@flyingrobots/bijou 7.2.0 → 8.0.0-rc.1

package/dist/core/text/grapheme.js

@@ -1,298 +1,6 @@
-/**
- * Grapheme cluster utilities for correct Unicode text measurement.
- *
- * Uses `Intl.Segmenter` for proper grapheme cluster iteration and a compact
- * lookup for East Asian Wide / emoji display widths.
- */
-// ---------------------------------------------------------------------------
-// Segmenter (lazy singleton)
-// ---------------------------------------------------------------------------
-/** Cached singleton `Intl.Segmenter` instance for grapheme-level segmentation. */
-let _segmenter;
-/**
- * Return the lazily-initialized grapheme segmenter singleton.
- *
- * @returns The shared `Intl.Segmenter` configured for English grapheme granularity.
- */
-function segmenter() {
-    _segmenter ??= new Intl.Segmenter('en', { granularity: 'grapheme' });
-    return _segmenter;
-}
-// ---------------------------------------------------------------------------
-// Wide character detection
-// ---------------------------------------------------------------------------
-/**
- * Determine whether a Unicode code point occupies two terminal columns.
- *
- * Covers:
- * - CJK Unified Ideographs (U+4E00–U+9FFF)
- * - CJK Extension A (U+3400–U+4DBF)
- * - CJK Extension B–I (U+20000–U+3FFFF)
- * - CJK Compatibility Ideographs (U+F900–U+FAFF)
- * - Fullwidth Forms (U+FF01–U+FF60, U+FFE0–U+FFE6)
- * - CJK Radicals / Kangxi (U+2E80–U+2FDF)
- * - CJK Symbols and Punctuation (U+3000–U+303F)
- * - Enclosed CJK Letters (U+3200–U+33FF)
- * - CJK Compatibility (U+FE30–U+FE4F)
- * - Hangul Syllables (U+AC00–U+D7A3)
- * - Emoji (most U+1F000+)
- *
- * @param cp - Unicode code point to test.
- * @returns `true` if the code point renders as two columns wide.
- */
-export function isWideChar(cp) {
-    // Fullwidth Forms
-    if (cp >= 0xFF01 && cp <= 0xFF60)
-        return true;
-    if (cp >= 0xFFE0 && cp <= 0xFFE6)
-        return true;
-    // CJK Radicals, Kangxi Radicals
-    if (cp >= 0x2E80 && cp <= 0x2FDF)
-        return true;
-    // CJK Symbols and Punctuation, Hiragana, Katakana
-    if (cp >= 0x3000 && cp <= 0x33FF)
-        return true;
-    // CJK Extension A
-    if (cp >= 0x3400 && cp <= 0x4DBF)
-        return true;
-    // CJK Unified Ideographs
-    if (cp >= 0x4E00 && cp <= 0x9FFF)
-        return true;
-    // Hangul Syllables (U+AC00–U+D7A3; excludes Jamo Extended-B which are narrow)
-    if (cp >= 0xAC00 && cp <= 0xD7A3)
-        return true;
-    // CJK Compatibility Ideographs
-    if (cp >= 0xF900 && cp <= 0xFAFF)
-        return true;
-    // CJK Compatibility Forms
-    if (cp >= 0xFE30 && cp <= 0xFE4F)
-        return true;
-    // CJK Extension B–I and beyond
-    if (cp >= 0x20000 && cp <= 0x3FFFF)
-        return true;
-    // Emoji blocks (most render as 2 columns)
-    // Miscellaneous Symbols and Pictographs
-    if (cp >= 0x1F300 && cp <= 0x1F9FF)
-        return true;
-    // Supplemental Symbols and Pictographs
-    if (cp >= 0x1FA00 && cp <= 0x1FA6F)
-        return true;
-    // Symbols and Pictographs Extended-A
-    if (cp >= 0x1FA70 && cp <= 0x1FAFF)
-        return true;
-    // Emoticons
-    if (cp >= 0x1F600 && cp <= 0x1F64F)
-        return true;
-    // Transport and Map Symbols
-    if (cp >= 0x1F680 && cp <= 0x1F6FF)
-        return true;
-    // Dingbats (U+2702–U+27B0) are NOT uniformly wide per Unicode
-    // East Asian Width. Most are Narrow text characters (✓ ✗ ✌).
-    // They only become wide with emoji presentation (U+FE0F), which
-    // is handled by graphemeClusterWidth checking for the variation
-    // selector in the grapheme cluster.
-    // Regional indicators (flags)
-    if (cp >= 0x1F1E0 && cp <= 0x1F1FF)
-        return true;
-    // Playing Cards, Mahjong Tiles
-    if (cp >= 0x1F000 && cp <= 0x1F02F)
-        return true;
-    return false;
-}
-// ---------------------------------------------------------------------------
-// Grapheme segmentation
-// ---------------------------------------------------------------------------
-/**
- * Split a string into an array of grapheme clusters.
- *
- * Each element is a single user-perceived character (handles combining
- * marks, ZWJ sequences, flag pairs, skin tones, etc.).
- *
- * @param str - Input string to segment.
- * @returns Array of grapheme cluster strings.
- */
-export function segmentGraphemes(str) {
-    const segments = segmenter().segment(str);
-    const result = [];
-    for (const { segment } of segments) {
-        result.push(segment);
-    }
-    return result;
-}
-// ---------------------------------------------------------------------------
-// ANSI stripping
-// ---------------------------------------------------------------------------
-/**
- * Pattern matching ANSI SGR escape sequences (e.g. `\x1b[31m`).
- *
- * Exported for reuse by text utilities that need to strip terminal
- * style escapes before measuring or clipping visible content.
- */
-// eslint-disable-next-line no-control-regex
-export const ANSI_SGR_RE = /\x1b\[[0-9;]*m/;
-/**
- * Pattern matching OSC 8 hyperlink control sequences.
- *
- * This strips the invisible open/close control segments while preserving the
- * visible linked text between them.
- */
-// eslint-disable-next-line no-control-regex
-export const ANSI_OSC8_RE = /\x1b]8;;[^\x1b\x07]*(?:\x1b\\|\x07)/g;
-/**
- * Pattern matching broad ANSI control sequences (CSI, OSC, DCS, APC, etc.).
- *
- * Used for sanitizing untrusted terminal text at string-entry boundaries.
- * SGR and OSC 8 sequences can be temporarily preserved by higher-level helpers
- * when a caller intentionally opts into styled parsing.
- */
-// eslint-disable-next-line no-control-regex
-export const ANSI_CONTROL_SEQUENCE_RE = /\x1b(?:\[[0-?]*[ -/]*[@-~]|\][^\x1b\x07]*(?:\x1b\\|\x07)|[PX^_][\s\S]*?(?:\x1b\\|\x07)|[@-_])/g;
-/**
- * Strip all ANSI SGR escape sequences from a string.
- *
- * @param str - Input string potentially containing ANSI codes.
- * @returns The string with all ANSI SGR sequences removed.
