summaryrefslogtreecommitdiff
path: root/crawler/shop_whiskyzone.go
diff options
context:
space:
mode:
authorMax2018-02-06 00:35:39 +0100
committerMax2018-02-06 00:35:39 +0100
commit71950479fbd6088f249e5fda3b180f294d1d745d (patch)
tree06f360a7e02b7e0011bda815fa102ec54ae8d0ec /crawler/shop_whiskyzone.go
parent13a807854bf4d0258723ec3152b217ed4cf8e051 (diff)
downloadalkobote-71950479fbd6088f249e5fda3b180f294d1d745d.tar.gz
Moves crawler to designated directory.
Diffstat (limited to 'crawler/shop_whiskyzone.go')
-rw-r--r--crawler/shop_whiskyzone.go56
1 files changed, 56 insertions, 0 deletions
diff --git a/crawler/shop_whiskyzone.go b/crawler/shop_whiskyzone.go
new file mode 100644
index 0000000..a9e73d0
--- /dev/null
+++ b/crawler/shop_whiskyzone.go
@@ -0,0 +1,56 @@
+package main
+
+import (
+ "log"
+ "regexp"
+
+ "github.com/gocolly/colly"
+)
+
+func ScrapeWhiskyzone(shop Shop) []Angebot {
+
+ Whiskys := []Angebot{}
+
+ c := colly.NewCollector(
+ colly.AllowedDomains("whiskyzone.de"),
+ colly.AllowedDomains("www.whiskyzone.de"),
+ )
+
+ c.OnHTML(".product--info", func(e *colly.HTMLElement) {
+
+ W := Angebot{}
+
+ W.Name = e.ChildAttr("a", "title")
+ W.Url = e.ChildAttr("a", "href")
+ price_discount_noisy := e.ChildText(".price--default")
+ price_regular_noisy := e.ChildText(".price--discount")
+
+ r, err := regexp.Compile("[0-9]+(,[0-9]{1,2})")
+ if err != nil {
+ log.Fatal(err)
+ }
+ W.Discounted_price, err = sanitize_price(r.FindString(price_discount_noisy))
+ if err != nil {
+ log.Fatal(err)
+ return
+ }
+ W.Original_price, err = sanitize_price(r.FindString(price_regular_noisy))
+ if err != nil {
+ log.Fatal(err)
+ return
+ }
+
+ e.ForEach(".image--media", func(i int, e *colly.HTMLElement) {
+ W.Image_url = e.ChildAttr("img", "src")
+ })
+
+ W.Shop = shop.Id
+ W.Spirit_type = "Whisky"
+
+ Whiskys = append(Whiskys, W)
+ })
+
+ c.Visit("https://www.whiskyzone.de/widgets/emotion/index/emotionId/248/controllerName/listing")
+
+ return Whiskys
+}