claude-code / utils / textHighlighting.ts

import {
  type AnsiCode,
  ansiCodesToString,
  reduceAnsiCodes,
  type Token,
  tokenize,
  undoAnsiCodes,
} from '@alcalzone/ansi-tokenize'
import type { Theme } from './theme.js'

export type TextHighlight = {
  start: number
  end: number
  color: keyof Theme | undefined
  dimColor?: boolean
  inverse?: boolean
  shimmerColor?: keyof Theme
  priority: number
}

export type TextSegment = {
  text: string
  start: number
  highlight?: TextHighlight
}

export function segmentTextByHighlights(
  text: string,
  highlights: TextHighlight[],
): TextSegment[] {
  if (highlights.length === 0) {
    return [{ text, start: 0 }]
  }

  const sortedHighlights = [...highlights].sort((a, b) => {
    if (a.start !== b.start) return a.start - b.start
    return b.priority - a.priority
  })

  const resolvedHighlights: TextHighlight[] = []
  const usedRanges: Array<{ start: number; end: number }> = []

  for (const highlight of sortedHighlights) {
    if (highlight.start === highlight.end) continue

    const overlaps = usedRanges.some(
      range =>
        (highlight.start >= range.start && highlight.start < range.end) ||
        (highlight.end > range.start && highlight.end <= range.end) ||
        (highlight.start <= range.start && highlight.end >= range.end),
    )

    if (!overlaps) {
      resolvedHighlights.push(highlight)
      usedRanges.push({ start: highlight.start, end: highlight.end })
    }
  }

  return new HighlightSegmenter(text).segment(resolvedHighlights)
}

class HighlightSegmenter {
  private readonly tokens: Token[]
  // Two position systems: "visible" (what the user sees, excluding ANSI codes)
  // and "string" (raw positions including ANSI codes for substring extraction)
  private visiblePos = 0
  private stringPos = 0
  private tokenIdx = 0
  private charIdx = 0 // offset within current text token (for partial consumption)
  private codes: AnsiCode[] = []

  constructor(private readonly text: string) {
    this.tokens = tokenize(text)
  }

  segment(highlights: TextHighlight[]): TextSegment[] {
    const segments: TextSegment[] = []

    for (const highlight of highlights) {
      const before = this.segmentTo(highlight.start)
      if (before) segments.push(before)

      const highlighted = this.segmentTo(highlight.end)
      if (highlighted) {
        highlighted.highlight = highlight
        segments.push(highlighted)
      }
    }

    const after = this.segmentTo(Infinity)
    if (after) segments.push(after)

    return segments
  }

  private segmentTo(targetVisiblePos: number): TextSegment | null {
    if (
      this.tokenIdx >= this.tokens.length ||
      targetVisiblePos <= this.visiblePos
    ) {
      return null
    }

    const visibleStart = this.visiblePos

    // Consume leading ANSI codes before first visible char
    while (this.tokenIdx < this.tokens.length) {
      const token = this.tokens[this.tokenIdx]!
      if (token.type !== 'ansi') break
      this.codes.push(token)
      this.stringPos += token.code.length
      this.tokenIdx++
    }

    const stringStart = this.stringPos
    const codesStart = [...this.codes]

    // Advance through tokens until we reach target
    while (
      this.visiblePos < targetVisiblePos &&
      this.tokenIdx < this.tokens.length
    ) {
      const token = this.tokens[this.tokenIdx]!

      if (token.type === 'ansi') {
        this.codes.push(token)
        this.stringPos += token.code.length
        this.tokenIdx++
      } else {
        const charsNeeded = targetVisiblePos - this.visiblePos
        const charsAvailable = token.value.length - this.charIdx
        const charsToTake = Math.min(charsNeeded, charsAvailable)

        this.stringPos += charsToTake
        this.visiblePos += charsToTake
        this.charIdx += charsToTake

        if (this.charIdx >= token.value.length) {
          this.tokenIdx++
          this.charIdx = 0
        }
      }
    }

    // Empty segment (can occur when only trailing ANSI codes remain)
    if (this.stringPos === stringStart) {
      return null
    }

    const prefixCodes = reduceCodes(codesStart)
    const suffixCodes = reduceCodes(this.codes)
    this.codes = suffixCodes

    const prefix = ansiCodesToString(prefixCodes)
    const suffix = ansiCodesToString(undoAnsiCodes(suffixCodes))

    return {
      text: prefix + this.text.substring(stringStart, this.stringPos) + suffix,
      start: visibleStart,
    }
  }
}

function reduceCodes(codes: AnsiCode[]): AnsiCode[] {
  return reduceAnsiCodes(codes).filter(c => c.code !== c.endCode)
}