From f6904aab20e2d09255fd0adabfd246165ff3cb02 Mon Sep 17 00:00:00 2001
From: Max
Date: Thu, 8 Feb 2018 18:26:41 +0100
Subject:  Crawler extracts volume, price per litre and abv. (MC Whisky, Rum &
 Co, Whic)

---
 crawler/utility.go | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)

(limited to 'crawler/utility.go')

diff --git a/crawler/utility.go b/crawler/utility.go
index a794c4b..9de7845 100644
--- a/crawler/utility.go
+++ b/crawler/utility.go
@@ -5,6 +5,8 @@ import (
 	"regexp"
 	"strconv"
 	"strings"
+
+	"github.com/gocolly/colly"
 )
 
 func detect_spirit_type(name string) string {
@@ -47,6 +49,7 @@ func extract_volume(volume string) (float32, error) {
 	if err != nil {
 		return 0, err
 	}
+
 	return float32(volume64), err
 }
 
@@ -61,5 +64,51 @@ func extract_abv(abv_noisy string) (float32, error) {
 	if err != nil {
 		return 0, err
 	}
+
 	return float32(abv64), nil
 }
+
+/*
+ * In litre, but float.
+ */
+func get_volume(e *colly.HTMLElement) float32 {
+
+	volume_noisy := e.Request.Ctx.Get("volume")
+
+	matched, err := regexp.MatchString(`[lL](iter)?`, volume_noisy)
+	if err != nil {
+		log.Fatal(err)
+	}
+	if !matched {
+		log.Println("get_volume: not matched: " + volume_noisy)
+		return 0
+	}
+
+	volume, err := extract_volume(volume_noisy)
+	if err != nil {
+		log.Println("get_volume: " + volume_noisy)
+		log.Fatal(err)
+	}
+
+	return volume
+}
+
+/*
+ * In procent. (float)
+ */
+func get_abv(e *colly.HTMLElement) float32 {
+
+	abv_noisy := e.Request.Ctx.Get("abv")
+
+	if abv_noisy == "" {
+		return 0
+	}
+
+	abv, err := extract_abv(abv_noisy)
+	if err != nil {
+		log.Println("get_abv: " + abv_noisy)
+		log.Fatal(err)
+	}
+
+	return abv
+}
-- 
cgit v1.2.3