@sushichan044/kg-core 0.6.1 → 0.6.2

package/dist/index.mjs

@@ -1062,6 +1062,83 @@ const GridCell = {
 		annotations: []
 	})
 };
+//#endregion
+//#region src/internal/japanese-brackets.ts
+/**
+* Brackets and quotes a Japanese paragraph may open with. Halfwidth forms are included because a
+* manuscript that uses them is still opening a paragraph, not writing prose.
+*/
+const OPENING_BRACKETS = "「『〖〈《（(“\"‘'［[〔｛{＜<";
+/**
+* The closing counterparts, in the same order.
+*/
+const CLOSING_BRACKETS = "」』〗〉》）)”\"’'］]〕｝}＞>";
+/**
+* Escapes the characters a regular expression character class would otherwise read as syntax, so a
+* set declared above can be dropped into a pattern.
+*/
+function characterClass(characters) {
+	return characters.replaceAll(/[\\\]^-]/gu, "\\$&");
+}
+//#endregion
+//#region src/internal/japanese-spaces.ts
+const LEADING_SPACES = /^[　 \t]+/u;
+/**
+* The run of spaces the text begins with, or an empty string. Halfwidth spaces and tabs count: a
+* rule has to see them to say they are the wrong space.
+*/
+function leadingSpaces(text) {
+	return LEADING_SPACES.exec(text)?.[0] ?? "";
+}
+//#endregion
+//#region src/internal/question-or-exclamation-spacing.ts
+/**
+* Runs, not single marks: `！？` is one mark and takes one gap, so the gap belongs after the run.
+*/
+const MARKS = /[？！]+/gu;
+function analyzeMatch(line, marks) {
+	const start = marks.index;
+	if (start === void 0) return void 0;
+	const marksRange = {
+		start,
+		end: start + marks[0].length
+	};
+	const next = line[marksRange.end];
+	if (next === void 0 || "」』〗〉》）)”\"’'］]〕｝}＞>".includes(next)) return void 0;
+	const spaces = leadingSpaces(line.slice(marksRange.end));
+	if (spaces === "") return {
+		kind: "missing",
+		marks: marksRange
+	};
+	const gap = {
+		start: marksRange.end,
+		end: marksRange.end + spaces.length
+	};
+	const following = line[gap.end];
+	if (following !== void 0 && "」』〗〉》）)”\"’'］]〕｝}＞>".includes(following)) return {
+		kind: "before-closing-bracket",
+		marks: marksRange,
+		gap
+	};
+	if (spaces !== "　") return {
+		kind: "invalid-space",
+		marks: marksRange,
+		gap
+	};
+	return {
+		kind: "valid",
+		marks: marksRange,
+		gap
+	};
+}
+/**
+* Classifies the spacing after each run of fullwidth question or exclamation marks in one display
+* line. Runs at the end of the line or immediately before a closing bracket need no gap and are
+* omitted from the result.
+*/
+function questionOrExclamationSpacings(line) {
+	return [...line.matchAll(MARKS)].map((marks) => analyzeMatch(line, marks)).filter((spacing) => spacing !== void 0);
+}
 const GridLine = {
 	schema: readonlyObject({
 		range: v.nullable(ManuscriptRange.schema),
@@ -1191,12 +1268,32 @@ function toCells(sourceLine, annotations) {
 		annotations: ManuscriptAnnotation.overlapping(annotations, grapheme.range)
 	}));
 }
+function hangingGapIndexes(sourceLine) {
+	const line = sourceLine.map(({ value }) => value).join("");
+	const gapStarts = new Set(questionOrExclamationSpacings(line).filter((spacing) => spacing.kind === "valid").map(({ gap }) => gap.start));
+	const indexes = /* @__PURE__ */ new Set();
+	let offset = 0;
+	for (const [index, grapheme] of sourceLine.entries()) {
+		if (gapStarts.has(offset)) indexes.add(index);
+		offset += grapheme.value.length;
+	}
+	return indexes;
+}
 /**
-* Wraps one source line across as many grid lines as its length requires.
+* Wraps one source line across as many grid lines as its length requires. A valid gap after ！ or ？
+* that would start a wrapped line is consumed as hanging punctuation instead of occupying a cell.
