter((entry) => entry.register === register).length;
|
|
38
|
+
/** 少ないほう。同数なら文書全体で少ないほう、それも同数なら丁寧体。 */
|
|
39
|
+
const minorityOf = (group, whole) => {
|
|
40
|
+
const difference = countIn(group, "polite") - countIn(group, "plain");
|
|
41
|
+
if (difference !== 0)
|
|
42
|
+
return difference < 0 ? "polite" : "plain";
|
|
43
|
+
return countIn(whole, "polite") <= countIn(whole, "plain") ? "polite" : "plain";
|
|
44
|
+
};
|
|
45
|
+
const slipsInGroup = (judged, group, limit) => {
|
|
46
|
+
const members = judged.filter((entry) => entry.group === group);
|
|
47
|
+
const minority = minorityOf(members, judged);
|
|
48
|
+
const slips = members.filter((entry) => entry.register === minority);
|
|
49
|
+
// 片方しか無ければ揃っている。少数派が閾値を超えて多ければ、混在ではなく別の文体。
|
|
50
|
+
if (slips.length === 0 || slips.length > limit)
|
|
51
|
+
return [];
|
|
52
|
+
return slips.map((entry) => ({ entry, count: slips.length }));
|
|
53
|
+
};
|
|
54
|
+
/**
|
|
55
|
+
* 調子が混ざった文を、judged の順で。本文は本文どうし、箇条書きは 1 つずつ、その中で揃っているかを見る。
|
|
56
|
+
* ですます調の本文に常体の箇条書きを置くのはよくある書き方で、箇条書きが丸ごと揃っていれば混在ではない。
|
|
57
|
+
*/
|
|
58
|
+
export const slipsOf = (judged, limit) => {
|
|
59
|
+
const slips = [...new Set(judged.map((entry) => entry.group))].flatMap((group) => slipsInGroup(judged, group, limit));
|
|
60
|
+
return judged.flatMap((entry) => slips.filter((slip) => slip.entry === entry));
|
|
61
|
+
};
|
|
62
|
+
//# sourceMappingURL=register.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"register.js","sourceRoot":"","sources":["../../src/detectors/register.ts"],"names":[],"mappings":"AAIA,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;AAElD,MAAM,WAAW,GAAG,CAAC,KAAY,EAAW,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;AAExE,MAAM,WAAW,GAAG,CAAC,KAAwB,EAAW,EAAE,CAAC,KAAK,EAAE,QAAQ,EAAE,CAAC,UAAU,CAAC,KAAK,WAAW,CAAC;AAEzG,oEAAoE;AACpE,MAAM,UAAU,GAAG,CAAC,SAA2B,EAAoB,EAAE;IACnE,MAAM,IAAI,GAAG,SAAS,CAAC,aAAa,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,KAAK,KAAK,CAAC,CAAC;IACrE,OAAO,SAAS,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC;AAC5C,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,aAAa,GAAG,CAAC,MAAwB,EAAE,SAA2B,EAAoB,EAAE,CAChG,WAAW,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,SAAS,CAAC,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;AAE1E;;;;;GAKG;AACH,MAAM,CAAC,MAAM,UAAU,GAAG,CAAC,MAAwB,EAAE,SAA2B,EAAE,MAAyB,EAAwB,EAAE;IACnI,MAAM,MAAM,GAAG,aAAa,CAAC,MAAM,EAAE,SAAS,CAAC,CAAC;IAChD,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,WAAW,CAAC;QAAE,OAAO,SAAS,CAAC;IAChD,MAAM,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,KAAK,SAAS,IAAI,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;IACvI,OAAO,QAAQ,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC;AACvC,CAAC,CAAC;AAEF,MAAM,SAAS,GAAG,UAAU,CAAC;AAE7B;;;GAGG;AACH,MAAM,CAAC,MAAM,aAAa,GAAG,CAAC,QAAkB,EAAE,IAA0B,EAAW,EAAE,CACvF,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,IAAI,KAAK,SAAS,IAAI,IAAI,CAAC,IAAI,CAAC,KAAK,KAAK,QAAQ,CAAC,IAAI,CAAC,GAAG,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,GAAG,KAAK,KAAK,CAAC;AAElI,sEAAsE;AACtE,MAAM,CAAC,MAAM,aAAa,GAAG,CAAC,MAAc,EAAE,KAAsB,EAAsB,EAAE,CAC1F,KAAK;KACF,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,MAAM,IAAI,IAAI,CAAC,KAAK,IAAI,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC;KAC3D,MAAM,CAAqB,CAAC,KAAK,EAAE,IAAI,EAAE,EAAE,CAAC,CAAC,KAAK,KAAK,SAAS,IAAI,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,EAAE,SAAS,CAAC,CAAC;AAQ9H,MAAM,OAAO,GAAG,CAAC,MAAyB,EAAE,QAAkB,EAAU,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC,MAAM,CAAC;AAExI,uCAAuC;AACvC,MAAM,UAAU,GAAG,CAAC,KAAwB,EAAE,KAAwB,EAAY,EAAE;IAClF,MAAM,UAAU,GAAG,OAAO,CAAC,KAAK,EAAE,QAAQ,CAAC,GAAG,OAAO,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtE,IAAI,UAAU,KAAK,CAAC;QAAE,OAAO,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC;IACjE,OAAO,OAAO,CAAC,KAAK,EAAE,QAAQ,CAAC,IAAI,OAAO,CAAC,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,OAAO,CAAC;AAClF,CAAC,CAAC;AAEF,MAAM,YAAY,GAAG,CAAmB,MAAoB,EAAE,KAAyB,EAAE,KAAa,EAAa,EAAE;IACnH,MAAM,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,KAAK,KAAK,CAAC,CAAC;IAChE,MAAM,QAAQ,GAAG,UAAU,CAAC,OAAO,EAAE,MAAM,CAAC,CAAC;IAC7C,MAAM,KAAK,GAAG,OAAO,CAAC,MAAM,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,QAAQ,KAAK,QAAQ,CAAC,CAAC;IACrE,2CAA2C;IAC3C,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,IAAI,KAAK,CAAC,MAAM,GAAG,KAAK;QAAE,OAAO,EAAE,CAAC;IAC1D,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC,CAAC;AAChE,CAAC,CAAC;AAEF;;;GAGG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,CAAmB,MAAoB,EAAE,KAAa,EAAa,EAAE;IAC1F,MAAM,KAAK,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,YAAY,CAAC,MAAM,EAAE,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC;IACtH,OAAO,MAAM,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC;AACjF,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-ending.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"sentence-ending.d.ts","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,QAAQ,EAAqB,MAAM,cAAc,CAAC;AAkBhE,eAAO,MAAM,cAAc,EAAE,QAgB5B,CAAC"}
|
|
@@ -1,38 +1,27 @@
|
|
|
1
|
-
import { endingTokens,
|
|
2
|
-
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
7
|
-
|
|
8
|
-
|
|
9
|
-
* 述語を持たないものは文として数えない。見出しの下の `MaaSサービス` のような
|
|
10
|
-
* 名前だけの行がそのまま「である調」の少数派になり、実文書の誤検知はすべてこれだった。
|
|
11
|
-
*/
|
|
12
|
-
const isSentence = (sentence) => isClosed(sentence) && hasPredicate(sentence) && endingTokens(sentence).length > 0;
|
|
13
|
-
/**
|
|
14
|
-
* 文末の語だけを見る。文の途中の引用(「ご覧ください」という表現を使う。)は文末の調子ではない。
|
|
15
|
-
* 書いた形でも原形でも当てる。「ください」の原形は「くださる」で、原形だけを見ると丁寧な文末を見落とす。
|
|
16
|
-
*/
|
|
17
|
-
const isPolite = (sentence, polite) => endingTokens(sentence).some((token) => polite.includes(token.surface) || (token.lemma !== undefined && polite.includes(token.lemma)));
|
|
1
|
+
import { endingTokens, isClosed } from "../sentence-shape.js";
|
|
2
|
+
import { continuesInto, outermostList, registerOf, slipsOf } from "./register.js";
|
|
3
|
+
const registerOfSentence = (sentence, polite) => {
|
|
4
|
+
const ending = endingTokens(sentence);
|
|
5
|
+
const tokens = sentence.tokens ?? [];
|
|
6
|
+
const head = ending[0] === undefined ? -1 : tokens.indexOf(ending[0]);
|
|
7
|
+
return registerOf(ending, head > 0 ? tokens.slice(0, head) : [], polite);
|
|
8
|
+
};
|
|
18
9
|
export const sentenceEnding = (doc, options) => {
|
|
19
10
|
const polite = (options.lexicon ?? []).map((entry) => entry.pattern);
|
|
20
|
-
const
|
|
21
|
-
const
|
|
22
|
-
|
|
23
|
-
|
|
24
|
-
|
|
25
|
-
|
|
26
|
-
|
|
27
|
-
return judged
|
|
28
|
-
.filter((entry) => entry.polite === minorityIsPolite)
|
|
29
|
-
.map(({ sentence }) => ({
|
|
11
|
+
const lists = doc.lists.map((list) => list.span);
|
|
12
