Structures the crawler for bottleworld and mcwhisky.

4 files changed, 73 insertions, 27 deletions

diff --git a/bottleworld.go b/bottleworld.go
index edba866..55ee47c 100644
--- a/bottleworld.go
+++ b/bottleworld.go
@@ -1,7 +1,6 @@
 package main
 
 import (
-	"fmt"
 	"log"
 	"regexp"
 	//	"strings"
@@ -10,13 +9,17 @@ import (
 	"github.com/gocolly/colly"
 )
 
-func ScrapeBottleWord() {
+func ScrapeBottleWord() []Angebot {
+	Whiskys := []Angebot{}
+
 	c := colly.NewCollector(
 		colly.AllowedDomains("bottleworld.de"),
 		colly.AllowedDomains("www.bottleworld.de"),
 	)
 
 	c.OnHTML("li.item", func(e *colly.HTMLElement) {
+		W := Angebot{}
+
 		whisky_name := e.ChildText("h2 > a")
 
 		matched, err := regexp.MatchString("Whiske?y", whisky_name)
@@ -24,25 +27,39 @@ func ScrapeBottleWord() {
 			log.Fatal(err)
 		}
 		if !matched {
+			//W.Spirit_type = "Anderes"
 			return
+		} else {
+			W.Spirit_type = "Whisky"
 		}
 
 		whisky_url := e.ChildAttr("a", "href")
-		log.Println(whisky_name)
-		log.Println(whisky_url)
+		W.Name = whisky_name
+		W.Url = whisky_url
 
 		e.ForEach(".price-box", func(i int, e *colly.HTMLElement) {
 			e.ForEach(".old-price", func(i int, e *colly.HTMLElement) {
-				log.Println(e.ChildText(".price"))
+				W.Original_price, err = sanitize_price(e.ChildText(".price"))
+				if err != nil {
+					log.Fatal(err)
+				}
 			})
 			e.ForEach(".special-price", func(i int, e *colly.HTMLElement) {
-				log.Println(e.ChildText(".price"))
+				W.Discounted_price, err = sanitize_price(e.ChildText(".price"))
+				if err != nil {
+					log.Fatal(err)
+				}
 			})
 		})
-		log.Println(e.ChildAttr("img", "src"))
+		W.Image_url = e.ChildAttr("img", "src")
 
-		fmt.Println("")
+		W.Shop = "bottleworld.de"
+		W.Spirit_type = "Whisky"
+
+		Whiskys = append(Whiskys, W)
 	})
 
 	c.Visit("https://www.bottleworld.de/aktuelle-sonderpreise/show/all")
+
+	return Whiskys
 }
diff --git a/main.go b/main.go
index 3a8efa3..f78a777 100644
--- a/main.go
+++ b/main.go
@@ -19,19 +19,19 @@ type Angebot struct {
 
 func main() {
 
-	/*
-		printName("BottleWorld")
-		ScrapeBottleWord()
+	W := ScrapeBottleWord()
+	printName(W, "BottleWorld")
 
-		printName("MC Whisky")
-		ScrapeMCWhisky()
+	W = ScrapeMCWhisky()
+	printName(W, "MC Whisky")
 
-		printName("Whic")
-		ScrapeWhic()
-	*/
+	/*
+			printName("Whic")
+			ScrapeWhic()
 
-	W := ScrapeWhiskyde()
-	printName(W, "Whisky.de")
+		W = ScrapeWhiskyde()
+		printName(W, "Whisky.de")
+	*/
 
 	/*
 		printName("Whiskysite.nl")
diff --git a/mcwhisky.go b/mcwhisky.go
index 142d65a..b50b4e5 100644
--- a/mcwhisky.go
+++ b/mcwhisky.go
@@ -1,7 +1,6 @@
 package main
 
 import (
-	"fmt"
 	"log"
 	//	"strings"
 
@@ -9,30 +8,47 @@ import (
 	"github.com/gocolly/colly"
 )
 
-func ScrapeMCWhisky() {
+func ScrapeMCWhisky() []Angebot {
+	Whiskys := []Angebot{}
+
 	c := colly.NewCollector(
 		colly.AllowedDomains("mcwhisky.com"),
 		colly.AllowedDomains("www.mcwhisky.com"),
 	)
 
 	c.OnHTML("li.item", func(e *colly.HTMLElement) {
-		whisky_name := e.ChildAttr("a", "title")
+		W := Angebot{}
+
+		whisky_name := sanitize_name(e.ChildAttr("a", "title"))
 		whisky_url := e.ChildAttr("a", "href")
-		log.Println(whisky_name)
-		log.Println(whisky_url)
+		W.Name = whisky_name
+		W.Url = whisky_url
+
+		var err error
 
 		e.ForEach(".price-box", func(i int, e *colly.HTMLElement) {
 			e.ForEach(".old-price", func(i int, e *colly.HTMLElement) {
-				log.Println(e.ChildText(".price"))
+				W.Original_price, err = sanitize_price(e.ChildText(".price"))
+				if err != nil {
+					log.Fatal(err)
+				}
 			})
 			e.ForEach(".special-price", func(i int, e *colly.HTMLElement) {
-				log.Println(e.ChildText(".price"))
+				W.Discounted_price, err = sanitize_price(e.ChildText(".price"))
+				if err != nil {
+					log.Fatal(err)
+				}
 			})
 		})
-		log.Println(e.ChildAttr("img", "src"))
+		W.Image_url = e.ChildAttr("img", "src")
 
-		fmt.Println("")
+		W.Shop = "MC Whisky"
+		W.Spirit_type = "Whisky"
+
+		Whiskys = append(Whiskys, W)
 	})
 
 	c.Visit("https://www.mcwhisky.com/whisky/whisky-sonderangebote.html")
+
+	return Whiskys
 }
diff --git a/sanitize_name.go b/sanitize_name.go
new file mode 100644
index 0000000..73b2714
--- /dev/null
+++ b/sanitize_name.go
@@ -0,0 +1,13 @@
+package main
+
+import (
+	"strings"
+)
+
+func sanitize_name(name string) string {
+	if strings.Contains(name, "y.o.") {
+		name = strings.Replace(name, "y.o.", "Jahre", 1)
+	}
+
+	return name
+}