 */
-function wrapLine(cells, charsPerLine) {
+function wrapLine(cells, charsPerLine, hangingGaps) {
 	if (cells.length === 0) return [GridLine.padded([], charsPerLine)];
-	return Array.from({ length: Math.ceil(cells.length / charsPerLine) }, (_unused, index) => GridLine.padded(cells.slice(index * charsPerLine, (index + 1) * charsPerLine), charsPerLine));
+	const lines = [];
+	let cursor = 0;
+	while (cursor < cells.length) {
+		if (lines.length > 0 && hangingGaps.has(cursor)) cursor += 1;
+		if (cursor >= cells.length) break;
+		lines.push(GridLine.padded(cells.slice(cursor, cursor + charsPerLine), charsPerLine));
+		cursor += charsPerLine;
+	}
+	return lines;
 }
 function blankLines(count, charsPerLine) {
 	return Array.from({ length: count }, () => GridLine.padded([], charsPerLine));
@@ -1237,8 +1334,7 @@ function buildPages(contentLines, settings) {
 function composeGrid(manuscript, settings) {
 	const { charsPerLine } = settings.grid;
 	const sourceLines = displayedLines(manuscript.graphemes);
-	const cellLines = sourceLines.map((line) => toCells(line, manuscript.annotations));
-	const manuscriptLines = cellLines.flatMap((cells) => wrapLine(cells, charsPerLine));
+	const manuscriptLines = sourceLines.flatMap((line) => wrapLine(toCells(line, manuscript.annotations), charsPerLine, hangingGapIndexes(line)));
 	const pages = buildPages([
 		...blankLines(settings.offsets.document.leading, charsPerLine),
 		...manuscriptLines,
@@ -1248,7 +1344,7 @@ function composeGrid(manuscript, settings) {
 		pages,
 		geometry: ManuscriptGeometry.of(settings.grid, settings.appearance),
 		stats: {
-			chars: cellLines.reduce((total, cells) => total + cells.length, 0),
+			chars: sourceLines.reduce((total, line) => total + line.length, 0),
 			sourceLines: sourceLines.length,
 			pages: pages.length
 		}
@@ -1544,7 +1640,7 @@ function defineMatchRule(options) {
 }
 //#endregion
 //#region src/proofreading/rules/consistent-kanji-opening.ts
-const RULE_ID$6 = "kg/consistent-kanji-opening";
+const RULE_ID$5 = "kg/consistent-kanji-opening";
 const MESSAGE = "「{{ closed }}」と「{{ opened }}」が混在しています。文脈上の使い分けか表記ゆれかを確認してください";
 const KanjiOpeningPairSchema = v.pipe(readonlyObject({
 	closed: v.pipe(v.string(), v.nonEmpty()),
@@ -1598,7 +1694,7 @@ const DEFAULT_PAIRS = [
 * two spellings are a deliberate distinction is the author's call, so the finding is a warning.
 */
 function build$3(pairs) {
-	return defineParsedRule(RULE_ID$6, MESSAGE, (manuscript, report) => {
+	return defineParsedRule(RULE_ID$5, MESSAGE, (manuscript, report) => {
 		for (const { closed, opened } of pairs) {
 			if (!manuscript.displayText.includes(opened)) continue;
 			const index = manuscript.displayText.indexOf(closed);
@@ -1619,7 +1715,7 @@ function createConsistentKanjiOpeningRule(options = {}) {
 	const pairs = v.safeParse(PairsSchema, options.pairs);
 	if (!pairs.success) return ManuscriptResult.fail({
 		kind: "InvalidRuleOptions",
-		ruleId: RULE_ID$6,
+		ruleId: RULE_ID$5,
 		option: "pairs",
 		issues: ValidationIssue.from(pairs.issues)
 	});
@@ -1664,7 +1760,7 @@ const consistentNumeralWidthRule = () => defineMixedWidthRule({
 });
 //#endregion
 //#region src/proofreading/rules/dash.ts
-const RULE_ID$5 = "kg/dash";
+const RULE_ID$4 = "kg/dash";
 const DEFAULT_PREFERRED = "―";
 const DashCharacterSchema = v.picklist([
 	"―",
@@ -1686,7 +1782,7 @@ function build$2(preferred) {
 	return {
 		kind: "parsed",
 		meta: {
-			id: RULE_ID$5,
+			id: RULE_ID$4,
 			messages
 		},
 		check: (manuscript, context) => {
@@ -1715,7 +1811,7 @@ function createDashRule(options = {}) {
