@sushichan044/kg-core 0.10.0 → 0.10.1

package/dist/index.mjs

@@ -1,5 +1,5 @@
-import { _ as readonlyObject, c as ValidationIssue, f as NamespacedId, l as ManuscriptResult, p as ManuscriptRange, s as graphemeSegmenter, u as ManuscriptDiagnostic } from "./question-or-exclamation-spacing-Bv-mF2ZE.mjs";
-import { a as PaperSize, d as ManuscriptOffsets, f as ManuscriptAppearanceSettings, g as ParsedManuscript, h as FontPresetId, i as ManuscriptGeometry, l as NovelCompositionSettings, m as FontSizePt, n as novelComposer, p as PaperSizeId, u as NovelFlowSettings } from "./novel-composer-C0EqiM0f.mjs";
+import { c as graphemeSegmenter, d as ManuscriptDiagnostic, l as ValidationIssue, m as ManuscriptRange, p as NamespacedId, u as ManuscriptResult, v as readonlyObject } from "./question-or-exclamation-spacing-Cp3eOGlA.mjs";
+import { a as PaperSize, d as ManuscriptOffsets, f as ManuscriptAppearanceSettings, g as ParsedManuscript, h as FontPresetId, i as ManuscriptGeometry, l as NovelCompositionSettings, m as FontSizePt, n as novelComposer, p as PaperSizeId, u as NovelFlowSettings } from "./novel-composer-D_orH6ET.mjs";
 import { n as assertNever, t as describeCause } from "./describe-cause-BV8mZ3Um.mjs";
 import * as v from "valibot";
 const Rejection = { schema: readonlyObject({ reason: v.string() }) };

package/dist/lint.mjs

@@ -1,4 +1,4 @@
-import { _ as readonlyObject, a as OPENING_BRACKETS, c as ValidationIssue, d as DiagnosticSeverity, f as NamespacedId, g as readonlyArray, i as CLOSING_BRACKETS, l as ManuscriptResult, n as leadingSpaces, o as characterClass, p as ManuscriptRange, s as graphemeSegmenter, t as questionOrExclamationSpacings, u as ManuscriptDiagnostic } from "./question-or-exclamation-spacing-Bv-mF2ZE.mjs";
+import { _ as readonlyArray, a as CLOSING_BRACKETS, c as graphemeSegmenter, d as ManuscriptDiagnostic, f as DiagnosticSeverity, l as ValidationIssue, m as ManuscriptRange, n as leadingSpaces, o as OPENING_BRACKETS, p as NamespacedId, s as characterClass, t as questionOrExclamationSpacings, u as ManuscriptResult, v as readonlyObject } from "./question-or-exclamation-spacing-Cp3eOGlA.mjs";
 import { n as assertNever, t as describeCause } from "./describe-cause-BV8mZ3Um.mjs";
 import * as v from "valibot";
 //#region src/proofreading/invalid-rule-options.ts