- */
-export function stripAnsi(str) {
-    return str
-        .replace(ANSI_OSC8_RE, '')
-        .replace(new RegExp(ANSI_SGR_RE, 'g'), '');
-}
-/**
- * Sanitize raw terminal text before it crosses into the surface model.
- *
- * Removes destructive ANSI/control sequences and normalizes tabs/newlines so
- * untrusted content cannot inject cursor movement or clear-screen behavior when
- * later rendered back out through {@link surfaceToString} or the diff writer.
- *
- * By default this removes all ANSI sequences. Callers that intentionally parse
- * styled content can preserve SGR and/or OSC 8 sequences with the option flags.
- */
-export function sanitizeTerminalText(str, options = {}) {
-    const allowAnsiStyling = options.allowAnsiStyling ?? false;
-    const allowHyperlinks = options.allowHyperlinks ?? false;
-    const tabWidth = Math.max(1, Math.floor(options.tabWidth ?? 2));
-    const placeholders = [];
-    let text = str
-        .replace(/\r\n?/g, '\n')
-        .replace(/\t/g, ' '.repeat(tabWidth));
-    if (allowHyperlinks) {
-        text = stashTerminalSequences(text, ANSI_OSC8_RE, placeholders);
-    }
-    if (allowAnsiStyling) {
-        text = stashTerminalSequences(text, new RegExp(ANSI_SGR_RE, 'g'), placeholders);
-    }
-    text = stripUnsafeControlChars(text.replace(ANSI_CONTROL_SEQUENCE_RE, ''));
-    return restoreTerminalSequences(text, placeholders);
-}
-/**
- * Default boundary for untrusted plain text that will become cells or inline
- * terminal payloads.
- *
- * This strips destructive control sequences, normalizes tabs, and flattens
- * multiline input unless the caller explicitly opts into preserving newlines.
- * Callers that intentionally preserve styling should use
- * {@link sanitizeTerminalText} with allow flags or route through
- * `parseAnsiToSurface()`.
- */
-export function sanitizePlainTerminalText(str, options = {}) {
-    const preserveNewlines = options.preserveNewlines ?? false;
-    const newlineReplacement = options.newlineReplacement ?? ' ';
-    const text = sanitizeTerminalText(str, { tabWidth: options.tabWidth });
-    if (preserveNewlines)
-        return text;
-    return text.replace(/\n+/g, newlineReplacement);
-}
-function stashTerminalSequences(text, pattern, placeholders) {
-    return text.replace(pattern, (match) => {
-        const index = placeholders.push(match) - 1;
-        return `\uE000${String(index)}\uE001`;
-    });
-}
-function restoreTerminalSequences(text, placeholders) {
-    return text.replace(/\uE000(\d+)\uE001/g, (_match, rawIndex) => {
-        const index = Number.parseInt(rawIndex, 10);
-        return placeholders[index] ?? '';
-    });
-}
-function stripUnsafeControlChars(text) {
-    let result = '';
-    for (const ch of text) {
-        const cp = ch.codePointAt(0);
-        if (cp === undefined)
-            continue;
-        if ((cp >= 0 && cp <= 8) || (cp >= 11 && cp <= 31) || cp === 127)
-            continue;
-        result += ch;
-    }
-    return result;
-}
-// ---------------------------------------------------------------------------
-// Display width
-// ---------------------------------------------------------------------------
-/**
- * Compute the terminal display width of a single grapheme cluster.
- *
- * The width is determined by the widest code point in the cluster:
- * - East Asian Wide / Fullwidth characters → 2
- * - Emoji (U+1F300+) → 2
- * - Everything else (ASCII, Latin, combining marks, etc.) → 1
- *
- * Zero-width characters (combining marks, ZWJ, variation selectors)
- * don't add width on their own — they're part of the cluster.
- *
- * @param grapheme - A single grapheme cluster string.
- * @returns Display width: 1 for narrow characters, 2 for wide/emoji.
- */
-export function graphemeClusterWidth(grapheme) {
-    let maxWidth = 1;
-    let hasEmojiPresentation = false;
-    for (const ch of grapheme) {
-        const cp = ch.codePointAt(0);
-        if (cp === undefined)
-            continue;
-        // Detect emoji presentation selector (U+FE0F) — forces base char to 2-wide
-        if (cp === 0xFE0F) {
-            hasEmojiPresentation = true;
-            continue;
-        }
-        // Skip zero-width joiners and other variation selectors
-        if (cp === 0x200D || (cp >= 0xFE00 && cp <= 0xFE0E) || (cp >= 0xE0100 && cp <= 0xE01EF)) {
-            continue;
-        }
-        if (isWideChar(cp)) {
-            maxWidth = 2;
-            break; // Can't be wider than 2
-        }
-    }
-    // Emoji presentation (U+FE0F) makes any base character 2-wide.
-    // This handles Dingbats (✂️ ✈️) and other text chars presented as emoji.
-    if (hasEmojiPresentation && maxWidth < 2)
-        maxWidth = 2;
-    return maxWidth;
-}
-/**
- * Compute the terminal display width of a string.
- *
- * Strip ANSI escape sequences, segment into grapheme clusters,
- * and sum display widths. Correctly handles:
- * - Multi-codepoint emoji (flags, ZWJ families, skin tones)
- * - East Asian Wide characters (CJK, fullwidth forms)
- * - Combining marks (accented characters)
- * - ANSI escape sequences (ignored)
- *
- * @param str - Input string, may contain ANSI escape sequences.
- * @returns Total display width in terminal columns.
- */
-export function graphemeWidth(str) {
-    // Strip ANSI escapes first
-    const clean = stripAnsi(str);
-    if (clean.length === 0)
-        return 0;
-    let width = 0;
-    for (const { segment } of segmenter().segment(clean)) {
-        width += graphemeClusterWidth(segment);
-    }
-    return width;
-}
+export { isWideChar } from './grapheme.part01.js';
+export { segmentGraphemes, ANSI_SGR_RE, ANSI_OSC8_RE, ANSI_CONTROL_SEQUENCE_RE, stripAnsi, sanitizeTerminalText, } from './grapheme.part02.js';
+export { sanitizePlainTerminalText } from './grapheme-plain.js';
+export { graphemeClusterWidth } from './grapheme.part03.js';
+export { graphemeWidth } from './grapheme.part04.js';
 //# sourceMappingURL=grapheme.js.map

