@pierre/diffs 1.3.0-beta.1 → 1.3.0-beta.10

package/dist/editor/textMeasure.js

@@ -1,38 +1,92 @@
-import { h, round } from "./utils.js";
-
+import { getGraphemeSegmenter, h, round } from "./utils.js";
 //#region src/editor/textMeasure.ts
+const TEXT_WIDTH_CACHE_LIMIT = 4096;
+const COMBINING_MARK_PATTERN = /\p{Mark}/u;
 var Metrics = class {
 	#root;
 	#canvasCtx;
 	#font;
+	#textWidthCache = /* @__PURE__ */ new Map();
 	/** Width of the '0' character. */
 	ch = -1;
 	/** Size of a tab(\t) character. */
 	tabSize = 2;
 	/** Height of the code line. */
 	lineHeight = 20;
+	/** Padding top of the root element. */
+	paddingTop = 0;
 	/** initialize the metrics */
 	init(root) {
 		if (this.#root === root && this.#canvasCtx !== void 0 && this.ch !== -1) return;
 		this.#root = root;
 		this.#canvasCtx ??= document.createElement("canvas").getContext("2d") ?? void 0;
 		if (this.#canvasCtx === void 0) throw new Error("Could not get canvas context");
+		const parent = root.parentElement;
+		if (parent !== null) {
+			const { paddingTop } = getComputedStyle(parent);
+			if (paddingTop.endsWith("px")) this.paddingTop = parseFloat(paddingTop.slice(0, -2));
+		}
 		const { fontSize, fontFamily, tabSize, lineHeight } = getComputedStyle(root);
-		if (lineHeight.endsWith("px")) this.lineHeight = Number(lineHeight.slice(0, -2));
-		else if (fontSize.endsWith("px")) this.lineHeight = round(Number(fontSize.slice(0, -2)) * Number(lineHeight));
+		if (lineHeight.endsWith("px")) this.lineHeight = parseFloat(lineHeight.slice(0, -2));
+		else if (fontSize.endsWith("px")) this.lineHeight = round(parseFloat(fontSize.slice(0, -2)) * parseFloat(lineHeight));
 		const font = fontSize + " " + fontFamily;
 		if (this.#font !== font || this.ch === -1) {
 			this.#font = font;
 			this.#canvasCtx.font = font;
 			this.ch = this.canvasMeasureTextWidth("0");
+			this.clearTextWidthCache();
 		}
-		this.tabSize = Number(tabSize);
+		const nextTabSize = parseInt(tabSize, 10);
+		if (!Number.isNaN(nextTabSize)) this.tabSize = nextTabSize;
+	}
+	/**
+	* Re-measure the '0' character width against the font that is loaded right
+	* now, returning true when measurement-dependent UI should refresh.
+	*
+	* A custom web font can finish loading after the editor first renders.
+	* Until then canvas measureText reports the fallback font's width, and
+	* getComputedStyle returns the same font-family string before and after the
+	* file arrives, so init()'s font guard never re-measures on its own. Call
+	* this once fonts have settled (e.g. on document.fonts.ready) to replace a
+	* width measured against the fallback font with the real glyph width. Font
+	* completion can also change DOM-measured emoji/ZWJ widths while leaving the
+	* ASCII width unchanged, and the computed font string stays the same in both
+	* cases. The boolean return lets the caller skip re-rendering when there is no
+	* measured state to refresh.
+	*/
+	remeasureCharacterWidth() {
+		if (this.#canvasCtx === void 0 || this.#font === void 0) return false;
+		this.#canvasCtx.font = this.#font;
+		const ch = this.canvasMeasureTextWidth("0");
+		const characterWidthChanged = ch !== this.ch;
+		if (characterWidthChanged) this.ch = ch;
+		const textWidthCacheCleared = this.#clearTextWidthCache();
+		return characterWidthChanged || textWidthCacheCleared;
 	}
 	/** measure the width of the text */
 	measureTextWidth(text) {
-		const textWithExpandedTabs = text.replaceAll("	", " ".repeat(this.tabSize));
-		if (needsDomTextMeasurement(textWithExpandedTabs)) return this.domMeasureTextWidth(textWithExpandedTabs);
-		return this.canvasMeasureTextWidth(textWithExpandedTabs);
+		if (!text.includes("	")) return this.#measureTextWidthWithoutTabs(text);
+		if (this.#canvasCtx === void 0) throw new Error("Metrics not initialized");
+		const asciiColumns = getExpandedAsciiTextColumns(text, this.tabSize);
+		if (asciiColumns !== -1) return asciiColumns * this.ch;
+		let width = 0;
+		let runStart = 0;
+		const tabStopWidth = this.tabSize * this.ch;
+		for (let i = 0; i < text.length; i++) {
+			if (text.charCodeAt(i) !== 9) continue;
+			if (i > runStart) width += this.#measureTextWidthWithoutTabs(text.slice(runStart, i));
+			if (tabStopWidth > 0) {
+				const remainder = width % tabStopWidth;
+				width += remainder === 0 ? tabStopWidth : tabStopWidth - remainder;
+			}
+			runStart = i + 1;
+		}
+		if (runStart < text.length) width += this.#measureTextWidthWithoutTabs(text.slice(runStart));
+		return round(width);
+	}
