@sushichan044/kg-core 0.6.0 → 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
 		}
@@ -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;
 /**
@@ -1880,16 +1958,6 @@ 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$1 = "kg/paragraph-opening";
 const MESSAGES$1 = {
@@ -1961,10 +2029,6 @@ const MESSAGES = {
 	"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.
 */
@@ -1981,20 +2045,18 @@ 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);
 		}
 	}
 });

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sushichan044/kg-core",
-  "version": "0.6.0",
+  "version": "0.6.2",
   "description": "Framework-independent parsing, composition, and proofreading for Japanese manuscripts",
   "keywords": [
     "japanese",