mev-beta/orig/tests/calculation-validation/parser.go

package calculation_validation

import (
	"bufio"
	"fmt"
	"math/big"
	"os"
	"regexp"
	"strconv"
	"strings"
	"time"
)

// LogParser parses arbitrage opportunity data from log files
type LogParser struct {
	executablePattern  *regexp.Regexp
	profitPattern      *regexp.Regexp
	thresholdPattern   *regexp.Regexp
	v3CalcPattern      *regexp.Regexp
	opportunityPattern *regexp.Regexp
}

// NewLogParser creates a new log parser
func NewLogParser() *LogParser {
	return &LogParser{
		executablePattern:  regexp.MustCompile(`EXECUTABLE OPPORTUNITY: ID=([^,]+), Profit=([0-9.]+) ETH \(threshold=([0-9.]+) ETH\)`),
		profitPattern:      regexp.MustCompile(`Estimated Profit: \$([0-9,.]+) USD`),
		thresholdPattern:   regexp.MustCompile(`Profit threshold check: netProfit=([0-9.]+) ETH, minThreshold=([0-9.]+) ETH`),
		v3CalcPattern:      regexp.MustCompile(`V3 calculation: amountIn=([0-9]+), amountOut=([0-9]+), fee=([0-9]+), finalOut=([0-9]+)`),
		opportunityPattern: regexp.MustCompile(`ARBITRAGE OPPORTUNITY DETECTED`),
	}
}

// ParseExecutableOpportunities parses executable opportunities from log file
func (lp *LogParser) ParseExecutableOpportunities(logFile string) ([]*OpportunityTestData, error) {
	file, err := os.Open(logFile)
	if err != nil {
		return nil, fmt.Errorf("failed to open log file: %w", err)
	}
	defer file.Close()

	var opportunities []*OpportunityTestData
	scanner := bufio.NewScanner(file)

	for scanner.Scan() {
		line := scanner.Text()

		// Parse executable opportunity
		if matches := lp.executablePattern.FindStringSubmatch(line); matches != nil {
			opp := &OpportunityTestData{
				ID:           matches[1],
				Timestamp:    time.Now(), // Would need to parse from log timestamp
				IsExecutable: true,
			}

			// Parse profit
			profit, err := strconv.ParseFloat(matches[2], 64)
			if err == nil {
				opp.NetProfitETH = big.NewFloat(profit)
			}

			// Parse threshold
			threshold, err := strconv.ParseFloat(matches[3], 64)
			if err == nil {
				opp.ThresholdCheck = &ThresholdCheck{
					NetProfit:    big.NewFloat(profit),
					MinThreshold: big.NewFloat(threshold),
					Passed:       profit >= threshold,
				}
			}

			opportunities = append(opportunities, opp)
		}
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("error reading log file: %w", err)
	}

	return opportunities, nil
}

// ParseV3Calculations parses V3 swap calculations from log file
func (lp *LogParser) ParseV3Calculations(logFile string) ([]SwapCalculation, error) {
	file, err := os.Open(logFile)
	if err != nil {
		return nil, fmt.Errorf("failed to open log file: %w", err)
	}
	defer file.Close()

	var calculations []SwapCalculation
	scanner := bufio.NewScanner(file)

	for scanner.Scan() {
		line := scanner.Text()

		if matches := lp.v3CalcPattern.FindStringSubmatch(line); matches != nil {
			amountIn, _ := new(big.Int).SetString(matches[1], 10)
			amountOut, _ := new(big.Int).SetString(matches[2], 10)
			fee, _ := strconv.ParseUint(matches[3], 10, 32)
			finalOut, _ := new(big.Int).SetString(matches[4], 10)

			calculations = append(calculations, SwapCalculation{
				AmountIn:  amountIn,
				AmountOut: amountOut,
				Fee:       uint32(fee),
				FinalOut:  finalOut,
			})
		}
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("error reading log file: %w", err)
	}

	return calculations, nil
}

// ParseThresholdChecks parses profit threshold validation checks
func (lp *LogParser) ParseThresholdChecks(logFile string) ([]*ThresholdCheck, error) {
	file, err := os.Open(logFile)
	if err != nil {
		return nil, fmt.Errorf("failed to open log file: %w", err)
	}
	defer file.Close()