package/dist/{novel-composer-C0EqiM0f.mjs → novel-composer-D_orH6ET.mjs}

@@ -1,4 +1,4 @@
-import { _ as readonlyObject, a as OPENING_BRACKETS, f as NamespacedId, g as readonlyArray, h as nonNegativeInteger, i as CLOSING_BRACKETS, l as ManuscriptResult, m as TextRange, p as ManuscriptRange, r as DIVIDING_PUNCTUATION, s as graphemeSegmenter, t as questionOrExclamationSpacings } from "./question-or-exclamation-spacing-Bv-mF2ZE.mjs";
+import { _ as readonlyArray, a as CLOSING_BRACKETS, c as graphemeSegmenter, g as nonNegativeInteger, h as TextRange, i as TATE_CHU_YOKO_MARKS, m as ManuscriptRange, o as OPENING_BRACKETS, p as NamespacedId, r as DIVIDING_PUNCTUATION, s as characterClass, t as questionOrExclamationSpacings, u as ManuscriptResult, v as readonlyObject } from "./question-or-exclamation-spacing-Cp3eOGlA.mjs";
 import * as v from "valibot";
 import { eastAsianWidth } from "get-east-asian-width";
 const BoldAnnotation = { schema: readonlyObject({
@@ -363,26 +363,46 @@ const BoundaryRule = { resolve: (left, right, profile, context) => {
 const ASCII_ALPHANUMERIC = /^[A-Za-z0-9]$/u;
 const ASCII_TWO_DIGITS = /^[0-9]{2}$/u;
 const UPRIGHT_LATIN_ABBREVIATION = /^(?:[A-Z]+|[A-Z][a-z]{1,2})$/u;
+const TATE_CHU_YOKO_MARK = new RegExp(`^[${characterClass(TATE_CHU_YOKO_MARKS)}]$`, "u");
 const FULLWIDTH_ALPHANUMERIC = /^[Ａ-Ｚａ-ｚ０-９]$/u;
+function stretchEnd(graphemes, start, pattern) {
+	let end = start + 1;
+	while (pattern.test(graphemes[end]?.value ?? "")) end += 1;
+	return end;
+}
+function presentationOf(text) {
+	if (ASCII_ALPHANUMERIC.test(text[0] ?? "")) {
+		if (ASCII_TWO_DIGITS.test(text)) return "tate-chu-yoko";
+		return text.length === 1 || UPRIGHT_LATIN_ABBREVIATION.test(text) ? "upright" : "sideways";
+	}
+	return FULLWIDTH_ALPHANUMERIC.test(text) ? "upright" : "mixed";
+}
 function recognize(graphemes) {
 	const runs = [];
-	let cursor = 0;
-	while (cursor < graphemes.length) {
-		const first = graphemes[cursor];
-		if (first === void 0) break;
-		const start = cursor;
-		cursor += 1;
-		const ascii = ASCII_ALPHANUMERIC.test(first.value);
-		if (ascii) while (ASCII_ALPHANUMERIC.test(graphemes[cursor]?.value ?? "")) cursor += 1;
-		const members = graphemes.slice(start, cursor);
-		const text = members.map(({ value }) => value).join("");
-		const presentation = ascii ? ASCII_TWO_DIGITS.test(text) ? "tate-chu-yoko" : text.length === 1 || UPRIGHT_LATIN_ABBREVIATION.test(text) ? "upright" : "sideways" : FULLWIDTH_ALPHANUMERIC.test(first.value) ? "upright" : "mixed";
+	const push = (members, presentation) => {
 		const range = ManuscriptRange.merge(members.map((member) => member.range));
 		if (range !== null) runs.push({
 			range,
 			presentation,
 			members
 		});
+	};
+	let cursor = 0;
+	while (cursor < graphemes.length) {
+		const first = graphemes[cursor];
+		if (first === void 0) break;
+		if (TATE_CHU_YOKO_MARK.test(first.value)) {
+			const end = stretchEnd(graphemes, cursor, TATE_CHU_YOKO_MARK);
+			const members = graphemes.slice(cursor, end);
+			if (members.length === 2 || members.length === 3) push(members, "tate-chu-yoko");
+			else for (const member of members) push([member], "mixed");
+			cursor = end;
+			continue;
+		}
+		const end = ASCII_ALPHANUMERIC.test(first.value) ? stretchEnd(graphemes, cursor, ASCII_ALPHANUMERIC) : cursor + 1;
+		const members = graphemes.slice(cursor, end);
+		push(members, presentationOf(members.map(({ value }) => value).join("")));
+		cursor = end;
 	}
 	return runs;
