Merge branch 'bnewbold-skate-tweaks' into 'master'

proposed changes and fixes to skate matching

See merge request martin/cgraph!3

1 files changed, 4 insertions, 33 deletions

diff --git a/skate/unstructured.go b/skate/unstructured.go
index f2c1d21..a172e8b 100644
--- a/skate/unstructured.go
+++ b/skate/unstructured.go
@@ -2,19 +2,12 @@ package skate
 
 import (
 	"regexp"
-	"strings"
 )
 
 var (
-	PatDOI          = regexp.MustCompile(`10[.][0-9]{1,8}/[^ ]*[\w]`)
-	PatDOINoHyphen  = regexp.MustCompile(`10[.][0-9]{1,8}/[^ -]*[\w]`)
-	PatArxiv        = regexp.MustCompile(`https?://arxiv.org/(pdf|abs)/([0-9]{4,4}[.][0-9]{1,8})(v[0-9]{1,2})?(.pdf)?`)
-	DOILinkPrefixes = []string{
-		"http://doi.org/",
-		"http://dx.doi.org/",
-		"https://doi.org/",
-		"https://dx.doi.org/",
-	}
+	PatDOI         = regexp.MustCompile(`10[.][0-9]{1,8}/[^ ]*[\w]`)
+	PatDOINoHyphen = regexp.MustCompile(`10[.][0-9]{1,8}/[^ -]*[\w]`)
+	PatArxiv       = regexp.MustCompile(`https?://arxiv.org/(pdf|abs)/([0-9]{4,4}[.][0-9]{1,8})(v[0-9]{1,2})?(.pdf)?`)
 )
 
 // ParseUnstructured will in-place augment missing DOI, arxiv id and so on.
@@ -24,32 +17,10 @@ func ParseUnstructured(ref *Ref) error {
 		v   string
 		vs  []string
 	)
-	// Handle things like: 10.1111/j.1550-7408.1968.tb02138.x-BIB5|cit5,
-	// 10.1111/j.1558-5646.1997.tb02431.x-BIB0008|evo02431-cit-0008, ...
-	if strings.Contains(strings.ToLower(ref.Key), "-bib") && ref.Biblio.DOI == "" {
-		parts := strings.Split(strings.ToLower(ref.Key), "-bib")
-		ref.Biblio.DOI = parts[0]
-	}
 	// DOI
 	v = PatDOI.FindString(uns)
 	if v != "" && ref.Biblio.DOI == "" {
-		ref.Biblio.DOI = v
-	}
-	// DOI in Key
-	v = PatDOINoHyphen.FindString(ref.Key)
-	if v != "" && ref.Biblio.DOI == "" {
-		ref.Biblio.DOI = v
-	}
-	// DOI in URL
-	for _, prefix := range DOILinkPrefixes {
-		if ref.Biblio.DOI != "" && strings.HasPrefix(ref.Biblio.Url, prefix) {
-			ref.Biblio.DOI = strings.Replace(ref.Biblio.Url, prefix, "", -1)
-		}
-	}
-	// Another DOI pattern.
-	v = PatDOINoHyphen.FindString(ref.Key)
-	if v != "" && ref.Biblio.DOI == "" {
-		ref.Biblio.DOI = v
+		ref.Biblio.DOI = SanitizeDOI(v)
 	}
 	// Arxiv
 	vs = PatArxiv.FindStringSubmatch(uns)