	var checks []*ThresholdCheck
	scanner := bufio.NewScanner(file)

	for scanner.Scan() {
		line := scanner.Text()

		if matches := lp.thresholdPattern.FindStringSubmatch(line); matches != nil {
			netProfit, _ := strconv.ParseFloat(matches[1], 64)
			minThreshold, _ := strconv.ParseFloat(matches[2], 64)

			checks = append(checks, &ThresholdCheck{
				NetProfit:    big.NewFloat(netProfit),
				MinThreshold: big.NewFloat(minThreshold),
				Passed:       netProfit >= minThreshold,
			})
		}
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("error reading log file: %w", err)
	}

	return checks, nil
}

// ExtractProfitValues extracts all profit values from opportunities
func ExtractProfitValues(opportunities []*OpportunityTestData) []*big.Float {
	var profits []*big.Float
	for _, opp := range opportunities {
		if opp.NetProfitETH != nil {
			profits = append(profits, opp.NetProfitETH)
		}
	}
	return profits
}

// CalculateStatistics calculates statistics from profit values
func CalculateStatistics(profits []*big.Float) (total, average, max, min *big.Float) {
	if len(profits) == 0 {
		return big.NewFloat(0), big.NewFloat(0), big.NewFloat(0), big.NewFloat(0)
	}

	total = big.NewFloat(0)
	max = new(big.Float).Set(profits[0])
	min = new(big.Float).Set(profits[0])

	for _, profit := range profits {
		total.Add(total, profit)
		if profit.Cmp(max) > 0 {
			max = new(big.Float).Set(profit)
		}
		if profit.Cmp(min) < 0 {
			min = new(big.Float).Set(profit)
		}
	}

	average = new(big.Float).Quo(total, big.NewFloat(float64(len(profits))))
	return total, average, max, min
}

// ParseOpportunityDetails parses detailed opportunity information from multi-line log entries
func (lp *LogParser) ParseOpportunityDetails(logFile string) ([]*OpportunityTestData, error) {
	file, err := os.Open(logFile)
	if err != nil {
		return nil, fmt.Errorf("failed to open log file: %w", err)
	}
	defer file.Close()

	var opportunities []*OpportunityTestData
	var currentOpp *OpportunityTestData
	scanner := bufio.NewScanner(file)

	for scanner.Scan() {
		line := scanner.Text()

		// Start of new opportunity
		if lp.opportunityPattern.MatchString(line) {
			if currentOpp != nil {
				opportunities = append(opportunities, currentOpp)
			}
			currentOpp = &OpportunityTestData{
				Timestamp: time.Now(),
			}
			continue
		}

		if currentOpp == nil {
			continue
		}

		// Parse various fields
		if strings.Contains(line, "Transaction:") {
			parts := strings.Split(line, ":")
			if len(parts) > 1 {
				currentOpp.TransactionHash = strings.TrimSpace(parts[1])
			}
		}

		if matches := lp.profitPattern.FindStringSubmatch(line); matches != nil {
			profitStr := strings.ReplaceAll(matches[1], ",", "")
			profitUSD, _ := strconv.ParseFloat(profitStr, 64)
			currentOpp.NetProfitUSD = big.NewFloat(profitUSD)
		}

		if strings.Contains(line, "netProfitETH:") {
			parts := strings.Split(line, ":")
			if len(parts) > 1 {
				profitStr := strings.TrimSpace(parts[1])
				profit, _ := strconv.ParseFloat(profitStr, 64)
				currentOpp.NetProfitETH = big.NewFloat(profit)
			}
		}

		if strings.Contains(line, "isExecutable:") {
			currentOpp.IsExecutable = strings.Contains(line, "isExecutable:true")
		}

		if strings.Contains(line, "rejectReason:") {
			parts := strings.SplitN(line, ":", 2)
			if len(parts) > 1 {
				currentOpp.RejectReason = strings.TrimSpace(parts[1])
			}
		}
	}

	// Don't forget the last opportunity
	if currentOpp != nil {
		opportunities = append(opportunities, currentOpp)
	}

	if err := scanner.Err(); err != nil {
		return nil, fmt.Errorf("error reading log file: %w", err)
	}

	return opportunities, nil
}