 }
@@ -914,6 +934,7 @@ const OPENING = /* @__PURE__ */ new Set(`${OPENING_BRACKETS}【〘〝｟«`);
 const CLOSING = /* @__PURE__ */ new Set(`${CLOSING_BRACKETS}】〙〟｠»`);
 const HYPHENS = /* @__PURE__ */ new Set("‐‑⁃–");
 const DIVIDING_MARKS = new Set(DIVIDING_PUNCTUATION);
+const COMBINED_DIVIDING_MARKS = new Set(TATE_CHU_YOKO_MARKS);
 const MIDDLE_DOTS = /* @__PURE__ */ new Set("・：；");
 const SMALL_KANA = /* @__PURE__ */ new Set("ァィゥェォッャュョヮヵヶぁぃぅぇぉっゃゅょゎゕゖ");
 const ITERATION_MARKS = /* @__PURE__ */ new Set("ヽヾゝゞ々〻");
@@ -928,7 +949,8 @@ const WESTERN = /^[\p{Script=Latin}\p{Script=Greek}\p{Script=Cyrillic}]$/u;
 /**
 * Order matters: a character inside a tate-chu-yoko group is cl-30 (3.2.5) and an upright Latin
 * letter or Arabic numeral is cl-19 (3.2.4), whatever either would be on its own; anything left
-* unmatched is cl-19 too, the class JLReq reserves for kanji and the like.
+* unmatched is cl-19 too, the class JLReq reserves for kanji and the like. A tate-chu-yoko `！？`
+* stays cl-04, the class JLReq gives `⁉`, so it still may not open a line.
 *
 * The quarter em 3.2.6 sets between Japanese and a Western character or numeral applies only to a
 * character rotated 90 degrees in vertical writing mode or mixed into horizontal writing mode —
@@ -937,7 +959,7 @@ const WESTERN = /^[\p{Script=Latin}\p{Script=Greek}\p{Script=Cyrillic}]$/u;
 * the same as a tate-chu-yoko run's 3.2.5.
 */
 function classify({ value, presentation }) {
-	if (presentation === "tate-chu-yoko") return "cl-30";
+	if (presentation === "tate-chu-yoko") return COMBINED_DIVIDING_MARKS.has(value) ? "cl-04" : "cl-30";
 	if (presentation === "upright") return "cl-19";
 	if (OPENING.has(value)) return "cl-01";
 	if (CLOSING.has(value)) return "cl-02";
@@ -1727,9 +1749,28 @@ const MeasurementSession = { create: (provider, settings) => {
 				renderSpan: cluster.renderSpan,
 				value: text.slice(cluster.textRange.start, cluster.textRange.end)
 			}));
+			const segments = [...graphemeSegmenter.segment(text)];
+			if (kind === "base" && presentation === "tate-chu-yoko") {
+				const advanceEm = full.advanceEm / segments.length;
+				return segments.map(({ segment, index }, position) => {
+					const span = {
+						offsetEm: position * advanceEm,
+						advanceEm
+					};
+					return {
+						value: segment,
+						textRange: MeasurementTextRange.of({
+							start: index,
+							end: index + segment.length
+						}),
+						layoutSpan: span,
+						renderSpan: span
+					};
+				});
+			}
 			const pieces = [];
 			let offsetEm = 0;
-			for (const { segment, index } of graphemeSegmenter.segment(text)) {
+			for (const { segment, index } of segments) {
 				const measured = get(segment, kind, presentation);
 				if (measured === void 0) return void 0;
 				const cluster = measured.kind === "clustered" ? measured.clusters[0] : void 0;
@@ -2772,7 +2813,7 @@ const logicalRunMeasurer = (request) => {
 	const segments = [...graphemeSegmenter.segment(request.text)];
 	return {
 		kind: "advance",
-		advanceEm: (request.kind === "base" && request.presentation === "upright" ? segments.length : segments.reduce((total, { segment }) => total + graphemeWidth(segment), 0)) * request.scale
+		advanceEm: (request.kind === "base" && request.presentation === "upright" ? segments.length : request.kind === "base" && request.presentation === "tate-chu-yoko" ? 1 : segments.reduce((total, { segment }) => total + graphemeWidth(segment), 0)) * request.scale
 	};
 };
 //#endregion