+	#measureTextWidthWithoutTabs(text) {
+		if (needsDomTextMeasurement(text)) return this.domMeasureTextWidth(text);
+		return this.canvasMeasureTextWidth(text);
 	}
 	/** measure the width of the text using the canvas measureText API */
 	canvasMeasureTextWidth(text) {
@@ -41,10 +95,14 @@ var Metrics = class {
 	}
 	/**
 	* measure the width of the text using the DOM
-	* this is slow because it cause a reflow, use it for non-ascii text
+	* this is slow because it cause a reflow, use it for non-ascii text;
+	* results are memoized per text so repeated measurements skip the reflow
 	*/
 	domMeasureTextWidth(text) {
 		if (this.#root === void 0) throw new Error("Metrics not initialized");
+		const cacheKey = text + "|" + this.#font;
+		const cached = this.#textWidthCache.get(cacheKey);
+		if (cached !== void 0) return cached;
 		const measureEl = h("span", {
 			style: {
 				position: "absolute",
@@ -57,18 +115,38 @@ var Metrics = class {
 			},
 			textContent: text
 		}, this.#root);
+		let width;
 		try {
-			return measureEl.getBoundingClientRect().width;
+			width = round(measureEl.getBoundingClientRect().width);
 		} finally {
 			measureEl.remove();
 		}
+		if (this.#textWidthCache.size >= TEXT_WIDTH_CACHE_LIMIT) {
+			const oldestKey = this.#textWidthCache.keys().next().value;
+			if (oldestKey !== void 0) this.#textWidthCache.delete(oldestKey);
+		}
+		this.#textWidthCache.set(cacheKey, width);
+		return width;
+	}
+	/**
+	* discard memoized DOM text widths
+	* call this when the inherited font may have changed without re-running
+	* init(), e.g. on a layout reflow, so stale widths are not reused
+	*/
+	clearTextWidthCache() {
+		this.#clearTextWidthCache();
+	}
+	#clearTextWidthCache() {
+		const hadCachedTextWidths = this.#textWidthCache.size > 0;
+		this.#textWidthCache.clear();
+		return hadCachedTextWidths;
 	}
 };
 /** Check if the text needs DOM text measurement. */
 function needsDomTextMeasurement(text) {
 	for (let i = 0; i < text.length; i++) {
 		const code = text.charCodeAt(i);
-		if (code >= 55296 && code <= 57343 || code === 8205 || code === 65038 || code === 65039) return true;
+		if (code >= 55296 && code <= 57343 || code === 8205 || code === 65038 || code === 65039 || code > 127 && COMBINING_MARK_PATTERN.test(text.charAt(i))) return true;
 	}
 	return false;
 }
@@ -76,12 +154,22 @@ function needsDomTextMeasurement(text) {
 function snapTextOffsetToUnicodeBoundary(text, offset) {
 	const boundedOffset = Math.max(0, Math.min(offset, text.length));
 	if (boundedOffset === 0 || boundedOffset === text.length || !needsDomTextMeasurement(text)) return boundedOffset;
-	const segmenter = new Intl.Segmenter(void 0, { granularity: "grapheme" });
-	for (const segment of segmenter.segment(text)) {
-		const segmentStart = segment.index;
-		const segmentEnd = segmentStart + segment.segment.length;
+	const segmenter = getGraphemeSegmenter();
+	if (segmenter !== void 0) {
+		for (const segment of segmenter.segment(text)) {
+			const segmentStart = segment.index;
+			const segmentEnd = segmentStart + segment.segment.length;
+			if (boundedOffset > segmentStart && boundedOffset < segmentEnd) return segmentEnd;
+			if (boundedOffset <= segmentStart) break;
+		}
+		return boundedOffset;
+	}
+	let segmentStart = 0;
+	for (const codePoint of text) {
+		const segmentEnd = segmentStart + codePoint.length;
 		if (boundedOffset > segmentStart && boundedOffset < segmentEnd) return segmentEnd;
 		if (boundedOffset <= segmentStart) break;
+		segmentStart = segmentEnd;
 	}
 	return boundedOffset;
 }
@@ -89,20 +177,33 @@ function snapTextOffsetToUnicodeBoundary(text, offset) {
 function getUnicodeMeasurementOffsets(text) {
 	if (!needsDomTextMeasurement(text)) return;
 	const offsets = [0];
-	const segmenter = new Intl.Segmenter(void 0, { granularity: "grapheme" });
-	for (const segment of segmenter.segment(text)) offsets.push(segment.index + segment.segment.length);
+	const segmenter = getGraphemeSegmenter();
+	if (segmenter !== void 0) {
+		for (const segment of segmenter.segment(text)) offsets.push(segment.index + segment.segment.length);
+		return offsets;
+	}
+	let offset = 0;
+	for (const codePoint of text) {
+		offset += codePoint.length;
+		offsets.push(offset);
+	}
 	return offsets;
 }
-/** get the number of columns of the ASCII text */
+/**
+* Count the rendered columns of ASCII text, advancing each tab to the next
+* fixed tab stop (a multiple of tabSize) to match CSS `tab-size`. Returns -1
+* for non-ASCII text, which must be measured glyph-by-glyph instead.