package/dist/core/text/grapheme.js.map

@@ -1 +1 @@
-{"version":3,"file":"grapheme.js","sourceRoot":"","sources":["../../../src/core/text/grapheme.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,8EAA8E;AAC9E,6BAA6B;AAC7B,8EAA8E;AAE9E,kFAAkF;AAClF,IAAI,UAAsC,CAAC;AAE3C;;;;GAIG;AACH,SAAS,SAAS;IAChB,UAAU,KAAK,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,EAAE,WAAW,EAAE,UAAU,EAAE,CAAC,CAAC;IACrE,OAAO,UAAU,CAAC;AACpB,CAAC;AAED,8EAA8E;AAC9E,2BAA2B;AAC3B,8EAA8E;AAE9E;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,UAAU,UAAU,CAAC,EAAU;IACnC,kBAAkB;IAClB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAC9C,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,gCAAgC;IAChC,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,kDAAkD;IAClD,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,kBAAkB;IAClB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,yBAAyB;IACzB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,8EAA8E;IAC9E,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,+BAA+B;IAC/B,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,0BAA0B;IAC1B,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,+BAA+B;IAC/B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAEhD,0CAA0C;IAC1C,wCAAwC;IACxC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,uCAAuC;IACvC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,qCAAqC;IACrC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,YAAY;IACZ,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,4BAA4B;IAC5B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,8DAA8D;IAC9D,6DAA6D;IAC7D,gEAAgE;IAChE,gEAAgE;IAChE,oCAAoC;IACpC,8BAA8B;IAC9B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,+BAA+B;IAC/B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAEhD,OAAO,KAAK,CAAC;AACf,CAAC;AAED,8EAA8E;AAC9E,wBAAwB;AACxB,8EAA8E;AAE9E;;;;;;;;GAQG;AACH,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC1C,MAAM,QAAQ,GAAG,SAAS,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAC1C,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,KAAK,MAAM,EAAE,OAAO,EAAE,IAAI,QAAQ,EAAE,CAAC;QACnC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACvB,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,8EAA8E;AAC9E,iBAAiB;AACjB,8EAA8E;AAE9E;;;;;GAKG;AACH,4CAA4C;AAC5C,MAAM,CAAC,MAAM,WAAW,GAAG,gBAAgB,CAAC;AAE5C;;;;;GAKG;AACH,4CAA4C;AAC5C,MAAM,CAAC,MAAM,YAAY,GAAG,sCAAsC,CAAC;AAEnE;;;;;;GAMG;AACH,4CAA4C;AAC5C,MAAM,CAAC,MAAM,wBAAwB,GAAG,gGAAgG,CAAC;AAWzI;;;;;GAKG;AACH,MAAM,UAAU,SAAS,CAAC,GAAW;IACnC,OAAO,GAAG;SACP,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC;SACzB,OAAO,CAAC,IAAI,MAAM,CAAC,WAAW,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC;AAC/C,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,oBAAoB,CAClC,GAAW,EACX,UAAuC,EAAE;IAEzC,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,KAAK,CAAC;IAC3D,MAAM,eAAe,GAAG,OAAO,CAAC,eAAe,IAAI,KAAK,CAAC;IACzD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC;IAChE,MAAM,YAAY,GAAa,EAAE,CAAC;IAElC,IAAI,IAAI,GAAG,GAAG;SACX,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC;SACvB,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC;IAExC,IAAI,eAAe,EAAE,CAAC;QACpB,IAAI,GAAG,sBAAsB,CAAC,IAAI,EAAE,YAAY,EAAE,YAAY,CAAC,CAAC;IAClE,CAAC;IACD,IAAI,gBAAgB,EAAE,CAAC;QACrB,IAAI,GAAG,sBAAsB,CAAC,IAAI,EAAE,IAAI,MAAM,CAAC,WAAW,EAAE,GAAG,CAAC,EAAE,YAAY,CAAC,CAAC;IAClF,CAAC;IAED,IAAI,GAAG,uBAAuB,CAAC,IAAI,CAAC,OAAO,CAAC,wBAAwB,EAAE,EAAE,CAAC,CAAC,CAAC;IAE3E,OAAO,wBAAwB,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;AACtD,CAAC;AAgBD;;;;;;;;;GASG;AACH,MAAM,UAAU,yBAAyB,CACvC,GAAW,EACX,UAA4C,EAAE;IAE9C,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,KAAK,CAAC;IAC3D,MAAM,kBAAkB,GAAG,OAAO,CAAC,kBAAkB,IAAI,GAAG,CAAC;IAC7D,MAAM,IAAI,GAAG,oBAAoB,CAAC,GAAG,EAAE,EAAE,QAAQ,EAAE,OAAO,CAAC,QAAQ,EAAE,CAAC,CAAC;IACvE,IAAI,gBAAgB;QAAE,OAAO,IAAI,CAAC;IAClC,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,EAAE,kBAAkB,CAAC,CAAC;AAClD,CAAC;AAED,SAAS,sBAAsB,CAC7B,IAAY,EACZ,OAAe,EACf,YAAsB;IAEtB,OAAO,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE,CAAC,KAAK,EAAE,EAAE;QACrC,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAC3C,OAAO,SAAS,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC;IACxC,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,wBAAwB,CAAC,IAAY,EAAE,YAA+B;IAC7E,OAAO,IAAI,CAAC,OAAO,CAAC,oBAAoB,EAAE,CAAC,MAAM,EAAE,QAAgB,EAAE,EAAE;QACrE,MAAM,KAAK,GAAG,MAAM,CAAC,QAAQ,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC;QAC5C,OAAO,YAAY,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC;IACnC,CAAC,CAAC,CAAC;AACL,CAAC;AAED,SAAS,uBAAuB,CAAC,IAAY;IAC3C,IAAI,MAAM,GAAG,EAAE,CAAC;IAChB,KAAK,MAAM,EAAE,IAAI,IAAI,EAAE,CAAC;QACtB,MAAM,EAAE,GAAG,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC;QAC7B,IAAI,EAAE,KAAK,SAAS;YAAE,SAAS;QAC/B,IAAI,CAAC,EAAE,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,CAAC,IAAI,EAAE,KAAK,GAAG;YAAE,SAAS;QAC3E,MAAM,IAAI,EAAE,CAAC;IACf,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AAED,8EAA8E;AAC9E,gBAAgB;AAChB,8EAA8E;AAE9E;;;;;;;;;;;;;GAaG;AACH,MAAM,UAAU,oBAAoB,CAAC,QAAgB;IACnD,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,oBAAoB,GAAG,KAAK,CAAC;IACjC,KAAK,MAAM,EAAE,IAAI,QAAQ,EAAE,CAAC;QAC1B,MAAM,EAAE,GAAG,EAAE,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC;QAC7B,IAAI,EAAE,KAAK,SAAS;YAAE,SAAS;QAC/B,2EAA2E;QAC3E,IAAI,EAAE,KAAK,MAAM,EAAE,CAAC;YAClB,oBAAoB,GAAG,IAAI,CAAC;YAC5B,SAAS;QACX,CAAC;QACD,wDAAwD;QACxD,IAAI,EAAE,KAAK,MAAM,IAAI,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC,IAAI,CAAC,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO,CAAC,EAAE,CAAC;YACxF,SAAS;QACX,CAAC;QACD,IAAI,UAAU,CAAC,EAAE,CAAC,EAAE,CAAC;YACnB,QAAQ,GAAG,CAAC,CAAC;YACb,MAAM,CAAE,wBAAwB;QAClC,CAAC;IACH,CAAC;IACD,+DAA+D;IAC/D,yEAAyE;IACzE,IAAI,oBAAoB,IAAI,QAAQ,GAAG,CAAC;QAAE,QAAQ,GAAG,CAAC,CAAC;IACvD,OAAO,QAAQ,CAAC;AAClB,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,aAAa,CAAC,GAAW;IACvC,2BAA2B;IAC3B,MAAM,KAAK,GAAG,SAAS,CAAC,GAAG,CAAC,CAAC;IAC7B,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAEjC,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,KAAK,MAAM,EAAE,OAAO,EAAE,IAAI,SAAS,EAAE,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QACrD,KAAK,IAAI,oBAAoB,CAAC,OAAO,CAAC,CAAC;IACzC,CAAC;IACD,OAAO,KAAK,CAAC;AACf,CAAC"}
+{"version":3,"file":"grapheme.js","sourceRoot":"","sources":["../../../src/core/text/grapheme.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAElD,OAAO,EACL,gBAAgB,EAChB,WAAW,EACX,YAAY,EACZ,wBAAwB,EACxB,SAAS,EACT,oBAAoB,GACrB,MAAM,sBAAsB,CAAC;AAE9B,OAAO,EAAE,yBAAyB,EAAE,MAAM,qBAAqB,CAAC;AAChE,OAAO,EAAE,oBAAoB,EAAE,MAAM,sBAAsB,CAAC;AAC5D,OAAO,EAAE,aAAa,EAAE,MAAM,sBAAsB,CAAC"}