package/dist/plugin.d.mts

@@ -1,5 +1,5 @@
 import { a as ManuscriptResult, i as ParsedManuscript, n as ValidationIssue, r as NamespacedId } from "./composed-manuscript-C0f3mxKs.mjs";
-import { a as RunMeasurer, c as MeasurementRequest, d as ManuscriptComposer, f as ManuscriptParser, o as logicalRunMeasurer, p as Rejection, r as createNovelComposer, s as RunMeasurement, t as PresentationKind } from "./presentation-kind-DCgXI574.mjs";
+import { a as RunMeasurer, c as MeasurementRequest, d as ManuscriptComposer, f as ManuscriptParser, o as logicalRunMeasurer, p as Rejection, r as createNovelComposer, s as RunMeasurement, t as PresentationKind } from "./presentation-kind-BwIZFSZS.mjs";
 import * as v from "valibot";
 //#region src/composer/measurement-text-range.d.ts
 declare const MeasurementTextRangeSchema: v.SchemaWithPipe<readonly [v.SchemaWithPipe<readonly [v.SchemaWithPipe<readonly [v.StrictObjectSchema<{

package/dist/plugin.mjs

@@ -1,5 +1,5 @@
-import { _ as readonlyObject, c as ValidationIssue, f as NamespacedId, l as ManuscriptResult } from "./question-or-exclamation-spacing-Bv-mF2ZE.mjs";
-import { c as MeasurementTextRange, g as ParsedManuscript, h as FontPresetId, m as FontSizePt, o as RunMeasurement, r as logicalRunMeasurer, s as MeasuredCluster, t as createNovelComposer } from "./novel-composer-C0EqiM0f.mjs";
+import { l as ValidationIssue, p as NamespacedId, u as ManuscriptResult, v as readonlyObject } from "./question-or-exclamation-spacing-Cp3eOGlA.mjs";
+import { c as MeasurementTextRange, g as ParsedManuscript, h as FontPresetId, m as FontSizePt, o as RunMeasurement, r as logicalRunMeasurer, s as MeasuredCluster, t as createNovelComposer } from "./novel-composer-D_orH6ET.mjs";
 import * as v from "valibot";
 const PresentationKind = { schema: v.picklist([
 	"mixed",

package/dist/{presentation-kind-DCgXI574.d.mts → presentation-kind-BwIZFSZS.d.mts}

@@ -12202,57 +12202,6 @@ declare const novelComposer: Readonly<{
           }> | null;
           inlineSizeEm: number;
           readonly items: readonly (Readonly<{
-            kind: "glue";
-            origin: "generated";
-            offsetEm: number;
-            widthEm: number;
-            naturalWidthEm: number;
-            adjustment: "natural" | "shrunk" | "stretched";
-          }> | Readonly<{
-            kind: "glue";
-            origin: "source";
-            value: string;
-            readonly range: Readonly<{
-              source: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"SourceRange">;
-              display: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"DisplayRange">;
-              graphemes: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"GraphemeRange">;
-            }>;
-            offsetEm: number;
-            widthEm: number;
-            naturalWidthEm: number;
-            adjustment: "natural" | "shrunk" | "stretched";
-          }> | Readonly<{
-            kind: "kern";
-            offsetEm: number;
-            widthEm: number;
-          }> | Readonly<{
-            kind: "suppressed";
-            value: string;
-            readonly range: Readonly<{
-              source: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"SourceRange">;
-              display: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"DisplayRange">;
-              graphemes: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"GraphemeRange">;
-            }>;
-            reason: "question-or-exclamation-gap";
-          }> | Readonly<{
             value: string;
             readonly range: Readonly<{
               source: Readonly<{
@@ -12388,6 +12337,57 @@ declare const novelComposer: Readonly<{
             kind: "combined-glyph";
             presentation: "sideways" | "tate-chu-yoko";
             disposition: "placed";
+          }> | Readonly<{
+            kind: "glue";
+            origin: "generated";
+            offsetEm: number;
+            widthEm: number;
+            naturalWidthEm: number;
+            adjustment: "natural" | "shrunk" | "stretched";
+          }> | Readonly<{
+            kind: "glue";
+            origin: "source";
+            value: string;
+            readonly range: Readonly<{
+              source: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"SourceRange">;
+              display: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"DisplayRange">;
+              graphemes: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"GraphemeRange">;
+            }>;
+            offsetEm: number;
+            widthEm: number;
+            naturalWidthEm: number;
+            adjustment: "natural" | "shrunk" | "stretched";
+          }> | Readonly<{
+            kind: "kern";
+            offsetEm: number;
+            widthEm: number;
+          }> | Readonly<{
+            kind: "suppressed";
+            value: string;
+            readonly range: Readonly<{
+              source: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"SourceRange">;
+              display: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"DisplayRange">;
+              graphemes: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"GraphemeRange">;
+            }>;
+            reason: "question-or-exclamation-gap";
           }>)[];
           break: Readonly<{
             kind: "natural";
@@ -12649,57 +12649,6 @@ declare const novelComposer: Readonly<{
           }> | null;
           inlineSizeEm: number;
           readonly items: readonly (Readonly<{
-            kind: "glue";
-            origin: "generated";
-            offsetEm: number;
-            widthEm: number;
-            naturalWidthEm: number;
-            adjustment: "natural" | "shrunk" | "stretched";
-          }> | Readonly<{
-            kind: "glue";
-            origin: "source";
-            value: string;
-            readonly range: Readonly<{
-              source: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"SourceRange">;
-              display: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"DisplayRange">;
-              graphemes: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"GraphemeRange">;
-            }>;
-            offsetEm: number;
-            widthEm: number;
-            naturalWidthEm: number;
-            adjustment: "natural" | "shrunk" | "stretched";
-          }> | Readonly<{
-            kind: "kern";
-            offsetEm: number;
-            widthEm: number;
-          }> | Readonly<{
-            kind: "suppressed";
-            value: string;
-            readonly range: Readonly<{
-              source: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"SourceRange">;
-              display: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"DisplayRange">;
-              graphemes: Readonly<{
-                start: number;
-                end: number;
-              }> & import("valibot").Brand<"GraphemeRange">;
-            }>;
-            reason: "question-or-exclamation-gap";
-          }> | Readonly<{
             value: string;
             readonly range: Readonly<{
               source: Readonly<{
@@ -12835,6 +12784,57 @@ declare const novelComposer: Readonly<{
             kind: "combined-glyph";
             presentation: "sideways" | "tate-chu-yoko";
             disposition: "placed";
+          }> | Readonly<{
+            kind: "glue";
+            origin: "generated";
+            offsetEm: number;
+            widthEm: number;
+            naturalWidthEm: number;
+            adjustment: "natural" | "shrunk" | "stretched";
+          }> | Readonly<{
+            kind: "glue";
+            origin: "source";
+            value: string;
+            readonly range: Readonly<{
+              source: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"SourceRange">;
+              display: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"DisplayRange">;
+              graphemes: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"GraphemeRange">;
+            }>;
+            offsetEm: number;
+            widthEm: number;
+            naturalWidthEm: number;
+            adjustment: "natural" | "shrunk" | "stretched";
+          }> | Readonly<{
+            kind: "kern";
+            offsetEm: number;
+            widthEm: number;
+          }> | Readonly<{
+            kind: "suppressed";
+            value: string;
+            readonly range: Readonly<{
+              source: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"SourceRange">;
+              display: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"DisplayRange">;
+              graphemes: Readonly<{
+                start: number;
+                end: number;
+              }> & import("valibot").Brand<"GraphemeRange">;
+            }>;
+            reason: "question-or-exclamation-gap";
           }>)[];
           break: Readonly<{
             kind: "natural";