 	const preferred = v.safeParse(DashCharacterSchema, options.preferred);
 	if (!preferred.success) return ManuscriptResult.fail({
 		kind: "InvalidRuleOptions",
-		ruleId: RULE_ID$5,
+		ruleId: RULE_ID$4,
 		option: "preferred",
 		issues: ValidationIssue.from(preferred.issues)
 	});
@@ -1742,8 +1838,8 @@ const evenEllipsisRule = () => defineMatchRule({
 });
 //#endregion
 //#region src/proofreading/rules/fullwidth-japanese-punctuation.ts
-const RULE_ID$4 = "kg/fullwidth-japanese-punctuation";
-const MESSAGES$3 = {
+const RULE_ID$3 = "kg/fullwidth-japanese-punctuation";
+const MESSAGES$2 = {
 	"halfwidth-japanese-punctuation": "半角の「{{ character }}」があります。日本語の句読点・括弧・感嘆符・疑問符は全角にしてください",
 	"halfwidth-near-japanese": "日本語の前後に半角の「{{ character }}」があります。全角にすべきか確認してください"
 };
@@ -1788,8 +1884,8 @@ function isEllipsisSubstitute(text, index) {
 const fullwidthJapanesePunctuationRule = () => ({
 	kind: "parsed",
 	meta: {
-		id: RULE_ID$4,
-		messages: MESSAGES$3
+		id: RULE_ID$3,
+		messages: MESSAGES$2
 	},
 	check: (manuscript, context) => {
 		for (const line of splitDisplayLines(manuscript.displayText)) for (const { pattern, messageId, severity, requiresJapanese } of CHECKS) for (const match of line.text.matchAll(pattern)) {
@@ -1807,12 +1903,12 @@ const fullwidthJapanesePunctuationRule = () => ({
 });
 //#endregion
 //#region src/proofreading/rules/max-arabic-numeral-digits.ts
-const RULE_ID$3 = "kg/max-arabic-numeral-digits";
+const RULE_ID$2 = "kg/max-arabic-numeral-digits";
 const DEFAULT_MAX_DIGITS = 2;
 const MaxDigitsSchema = v.pipe(v.number(), v.finite(), v.integer(), v.minValue(1));
 function build$1(maxDigits) {
 	return defineMatchRule({
-		id: RULE_ID$3,
+		id: RULE_ID$2,
 		pattern: /([0-9０-９]+)(?:[.．]([0-9０-９]+))?/gu,
 		accept: (match) => (match[1]?.length ?? 0) > maxDigits || (match[2]?.length ?? 0) > maxDigits,
 		message: `${maxDigits}桁を超えるアラビア数字が使われています`
@@ -1824,7 +1920,7 @@ function createMaxArabicNumeralDigitsRule(options = {}) {
 	const maxDigits = v.safeParse(MaxDigitsSchema, options.maxDigits);
 	if (!maxDigits.success) return ManuscriptResult.fail({
 		kind: "InvalidRuleOptions",
-		ruleId: RULE_ID$3,
+		ruleId: RULE_ID$2,
 		option: "maxDigits",
 		issues: ValidationIssue.from(maxDigits.issues)
 	});
@@ -1852,24 +1948,6 @@ const noConsecutivePunctuationRule = () => defineMatchRule({
 	message: "句読点が連続しています"
 });
 //#endregion
-//#region src/proofreading/rules/internal/brackets.ts
-/**
-* Brackets and quotes a Japanese paragraph may open with. Halfwidth forms are included because a
-* manuscript that uses them is still opening a paragraph, not writing prose.
-*/
-const OPENING_BRACKETS = "「『〖〈《（(“\"‘'［[〔｛{＜<";
-/**
-* The closing counterparts, in the same order.
-*/
-const CLOSING_BRACKETS = "」』〗〉》）)”\"’'］]〕｝}＞>";
-/**
-* Escapes the characters a regular expression character class would otherwise read as syntax, so a
-* set declared above can be dropped into a pattern.
-*/
-function characterClass(characters) {
-	return characters.replaceAll(/[\\\]^-]/gu, "\\$&");
-}
-//#endregion
 //#region src/proofreading/rules/internal/decoration-line.ts
 const DECORATION_LINE = /^[＊*◆◇■□●○▽▼△▲☆★・…―─━ー\-=＝~～\s]+$/u;
 /**
@@ -1880,19 +1958,9 @@ function isDecorationLine(text) {
 	return DECORATION_LINE.test(text);
 }
 //#endregion
-//#region src/proofreading/rules/internal/spaces.ts
-const LEADING_SPACES = /^[　 \t]+/u;
-/**
-* The run of spaces the text begins with, or an empty string. Halfwidth spaces and tabs count: a
-* rule has to see them to say they are the wrong space.
