143 lines
3.6 KiB
Go

package crawler
import (
"context"
"fmt"
"log/slog"
"os"
"time"
"github.com/maximotejeda/us_dop_scrapper/helpers"
"github.com/maximotejeda/us_dop_scrapper/internal/application/core/domain"
"github.com/maximotejeda/us_dop_scrapper/internal/ports"
"github.com/playwright-community/playwright-go"
)
type scotia struct {
client ports.DollarPort
}
func NewScotia(client ports.DollarPort) ports.APIPorts {
return &scotia{
client: client,
}
}
func (sct scotia) Scrape(ctx context.Context, page playwright.Page, log *slog.Logger) (insts []*domain.History, err error) {
tout := 120000.00
uri := os.Getenv("SCOTIA")
log = log.With("scrapper", "scotia")
if _, err := page.Goto(uri, playwright.PageGotoOptions{
Timeout: &tout,
WaitUntil: playwright.WaitUntilStateLoad,
}); err != nil {
log.Error("could not get info", "error", err)
return nil, err
}
currencyTable := page.Locator(".bns--table")
currencyTable.WaitFor()
firstRow := page.Locator(".bns--table > tbody:nth-child(1) > tr:nth-child(2)")
secondRow := page.Locator(".bns--table > tbody:nth-child(1) > tr:nth-child(3)")
// the same institution has 2 prices for dollar sell
// first row is onsite selling dollars
// second row is online selling dollars
buyOnsite := firstRow.Locator("td:nth-child(3)")
sellOnsite := firstRow.Locator("td:nth-child(4)")
// the first row has 4 elements
// but the second row only has 3
buyOnline := secondRow.Locator("td:nth-child(2)")
sellOnline := secondRow.Locator("td:nth-child(3)")
compraOnsiteSTR, err := buyOnsite.InnerText()
if err != nil {
log.Error("could not get compra str", "err", err)
return nil, err
}
ventaOnsiteSTR, err := sellOnsite.InnerText()
if err != nil {
log.Error("could not get venta string", "err", err)
return nil, err
}
instOnsite := &domain.History{
Name: "scotiabank",
Parser: "scotia",
Parsed: time.Now().Unix(),
}
instOnsite.Venta = helpers.Normalize(ventaOnsiteSTR)
instOnsite.Compra = helpers.Normalize(compraOnsiteSTR)
if instOnsite.Compra == 0 || instOnsite.Venta == 0 {
return nil, fmt.Errorf("scotia: institution not parsed: %v", instOnsite)
}
compraOnlineSTR, err := buyOnline.InnerText()
if err != nil {
log.Error("could not get compra onlie str", "err", err)
return nil, err
}
ventaOnlineSTR, err := sellOnline.InnerText()
if err != nil {
log.Error("could not get venta online string", "err", err)
return nil, err
}
instOnline := &domain.History{
Name: "scotiabank cambio online",
Parser: "scotia",
Parsed: time.Now().Unix(),
}
instOnline.Venta = helpers.Normalize(ventaOnlineSTR)
instOnline.Compra = helpers.Normalize(compraOnlineSTR)
if instOnline.Compra == 0 || instOnline.Venta == 0 {
return nil, fmt.Errorf("scotia: institution not parsed: %v", instOnline)
}
insts = append(insts, instOnline, instOnsite)
return insts, nil
}
func (sct scotia) ExecParser(
ctx context.Context,
browser *playwright.Browser,
log *slog.Logger) (err error) {
t := true
ua := helpers.NewMobileUA()
b := *browser
page, err := b.NewPage(playwright.BrowserNewPageOptions{
UserAgent: &ua,
// IsMobile: &t,
HasTouch: &t,
Viewport: &playwright.Size{
Width: 412,
Height: 915,
},
Screen: &playwright.Size{
Width: 412,
Height: 915,
},
})
if err != nil {
log.Error("creating page", "error", err)
os.Exit(1)
}
ctx, cancel := context.WithTimeout(ctx, 6*time.Minute)
defer page.Close()
defer cancel()
insts, err := sct.Scrape(ctx, page, log)
// here we execute db operations
if err != nil {
return err
}
for _, inst := range insts {
sct.client.NewHistory(inst)
}
return err
}