From 0396db2bc11eaee8f2058e82cc1b5ea0af9f0956 Mon Sep 17 00:00:00 2001
From: Martin Czygan <martin.czygan@gmail.com>
Date: Tue, 27 Apr 2021 23:38:52 +0200
Subject: ref: parse out isbn

---
 skate/fixtures/ref_with_isbn.json | 13 ++++++++++++
 skate/schema.go                   | 44 +++++++++++++++++++++++++++++++++++++++
 skate/schema_test.go              |  1 +
 3 files changed, 58 insertions(+)
 create mode 100644 skate/fixtures/ref_with_isbn.json

(limited to 'skate')

diff --git a/skate/fixtures/ref_with_isbn.json b/skate/fixtures/ref_with_isbn.json
new file mode 100644
index 0000000..2cd8480
--- /dev/null
+++ b/skate/fixtures/ref_with_isbn.json
@@ -0,0 +1,13 @@
+{
+  "biblio": {
+    "title": "Antibiotic Resistant Bacteria -A Continuous Challenge in the New Millennium Edited by Dr. Marina Pana ISBN",
+    "unstructured": "www.intechopen.com Antibiotic Resistant Bacteria -A Continuous Challenge in the New Millennium Edited by Dr. Marina Pana ISBN 978-953-51-0472-8"
+  },
+  "index": 443,
+  "key": "b443",
+  "ref_source": "grobid",
+  "release_ident": "n4zvrgchmfexdb6gesxfgxykxi",
+  "release_year": 2012,
+  "work_ident": "aaan6iujevgpnmcif2hb62uaai"
+}
+
diff --git a/skate/schema.go b/skate/schema.go
index a9b1e8a..6c96bb8 100644
--- a/skate/schema.go
+++ b/skate/schema.go
@@ -2,6 +2,7 @@ package skate
 
 import (
 	"fmt"
+	"regexp"
 	"strconv"
 	"strings"
 
@@ -9,6 +10,11 @@ import (
 	"git.archive.org/martin/cgraph/skate/set"
 )
 
+var (
+	isbn10Regex = regexp.MustCompile(`[0-9xX -]{10,18}`)
+	isbn13Regex = regexp.MustCompile(`9[0-9xX -]{12,20}`)
+)
+
 // RefToRelease converts a ref to a release. Set a extra.skate.status flag to
 // be able to distinguish converted entities later.
 func RefToRelease(ref *Ref) (*Release, error) {
@@ -41,6 +47,44 @@ func RefToRelease(ref *Ref) (*Release, error) {
 		contribs[i].RawName = name
 	}
 	release.Contribs = contribs
+	// XXX: Find ISBN in unstructured. Might be expensive, do we need a flag?
+	unlo := strings.ToLower(ref.Biblio.Unstructured)
+	if strings.Contains(unlo, "isbn") {
+		// ISBN: 10: 0137822693, pp: 373
+		// Robotec, E. (1996). Scorbot ER VII, User's Manual, Eshed Robotec,
+		// ISBN9652910333. Shannon, C. (1948). A Mathematical Theory of
+		// Communication. The Bell System Technical Journal. July; October,
+		// Vol. 27, pp. 379-423; 623-656.
+		// Artech House, ISBN: 978-1-60807-201-9, 2011.
+		// ...
+		var (
+			candidates10 = isbn10Regex.FindAllString(ref.Biblio.Unstructured, -1)
+			candidates13 = isbn13Regex.FindAllString(ref.Biblio.Unstructured, -1)
+			valid        = set.New()
+		)
+		for _, v := range append(candidates10, candidates13...) {
+			var u []rune
+			for _, c := range v {
+				if c >= '0' && c <= '9' || c == 'x' || c == 'X' {
+					u = append(u, c)
+				}
+			}
+			s := string(u)
+			if !isbn.Validate(s) {
+				continue
+			}
+			if len(s) < 12 {
+				w, err := isbn.To13(s)
+				if err != nil {
+					continue
+				}
+				valid.Add(w)
+			} else {
+				valid.Add(s)
+			}
+		}
+		release.ExtIDs.ISBN = valid.Slice()
+	}
 	return &release, nil
 }
 
diff --git a/skate/schema_test.go b/skate/schema_test.go
index 6a95115..c1cec35 100644
--- a/skate/schema_test.go
+++ b/skate/schema_test.go
@@ -9,6 +9,7 @@ import (
 	"github.com/nsf/jsondiff"
 )
 
+// XXX: Work on JSON directly, as structs can get unwieldy.
 func TestOpenLibraryToRelease(t *testing.T) {
 	var cases = []struct {
 		work    OpenLibraryWork
-- 
cgit v1.2.3