diff options
Diffstat (limited to 'skate/url.go')
| -rw-r--r-- | skate/url.go | 13 |
1 files changed, 5 insertions, 8 deletions
diff --git a/skate/url.go b/skate/url.go index 0f7ad7f..af4ac0c 100644 --- a/skate/url.go +++ b/skate/url.go @@ -19,24 +19,23 @@ var ( // http://agingmind.utdallas.edu/facedb/view/neutral-faces.doi:10.3758/BF03206543. // http://10.3386/w20634https://doi.org/10.3386/w20634 func SanitizeURL(s string) string { - index := suffixarray.New([]byte(s)) - indices := index.Lookup([]byte("http"), -1) + var ( + index = suffixarray.New([]byte(s)) + indices = index.Lookup([]byte("http"), -1) + ) if len(indices) > 1 { s = s[0:indices[1]] // only use the first s = strings.TrimRight(s, ":") + s = strings.TrimRight(s, ";") } - // http://!!!: // http://! // http://" s = patNonWordDomain.ReplaceAllString(s, `$1$3`) - // http:///en.m.wikipedia.org/ChenLong s = patRepeatedHttpSlashes.ReplaceAllString(s, `$1$3`) - // http://10.1113/jphysiol.2002.026047 s = patHttpDOI.ReplaceAllString(s, `https://doi.org/$2`) - // .diaksestanggal27-03-2017.10.30Wib // accessedon15 // .Accessed @@ -45,11 +44,9 @@ func SanitizeURL(s string) string { if patAccessedOn.MatchString(s) { s = patAccessedOn.ReplaceAllString(s, `$1`) } - // http://140.120.197.173/Ecology/Download/Timing-MSChart.zipJournalofInsectScience if patFileExtraSuffix.MatchString(s) { s = patFileExtraSuffix.ReplaceAllString(s, `$1`) } - return s } |