-*/
-function leadingSpaces(text) {
-	return LEADING_SPACES.exec(text)?.[0] ?? "";
-}
-//#endregion
 //#region src/proofreading/rules/paragraph-opening.ts
-const RULE_ID$2 = "kg/paragraph-opening";
-const MESSAGES$2 = {
+const RULE_ID$1 = "kg/paragraph-opening";
+const MESSAGES$1 = {
 	"leading-character": "段落の先頭には全角スペースまたは開き括弧が必要です",
 	"indent-width": "段落冒頭の字下げは全角スペース1字にしてください",
 	"unexpected-indent": "括弧から始まる段落を字下げすることはできません"
@@ -1907,8 +1975,8 @@ function build(openingBrackets) {
 	return {
 		kind: "parsed",
 		meta: {
-			id: RULE_ID$2,
-			messages: MESSAGES$2
+			id: RULE_ID$1,
+			messages: MESSAGES$1
 		},
 		check: (manuscript, context) => {
 			const report = (start, end, messageId) => {
@@ -1938,7 +2006,7 @@ function createParagraphOpeningRule(options = {}) {
 	const brackets = v.safeParse(OpeningBracketsSchema, options.openingBrackets);
 	if (!brackets.success) return ManuscriptResult.fail({
 		kind: "InvalidRuleOptions",
-		ruleId: RULE_ID$2,
+		ruleId: RULE_ID$1,
 		option: "openingBrackets",
 		issues: ValidationIssue.from(brackets.issues)
 	});
@@ -1954,25 +2022,21 @@ const punctuationBeforeClosingQuoteRule = () => defineMatchRule({
 });
 //#endregion
 //#region src/proofreading/rules/space-after-question-or-exclamation.ts
-const RULE_ID$1 = "kg/space-after-question-or-exclamation";
-const MESSAGES$1 = {
+const RULE_ID = "kg/space-after-question-or-exclamation";
+const MESSAGES = {
 	default: "感嘆符または疑問符の直後には全角スペースか閉じ括弧が必要です",
 	"space-width": "感嘆符または疑問符の直後の空白は全角スペース1字にしてください",
 	"before-closing-bracket": "閉じ括弧の直前に空白を置くことはできません"
 };
 /**
-* Runs, not single marks: `！？` is one mark and takes one gap, so the gap belongs after the run.
-*/
-const MARKS = /[？！]+/gu;
-/**
 * Checks the gap that follows every run of ！ or ？. A run at the end of a line or right before a
 * closing bracket needs no gap; anywhere else it takes exactly one ideographic space.
 */
 const spaceAfterQuestionOrExclamationRule = () => ({
 	kind: "parsed",
 	meta: {
-		id: RULE_ID$1,
-		messages: MESSAGES$1
+		id: RULE_ID,
+		messages: MESSAGES
 	},
 	check: (manuscript, context) => {
 		const report = (start, end, messageId) => {
@@ -1981,93 +2045,39 @@ const spaceAfterQuestionOrExclamationRule = () => ({
 				messageId
 			});
 		};
-		for (const line of splitDisplayLines(manuscript.displayText)) for (const marks of line.text.matchAll(MARKS)) {
-			const marksEnd = marks.index + marks[0].length;
-			const next = line.text[marksEnd];
-			if (next === void 0 || "」』〗〉》）)”\"’'］]〕｝}＞>".includes(next)) continue;
-			const spaces = leadingSpaces(line.text.slice(marksEnd));
-			if (spaces === "") {
-				report(line.start + marks.index, line.start + marksEnd, "default");
-				continue;
-			}
-			const following = line.text[marksEnd + spaces.length];
-			const spacesStart = line.start + marksEnd;
-			const spacesEnd = spacesStart + spaces.length;
-			if (following !== void 0 && "」』〗〉》）)”\"’'］]〕｝}＞>".includes(following)) report(spacesStart, spacesEnd, "before-closing-bracket");
-			else if (spaces !== "　") report(spacesStart, spacesEnd, "space-width");
+		for (const line of splitDisplayLines(manuscript.displayText)) for (const spacing of questionOrExclamationSpacings(line.text)) switch (spacing.kind) {
+			case "valid": break;
+			case "missing":
+				report(line.start + spacing.marks.start, line.start + spacing.marks.end, "default");
+				break;
+			case "invalid-space":
+				report(line.start + spacing.gap.start, line.start + spacing.gap.end, "space-width");
+				break;
+			case "before-closing-bracket":
+				report(line.start + spacing.gap.start, line.start + spacing.gap.end, "before-closing-bracket");
+				break;
+			default: assertNever(spacing);
 		}
 	}
 });
 //#endregion
 //#region src/proofreading/rules/variant-character.ts
-const RULE_ID = "kg/variant-character";
-const MESSAGES = {
-	default: "異体字または字形選択子が使われています",
-	"with-suggestion": "異体字または字形選択子が使われています。「{{ suggestion }}」ではありませんか？"
-};
-/**
-* Mongolian free variation selectors, variation selectors (incl. emoji presentation), IVSes: all
-* combine with a preceding base character rather than standing on their own.