+*/
 function getExpandedAsciiTextColumns(text, tabSize) {
 	let columns = 0;
 	for (let i = 0; i < text.length; i++) {
-		if (text.charCodeAt(i) > 127) return -1;
-		columns += text.charCodeAt(i) === 9 ? tabSize : 1;
+		const code = text.charCodeAt(i);
+		if (code > 127) return -1;
+		columns += code === 9 ? tabSize - columns % tabSize : 1;
 	}
 	return columns;
 }
-
 //#endregion
 export { Metrics, getExpandedAsciiTextColumns, getUnicodeMeasurementOffsets, needsDomTextMeasurement, snapTextOffsetToUnicodeBoundary };
+
 //# sourceMappingURL=textMeasure.js.map

package/dist/editor/textMeasure.js.map

@@ -1 +1 @@
-{"version":3,"file":"textMeasure.js","names":["#root","#canvasCtx","#font"],"sources":["../../src/editor/textMeasure.ts"],"sourcesContent":["import { h, round } from './utils';\n\nexport class Metrics {\n  #root?: HTMLElement;\n  #canvasCtx?: CanvasRenderingContext2D;\n  #font?: string;\n\n  /** Width of the '0' character. */\n  ch: number = -1;\n  /** Size of a tab(\\t) character. */\n  tabSize: number = 2;\n  /** Height of the code line. */\n  lineHeight: number = 20;\n\n  /** initialize the metrics */\n  init(root: HTMLElement): void {\n    if (\n      this.#root === root &&\n      this.#canvasCtx !== undefined &&\n      this.ch !== -1\n    ) {\n      // already initialized\n      return;\n    }\n\n    this.#root = root;\n    this.#canvasCtx ??=\n      document.createElement('canvas').getContext('2d') ?? undefined;\n    if (this.#canvasCtx === undefined) {\n      throw new Error('Could not get canvas context');\n    }\n\n    const { fontSize, fontFamily, tabSize, lineHeight } =\n      getComputedStyle(root);\n    if (lineHeight.endsWith('px')) {\n      this.lineHeight = Number(lineHeight.slice(0, -2));\n    } else if (fontSize.endsWith('px')) {\n      this.lineHeight = round(\n        Number(fontSize.slice(0, -2)) * Number(lineHeight)\n      );\n    }\n    const font = fontSize + ' ' + fontFamily;\n    if (this.#font !== font || this.ch === -1) {\n      this.#font = font;\n      this.#canvasCtx.font = font;\n      this.ch = this.canvasMeasureTextWidth('0');\n    }\n    this.tabSize = Number(tabSize);\n  }\n\n  /** measure the width of the text */\n  measureTextWidth(text: string): number {\n    const textWithExpandedTabs = text.replaceAll(\n      '\\t',\n      ' '.repeat(this.tabSize)\n    );\n    if (needsDomTextMeasurement(textWithExpandedTabs)) {\n      return this.domMeasureTextWidth(textWithExpandedTabs);\n    }\n    return this.canvasMeasureTextWidth(textWithExpandedTabs);\n  }\n\n  /** measure the width of the text using the canvas measureText API */\n  canvasMeasureTextWidth(text: string): number {\n    if (this.#canvasCtx === undefined) {\n      throw new Error('Metrics not initialized');\n    }\n    return round(this.#canvasCtx.measureText(text).width);\n  }\n\n  /**\n   * measure the width of the text using the DOM\n   * this is slow because it cause a reflow, use it for non-ascii text\n   */\n  domMeasureTextWidth(text: string): number {\n    if (this.#root === undefined) {\n      throw new Error('Metrics not initialized');\n    }\n    const measureEl = h(\n      'span',\n      {\n        style: {\n          position: 'absolute',\n          top: '0',\n          left: '0',\n          visibility: 'hidden',\n          pointerEvents: 'none',\n          whiteSpace: 'pre',\n          font: 'inherit',\n        },\n        textContent: text,\n      },\n      this.#root\n    );\n    try {\n      return measureEl.getBoundingClientRect().width;\n    } finally {\n      measureEl.remove();\n    }\n  }\n}\n\n/** Check if the text needs DOM text measurement. */\nexport function needsDomTextMeasurement(text: string): boolean {\n  for (let i = 0; i < text.length; i++) {\n    const code = text.charCodeAt(i);\n    if (\n      (code >= 0xd800 && code <= 0xdfff) ||\n      code === 0x200d ||\n      code === 0xfe0e ||\n      code === 0xfe0f\n    ) {\n      return true;\n    }\n  }\n  return false;\n}\n\n/** snap the text offset to the Unicode boundary */\nexport function snapTextOffsetToUnicodeBoundary(\n  text: string,\n  offset: number\n): number {\n  const boundedOffset = Math.max(0, Math.min(offset, text.length));\n  if (\n    boundedOffset === 0 ||\n    boundedOffset === text.length ||\n    !needsDomTextMeasurement(text)\n  ) {\n    return boundedOffset;\n  }\n  // Avoid measuring a caret position inside one visual emoji/grapheme.