package/dist/core/text/grapheme.part01.d.ts

@@ -0,0 +1,35 @@
+/**
+ * Grapheme cluster utilities for correct Unicode text measurement.
+ *
+ * Uses `Intl.Segmenter` for proper grapheme cluster iteration and a compact
+ * lookup for East Asian Wide / emoji display widths.
+ */
+/** Cached singleton `Intl.Segmenter` instance for grapheme-level segmentation. */
+export declare let _segmenter: Intl.Segmenter | undefined;
+/**
+ * Return the lazily-initialized grapheme segmenter singleton.
+ *
+ * @returns The shared `Intl.Segmenter` configured for English grapheme granularity.
+ */
+export declare function segmenter(): Intl.Segmenter;
+/**
+ * Determine whether a Unicode code point occupies two terminal columns.
+ *
+ * Covers:
+ * - CJK Unified Ideographs (U+4E00–U+9FFF)
+ * - CJK Extension A (U+3400–U+4DBF)
+ * - CJK Extension B–I (U+20000–U+3FFFF)
+ * - CJK Compatibility Ideographs (U+F900–U+FAFF)
+ * - Fullwidth Forms (U+FF01–U+FF60, U+FFE0–U+FFE6)
+ * - CJK Radicals / Kangxi (U+2E80–U+2FDF)
+ * - CJK Symbols and Punctuation (U+3000–U+303F)
+ * - Enclosed CJK Letters (U+3200–U+33FF)
+ * - CJK Compatibility (U+FE30–U+FE4F)
+ * - Hangul Syllables (U+AC00–U+D7A3)
+ * - Emoji (most U+1F000+)
+ *
+ * @param cp - Unicode code point to test.
+ * @returns `true` if the code point renders as two columns wide.
+ */
+export declare function isWideChar(cp: number): boolean;
+//# sourceMappingURL=grapheme.part01.d.ts.map

package/dist/core/text/grapheme.part01.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"grapheme.part01.d.ts","sourceRoot":"","sources":["../../../src/core/text/grapheme.part01.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAMH,kFAAkF;AAClF,eAAO,IAAI,UAAU,EAAE,IAAI,CAAC,SAAS,GAAG,SAAS,CAAC;AAClD;;;;GAIG;AACH,wBAAgB,SAAS,IAAI,IAAI,CAAC,SAAS,CAG1C;AAKD;;;;;;;;;;;;;;;;;;GAkBG;AACH,wBAAgB,UAAU,CAAC,EAAE,EAAE,MAAM,GAAG,OAAO,CAmD9C"}

package/dist/core/text/grapheme.part01.js

@@ -0,0 +1,102 @@
+/**
+ * Grapheme cluster utilities for correct Unicode text measurement.
+ *
+ * Uses `Intl.Segmenter` for proper grapheme cluster iteration and a compact
+ * lookup for East Asian Wide / emoji display widths.
+ */
+// ---------------------------------------------------------------------------
+// Segmenter (lazy singleton)
+// ---------------------------------------------------------------------------
+/** Cached singleton `Intl.Segmenter` instance for grapheme-level segmentation. */
+export let _segmenter;
+/**
+ * Return the lazily-initialized grapheme segmenter singleton.
+ *
+ * @returns The shared `Intl.Segmenter` configured for English grapheme granularity.
+ */
+export function segmenter() {
+    _segmenter ??= new Intl.Segmenter('en', { granularity: 'grapheme' });
+    return _segmenter;
+}
+// ---------------------------------------------------------------------------
+// Wide character detection
+// ---------------------------------------------------------------------------
+/**
+ * Determine whether a Unicode code point occupies two terminal columns.
+ *
+ * Covers:
+ * - CJK Unified Ideographs (U+4E00–U+9FFF)
+ * - CJK Extension A (U+3400–U+4DBF)
+ * - CJK Extension B–I (U+20000–U+3FFFF)
+ * - CJK Compatibility Ideographs (U+F900–U+FAFF)
+ * - Fullwidth Forms (U+FF01–U+FF60, U+FFE0–U+FFE6)
+ * - CJK Radicals / Kangxi (U+2E80–U+2FDF)
+ * - CJK Symbols and Punctuation (U+3000–U+303F)
+ * - Enclosed CJK Letters (U+3200–U+33FF)
+ * - CJK Compatibility (U+FE30–U+FE4F)
+ * - Hangul Syllables (U+AC00–U+D7A3)
+ * - Emoji (most U+1F000+)
+ *
+ * @param cp - Unicode code point to test.
+ * @returns `true` if the code point renders as two columns wide.
+ */
+export function isWideChar(cp) {
+    // Fullwidth Forms
+    if (cp >= 0xff01 && cp <= 0xff60)
+        return true;
+    if (cp >= 0xffe0 && cp <= 0xffe6)
+        return true;
+    // CJK Radicals, Kangxi Radicals
+    if (cp >= 0x2e80 && cp <= 0x2fdf)
+        return true;
+    // CJK Symbols and Punctuation, Hiragana, Katakana
+    if (cp >= 0x3000 && cp <= 0x33ff)
+        return true;
+    // CJK Extension A
+    if (cp >= 0x3400 && cp <= 0x4dbf)
+        return true;
+    // CJK Unified Ideographs
+    if (cp >= 0x4e00 && cp <= 0x9fff)
+        return true;
+    // Hangul Syllables (U+AC00–U+D7A3; excludes Jamo Extended-B which are narrow)
+    if (cp >= 0xac00 && cp <= 0xd7a3)
+        return true;
+    // CJK Compatibility Ideographs
+    if (cp >= 0xf900 && cp <= 0xfaff)
+        return true;
+    // CJK Compatibility Forms
+    if (cp >= 0xfe30 && cp <= 0xfe4f)
+        return true;
+    // CJK Extension B–I and beyond
+    if (cp >= 0x20000 && cp <= 0x3ffff)
+        return true;
+    // Emoji blocks (most render as 2 columns)
+    // Miscellaneous Symbols and Pictographs
+    if (cp >= 0x1f300 && cp <= 0x1f9ff)
+        return true;
+    // Supplemental Symbols and Pictographs
+    if (cp >= 0x1fa00 && cp <= 0x1fa6f)
+        return true;
+    // Symbols and Pictographs Extended-A
+    if (cp >= 0x1fa70 && cp <= 0x1faff)
+        return true;
+    // Emoticons
+    if (cp >= 0x1f600 && cp <= 0x1f64f)
+        return true;
+    // Transport and Map Symbols
+    if (cp >= 0x1f680 && cp <= 0x1f6ff)
+        return true;
+    // Dingbats (U+2702–U+27B0) are NOT uniformly wide per Unicode
+    // East Asian Width. Most are Narrow text characters (✓ ✗ ✌).
+    // They only become wide with emoji presentation (U+FE0F), which
+    // is handled by graphemeClusterWidth checking for the variation
+    // selector in the grapheme cluster.
+    // Regional indicators (flags)
+    if (cp >= 0x1f1e0 && cp <= 0x1f1ff)
+        return true;
+    // Playing Cards, Mahjong Tiles
+    if (cp >= 0x1f000 && cp <= 0x1f02f)
+        return true;
+    return false;
+}
+//# sourceMappingURL=grapheme.part01.js.map

