From a25368ce25e3de3add81e4347639a9b0401750a7 Mon Sep 17 00:00:00 2001 From: horus Date: Sat, 16 Jun 2018 13:52:50 +0200 Subject: Improves sanitizing function. (crawler) --- crawler/sanitize.go | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'crawler/sanitize.go') diff --git a/crawler/sanitize.go b/crawler/sanitize.go index 960d5f6..4a5197b 100644 --- a/crawler/sanitize.go +++ b/crawler/sanitize.go @@ -82,6 +82,10 @@ func sanitize_name(name string) string { name = strings.Replace(name, "years old", "Jahre", 1) } + if strings.Contains(name, "years") { + name = strings.Replace(name, "years", "Jahre", 1) + } + if strings.Contains(name, "Years Old") { name = strings.Replace(name, "Years Old", "Jahre", 1) } @@ -201,13 +205,15 @@ func sanitize_base_price(price_noisy string) (price int, err error) { 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) } + if strings.Contains(price_noisy, "/L") { + price_noisy = strings.Replace(price_noisy, "/L", "", -1) + } + price_noisy = strings.TrimSpace(price_noisy) return convert_price(price_noisy) } -- cgit v1.2.3