package/dist/{question-or-exclamation-spacing-Bv-mF2ZE.mjs → question-or-exclamation-spacing-Cp3eOGlA.mjs}

@@ -195,6 +195,10 @@ function characterClass(characters) {
 * neutral.
 */
 const DIVIDING_PUNCTUATION = "！？‼⁇⁈⁉";
+/**
+* The marks a short run of which is set as one tate-chu-yoko cell, `！？` reading as `⁉`.
+*/
+const TATE_CHU_YOKO_MARKS = "!?！？";
 //#endregion
 //#region src/internal/japanese-spaces.ts
 const LEADING_SPACES = /^[　 \t]+/u;
@@ -255,4 +259,4 @@ function questionOrExclamationSpacings(line) {
 	return [...line.matchAll(MARKS)].map((marks) => analyzeMatch(line, marks)).filter((spacing) => spacing !== void 0);
 }
 //#endregion
-export { readonlyObject as _, OPENING_BRACKETS as a, ValidationIssue as c, DiagnosticSeverity as d, NamespacedId as f, readonlyArray as g, nonNegativeInteger as h, CLOSING_BRACKETS as i, ManuscriptResult as l, TextRange as m, leadingSpaces as n, characterClass as o, ManuscriptRange as p, DIVIDING_PUNCTUATION as r, graphemeSegmenter as s, questionOrExclamationSpacings as t, ManuscriptDiagnostic as u };
+export { readonlyArray as _, CLOSING_BRACKETS as a, graphemeSegmenter as c, ManuscriptDiagnostic as d, DiagnosticSeverity as f, nonNegativeInteger as g, TextRange as h, TATE_CHU_YOKO_MARKS as i, ValidationIssue as l, ManuscriptRange as m, leadingSpaces as n, OPENING_BRACKETS as o, NamespacedId as p, DIVIDING_PUNCTUATION as r, characterClass as s, questionOrExclamationSpacings as t, ManuscriptResult as u, readonlyObject as v };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sushichan044/kg-core",
-  "version": "0.10.0",
+  "version": "0.10.1",
   "description": "Framework-independent parsing, composition, and proofreading for Japanese manuscripts",
   "keywords": [
     "japanese",