wip: ol conversion and tests

2 files changed, 184 insertions, 11 deletions

diff --git a/skate/schema.go b/skate/schema.go
index 1dd02d9..a9b1e8a 100644
--- a/skate/schema.go
+++ b/skate/schema.go
@@ -4,6 +4,9 @@ import (
 	"fmt"
 	"strconv"
 	"strings"
+
+	"git.archive.org/martin/cgraph/skate/isbn"
+	"git.archive.org/martin/cgraph/skate/set"
 )
 
 // RefToRelease converts a ref to a release. Set a extra.skate.status flag to
@@ -81,15 +84,15 @@ type Release struct {
 		Role    string `json:"role,omitempty"`
 	} `json:"contribs,omitempty"`
 	ExtIDs struct {
-		DOI         string `json:"doi,omitempty"`
-		PMID        string `json:"pmid,omitempty"`
-		PMCID       string `json:"pmcid,omitempty"`
-		Arxiv       string `json:"arxiv,omitempty"`
-		Core        string `json:"core,omitempty"`
-		WikidataQID string `json:"wikidata_qid,omitempty"`
-		Jstor       string `json:"jstor,omitempty"`
-		ISBN10      string `json:"isbn10,omitempty"`
-		ISBN13      string `json:"isbn13,omitempty"`
+		Arxiv       string   `json:"arxiv,omitempty"`
+		Core        string   `json:"core,omitempty"`
+		DOI         string   `json:"doi,omitempty"`
+		ISBN        []string `json:"isbn,omitempty"` // should be isbn13
+		Jstor       string   `json:"jstor,omitempty"`
+		OLID        string   `json:"olid,omitempty"`
+		PMCID       string   `json:"pmcid,omitempty"`
+		PMID        string   `json:"pmid,omitempty"`
+		WikidataQID string   `json:"wikidata_qid,omitempty"`
 	} `json:"ext_ids,omitempty"`
 	Ident     string `json:"ident,omitempty"`
 	Publisher string `json:"publisher,omitempty"`
@@ -140,6 +143,9 @@ type Release struct {
 				URL string `json:"url,omitempty"`
 			} `json:"rg,omitempty"`
 		} `json:"skate,omitempty"`
+		OpenLibrary struct {
+			HasFulltext bool `json:"has_fulltext,omitempty"`
+		} `json:"ol,omitempty"`
 	} `json:"extra,omitempty"`
 }
 
