summaryrefslogtreecommitdiff
path: root/crawler/sanitize.go
diff options
context:
space:
mode:
authorhorus_arch2018-02-07 19:09:22 +0100
committerhorus_arch2018-02-07 19:09:22 +0100
commit14a89c34096d2ddb2a7750feda143207110e838b (patch)
treea3edce53bf431977b196fbe4e7fcd1027c11dc41 /crawler/sanitize.go
parent213317c558142f18b213d3d7b46231817138ca20 (diff)
downloadalkobote-14a89c34096d2ddb2a7750feda143207110e838b.tar.gz
Adds sanitizer.
Diffstat (limited to 'crawler/sanitize.go')
-rw-r--r--crawler/sanitize.go30
1 files changed, 30 insertions, 0 deletions
diff --git a/crawler/sanitize.go b/crawler/sanitize.go
new file mode 100644
index 0000000..ddcd4f6
--- /dev/null
+++ b/crawler/sanitize.go
@@ -0,0 +1,30 @@
+package main
+
+import (
+ "log"
+ "regexp"
+ "strings"
+)
+
+func sanitize_offer(angebote []Angebot) []Angebot {
+
+ for _, offer := range angebote {
+ offer.Name = sanitize_name(offer.Name)
+ }
+
+ return angebote
+}
+
+func sanitize_name(name string) string {
+ if strings.Contains(name, "y.o.") {
+ name = strings.Replace(name, "y.o.", "Jahre", 1)
+ }
+ r_liter, err := regexp.Compile("[0-9]+([,.][0-9](([lL])| ([Ll]iter))?")
+ if err != nil {
+ log.Fatal(err)
+ }
+ name_liter := r_liter.FindString(name)
+ name = strings.Replace(name, name_liter, "", 1)
+
+ return name
+}