From ff88caff491350c6acac6f9e3d139506a9e155ef Mon Sep 17 00:00:00 2001
From: Joellensilva <joellen.absilva@gmail.com>
Date: Tue, 27 Feb 2024 15:52:49 -0300
Subject: [PATCH] atualizando url e adicionando user-agent

---
 crawler.go | 16 +++++++++-------
 main.go    | 14 ++++++++++++--
 2 files changed, 21 insertions(+), 9 deletions(-)

diff --git a/crawler.go b/crawler.go
index 7985d42..8570357 100644
--- a/crawler.go
+++ b/crawler.go
@@ -16,6 +16,7 @@ import (
 type crawler struct {
 	collectionTimeout time.Duration
 	timeBetweenSteps  time.Duration
+	downloadTimeout   time.Duration
 	year              string
 	month             string
 	output            string
@@ -34,6 +35,7 @@ func (c crawler) crawl() ([]string, error) {
 			chromedp.Flag("headless", true), // mude para false para executar com navegador visível.
 			chromedp.NoSandbox,
 			chromedp.DisableGPU,
+			chromedp.UserAgent("Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/121.0.0.0 Mobile Safari/537.36"),
 		)...,
 	)
 	defer allocCancel()
@@ -108,11 +110,11 @@ func (c crawler) abreCaixaDialogo(ctx context.Context, tipo string) error {
 	var baseURL string
 	var selectYear string
 	if tipo == "contra" {
-		baseURL = "http://www.mpap.mp.br/transparencia/index.php?pg=consulta_folha_membros_ativos"
-		selectYear = `//select[@id="ano"]`
+		baseURL = "https://portal.mpap.mp.br/transparencia/index.php?pg=consulta_folha_membros_ativos"
+		selectYear = `//*[@id="ano"]`
 	} else {
-		baseURL = "https://www.mpap.mp.br/transparencia/index.php?pg=consulta_verbas_indenizatorias"
-		selectYear = `//select[@id="ano_verbas"]`
+		baseURL = "https://portal.mpap.mp.br/transparencia/index.php?pg=consulta_verbas_indenizatorias"
+		selectYear = `//*[@id="ano_verbas"]`
 	}
 
 	return chromedp.Run(ctx,
@@ -124,7 +126,7 @@ func (c crawler) abreCaixaDialogo(ctx context.Context, tipo string) error {
 		chromedp.Sleep(c.timeBetweenSteps),
 
 		// Seleciona mes
-		chromedp.SetValue(`//select[@id="mes"]`, strings.TrimPrefix(c.month, "0"), chromedp.BySearch, chromedp.NodeVisible),
+		chromedp.SetValue(`//*[@id="mes"]`, strings.TrimPrefix(c.month, "0"), chromedp.BySearch, chromedp.NodeVisible),
 		chromedp.Sleep(c.timeBetweenSteps),
 
 		// Busca
@@ -144,13 +146,13 @@ func (c crawler) exportaPlanilha(ctx context.Context, fName string) error {
 		chromedp.Run(ctx,
 			// Clica no botão de download
 			chromedp.Click(`/html/body/div[1]/center/div/fieldset/div[3]/form/button[2]`, chromedp.BySearch, chromedp.NodeVisible),
-			chromedp.Sleep(c.timeBetweenSteps),
+			chromedp.Sleep(c.downloadTimeout),
 		)
 	} else {
 		chromedp.Run(ctx,
 			// Clica no botão de download
 			chromedp.Click(`/html/body/div[1]/center/div/fieldset/div/form/button[2]`, chromedp.BySearch, chromedp.NodeVisible),
-			chromedp.Sleep(c.timeBetweenSteps),
+			chromedp.Sleep(c.downloadTimeout),
 		)
 	}
 
diff --git a/main.go b/main.go
index 3f9aa07..c3afd35 100644
--- a/main.go
+++ b/main.go
@@ -11,7 +11,8 @@ import (
 
 const (
 	defaultGeneralTimeout      = 4 * time.Minute  // Duração máxima total da coleta de todos os arquivos. Valor padrão calculado a partir de uma média de execuções ~4.5min
-	defaulTimeBetweenSteps     = 5 * time.Second //Tempo de espera entre passos do coletor."
+	defaulTimeBetweenSteps     = 5 * time.Second  //Tempo de espera entre passos do coletor."
+	defaultFileDownloadTimeout = 20 * time.Second // Duração que o coletor deve esperar até que o download de cada um dos arquivos seja concluído
 )
 
 func main() {
@@ -51,9 +52,18 @@ func main() {
 			log.Fatalf("Invalid TIME_BETWEEN_STEPS (\"%s\"): %q", os.Getenv("TIME_BETWEEN_STEPS"), err)
 		}
 	}
+	downloadTimeout := defaultFileDownloadTimeout
+	if os.Getenv("DOWNLOAD_TIMEOUT") != "" {
+		var err error
+		downloadTimeout, err = time.ParseDuration(os.Getenv("DOWNLOAD_TIMEOUT"))
+		if err != nil {
+			log.Fatalf("Invalid DOWNLOAD_TIMEOUT (\"%s\"): %q", os.Getenv("DOWNLOAD_TIMEOUT"), err)
+		}
+	}
 	c := crawler{
 		collectionTimeout: generalTimeout,
 		timeBetweenSteps:  timeBetweenSteps,
+		downloadTimeout:   downloadTimeout,
 		year:              year,
 		month:             month,
 		output:            outputFolder,
@@ -66,4 +76,4 @@ func main() {
 	// O parser do MPAP espera os arquivos separados por \n. Mudanças aqui tem
 	// refletir as expectativas lá.
 	fmt.Println(strings.Join(downloads, "\n"))
-}
\ No newline at end of file
+}