claude-code/src/utils/sliceAnsi.ts

import {
  type AnsiCode,
  type Char,
  ansiCodesToString,
  reduceAnsiCodes,
  tokenize,
  undoAnsiCodes,
} from '@alcalzone/ansi-tokenize'
import { stringWidth } from '@anthropic/ink'

// A code is an "end code" if its code equals its endCode (e.g., hyperlink close)
function isEndCode(code: AnsiCode): boolean {
  return code.code === code.endCode
}

// Filter to only include "start codes" (not end codes)
function filterStartCodes(codes: AnsiCode[]): AnsiCode[] {
  return codes.filter(c => !isEndCode(c))
}

/**
 * Slice a string containing ANSI escape codes.
 *
 * Unlike the slice-ansi package, this properly handles OSC 8 hyperlink
 * sequences because @alcalzone/ansi-tokenize tokenizes them correctly.
 */
export default function sliceAnsi(
  str: string,
  start: number,
  end?: number,
): string {
  // Don't pass `end` to tokenize — it counts code units, not display cells,
  // so it drops tokens early for text with zero-width combining marks.
  const tokens = tokenize(str)
  let activeCodes: AnsiCode[] = []
  let position = 0
  let result = ''
  let include = false

  for (const token of tokens) {
    // Advance by display width, not code units. Combining marks (Devanagari
    // matras, virama, diacritics) are width 0 — counting them via .length
    // advanced position past `end` early and truncated the slice. Callers
    // pass start/end in display cells (via stringWidth), so position must
    // track the same units.
    const width =
      token.type === 'ansi'
        ? 0
        : token.type === 'char'
          ? token.fullWidth
            ? 2
            : stringWidth(token.value)
          : 0

    // Break AFTER trailing zero-width marks — a combining mark attaches to
    // the preceding base char, so "भा" (भ + ा, 1 display cell) sliced at
    // end=1 must include the ा. Breaking on position >= end BEFORE the
    // zero-width check would drop it and render भ bare. ANSI codes are
    // width 0 but must NOT be included past end (they open new style runs
    // that leak into the undo sequence), so gate on char type too. The
    // !include guard ensures empty slices (start===end) stay empty even
    // when the string starts with a zero-width char (BOM, ZWJ).
    if (end !== undefined && position >= end) {
      if (token.type === 'ansi' || width > 0 || !include) break
    }

    if (token.type === 'ansi') {
      activeCodes.push(token)
      if (include) {
        // Emit all ANSI codes during the slice
        result += token.code
      }
    } else {
      if (!include && position >= start) {
        // Skip leading zero-width marks at the start boundary — they belong
        // to the preceding base char in the left half. Without this, the
        // mark appears in BOTH halves: left+right ≠ original. Only applies
        // when start > 0 (otherwise there's no preceding char to own it).
        if (start > 0 && width === 0) continue
        include = true
        // Reduce and filter to only active start codes
        activeCodes = filterStartCodes(reduceAnsiCodes(activeCodes))
        result = ansiCodesToString(activeCodes)
      }

      if (include) {
        result += (token as Char).value
      }

      position += width
    }
  }

  // Only undo start codes that are still active
  const activeStartCodes = filterStartCodes(reduceAnsiCodes(activeCodes))
  result += ansiCodesToString(undoAnsiCodes(activeStartCodes))
  return result
}