\n  // Browser caret movement can report offsets around UTF-16 surrogate\n  // pairs and emoji joiners; measuring a partial sequence gives a\n  // replacement-glyph width.\n  const segmenter = new Intl.Segmenter(undefined, {\n    granularity: 'grapheme',\n  });\n  for (const segment of segmenter.segment(text)) {\n    const segmentStart = segment.index;\n    const segmentEnd = segmentStart + segment.segment.length;\n    if (boundedOffset > segmentStart && boundedOffset < segmentEnd) {\n      return segmentEnd;\n    }\n    if (boundedOffset <= segmentStart) {\n      break;\n    }\n  }\n  return boundedOffset;\n}\n\n/** get the offsets of the Unicode grapheme clusters in the text */\nexport function getUnicodeMeasurementOffsets(\n  text: string\n): number[] | undefined {\n  if (!needsDomTextMeasurement(text)) {\n    return undefined;\n  }\n  const offsets = [0];\n  const segmenter = new Intl.Segmenter(undefined, {\n    granularity: 'grapheme',\n  });\n  for (const segment of segmenter.segment(text)) {\n    offsets.push(segment.index + segment.segment.length);\n  }\n  return offsets;\n}\n\n/** get the number of columns of the ASCII text */\nexport function getExpandedAsciiTextColumns(\n  text: string,\n  tabSize: number\n): number {\n  let columns = 0;\n  for (let i = 0; i < text.length; i++) {\n    if (text.charCodeAt(i) > 127) {\n      return -1;\n    }\n    columns += text.charCodeAt(i) === /* '\\t' */ 9 ? tabSize : 1;\n  }\n  return columns;\n}\n"],"mappings":";;;AAEA,IAAa,UAAb,MAAqB;CACnB;CACA;CACA;;CAGA,KAAa;;CAEb,UAAkB;;CAElB,aAAqB;;CAGrB,KAAK,MAAyB;AAC5B,MACE,MAAKA,SAAU,QACf,MAAKC,cAAe,UACpB,KAAK,OAAO,GAGZ;AAGF,QAAKD,OAAQ;AACb,QAAKC,cACH,SAAS,cAAc,SAAS,CAAC,WAAW,KAAK,IAAI;AACvD,MAAI,MAAKA,cAAe,OACtB,OAAM,IAAI,MAAM,+BAA+B;EAGjD,MAAM,EAAE,UAAU,YAAY,SAAS,eACrC,iBAAiB,KAAK;AACxB,MAAI,WAAW,SAAS,KAAK,CAC3B,MAAK,aAAa,OAAO,WAAW,MAAM,GAAG,GAAG,CAAC;WACxC,SAAS,SAAS,KAAK,CAChC,MAAK,aAAa,MAChB,OAAO,SAAS,MAAM,GAAG,GAAG,CAAC,GAAG,OAAO,WAAW,CACnD;EAEH,MAAM,OAAO,WAAW,MAAM;AAC9B,MAAI,MAAKC,SAAU,QAAQ,KAAK,OAAO,IAAI;AACzC,SAAKA,OAAQ;AACb,SAAKD,UAAW,OAAO;AACvB,QAAK,KAAK,KAAK,uBAAuB,IAAI;;AAE5C,OAAK,UAAU,OAAO,QAAQ;;;CAIhC,iBAAiB,MAAsB;EACrC,MAAM,uBAAuB,KAAK,WAChC,KACA,IAAI,OAAO,KAAK,QAAQ,CACzB;AACD,MAAI,wBAAwB,qBAAqB,CAC/C,QAAO,KAAK,oBAAoB,qBAAqB;AAEvD,SAAO,KAAK,uBAAuB,qBAAqB;;;CAI1D,uBAAuB,MAAsB;AAC3C,MAAI,MAAKA,cAAe,OACtB,OAAM,IAAI,MAAM,0BAA0B;AAE5C,SAAO,MAAM,MAAKA,UAAW,YAAY,KAAK,CAAC,MAAM;;;;;;CAOvD,oBAAoB,MAAsB;AACxC,MAAI,MAAKD,SAAU,OACjB,OAAM,IAAI,MAAM,0BAA0B;EAE5C,MAAM,YAAY,EAChB,QACA;GACE,OAAO;IACL,UAAU;IACV,KAAK;IACL,MAAM;IACN,YAAY;IACZ,eAAe;IACf,YAAY;IACZ,MAAM;IACP;GACD,aAAa;GACd,EACD,MAAKA,KACN;AACD,MAAI;AACF,UAAO,UAAU,uBAAuB,CAAC;YACjC;AACR,aAAU,QAAQ;;;;;AAMxB,SAAgB,wBAAwB,MAAuB;AAC7D,MAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;EACpC,MAAM,OAAO,KAAK,WAAW,EAAE;AAC/B,MACG,QAAQ,SAAU,QAAQ,SAC3B,SAAS,QACT,SAAS,SACT,SAAS,MAET,QAAO;;AAGX,QAAO;;;AAIT,SAAgB,gCACd,MACA,QACQ;CACR,MAAM,gBAAgB,KAAK,IAAI,GAAG,KAAK,IAAI,QAAQ,KAAK,OAAO,CAAC;AAChE,KACE,kBAAkB,KAClB,kBAAkB,KAAK,UACvB,CAAC,wBAAwB,KAAK,CAE9B,QAAO;CAMT,MAAM,YAAY,IAAI,KAAK,UAAU,QAAW,EAC9C,aAAa,YACd,CAAC;AACF,MAAK,MAAM,WAAW,UAAU,QAAQ,KAAK,EAAE;EAC7C,MAAM,eAAe,QAAQ;EAC7B,MAAM,aAAa,eAAe,QAAQ,QAAQ;AAClD,MAAI,gBAAgB,gBAAgB,gBAAgB,WAClD,QAAO;AAET,MAAI,iBAAiB,aACnB;;AAGJ,QAAO;;;AAIT,SAAgB,6BACd,MACsB;AACtB,KAAI,CAAC,wBAAwB,KAAK,CAChC;CAEF,MAAM,UAAU,CAAC,EAAE;CACnB,MAAM,YAAY,IAAI,KAAK,UAAU,QAAW,EAC9C,aAAa,YACd,CAAC;AACF,MAAK,MAAM,WAAW,UAAU,QAAQ,KAAK,CAC3C,SAAQ,KAAK,QAAQ,QAAQ,QAAQ,QAAQ,OAAO;AAEtD,QAAO;;;AAIT,SAAgB,4BACd,MACA,SACQ;CACR,IAAI,UAAU;AACd,MAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;AACpC,MAAI,KAAK,WAAW,EAAE,GAAG,IACvB,QAAO;AAET,aAAW,KAAK,WAAW,EAAE,KAAgB,IAAI,UAAU;;AAE7D,QAAO"}
+{"version":3,"file":"textMeasure.js","names":["#root","#canvasCtx","#font","#clearTextWidthCache","#measureTextWidthWithoutTabs","#textWidthCache"],"sources":["../../src/editor/textMeasure.ts"],"sourcesContent":["import { getGraphemeSegmenter, h, round } from './utils';\n\n// Upper bound on cached DOM text-width measurements. The cache only holds\n// non-ASCII runs (emoji, ZWJ sequences, variation selectors, combining marks),\n// so it stays small for ordinary code, but capping it prevents unbounded growth\n// on emoji-heavy documents. Past the cap the oldest entry is evicted.