parse out isbn from unstructured

2 files changed, 22 insertions, 0 deletions

diff --git a/skate/unstructured.go b/skate/unstructured.go
index 85ef0fe..7fda9d7 100644
--- a/skate/unstructured.go
+++ b/skate/unstructured.go
@@ -47,6 +47,7 @@ func ParseUnstructured(ref *Ref) error {
 			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
@@ -61,5 +62,15 @@ func ParseUnstructured(ref *Ref) error {
 			ref.Biblio.ArxivId = vs[1]
 		}
 	}
+	// XXX: ISBN
+	ref.Biblio.Extra.ISBN = ParseIsbn(uns)
+	// Some more examples:
+	// - 2005 Sep;95(9):1545-51
+	// - Pattern Anal. Mach. Intell., 36(2):346-360, 2014. 4, 6, 7"
+	//
+	// There seems to be a number tiny of data issues; e.g. one off dates. Can
+	// we fix them here?
+	// Include: https://images.webofknowledge.com/images/help/WOS/J_abrvjt.html, ...
+
 	return nil
 }
diff --git a/skate/unstructured_test.go b/skate/unstructured_test.go
index 41ff471..92f1d80 100644
--- a/skate/unstructured_test.go
+++ b/skate/unstructured_test.go
@@ -52,3 +52,14 @@ func TestParseUnstructured(t *testing.T) {
 		}
 	}
 }
+
+func BenchmarkParseUnstructured(b *testing.B) {
+	ref := Ref{
+		Biblio: Biblio{
+			Unstructured: "https://arxiv.org/pdf/0808.3320v3.pdf Hello 10.1111/j.1550-7408.1968.tb02138.x-BIB5",
+		},
+	}
+	for n := 0; n < b.N; n++ {
+		ParseUnstructured(&ref)
+	}
+}