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/sanitize.go | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-) (limited to 'crawler/sanitize.go') diff --git a/crawler/sanitize.go b/crawler/sanitize.go index fc4ee81..a40745a 100644 --- a/crawler/sanitize.go +++ b/crawler/sanitize.go @@ -92,10 +92,32 @@ func sanitize_name(name string) string { return name } -func sanitize_price_per(price_noisy string) (price int, err error) { +func sanitize_base_price(price_noisy string) (price int, err error) { if strings.Contains(price_noisy, "Preis pro Liter") { price_noisy = strings.Replace(price_noisy, "Preis pro Liter", "", -1) } + if strings.Contains(price_noisy, " pro 1 l") { + price_noisy = strings.Replace(price_noisy, " pro 1 l", "", -1) + } + + if strings.Contains(price_noisy, " pro 1 stück") { + price_noisy = strings.Replace(price_noisy, " pro 1 stück", "", -1) + } + + if strings.Contains(price_noisy, " pro 1 Stück") { + price_noisy = strings.Replace(price_noisy, " pro 1 Stück", "", -1) + } + + if strings.Contains(price_noisy, "Grundpreis:") { + price_noisy = strings.Replace(price_noisy, "Grundpreis", "", -1) + price_noisy = strings.TrimSpace(price_noisy) + } + + if strings.Contains(price_noisy, "/Liter") { + price_noisy = strings.Replace(price_noisy, "/Liter", "", -1) + price_noisy = strings.TrimSpace(price_noisy) + } + return convert_price(price_noisy) } -- cgit v1.2.3