\nconst TEXT_WIDTH_CACHE_LIMIT = 4096;\n\nconst COMBINING_MARK_PATTERN = /\\p{Mark}/u;\n\nexport class Metrics {\n  #root?: HTMLElement;\n  #canvasCtx?: CanvasRenderingContext2D;\n  #font?: string;\n\n  // Memoizes domMeasureTextWidth() results\n  #textWidthCache = new Map<string, number>();\n\n  /** Width of the '0' character. */\n  ch: number = -1;\n  /** Size of a tab(\\t) character. */\n  tabSize: number = 2;\n  /** Height of the code line. */\n  lineHeight: number = 20;\n  /** Padding top of the root element. */\n  paddingTop: number = 0;\n\n  /** initialize the metrics */\n  init(root: HTMLElement): void {\n    if (\n      this.#root === root &&\n      this.#canvasCtx !== undefined &&\n      this.ch !== -1\n    ) {\n      // already initialized\n      return;\n    }\n\n    this.#root = root;\n    this.#canvasCtx ??=\n      document.createElement('canvas').getContext('2d') ?? undefined;\n    if (this.#canvasCtx === undefined) {\n      throw new Error('Could not get canvas context');\n    }\n\n    const parent = root.parentElement;\n    if (parent !== null) {\n      const { paddingTop } = getComputedStyle(parent);\n      if (paddingTop.endsWith('px')) {\n        this.paddingTop = parseFloat(paddingTop.slice(0, -2));\n      }\n    }\n\n    const { fontSize, fontFamily, tabSize, lineHeight } =\n      getComputedStyle(root);\n    if (lineHeight.endsWith('px')) {\n      this.lineHeight = parseFloat(lineHeight.slice(0, -2));\n    } else if (fontSize.endsWith('px')) {\n      this.lineHeight = round(\n        parseFloat(fontSize.slice(0, -2)) * parseFloat(lineHeight)\n      );\n    }\n    const font = fontSize + ' ' + fontFamily;\n    if (this.#font !== font || this.ch === -1) {\n      this.#font = font;\n      this.#canvasCtx.font = font;\n      this.ch = this.canvasMeasureTextWidth('0');\n      // Cached DOM widths were measured against the previous font.\n      this.clearTextWidthCache();\n    }\n    const nextTabSize = parseInt(tabSize, 10);\n    if (!Number.isNaN(nextTabSize)) {\n      this.tabSize = nextTabSize;\n    }\n  }\n\n  /**\n   * Re-measure the '0' character width against the font that is loaded right\n   * now, returning true when measurement-dependent UI should refresh.\n   *\n   * A custom web font can finish loading after the editor first renders.\n   * Until then canvas measureText reports the fallback font's width, and\n   * getComputedStyle returns the same font-family string before and after the\n   * file arrives, so init()'s font guard never re-measures on its own. Call\n   * this once fonts have settled (e.g. on document.fonts.ready) to replace a\n   * width measured against the fallback font with the real glyph width. Font\n   * completion can also change DOM-measured emoji/ZWJ widths while leaving the\n   * ASCII width unchanged, and the computed font string stays the same in both\n   * cases. The boolean return lets the caller skip re-rendering when there is no\n   * measured state to refresh.\n   */\n  remeasureCharacterWidth(): boolean {\n    if (this.#canvasCtx === undefined || this.#font === undefined) {\n      return false;\n    }\n    this.#canvasCtx.font = this.#font;\n    const ch = this.canvasMeasureTextWidth('0');\n    const characterWidthChanged = ch !== this.ch;\n    if (characterWidthChanged) {\n      this.ch = ch;\n    }\n    const textWidthCacheCleared = this.#clearTextWidthCache();\n    return characterWidthChanged || textWidthCacheCleared;\n  }\n\n  /** measure the width of the text */\n  measureTextWidth(text: string): number {\n    if (!text.includes('\\t')) {\n      return this.#measureTextWidthWithoutTabs(text);\n    }\n    if (this.#canvasCtx === undefined) {\n      throw new Error('Metrics not initialized');\n    }\n    const asciiColumns = getExpandedAsciiTextColumns(text, this.tabSize);\n    if (asciiColumns !== -1) {\n      return asciiColumns * this.ch;\n    }\n\n    let width = 0;\n    let runStart = 0;\n    const tabStopWidth = this.tabSize * this.ch;\n    for (let i = 0; i < text.length; i++) {\n      if (text.charCodeAt(i) !== /* '\\t' */ 9) {\n        continue;\n      }\n\n      if (i > runStart) {\n        width += this.#measureTextWidthWithoutTabs(text.slice(runStart, i));\n      }\n      // CSS tab stops are pixel positions, so wide glyphs before a tab must\n      // contribute their measured width rather than one logical column.\n      if (tabStopWidth > 0) {\n        const remainder = width % tabStopWidth;\n        width += remainder === 0 ? tabStopWidth : tabStopWidth - remainder;\n      }\n      runStart = i + 1;\n    }\n\n    if (runStart < text.length) {\n      width += this.