package/dist/core/text/grapheme.part01.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"grapheme.part01.js","sourceRoot":"","sources":["../../../src/core/text/grapheme.part01.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,8EAA8E;AAC9E,6BAA6B;AAC7B,8EAA8E;AAE9E,kFAAkF;AAClF,MAAM,CAAC,IAAI,UAAsC,CAAC;AAClD;;;;GAIG;AACH,MAAM,UAAU,SAAS;IACvB,UAAU,KAAK,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,EAAE,WAAW,EAAE,UAAU,EAAE,CAAC,CAAC;IACrE,OAAO,UAAU,CAAC;AACpB,CAAC;AACD,8EAA8E;AAC9E,2BAA2B;AAC3B,8EAA8E;AAE9E;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,UAAU,UAAU,CAAC,EAAU;IACnC,kBAAkB;IAClB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAC9C,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,gCAAgC;IAChC,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,kDAAkD;IAClD,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,kBAAkB;IAClB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,yBAAyB;IACzB,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,8EAA8E;IAC9E,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,+BAA+B;IAC/B,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,0BAA0B;IAC1B,IAAI,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM;QAAE,OAAO,IAAI,CAAC;IAE9C,+BAA+B;IAC/B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAEhD,0CAA0C;IAC1C,wCAAwC;IACxC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,uCAAuC;IACvC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,qCAAqC;IACrC,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,YAAY;IACZ,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,4BAA4B;IAC5B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,8DAA8D;IAC9D,6DAA6D;IAC7D,gEAAgE;IAChE,gEAAgE;IAChE,oCAAoC;IACpC,8BAA8B;IAC9B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAChD,+BAA+B;IAC/B,IAAI,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO;QAAE,OAAO,IAAI,CAAC;IAEhD,OAAO,KAAK,CAAC;AACf,CAAC"}

package/dist/core/text/grapheme.part02.d.ts

@@ -0,0 +1,59 @@
+/**
+ * Split a string into an array of grapheme clusters.
+ *
+ * Each element is a single user-perceived character (handles combining
+ * marks, ZWJ sequences, flag pairs, skin tones, etc.).
+ *
+ * @param str - Input string to segment.
+ * @returns Array of grapheme cluster strings.
+ */
+export declare function segmentGraphemes(str: string): string[];
+/**
+ * Pattern matching ANSI SGR escape sequences (e.g. `\x1b[31m`).
+ *
+ * Exported for reuse by text utilities that need to strip terminal
+ * style escapes before measuring or clipping visible content.
+ */
+export declare const ANSI_SGR_RE: RegExp;
+/**
+ * Pattern matching OSC 8 hyperlink control sequences.
+ *
+ * This strips the invisible open/close control segments while preserving the
+ * visible linked text between them.
+ */
+export declare const ANSI_OSC8_RE: RegExp;
+/**
+ * Pattern matching broad ANSI control sequences (CSI, OSC, DCS, APC, etc.).
+ *
+ * Used for sanitizing untrusted terminal text at string-entry boundaries.
+ * SGR and OSC 8 sequences can be temporarily preserved by higher-level helpers
+ * when a caller intentionally opts into styled parsing.
+ */
+export declare const ANSI_CONTROL_SEQUENCE_RE: RegExp;
+export interface SanitizeTerminalTextOptions {
+    /** Preserve ANSI SGR styling escapes such as `\x1b[31m`. */
+    readonly allowAnsiStyling?: boolean;
+    /** Preserve OSC 8 hyperlink control sequences. */
+    readonly allowHyperlinks?: boolean;
+    /** Replacement width for literal tabs. Defaults to 2 spaces. */
+    readonly tabWidth?: number;
+}
+/**
+ * Strip all ANSI SGR escape sequences from a string.
+ *
+ * @param str - Input string potentially containing ANSI codes.
+ * @returns The string with all ANSI SGR sequences removed.
+ */
+export declare function stripAnsi(str: string): string;
+/**
+ * Sanitize raw terminal text before it crosses into the surface model.
+ *
+ * Removes destructive ANSI/control sequences and normalizes tabs/newlines so
+ * untrusted content cannot inject cursor movement or clear-screen behavior when
+ * later rendered back out through {@link surfaceToString} or the diff writer.
+ *
+ * By default this removes all ANSI sequences. Callers that intentionally parse
+ * styled content can preserve SGR and/or OSC 8 sequences with the option flags.
+ */
+export declare function sanitizeTerminalText(str: string, options?: SanitizeTerminalTextOptions): string;
+//# sourceMappingURL=grapheme.part02.d.ts.map

