@quandev104/pi-style 0.2.1 → 0.2.2

package/extension-src/pi-style/shared/ansi.ts

@@ -1,3 +1,5 @@
+import { visibleWidth as tuiVisibleWidth } from "@earendil-works/pi-tui";
+
 function isFinal(byte: string): boolean {
 	return byte >= "@" && byte <= "~";
 }
@@ -91,8 +93,12 @@ export function stripAnsi(value: string): string {
 	}
 	return output;
 }
+/**
+ * Terminal-correct visible width: delegates to pi-tui (ANSI-stripping,
+ * ASCII fast path, per-string cache, wide chars = 2 columns, tabs = 3).
+ */
 export function visibleWidth(value: string): number {
-	return [...stripAnsi(value)].length;
+	return tuiVisibleWidth(value);
 }
 export function resetAnsi(value: string): string {
 	return `${value}\x1b[0m`;
@@ -105,9 +111,10 @@ export function fitAnsiWidth(value: string, width: number, ellipsis = "…"): st
 export function truncateAnsi(value: string, width: number, ellipsis = "…"): string {
 	if (width <= 0) return "";
 	if (visibleWidth(value) <= width) return resetAnsi(value);
+	const ellipsisWidth = visibleWidth(ellipsis);
 	let output = "";
 	let visible = 0;
-	for (let i = 0; i < value.length && visible < width - visibleWidth(ellipsis); i++) {
+	for (let i = 0; i < value.length && visible < width - ellipsisWidth; i++) {
 		if (value.charCodeAt(i) === 27) {
 			const start = i;
 			i++;
@@ -125,12 +132,17 @@ export function wrapAnsi(value: string, width: number): string[] {
 	if (width <= 0) return [""];
 	const lines: string[] = [];
 	let line = "";
+	let lineWidth = 0;
 	for (const word of value.split(/\s+/)) {
-		const next = line ? `${line} ${word}` : word;
-		if (visibleWidth(next) <= width) line = next;
-		else {
+		const wordWidth = visibleWidth(word);
+		const nextWidth = line ? lineWidth + 1 + wordWidth : wordWidth;
+		if (nextWidth <= width) {
+			line = line ? `${line} ${word}` : word;
+			lineWidth = nextWidth;
+		} else {
 			if (line) lines.push(resetAnsi(line));
 			line = truncateAnsi(word, width);
+			lineWidth = visibleWidth(line);
 		}
 	}
 	if (line || lines.length === 0) lines.push(resetAnsi(line));

package/extension-src/pi-style/shared/box.ts

@@ -149,12 +149,63 @@ export function countLines(text: string): number {
 	return normalized.split("\n").length;
 }
 
+// Word-character membership table powering countWords. ASCII and the UTF-16
+// surrogate range are initialized eagerly; every other BMP code point is
+// resolved through the Unicode letter/digit class on first sight and memoized,
+// so repeat scans are pure table lookups. The per-code-point regex dispatch
+// this replaces measured ~1.9ms per 90KB output on every boxed footer render;
+// note that String.match with a \p{L}\p{N} class is no faster on V8 — the
+// memoized scan is the only variant that hit the <0.2ms budget.
+const WORD_CP_UNKNOWN = 255;
+const WORD_CP_SURROGATE = 254;
+const WORD_CP_CLASS = new Uint8Array(0x10000).fill(WORD_CP_UNKNOWN);
+for (let code = 0x30; code <= 0x39; code++) WORD_CP_CLASS[code] = 1; // 0-9
+for (let code = 0x41; code <= 0x5a; code++) WORD_CP_CLASS[code] = 1; // A-Z
+for (let code = 0x61; code <= 0x7a; code++) WORD_CP_CLASS[code] = 1; // a-z
+WORD_CP_CLASS[0x27] = 1; // '
+WORD_CP_CLASS[0x2d] = 1; // -
+WORD_CP_CLASS[0x5f] = 1; // _
+WORD_CP_CLASS.fill(WORD_CP_SURROGATE, 0xd800, 0xe000);
+const NON_ASCII_WORD_RE = /[\p{L}\p{N}]/u;
+const ASTRAL_WORD_CP = new Map<number, 0 | 1>();
+
+function astralWordMembership(codePoint: number): 0 | 1 {
+	const cached = ASTRAL_WORD_CP.get(codePoint);
+	if (cached !== undefined) return cached;
+	const membership: 0 | 1 = NON_ASCII_WORD_RE.test(String.fromCodePoint(codePoint)) ? 1 : 0;
+	ASTRAL_WORD_CP.set(codePoint, membership);
+	return membership;
+}
+
+/** Counts words as maximal runs of word characters (letters, digits,
+ *  underscore, apostrophe, hyphen) — identical counts to a per-code-point
+ *  `\p{L}\p{N}_'-` class test, in one table-driven pass with no allocations. */
 export function countWords(text: string): number {
+	const len = text.length;
 	let count = 0;
-	let inWord = false;
-	for (const char of text) {
-		const isWord = /[\p{L}\p{N}_'-]/u.test(char);
-		if (isWord && !inWord) count++;
+	let inWord: number = 0;
+	for (let i = 0; i < len; i++) {
+		const code = text.charCodeAt(i);
+		const membership = WORD_CP_CLASS[code] ?? 0;
+		let isWord: number;
+		if (membership <= 1) {
+			isWord = membership;
+		} else if (membership === WORD_CP_SURROGATE) {
+			const next = i + 1 < len ? text.charCodeAt(i + 1) : 0;
+			if (code <= 0xdbff && next >= 0xdc00 && next <= 0xdfff) {
+				// Astral word characters (e.g. mathematical alphanumerics) count as
+				// one code point, exactly like the previous code-point iteration.
+				isWord = astralWordMembership(0x10000 + ((code - 0xd800) << 10) + (next - 0xdc00));
+				i++; // consumed the pair's low surrogate half
+			} else {
+				isWord = 0; // lone surrogate half: never a word character
+			}
+		} else {
+			// First sight of this non-ASCII BMP code point: resolve once, memoize.
+			isWord = NON_ASCII_WORD_RE.test(String.fromCharCode(code)) ? 1 : 0;
+			WORD_CP_CLASS[code] = isWord;
+		}
+		count += isWord & (inWord ^ 1);
 		inWord = isWord;
 	}
 	return count;
@@ -210,6 +261,9 @@ const BOX_DIVIDER_RIGHT = "┤";
 /** Dash run before the right corner when a right-side border label is present. */
 const BOX_LABELED_RIGHT_DASH_MIN = 3;
 const BOX_WIDTH_CACHE = new Map<string, number>();
+/** Hard cap for BOX_WIDTH_CACHE; the oldest entry is evicted beyond this. Keys
+ *  embed full bash commands, so the cache must stay bounded across a session. */
+const BOX_WIDTH_CACHE_MAX_ENTRIES = 512;
 