-*/
-function isSelector(codePoint) {
-	return codePoint >= 6155 && codePoint <= 6157 || codePoint === 6159 || codePoint >= 65024 && codePoint <= 65039 || codePoint >= 917760 && codePoint <= 917999;
-}
 /**
-* CJK compatibility ideographs stand alone; each has a canonical unified form reachable via NFKC.
+* Mongolian free variation selectors, variation selectors, CJK compatibility ideographs, IVSes.
 */
-function isCompatibilityIdeograph(codePoint) {
-	return codePoint >= 63744 && codePoint <= 64255 || codePoint >= 194560 && codePoint <= 195103;
-}
 function isVariantCharacter(character) {
 	const codePoint = character.codePointAt(0);
 	if (codePoint === void 0) return false;
-	return isSelector(codePoint) || isCompatibilityIdeograph(codePoint);
-}
-/**
-* The plain form a reader would expect: the selector stripped away, or the compatibility
-* ideograph's canonical unified form. Undefined when that plain form is identical to the original,
-* since NFKC does not touch selectors and a suggestion equal to the input helps no one.
-*/
-function suggestPlainForm(segment) {
-	const withoutSelectors = Array.from(segment).filter((character) => {
-		const codePoint = character.codePointAt(0);
-		return codePoint !== void 0 && !isSelector(codePoint);
-	}).join("");
-	if (withoutSelectors === "") return void 0;
-	const normalized = Array.from(withoutSelectors).map((character) => {
-		const codePoint = character.codePointAt(0);
-		return codePoint !== void 0 && isCompatibilityIdeograph(codePoint) ? character.normalize("NFKC") : character;
-	}).join("");
-	return normalized === segment ? void 0 : normalized;
+	return codePoint >= 6155 && codePoint <= 6157 || codePoint === 6159 || codePoint >= 65024 && codePoint <= 65039 || codePoint >= 63744 && codePoint <= 64255 || codePoint >= 194560 && codePoint <= 195103 || codePoint >= 917760 && codePoint <= 917999;
 }
 /**
 * Scans the source rather than the display text: notation removal must not hide a variant.
 */
-const variantCharacterRule = () => ({
-	kind: "parsed",
-	meta: {
-		id: RULE_ID,
-		messages: MESSAGES
-	},
-	check: (manuscript, context) => {
-		for (const { index, segment } of graphemeSegmenter.segment(manuscript.source)) {
-			if (!Array.from(segment).some(isVariantCharacter)) continue;
-			const range = sourceRange(manuscript, {
-				start: index,
-				end: index + segment.length
-			});
-			const suggestion = suggestPlainForm(segment);
-			if (suggestion === void 0) context.report({
-				range,
-				messageId: "default"
-			});
-			else context.report({
-				range,
-				messageId: "with-suggestion",
-				data: { suggestion }
-			});
-		}
-	}
+const variantCharacterRule = () => defineParsedRule("kg/variant-character", "異体字または字形選択子が使われています", (manuscript, report) => {
+	for (const { index, segment } of graphemeSegmenter.segment(manuscript.source)) if (Array.from(segment).some(isVariantCharacter)) report(sourceRange(manuscript, {
+		start: index,
+		end: index + segment.length
+	}));
 });
 //#endregion
 //#region src/proofreading/rules/default-rules.ts

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sushichan044/kg-core",
-  "version": "0.6.1",
+  "version": "0.6.2",
   "description": "Framework-independent parsing, composition, and proofreading for Japanese manuscripts",
   "keywords": [
     "japanese",