#measureTextWidthWithoutTabs(text.slice(runStart));\n    }\n    return round(width);\n  }\n\n  #measureTextWidthWithoutTabs(text: string): number {\n    if (needsDomTextMeasurement(text)) {\n      return this.domMeasureTextWidth(text);\n    }\n    return this.canvasMeasureTextWidth(text);\n  }\n\n  /** measure the width of the text using the canvas measureText API */\n  canvasMeasureTextWidth(text: string): number {\n    if (this.#canvasCtx === undefined) {\n      throw new Error('Metrics not initialized');\n    }\n    return round(this.#canvasCtx.measureText(text).width);\n  }\n\n  /**\n   * measure the width of the text using the DOM\n   * this is slow because it cause a reflow, use it for non-ascii text;\n   * results are memoized per text so repeated measurements skip the reflow\n   */\n  domMeasureTextWidth(text: string): number {\n    if (this.#root === undefined) {\n      throw new Error('Metrics not initialized');\n    }\n    const cacheKey = text + '|' + this.#font;\n    const cached = this.#textWidthCache.get(cacheKey);\n    if (cached !== undefined) {\n      return cached;\n    }\n    const measureEl = h(\n      'span',\n      {\n        style: {\n          position: 'absolute',\n          top: '0',\n          left: '0',\n          visibility: 'hidden',\n          pointerEvents: 'none',\n          whiteSpace: 'pre',\n          font: 'inherit',\n        },\n        textContent: text,\n      },\n      this.#root\n    );\n    let width: number;\n    try {\n      // round() to match canvasMeasureTextWidth and ch; otherwise the DOM path\n      // returns raw sub-pixel widths and caret/selection offsets drift between\n      // ASCII and non-ASCII runs on the same line.\n      width = round(measureEl.getBoundingClientRect().width);\n    } finally {\n      measureEl.remove();\n    }\n    if (this.#textWidthCache.size >= TEXT_WIDTH_CACHE_LIMIT) {\n      const oldestKey = this.#textWidthCache.keys().next().value;\n      if (oldestKey !== undefined) {\n        this.#textWidthCache.delete(oldestKey);\n      }\n    }\n    this.#textWidthCache.set(cacheKey, width);\n    return width;\n  }\n\n  /**\n   * discard memoized DOM text widths\n   * call this when the inherited font may have changed without re-running\n   * init(), e.g. on a layout reflow, so stale widths are not reused\n   */\n  clearTextWidthCache(): void {\n    this.#clearTextWidthCache();\n  }\n\n  #clearTextWidthCache(): boolean {\n    const hadCachedTextWidths = this.#textWidthCache.size > 0;\n    this.#textWidthCache.clear();\n    return hadCachedTextWidths;\n  }\n}\n\n/** Check if the text needs DOM text measurement. */\nexport function needsDomTextMeasurement(text: string): boolean {\n  for (let i = 0; i < text.length; i++) {\n    const code = text.charCodeAt(i);\n    if (\n      (code >= 0xd800 && code <= 0xdfff) ||\n      code === 0x200d ||\n      code === 0xfe0e ||\n      code === 0xfe0f ||\n      (code > 0x7f && COMBINING_MARK_PATTERN.test(text.charAt(i)))\n    ) {\n      return true;\n    }\n  }\n  return false;\n}\n\n/** snap the text offset to the Unicode boundary */\nexport function snapTextOffsetToUnicodeBoundary(\n  text: string,\n  offset: number\n): number {\n  const boundedOffset = Math.max(0, Math.min(offset, text.length));\n  if (\n    boundedOffset === 0 ||\n    boundedOffset === text.length ||\n    !needsDomTextMeasurement(text)\n  ) {\n    return boundedOffset;\n  }\n  // Avoid measuring a caret position inside one visual emoji/grapheme.\n  // Browser caret movement can report offsets around UTF-16 surrogate\n  // pairs and emoji joiners; measuring a partial sequence gives a\n  // replacement-glyph width.\n  const segmenter = getGraphemeSegmenter();\n  if (segmenter !== undefined) {\n    for (const segment of segmenter.segment(text)) {\n      const segmentStart = segment.index;\n      const segmentEnd = segmentStart + segment.segment.length;\n      if (boundedOffset > segmentStart && boundedOffset < segmentEnd) {\n        return segmentEnd;\n      }\n      if (boundedOffset <= segmentStart) {\n        break;\n      }\n    }\n    return boundedOffset;\n  }\n  // Degraded path for engines lacking Intl.Segmenter: snap out of a\n  // surrogate pair by stepping over code points.