package/dist/core/text/grapheme.part02.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"grapheme.part02.d.ts","sourceRoot":"","sources":["../../../src/core/text/grapheme.part02.ts"],"names":[],"mappings":"AAWA;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,EAAE,CAOtD;AAKD;;;;;GAKG;AAEH,eAAO,MAAM,WAAW,QAAmB,CAAC;AAC5C;;;;;GAKG;AAEH,eAAO,MAAM,YAAY,QAAyC,CAAC;AACnE;;;;;;GAMG;AACH,eAAO,MAAM,wBAAwB,QAE6D,CAAC;AACnG,MAAM,WAAW,2BAA2B;IAC1C,4DAA4D;IAC5D,QAAQ,CAAC,gBAAgB,CAAC,EAAE,OAAO,CAAC;IACpC,kDAAkD;IAClD,QAAQ,CAAC,eAAe,CAAC,EAAE,OAAO,CAAC;IACnC,gEAAgE;IAChE,QAAQ,CAAC,QAAQ,CAAC,EAAE,MAAM,CAAC;CAC5B;AACD;;;;;GAKG;AACH,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAI7C;AACD;;;;;;;;;GASG;AACH,wBAAgB,oBAAoB,CAClC,GAAG,EAAE,MAAM,EACX,OAAO,GAAE,2BAAgC,GACxC,MAAM,CAsBR"}

package/dist/core/text/grapheme.part02.js

@@ -0,0 +1,88 @@
+import { segmenter } from './grapheme.part01.js';
+import { restoreTerminalSequences, stashTerminalSequences, stripUnsafeControlChars, } from './grapheme.part03.js';
+// ---------------------------------------------------------------------------
+// Grapheme segmentation
+// ---------------------------------------------------------------------------
+/**
+ * Split a string into an array of grapheme clusters.
+ *
+ * Each element is a single user-perceived character (handles combining
+ * marks, ZWJ sequences, flag pairs, skin tones, etc.).
+ *
+ * @param str - Input string to segment.
+ * @returns Array of grapheme cluster strings.
+ */
+export function segmentGraphemes(str) {
+    const segments = segmenter().segment(str);
+    const result = [];
+    for (const { segment } of segments) {
+        result.push(segment);
+    }
+    return result;
+}
+// ---------------------------------------------------------------------------
+// ANSI stripping
+// ---------------------------------------------------------------------------
+/**
+ * Pattern matching ANSI SGR escape sequences (e.g. `\x1b[31m`).
+ *
+ * Exported for reuse by text utilities that need to strip terminal
+ * style escapes before measuring or clipping visible content.
+ */
+// eslint-disable-next-line no-control-regex
+export const ANSI_SGR_RE = /\x1b\[[0-9;]*m/;
+/**
+ * Pattern matching OSC 8 hyperlink control sequences.
+ *
+ * This strips the invisible open/close control segments while preserving the
+ * visible linked text between them.
+ */
+// eslint-disable-next-line no-control-regex
+export const ANSI_OSC8_RE = /\x1b]8;;[^\x1b\x07]*(?:\x1b\\|\x07)/g;
+/**
+ * Pattern matching broad ANSI control sequences (CSI, OSC, DCS, APC, etc.).
+ *
+ * Used for sanitizing untrusted terminal text at string-entry boundaries.
+ * SGR and OSC 8 sequences can be temporarily preserved by higher-level helpers
+ * when a caller intentionally opts into styled parsing.
+ */
+export const ANSI_CONTROL_SEQUENCE_RE = 
+// eslint-disable-next-line no-control-regex
+/\x1b(?:\[[0-?]*[ -/]*[@-~]|\][^\x1b\x07]*(?:\x1b\\|\x07)|[PX^_][\s\S]*?(?:\x1b\\|\x07)|[@-_])/g;
+/**
+ * Strip all ANSI SGR escape sequences from a string.
+ *
+ * @param str - Input string potentially containing ANSI codes.
+ * @returns The string with all ANSI SGR sequences removed.
+ */
+export function stripAnsi(str) {
+    return str
+        .replace(ANSI_OSC8_RE, '')
+        .replace(new RegExp(ANSI_SGR_RE, 'g'), '');
+}
+/**
+ * Sanitize raw terminal text before it crosses into the surface model.
+ *
+ * Removes destructive ANSI/control sequences and normalizes tabs/newlines so
+ * untrusted content cannot inject cursor movement or clear-screen behavior when
+ * later rendered back out through {@link surfaceToString} or the diff writer.
+ *
+ * By default this removes all ANSI sequences. Callers that intentionally parse
+ * styled content can preserve SGR and/or OSC 8 sequences with the option flags.
+ */
+export function sanitizeTerminalText(str, options = {}) {
+    const allowAnsiStyling = options.allowAnsiStyling ?? false;
+    const allowHyperlinks = options.allowHyperlinks ?? false;
+    const tabWidth = Math.max(1, Math.floor(options.tabWidth ?? 2));
+    const placeholders = [];
+    let text = str.replace(/\r\n?/g, '\n').replace(/\t/g, ' '.repeat(tabWidth));
+    if (allowHyperlinks) {
+        text = stashTerminalSequences(text, ANSI_OSC8_RE, placeholders);
+    }
+    if (allowAnsiStyling) {
+        text = stashTerminalSequences(text, new RegExp(ANSI_SGR_RE, 'g'), placeholders);
+    }
+    text = stripUnsafeControlChars(text.replace(ANSI_CONTROL_SEQUENCE_RE, ''));
+    return restoreTerminalSequences(text, placeholders);
+}
+//# sourceMappingURL=grapheme.part02.js.map

