Momento/memento-note/lib/chart/parser.ts

/**
 * Fast chart data parser - NO AI needed
 * Extracts structured data from text using regex patterns
 */

export interface ChartDataPoint {
  label: string
  value: number
}

export interface ParsedChartData {
  hasData: boolean
  data: ChartDataPoint[]
  suggestedType: 'bar' | 'horizontal-bar' | 'line' | 'area' | 'pie'
  confidence: number
}

/**
 * Parse text for chart data using multiple regex patterns
 * Runs in < 50ms - NO AI calls
 */
export function parseChartData(text: string): ParsedChartData {
  const patterns = [
    // Pattern 1: "Label : Value | Label2 : Value2"
    {
      regex: /([^|:|\n]+?)\s*[:|=]\s*(\d+(?:[.,]\d+)?)(?:\s*\||\n|,|$)/gi,
      priority: 1,
    },
    // Pattern 2: "Label: Value, Label2: Value2"
    {
      regex: /([^,:;\n]+?)\s*[:=]\s*(\d+(?:[.,]\d+)?)(?:\s*,|;|\n|$)/gi,
      priority: 2,
    },
    // Pattern 3: "Label - Value" or "Label — Value"
    {
      regex: /([^-—\n]+?)\s*[—-]\s*(\d+(?:[.,]\d+)?)(?:\s*,|\n|$)/gi,
      priority: 3,
    },
    // Pattern 4: Lists with percentages "Label • 45%"
    {
      regex: /([^•%\n]+?)\s*[•·]\s*(\d+)(?:\s*%)?(?:\s*,|\n|$)/gi,
      priority: 4,
    },
  ]

  const seen = new Set<string>()
  let bestMatch: { data: ChartDataPoint[], score: number, type: ParsedChartData['suggestedType'] } | null = null

  for (const pattern of patterns) {
    const matches: ChartDataPoint[] = []
    let match: RegExpExecArray | null

    // Reset regex state
    pattern.regex.lastIndex = 0

    while ((match = pattern.regex.exec(text)) !== null) {
      const label = match[1].trim()
      const valueStr = match[2].trim().replace(',', '.')
      const value = parseFloat(valueStr)

      if (!isNaN(value) && value > 0 && label && !seen.has(label)) {
        matches.push({ label, value })
        seen.add(label)
      }
    }

    if (matches.length >= 2) {
      const score = matches.length * pattern.priority
      if (!bestMatch || score > bestMatch.score) {
        bestMatch = {
          data: matches,
          score,
          type: suggestChartType(matches, text),
        }
      }
    }
  }

  if (!bestMatch) {
    return { hasData: false, data: [], suggestedType: 'bar', confidence: 0 }
  }

  return {
    hasData: true,
    data: bestMatch.data,
    suggestedType: bestMatch.type,
    confidence: Math.min(bestMatch.score / 10, 1),
  }
}

/**
 * Suggest chart type based on data and text context
 */
function suggestChartType(
  data: ChartDataPoint[],
  text: string
): ParsedChartData['suggestedType'] {
  const textLower = text.toLowerCase()

  // Time series indicators
  const timeKeywords = ['jan', 'feb', 'mar', 'apr', 'may', 'jun', 'jul', 'aug', 'sep', 'oct', 'nov', 'dec',
    'janvier', 'février', 'mars', 'avril', 'mai', 'juin', 'juillet', 'août', 'septembre', 'octobre', 'novembre', 'décembre',
    'q1', 'q2', 'q3', 'q4', 'month', 'mois', 'week', 'semaine', 'year', 'année', 'trend', 'tendance', 'growth', 'croissance']

  const hasTimeLabels = data.some(d => timeKeywords.some(k => d.label.toLowerCase().includes(k)))
  const hasTimeInText = timeKeywords.some(k => textLower.includes(k))

  // Proportion indicators
  const proportionKeywords = ['%', 'percent', 'part', 'share', 'répartition', 'distribution', 'segment', 'total']
  const hasProportion = proportionKeywords.some(k => textLower.includes(k))

  // Check if values sum to ~100 (percentage data)
  const sum = data.reduce((a, b) => a + b.value, 0)
  const isPercentage = Math.abs(sum - 100) < 10 || data.some(d => d.value > 100)

  // Decision tree
  if (hasProportion || isPercentage) {
    return data.length <= 5 ? 'pie' : 'bar'
  }

  if (hasTimeLabels || hasTimeInText) {
    return 'line'
  }

  // Long labels favor horizontal
  if (data.some(d => d.label.length > 15)) {
    return 'horizontal-bar'
  }

  return 'bar'
}

/**
 * Generate chart suggestions from parsed data
 * NO AI - instant generation
 */
export function generateChartSuggestions(data: ChartDataPoint[]): Array<{
  type: 'bar' | 'horizontal-bar' | 'line' | 'area' | 'pie'
  title: string
  data: ChartDataPoint[]
  description: string
}> {
  const suggestions: ReturnType<typeof generateChartSuggestions> = []

  // Always include bar chart as fallback
  suggestions.push({
    type: 'bar',
    title: 'Comparaison',
    data,
    description: 'Graphique à barres pour comparer les valeurs',
  })

  // Add line if we have enough data points
  if (data.length >= 3) {
    suggestions.push({
      type: 'line',
      title: 'Tendance',
      data,
      description: 'Graphique linéaire pour visualiser la progression',
    })
  }

  // Add area for magnitude
  if (data.length >= 3) {
    suggestions.push({
      type: 'area',
      title: 'Évolution',
      data,
      description: 'Graphique en aires pour souligner le volume',
    })
  }

  // Add pie if data looks proportional
  const sum = data.reduce((a, b) => a + b.value, 0)
  if (data.length >= 2 && data.length <= 6 && (sum > 90 || data.some(d => d.label.includes('%')))) {
    suggestions.push({
      type: 'pie',
      title: 'Répartition',
      data,
      description: 'Graphique circulaire pour montrer les proportions',
    })
  }

  // Add horizontal-bar for long labels
  if (data.some(d => d.label.length > 12)) {
    suggestions.push({
      type: 'horizontal-bar',
      title: 'Comparaison',
      data,
      description: 'Barres horizontales pour les labels longs',
    })
  }

  // Return max 3 suggestions
  return suggestions.slice(0, 3)
}