\n  let segmentStart = 0;\n  for (const codePoint of text) {\n    const segmentEnd = segmentStart + codePoint.length;\n    if (boundedOffset > segmentStart && boundedOffset < segmentEnd) {\n      return segmentEnd;\n    }\n    if (boundedOffset <= segmentStart) {\n      break;\n    }\n    segmentStart = segmentEnd;\n  }\n  return boundedOffset;\n}\n\n/** get the offsets of the Unicode grapheme clusters in the text */\nexport function getUnicodeMeasurementOffsets(\n  text: string\n): number[] | undefined {\n  if (!needsDomTextMeasurement(text)) {\n    return undefined;\n  }\n  const offsets = [0];\n  const segmenter = getGraphemeSegmenter();\n  if (segmenter !== undefined) {\n    for (const segment of segmenter.segment(text)) {\n      offsets.push(segment.index + segment.segment.length);\n    }\n    return offsets;\n  }\n  // Degraded path for engines lacking Intl.Segmenter: step by code point.\n  let offset = 0;\n  for (const codePoint of text) {\n    offset += codePoint.length;\n    offsets.push(offset);\n  }\n  return offsets;\n}\n\n/**\n * Count the rendered columns of ASCII text, advancing each tab to the next\n * fixed tab stop (a multiple of tabSize) to match CSS `tab-size`. Returns -1\n * for non-ASCII text, which must be measured glyph-by-glyph instead.\n */\nexport function getExpandedAsciiTextColumns(\n  text: string,\n  tabSize: number\n): number {\n  let columns = 0;\n  for (let i = 0; i < text.length; i++) {\n    const code = text.charCodeAt(i);\n    if (code > 127) {\n      return -1;\n    }\n    columns += code === /* '\\t' */ 9 ? tabSize - (columns % tabSize) : 1;\n  }\n  return columns;\n}\n"],"mappings":";;AAMA,MAAM,yBAAyB;AAE/B,MAAM,yBAAyB;AAE/B,IAAa,UAAb,MAAqB;CACnB;CACA;CACA;CAGA,kCAAkB,IAAI,IAAoB;;CAG1C,KAAa;;CAEb,UAAkB;;CAElB,aAAqB;;CAErB,aAAqB;;CAGrB,KAAK,MAAyB;EAC5B,IACE,KAAKA,UAAU,QACf,KAAKC,eAAe,KAAA,KACpB,KAAK,OAAO,IAGZ;EAGF,KAAKD,QAAQ;EACb,KAAKC,eACH,SAAS,cAAc,QAAQ,CAAC,CAAC,WAAW,IAAI,KAAK,KAAA;EACvD,IAAI,KAAKA,eAAe,KAAA,GACtB,MAAM,IAAI,MAAM,8BAA8B;EAGhD,MAAM,SAAS,KAAK;EACpB,IAAI,WAAW,MAAM;GACnB,MAAM,EAAE,eAAe,iBAAiB,MAAM;GAC9C,IAAI,WAAW,SAAS,IAAI,GAC1B,KAAK,aAAa,WAAW,WAAW,MAAM,GAAG,EAAE,CAAC;EAExD;EAEA,MAAM,EAAE,UAAU,YAAY,SAAS,eACrC,iBAAiB,IAAI;EACvB,IAAI,WAAW,SAAS,IAAI,GAC1B,KAAK,aAAa,WAAW,WAAW,MAAM,GAAG,EAAE,CAAC;OAC/C,IAAI,SAAS,SAAS,IAAI,GAC/B,KAAK,aAAa,MAChB,WAAW,SAAS,MAAM,GAAG,EAAE,CAAC,IAAI,WAAW,UAAU,CAC3D;EAEF,MAAM,OAAO,WAAW,MAAM;EAC9B,IAAI,KAAKC,UAAU,QAAQ,KAAK,OAAO,IAAI;GACzC,KAAKA,QAAQ;GACb,KAAKD,WAAW,OAAO;GACvB,KAAK,KAAK,KAAK,uBAAuB,GAAG;GAEzC,KAAK,oBAAoB;EAC3B;EACA,MAAM,cAAc,SAAS,SAAS,EAAE;EACxC,IAAI,CAAC,OAAO,MAAM,WAAW,GAC3B,KAAK,UAAU;CAEnB;;;;;;;;;;;;;;;;CAiBA,0BAAmC;EACjC,IAAI,KAAKA,eAAe,KAAA,KAAa,KAAKC,UAAU,KAAA,GAClD,OAAO;EAET,KAAKD,WAAW,OAAO,KAAKC;EAC5B,MAAM,KAAK,KAAK,uBAAuB,GAAG;EAC1C,MAAM,wBAAwB,OAAO,KAAK;EAC1C,IAAI,uBACF,KAAK,KAAK;EAEZ,MAAM,wBAAwB,KAAKC,qBAAqB;EACxD,OAAO,yBAAyB;CAClC;;CAGA,iBAAiB,MAAsB;EACrC,IAAI,CAAC,KAAK,SAAS,GAAI,GACrB,OAAO,KAAKC,6BAA6B,IAAI;EAE/C,IAAI,KAAKH,eAAe,KAAA,GACtB,MAAM,IAAI,MAAM,yBAAyB;EAE3C,MAAM,eAAe,4BAA4B,MAAM,KAAK,OAAO;EACnE,IAAI,iBAAiB,IACnB,OAAO,eAAe,KAAK;EAG7B,IAAI,QAAQ;EACZ,IAAI,WAAW;EACf,MAAM,eAAe,KAAK,UAAU,KAAK;EACzC,KAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;GACpC,IAAI,KAAK,WAAW,CAAC,MAAiB,GACpC;GAGF,IAAI,IAAI,UACN,SAAS,KAAKG,6BAA6B,KAAK,MAAM,UAAU,CAAC,CAAC;GAIpE,IAAI,eAAe,GAAG;IACpB,MAAM,YAAY,QAAQ;IAC1B,SAAS,cAAc,IAAI,eAAe,eAAe;GAC3D;GACA,WAAW,IAAI;EACjB;EAEA,IAAI,WAAW,KAAK,QAClB,SAAS,KAAKA,6BAA6B,KAAK,MAAM,QAAQ,CAAC;EAEjE,OAAO,MAAM,KAAK;CACpB;CAEA,6BAA6B,MAAsB;EACjD,IAAI,wBAAwB,IAAI,GAC9B,OAAO,KAAK,oBAAoB,IAAI;EAEtC,OAAO,KAAK,uBAAuB,IAAI;CACzC;;CAGA,uBAAuB,MAAsB;EAC3C,IAAI,KAAKH,eAAe,KAAA,GACtB,MAAM,IAAI,MAAM,yBAAyB;EAE3C,OAAO,MAAM,KAAKA,WAAW,YAAY,IAAI,CAAC,CAAC,KAAK;CACtD;;;;;;CAOA,oBAAoB,MAAsB;EACxC,IAAI,KAAKD,UAAU,KAAA,GACjB,MAAM,IAAI,MAAM,yBAAyB;EAE3C,MAAM,WAAW,OAAO,MAAM,KAAKE;EACnC,MAAM,SAAS,KAAKG,gBAAgB,IAAI,QAAQ;EAChD,IAAI,WAAW,KAAA,GACb,OAAO;EAET,MAAM,YAAY,EAChB,QACA;GACE,OAAO;IACL,UAAU;IACV,KAAK;IACL,MAAM;IACN,YAAY;IACZ,eAAe;IACf,YAAY;IACZ,MAAM;GACR;GACA,aAAa;EACf,GACA,KAAKL,KACP;EACA,IAAI;EACJ,IAAI;GAIF,QAAQ,MAAM,UAAU,sBAAsB,CAAC,CAAC,KAAK;EACvD,UAAU;GACR,UAAU,OAAO;EACnB;EACA,IAAI,KAAKK,gBAAgB,QAAQ,wBAAwB;GACvD,MAAM,YAAY,KAAKA,gBAAgB,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC;GACrD,IAAI,cAAc,KAAA,GAChB,KAAKA,gBAAgB,OAAO,SAAS;EAEzC;EACA,KAAKA,gBAAgB,IAAI,UAAU,KAAK;EACxC,OAAO;CACT;;;;;;CAOA,sBAA4B;EAC1B,KAAKF,qBAAqB;CAC5B;CAEA,uBAAgC;EAC9B,MAAM,sBAAsB,KAAKE,gBAAgB,OAAO;EACxD,KAAKA,gBAAgB,MAAM