package/dist/core/text/grapheme.part02.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"grapheme.part02.js","sourceRoot":"","sources":["../../../src/core/text/grapheme.part02.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,SAAS,EAAE,MAAM,sBAAsB,CAAC;AACjD,OAAO,EACL,wBAAwB,EACxB,sBAAsB,EACtB,uBAAuB,GACxB,MAAM,sBAAsB,CAAC;AAE9B,8EAA8E;AAC9E,wBAAwB;AACxB,8EAA8E;AAE9E;;;;;;;;GAQG;AACH,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC1C,MAAM,QAAQ,GAAG,SAAS,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC;IAC1C,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,KAAK,MAAM,EAAE,OAAO,EAAE,IAAI,QAAQ,EAAE,CAAC;QACnC,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACvB,CAAC;IACD,OAAO,MAAM,CAAC;AAChB,CAAC;AACD,8EAA8E;AAC9E,iBAAiB;AACjB,8EAA8E;AAE9E;;;;;GAKG;AACH,4CAA4C;AAC5C,MAAM,CAAC,MAAM,WAAW,GAAG,gBAAgB,CAAC;AAC5C;;;;;GAKG;AACH,4CAA4C;AAC5C,MAAM,CAAC,MAAM,YAAY,GAAG,sCAAsC,CAAC;AACnE;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,wBAAwB;AACnC,4CAA4C;AAC5C,gGAAgG,CAAC;AASnG;;;;;GAKG;AACH,MAAM,UAAU,SAAS,CAAC,GAAW;IACnC,OAAO,GAAG;SACP,OAAO,CAAC,YAAY,EAAE,EAAE,CAAC;SACzB,OAAO,CAAC,IAAI,MAAM,CAAC,WAAW,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC;AAC/C,CAAC;AACD;;;;;;;;;GASG;AACH,MAAM,UAAU,oBAAoB,CAClC,GAAW,EACX,UAAuC,EAAE;IAEzC,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,IAAI,KAAK,CAAC;IAC3D,MAAM,eAAe,GAAG,OAAO,CAAC,eAAe,IAAI,KAAK,CAAC;IACzD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,IAAI,CAAC,CAAC,CAAC,CAAC;IAChE,MAAM,YAAY,GAAa,EAAE,CAAC;IAElC,IAAI,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAC,OAAO,CAAC,KAAK,EAAE,GAAG,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC;IAE5E,IAAI,eAAe,EAAE,CAAC;QACpB,IAAI,GAAG,sBAAsB,CAAC,IAAI,EAAE,YAAY,EAAE,YAAY,CAAC,CAAC;IAClE,CAAC;IACD,IAAI,gBAAgB,EAAE,CAAC;QACrB,IAAI,GAAG,sBAAsB,CAC3B,IAAI,EACJ,IAAI,MAAM,CAAC,WAAW,EAAE,GAAG,CAAC,EAC5B,YAAY,CACb,CAAC;IACJ,CAAC;IAED,IAAI,GAAG,uBAAuB,CAAC,IAAI,CAAC,OAAO,CAAC,wBAAwB,EAAE,EAAE,CAAC,CAAC,CAAC;IAE3E,OAAO,wBAAwB,CAAC,IAAI,EAAE,YAAY,CAAC,CAAC;AACtD,CAAC"}

package/dist/core/text/grapheme.part03.d.ts

@@ -0,0 +1,19 @@
+export declare function stashTerminalSequences(text: string, pattern: RegExp, placeholders: string[]): string;
+export declare function restoreTerminalSequences(text: string, placeholders: readonly string[]): string;
+export declare function stripUnsafeControlChars(text: string): string;
+/**
+ * Compute the terminal display width of a single grapheme cluster.
+ *
+ * The width is determined by the widest code point in the cluster:
+ * - East Asian Wide / Fullwidth characters → 2
+ * - Emoji (U+1F300+) → 2
+ * - Everything else (ASCII, Latin, combining marks, etc.) → 1
+ *
+ * Zero-width characters (combining marks, ZWJ, variation selectors)
+ * don't add width on their own — they're part of the cluster.
+ *
+ * @param grapheme - A single grapheme cluster string.
+ * @returns Display width: 1 for narrow characters, 2 for wide/emoji.
+ */
+export declare function graphemeClusterWidth(grapheme: string): number;
+//# sourceMappingURL=grapheme.part03.d.ts.map

package/dist/core/text/grapheme.part03.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"grapheme.part03.d.ts","sourceRoot":"","sources":["../../../src/core/text/grapheme.part03.ts"],"names":[],"mappings":"AACA,wBAAgB,sBAAsB,CACpC,IAAI,EAAE,MAAM,EACZ,OAAO,EAAE,MAAM,EACf,YAAY,EAAE,MAAM,EAAE,GACrB,MAAM,CAKR;AACD,wBAAgB,wBAAwB,CACtC,IAAI,EAAE,MAAM,EACZ,YAAY,EAAE,SAAS,MAAM,EAAE,GAC9B,MAAM,CAKR;AACD,wBAAgB,uBAAuB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAS5D;AAKD;;;;;;;;;;;;;GAaG;AACH,wBAAgB,oBAAoB,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,CA4B7D"}