 export function boxWidth(width: number): number {
 	return Math.max(BOX_MIN_WIDTH, width);
@@ -236,6 +290,13 @@ function _tightBoxWidth(
 	if (!widthKey) return measuredWidth;
 	const cachedWidth = BOX_WIDTH_CACHE.get(widthKey) ?? 0;
 	const nextWidth = Math.min(boxWidth(availableWidth), Math.max(cachedWidth, measuredWidth));
+	// Bounded LRU: Map preserves insertion order, so delete+set refreshes the
+	// key's recency and the first key is the oldest evict candidate.
+	BOX_WIDTH_CACHE.delete(widthKey);
+	if (BOX_WIDTH_CACHE.size >= BOX_WIDTH_CACHE_MAX_ENTRIES) {
+		const oldestKey = BOX_WIDTH_CACHE.keys().next().value;
+		if (oldestKey !== undefined) BOX_WIDTH_CACHE.delete(oldestKey);
+	}
 	BOX_WIDTH_CACHE.set(widthKey, nextWidth);
 	return nextWidth;
 }
@@ -244,6 +305,11 @@ export function boxedToolWidthKey(toolName: string, detail: string): string {
 	return `${toolName}:${detail}`;
 }
 
+/** Test-only debug view of the bounded box width cache. */
+export function __getBoxWidthCacheDebugState(): { size: number } {
+	return { size: BOX_WIDTH_CACHE.size };
+}
+
 export function formatToolName(toolName: string): string {
 	const spaced = toolName
 		.replace(/[_-]+/g, " ")

package/extension-src/pi-style/shared/split-diff.ts

@@ -56,7 +56,13 @@ type DiffEntry =
 
 const ESC = "\x1b";
 const BG_ANSI_PATTERN = new RegExp(`${ESC}\\[(?:4\\d|10\\d|48;5;\\d{1,3}|48;2;\\d{1,3};\\d{1,3};\\d{1,3}|49)m`, "g");
+/** Any SGR escape (capture: parameter bytes). Shared with String.replace only —
+ *  replace resets lastIndex, so the global flag is safe here. */
+const ANSI_SGR_PATTERN = new RegExp(`${ESC}\\[([0-9;]*)m`, "g");
 const CONTROL_CHARS = "\u0000-\u0008\u000B\u000C\u000E-\u001F\u007F";
+/** Control characters (minus \r/\n, handled separately) stripped from diff
+ *   text. Replace-only usage, so the global flag is safe (see above). */
+const CONTROL_CHARS_PATTERN = new RegExp(`[${CONTROL_CHARS}]`, "g");
 
 const ADD_ROW_BACKGROUND_MIX_RATIO = 0.24;
 const REMOVE_ROW_BACKGROUND_MIX_RATIO = 0.12;
@@ -173,7 +179,7 @@ function resolveDiffPalette(theme: SplitDiffTheme): DiffPalette {
 function keepBackgroundAcrossResets(text: string, rowBgAnsi: string): string {
 	if (!text) return text;
 
-	return text.replace(new RegExp(`${ESC}\\[([0-9;]*)m`, "g"), (sequence, rawCodes) => {
+	return text.replace(ANSI_SGR_PATTERN, (sequence, rawCodes) => {
 		const split = String(rawCodes ?? "")
 			.split(";")
 			.filter(Boolean);
@@ -233,10 +239,7 @@ function applyBackgroundToVisibleRange(
 // ── Text utilities ─────────────────────────────────────────────────
 
 function sanitizeSingleLineText(value: string): string {
-	return value
-		.replace(/\r/g, "")
-		.replace(/\n/g, "")
-		.replace(new RegExp(`[${CONTROL_CHARS}]`, "g"), "");
+	return value.replace(/\r/g, "").replace(/\n/g, "").replace(CONTROL_CHARS_PATTERN, "");
 }
 
 function stripInlineBreaksPreserveAnsi(value: string): string {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@quandev104/pi-style",
-  "version": "0.2.1",
+  "version": "0.2.2",
   "description": "A native-layout, cohesive visual style package for Pi.",
   "license": "MIT",
   "type": "module",