;EAC3B,OAAO;CACT;AACF;;AAGA,SAAgB,wBAAwB,MAAuB;CAC7D,KAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;EACpC,MAAM,OAAO,KAAK,WAAW,CAAC;EAC9B,IACG,QAAQ,SAAU,QAAQ,SAC3B,SAAS,QACT,SAAS,SACT,SAAS,SACR,OAAO,OAAQ,uBAAuB,KAAK,KAAK,OAAO,CAAC,CAAC,GAE1D,OAAO;CAEX;CACA,OAAO;AACT;;AAGA,SAAgB,gCACd,MACA,QACQ;CACR,MAAM,gBAAgB,KAAK,IAAI,GAAG,KAAK,IAAI,QAAQ,KAAK,MAAM,CAAC;CAC/D,IACE,kBAAkB,KAClB,kBAAkB,KAAK,UACvB,CAAC,wBAAwB,IAAI,GAE7B,OAAO;CAMT,MAAM,YAAY,qBAAqB;CACvC,IAAI,cAAc,KAAA,GAAW;EAC3B,KAAK,MAAM,WAAW,UAAU,QAAQ,IAAI,GAAG;GAC7C,MAAM,eAAe,QAAQ;GAC7B,MAAM,aAAa,eAAe,QAAQ,QAAQ;GAClD,IAAI,gBAAgB,gBAAgB,gBAAgB,YAClD,OAAO;GAET,IAAI,iBAAiB,cACnB;EAEJ;EACA,OAAO;CACT;CAGA,IAAI,eAAe;CACnB,KAAK,MAAM,aAAa,MAAM;EAC5B,MAAM,aAAa,eAAe,UAAU;EAC5C,IAAI,gBAAgB,gBAAgB,gBAAgB,YAClD,OAAO;EAET,IAAI,iBAAiB,cACnB;EAEF,eAAe;CACjB;CACA,OAAO;AACT;;AAGA,SAAgB,6BACd,MACsB;CACtB,IAAI,CAAC,wBAAwB,IAAI,GAC/B;CAEF,MAAM,UAAU,CAAC,CAAC;CAClB,MAAM,YAAY,qBAAqB;CACvC,IAAI,cAAc,KAAA,GAAW;EAC3B,KAAK,MAAM,WAAW,UAAU,QAAQ,IAAI,GAC1C,QAAQ,KAAK,QAAQ,QAAQ,QAAQ,QAAQ,MAAM;EAErD,OAAO;CACT;CAEA,IAAI,SAAS;CACb,KAAK,MAAM,aAAa,MAAM;EAC5B,UAAU,UAAU;EACpB,QAAQ,KAAK,MAAM;CACrB;CACA,OAAO;AACT;;;;;;AAOA,SAAgB,4BACd,MACA,SACQ;CACR,IAAI,UAAU;CACd,KAAK,IAAI,IAAI,GAAG,IAAI,KAAK,QAAQ,KAAK;EACpC,MAAM,OAAO,KAAK,WAAW,CAAC;EAC9B,IAAI,OAAO,KACT,OAAO;EAET,WAAW,SAAoB,IAAI,UAAW,UAAU,UAAW;CACrE;CACA,OAAO;AACT"}

package/dist/editor/{tokenzier.d.ts → tokenizer.d.ts}

@@ -1,37 +1,41 @@
 import { BaseCodeOptions, DiffsHighlighter, HighlightedToken, RenderRange } from "../types.js";
 import { TextDocument, TextDocumentChange } from "./textDocument.js";
-import { IGrammar, StateStack } from "shiki/textmate";
 
-//#region src/editor/tokenzier.d.ts
+//#region src/editor/tokenizer.d.ts
 interface EditorTokenizerProps {
   highlighter: DiffsHighlighter;
   textDocument: TextDocument<unknown>;
   codeOptions: BaseCodeOptions;
+  matchBrackets?: boolean;
   setStyle: (style: string) => void;
   onDeferTokenize: (lines: Map<number, Array<HighlightedToken>>, themeType: 'dark' | 'light') => void;
+  onThemeChange?: () => void;
+  __debug?: boolean;
 }
 /** Stoppable code tokenizer for the editor */
 declare class EditorTokenizer {
   #private;
-  static TOKENIZE_TIME_LIMIT: number;
   get themeType(): 'light' | 'dark';
+  getStringCommentRegexpRangesInLine(lineIndex: number): [number, number][] | null;
   constructor({
     codeOptions,
     highlighter,
     textDocument,
+    matchBrackets,
     setStyle,
-    onDeferTokenize
+    onDeferTokenize,
+    onThemeChange,
+    __debug
   }: EditorTokenizerProps);
+  syncTheme(codeOptions: BaseCodeOptions): void;
   cleanUp(): void;
   tokenize(change: TextDocumentChange, renderRange?: RenderRange): Map<number, Array<HighlightedToken>>;
-  prebuildStateStackMap(renderRange?: RenderRange): void;
+  prebuildStateStack(renderRange?: RenderRange): void;
   stopBackgroundTokenize(): void;
+  pauseBackgroundTokenize(): void;
+  resumeBackgroundTokenize(): void;
 }
-declare function tokenizeLine(grammar: IGrammar, colorMap: string[], lineText: string, stateStack: StateStack, timeLimit?: number): {
-  ruleStack: StateStack;
-  resolvedTokens: Array<HighlightedToken>;
-};
-declare function renderLineTokens(tokens: Array<HighlightedToken>, themeType: 'light' | 'dark'): (HTMLElement | string)[];
+declare function renderLineTokens(tokens: Array<HighlightedToken>): (HTMLElement | string)[];
 //#endregion
-export { EditorTokenizer, EditorTokenizerProps, renderLineTokens, tokenizeLine };
-//# sourceMappingURL=tokenzier.d.ts.map
+export { EditorTokenizer, EditorTokenizerProps, renderLineTokens };
+//# sourceMappingURL=tokenizer.d.ts.map

package/dist/editor/tokenizer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"tokenizer.d.ts","names":[],"sources":["../../src/editor/tokenizer.ts"],"mappings":";;;;UAmBiB,oBAAA;EACf,WAAA,EAAa,gBAAA;EACb,YAAA,EAAc,YAAA;EACd,WAAA,EAAa,eAAA;EACb,aAAA;EACA,QAAA,GAAW,KAAA;EACX,eAAA,GACE,KAAA,EAAO,GAAA,SAAY,KAAA,CAAM,gBAAA,IACzB,SAAA;EAKF,aAAA;EACA,OAAA;AAAA;;cAIW,eAAA;EAAA;MA4EP,SAAA;EAIJ,kCAAA,CACE,SAAA;EAsBF,WAAA;IACE,WAAA;IACA,WAAA;IACA,YAAA;IACA,aAAA;IACA,QAAA;IACA,eAAA;IACA,aAAA;IACA;EAAA,GACC,oBAAA;EA6GH,SAAA,CAAU,WAAA,EAAa,eAAA;EA+DvB,OAAA;EAUA,QAAA,CACE,MAAA,EAAQ,kBAAA,EACR,WAAA,GAAc,WAAA,GACb,GAAA,SAAY,KAAA,CAAM,gBAAA;EA+LrB,kBAAA,CAAmB,WAAA,GAAc,WAAA;EAKjC,sBAAA;EAaA,uBAAA;EAYA,wBAAA;AAAA;AAAA,iBAuXc,gBAAA,CACd,MAAA,EAAQ,KAAA,CAAM,gBAAA,KACZ,WAAA"}