From c7105fe21f872295a8e773f353bc47fda2d0c292 Mon Sep 17 00:00:00 2001
From: Max
Date: Sat, 10 Feb 2018 01:52:59 +0100
Subject: Better detection of spirit type. (bottleworld)

---
 crawler/shop_bottleworld.go | 23 +++++++++++++----------
 1 file changed, 13 insertions(+), 10 deletions(-)

(limited to 'crawler/shop_bottleworld.go')

diff --git a/crawler/shop_bottleworld.go b/crawler/shop_bottleworld.go
index fdf1cd8..3eb151d 100644
--- a/crawler/shop_bottleworld.go
+++ b/crawler/shop_bottleworld.go
@@ -10,6 +10,7 @@ import (
 )
 
 func ScrapeBottleWord(shop Shop) []Angebot {
+	Shop_url := "https://www.bottleworld.de/aktuelle-sonderpreise/show/all"
 	Whiskys := []Angebot{}
 
 	c := colly.NewCollector(
@@ -22,15 +23,7 @@ func ScrapeBottleWord(shop Shop) []Angebot {
 
 		whisky_name := e.ChildText("h2 > a")
 
-		matched, err := regexp.MatchString("Whiske?y", whisky_name)
-		if err != nil {
-			log.Fatal(err)
-		}
-		if !matched {
-			W.Spirit_type = detect_spirit_type(whisky_name)
-		} else {
-			W.Spirit_type = "Whisky"
-		}
+		var err error
 
 		whisky_url := e.ChildAttr("a", "href")
 		W.Name = whisky_name
@@ -67,6 +60,8 @@ func ScrapeBottleWord(shop Shop) []Angebot {
 		W.Volume = get_volume(e)
 		W.Abv = get_abv(e)
 
+		W.Spirit_type = e.Request.Ctx.Get("spirit_type")
+
 		Whiskys = append(Whiskys, W)
 	})
 
@@ -85,7 +80,15 @@ func ScrapeBottleWord(shop Shop) []Angebot {
 		})
 	})
 
-	c.Visit("https://www.bottleworld.de/aktuelle-sonderpreise/show/all")
+	c.OnHTML(".short-description", func(e *colly.HTMLElement) {
+		if e.Request.URL.String() == Shop_url {
+			return
+		}
+		text_noisy := e.ChildText(".std")
+		e.Request.Ctx.Put("spirit_type", detect_spirit_type(text_noisy))
+	})
+
+	c.Visit(Shop_url)
 
 	return Whiskys
 }
-- 
cgit v1.2.3