|
+
const judged = doc.sentences.flatMap((sentence, index) => {
|
|
13
|
+
if (!isClosed(sentence) || continuesInto(sentence, doc.sentences[index + 1]))
|
|
14
|
+
return [];
|
|
15
|
+
const register = registerOfSentence(sentence, polite);
|
|
16
|
+
return register === undefined ? [] : [{ sentence, register, group: outermostList(sentence.span.start, lists) }];
|
|
17
|
+
});
|
|
18
|
+
return slipsOf(judged, options.limit).map(({ entry: { sentence }, count }) => ({
|
|
30
19
|
rule: "no-mixed-desumasu",
|
|
31
20
|
severity: "warning",
|
|
32
21
|
line: 0,
|
|
33
22
|
column: 0,
|
|
34
23
|
quote: sentence.text.trim(),
|
|
35
|
-
values: { count
|
|
24
|
+
values: { count, limit: options.limit, offset: sentence.span.start },
|
|
36
25
|
}));
|
|
37
26
|
};
|
|
38
27
|
//# sourceMappingURL=sentence-ending.js.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"sentence-ending.js","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,
|
|
1
|
+
{"version":3,"file":"sentence-ending.js","sourceRoot":"","sources":["../../src/detectors/sentence-ending.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAE,QAAQ,EAAE,MAAM,sBAAsB,CAAC;AAC9D,OAAO,EAAE,aAAa,EAAE,aAAa,EAAE,UAAU,EAAE,OAAO,EAAiB,MAAM,eAAe,CAAC;AAYjG,MAAM,kBAAkB,GAAG,CAAC,QAAkB,EAAE,MAAyB,EAAwB,EAAE;IACjG,MAAM,MAAM,GAAG,YAAY,CAAC,QAAQ,CAAC,CAAC;IACtC,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC;IACrC,MAAM,IAAI,GAAG,MAAM,CAAC,CAAC,CAAC,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;IACtE,OAAO,UAAU,CAAC,MAAM,EAAE,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;AAC3E,CAAC,CAAC;AAEF,MAAM,CAAC,MAAM,cAAc,GAAa,CAAC,GAAG,EAAE,OAAO,EAAa,EAAE;IAClE,MAAM,MAAM,GAAG,CAAC,OAAO,CAAC,OAAO,IAAI,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IACrE,MAAM,KAAK,GAAG,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;IACjD,MAAM,MAAM,GAAG,GAAG,CAAC,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,KAAK,EAAW,EAAE;QAChE,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,aAAa,CAAC,QAAQ,EAAE,GAAG,CAAC,SAAS,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC;YAAE,OAAO,EAAE,CAAC;QACxF,MAAM,QAAQ,GAAG,kBAAkB,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAC;QACtD,OAAO,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,QAAQ,EAAE,QAAQ,EAAE,KAAK,EAAE,aAAa,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,KAAK,CAAC,EAAE,CAAC,CAAC;IAClH,CAAC,CAAC,CAAC;IACH,OAAO,OAAO,CAAC,MAAM,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC,GAAG,CAAC,CAAC,EAAE,KAAK,EAAE,EAAE,QAAQ,EAAE,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC,CAAC;QAC7E,IAAI,EAAE,mBAAmB;QACzB,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,CAAC;QACP,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,QAAQ,CAAC,IAAI,CAAC,IAAI,EAAE;QAC3B,MAAM,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE;KACrE,CAAC,CAAC,CAAC;AACN,CAAC,CAAC"}
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"signals.d.ts","sourceRoot":"","sources":["../../src/detectors/signals.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"signals.d.ts","sourceRoot":"","sources":["../../src/detectors/signals.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,QAAQ,EAAoD,MAAM,cAAc,CAAC;AA4B/F,eAAO,MAAM,YAAY,EAAE,QAe1B,CAAC;AA4EF,eAAO,MAAM,eAAe,EAAE,QAoB7B,CAAC;AA4EF,eAAO,MAAM,gBAAgB,EAAE,QAkB9B,CAAC;AA4BF,eAAO,MAAM,gBAAgB,EAAE,QAkB9B,CAAC;AAQF,eAAO,MAAM,WAAW,EAAE,QAezB,CAAC"}
|