price scrapper

internal/scrapper/service.go

@@ -0,0 +1,167 @@
+package scrapper
+
+import (
+	"context"
+	"github.com/chromedp/chromedp"
+	log "github.com/sirupsen/logrus"
+	"regexp"
+	"scrapper-mandarake/internal/common"
+	"slices"
+	"strconv"
+	"strings"
+)
+
+func (s *Scrapper) getPrice(ctx context.Context, task common.Task) (int32, error) {
+	var (
+		singlePrice string
+		rangedPrice string
+		prices      []int32
+	)
+
+	if err := chromedp.Run(ctx,
+		chromedp.Navigate(task.Link),
+		chromedp.WaitReady("body"),
+		chromedp.Text(`div.price`, &singlePrice, chromedp.ByQuery, chromedp.AtLeast(0)),
+		chromedp.Text(`div.price_range`, &rangedPrice, chromedp.ByQuery, chromedp.AtLeast(0)),
+	); err != nil {
+		log.WithError(err).Error(pkgLogHeader + logGetPrice + "failed to get single price tag")
+		return zeroPrice, err
+	}
+	singlePrice = strings.TrimSpace(singlePrice)
+	prices = append(prices, s.getSinglePriceWithTax(singlePrice))
+
+	rangedPrice = strings.TrimSpace(rangedPrice)
+	if rangedPrice != "" {
+		prices = append(prices, s.getMinimalPriceFromRangeWithTax(rangedPrice))
+	}
+
+	minimal := slices.Min(prices)
+	log.Infof(pkgLogHeader+"uuid: %s, price: %d", task.MerchUuid, minimal)
+
+	return minimal, nil
+}
+
+func (s *Scrapper) getMinimalPrice(sessionCtx context.Context, task common.Task) int32 {
+	var (
+		singlePrice string
+		rangedPrice string
+	)
+
+	if err := chromedp.Run(sessionCtx,
+		chromedp.Navigate(task.Link),
+		chromedp.WaitVisible("body", chromedp.ByQuery),
+		chromedp.Evaluate(`(document.querySelector('div.price')?.innerText || '').trim()`, &singlePrice),
+		chromedp.Evaluate(`(document.querySelector('div.price_range')?.innerText || '').trim()`, &rangedPrice),
+	); err != nil {
+		return zeroPrice
+	}
+
+	minimal := s.processPrices(singlePrice, rangedPrice)
+	log.Infof(pkgLogHeader+"uuid: %s, price: %d", task.MerchUuid, minimal)
+	return minimal
+}
+
+func (s *Scrapper) processPrices(singlePrice, rangedPrice string) int32 {
+	var prices []int32
+
+	//in case of any errors or no price return zeroPrice const
+	//if success add to prices slice
+	if singlePrice != "" {
+		singlePrice = strings.TrimSpace(singlePrice)
+		counted, err := s.parseSinglePrice(singlePrice)
+		if err != nil {
+			log.WithFields(log.Fields{
+				"err":         err.Error(),
+				"singlePrice": singlePrice,
+			}).Error(pkgLogHeader + logGetPrice + "failed to parse single price, returning zero price")
+			return zeroPrice
+		}
+		prices = append(prices, counted)
+	} else {
+		log.Warn(pkgLogHeader + logGetPrice + "single price not found")
+		return zeroPrice
+	}
+
+	//optional, adds price only if no errors and has non zero value
+	if rangedPrice != "" {
+		rangedPrice = strings.TrimSpace(rangedPrice)
+		counted, err := s.parseRangedPrice(rangedPrice)
+		if err != nil {
+			log.WithFields(log.Fields{
+				"err":         err.Error(),
+				"rangedPrice": rangedPrice,
+			}).Error(pkgLogHeader + logGetPrice + "failed to parse ranged price")
+		} else {
+			if counted > 0 {
+				prices = append(prices, counted)
+			}
+		}
+	}
+	return slices.Min(prices)
+}
+
+func (s *Scrapper) getSinglePriceWithTax(rawPrice string) int32 {
+	re := regexp.MustCompile(`(\d+)\s*円`)
+	matches := re.FindStringSubmatch(rawPrice)
+	if len(matches) < 2 {
+		log.Error("Mandarake | Single price not found, returning zero price")
+		return zeroPrice
+	}
+
+	priceStr := matches[1]
+	price, err := strconv.Atoi(priceStr)
+	if err != nil {
+		log.Error("Mandarake | Failed to convert single price, returning zero price")
+		return zeroPrice
+	}
+	return int32(price)
+}
+
+func (s *Scrapper) getMinimalPriceFromRangeWithTax(priceRange string) int32 {
+	re := regexp.MustCompile(`他([\d,]+)円`)
+	matches := re.FindStringSubmatch(priceRange)
+	if len(matches) < 2 {
+		log.Error("Price not found in range, returning zero price")
+		return zeroPrice
+	}
+
+	priceStr := strings.ReplaceAll(matches[1], ",", "")
+	price, err := strconv.Atoi(priceStr)
+	if err != nil {
+		log.Error("Failed to convert minimal price in range, returning zero price")
+		return zeroPrice
+	}
+
+	return int32(float64(price) * taxMultiplier)
+}
+
+func (s *Scrapper) parseSinglePrice(rawPrice string) (int32, error) {
+	deCommaStr := strings.ReplaceAll(rawPrice, ",", "")
+	split := strings.Split(deCommaStr, "円")
+	finalPrice, err := s.countTax(split[0])
+	if err != nil {
+		return zeroPrice, err
+	}
+	return finalPrice, nil
+}
+
+func (s *Scrapper) parseRangedPrice(rawPrice string) (int32, error) {
+	deCommaStr := strings.ReplaceAll(rawPrice, ",", "")
+	split := strings.Split(deCommaStr, "円")
+	rm1 := strings.ReplaceAll(split[0], "(", "")
+	rm2 := strings.ReplaceAll(rm1, "他", "")
+
+	finalPrice, err := s.countTax(rm2)
+	if err != nil {
+		return zeroPrice, err
+	}
+	return finalPrice, nil
+}
+
+func (s *Scrapper) countTax(priceStr string) (int32, error) {
+	intPrice, err := strconv.Atoi(priceStr)
+	if err != nil {
+		return zeroPrice, err
+	}
+	return int32(float64(intPrice) * taxMultiplier), nil
+}