@@ -356,6 +362,7 @@ func OpenLibraryToRelease(w *OpenLibraryWork) (*Release, error) {
 			RawName string `json:"raw_name,omitempty"`
 			Role    string `json:"role,omitempty"`
 		}, len(w.AuthorName))
+		s = set.New()
 	)
 	for i, author := range w.AuthorName {
 		contribs[i].RawName = author
@@ -364,6 +371,25 @@ func OpenLibraryToRelease(w *OpenLibraryWork) (*Release, error) {
 	if len(w.PublishYear) > 0 {
 		release.ReleaseYearValue = w.FirstPublishYear
 	}
-	// XXX: isbn normalization
-	return nil, nil
+	for _, v := range w.Isbn {
+		switch {
+		case len(v) == 9 || len(v) == 10:
+			if w, err := isbn.To13(v); err == nil {
+				s.Add(w)
+			}
+		case len(v) == 12 || len(v) == 13:
+			if w, err := isbn.To13(v); err == nil {
+				s.Add(w)
+			}
+		default:
+			continue
+		}
+	}
+	if len(w.Publisher) > 0 {
+		release.Publisher = w.Publisher[0]
+	}
+	release.ExtIDs.ISBN = s.Slice()
+	release.ExtIDs.OLID = strings.Replace(w.Key, "/works/", "", 1)
+	release.Extra.OpenLibrary.HasFulltext = w.HasFulltext
+	return &release, nil
 }
diff --git a/skate/schema_test.go b/skate/schema_test.go
new file mode 100644
index 0000000..6a95115
--- /dev/null
+++ b/skate/schema_test.go
@@ -0,0 +1,147 @@
+package skate
+
+import (
+	"encoding/json"
+	"fmt"
+	"reflect"
+	"testing"
+
+	"github.com/nsf/jsondiff"
+)
+
+func TestOpenLibraryToRelease(t *testing.T) {
+	var cases = []struct {
+		work    OpenLibraryWork
+		release Release
+		err     error
+	}{
+		{
+			work: OpenLibraryWork{},
+			release: Release{
+				ExtIDs: struct {
+					Arxiv       string   `json:"arxiv,omitempty"`
+					Core        string   `json:"core,omitempty"`
+					DOI         string   `json:"doi,omitempty"`
+					ISBN        []string `json:"isbn,omitempty"`
+					Jstor       string   `json:"jstor,omitempty"`
+					OLID        string   `json:"olid,omitempty"`
+					PMCID       string   `json:"pmcid,omitempty"`
+					PMID        string   `json:"pmid,omitempty"`
+					WikidataQID string   `json:"wikidata_qid,omitempty"`
+				}{},
+			},
+			err: nil,
+		},
+		{
+			work: OpenLibraryWork{
+				Title: "Hello World",
+				Isbn: []string{
+					"2844273386",
+					"9782844273383",
+				},
+			},
+			release: Release{
+				ExtIDs: struct {
+					Arxiv       string   `json:"arxiv,omitempty"`
+					Core        string   `json:"core,omitempty"`
+					DOI         string   `json:"doi,omitempty"`
+					ISBN        []string `json:"isbn,omitempty"`
+					Jstor       string   `json:"jstor,omitempty"`
+					OLID        string   `json:"olid,omitempty"`
+					PMCID       string   `json:"pmcid,omitempty"`
+					PMID        string   `json:"pmid,omitempty"`
+					WikidataQID string   `json:"wikidata_qid,omitempty"`
+				}{
+					ISBN: []string{"9782844273383"},
+				},
+				Title: "Hello World",
+			},
+			err: nil,
+		},
+		{
+			work: OpenLibraryWork{
+				Key:   "/works/OL10000003W",
+				Title: "Hello World",
+				Isbn: []string{
+					"2844273386",
+					"9782844273383",
+				},
+				HasFulltext: false,
+			},
+			release: Release{
+				ExtIDs: struct {
+					Arxiv       string   `json:"arxiv,omitempty"`
+					Core        string   `json:"core,omitempty"`
+					DOI         string   `json:"doi,omitempty"`
+					ISBN        []string `json:"isbn,omitempty"`
+					Jstor       string   `json:"jstor,omitempty"`
+					OLID        string   `json:"olid,omitempty"`
+					PMCID       string   `json:"pmcid,omitempty"`
+					PMID        string   `json:"pmid,omitempty"`
+					WikidataQID string   `json:"wikidata_qid,omitempty"`
+				}{
+					ISBN: []string{"9782844273383"},
+					OLID: "OL10000003W",
+				},
+				Title: "Hello World",
+				Extra: struct {
+					ContainerName string      `json:"container_name,omitempty"`
+					SubtitleValue interface{} `json:"subtitle,omitempty"` // []str or str
+					Crossref      struct {
+						Type string `json:"type,omitempty"`
+					} `json:"crossref,omitempty"`
+					DataCite struct {
+						MetadataVersion int                `json:"metadataVersion,omitempty"`
+						Relations       []DataCiteRelation `json:"relations,omitempty"`
+					} `json:"datacite,omitempty"`
+					Skate struct {
+						// Mark as converted (e.g. by setting status to "ref")
+						Status string `json:"status,omitempty"`
+						// Carry the ref index and key around.
+						Ref struct {
+							Index   int64  `json:"index,omitempty"`
+							Key     string `json:"key,omitempty"`
+							Locator string `json:"locator,omitempty"`
+						} `json:"ref,omitempty"`
+						ResearchGate struct {
+							URL string `json:"url,omitempty"`
+						} `json:"rg,omitempty"`
+					} `json:"skate,omitempty"`
+					OpenLibrary struct {
+						HasFulltext bool `json:"has_fulltext,omitempty"`
+					} `json:"ol,omitempty"`
+				}{
+					OpenLibrary: struct {
+						HasFulltext bool `json:"has_fulltext,omitempty"`
+					}{
+						HasFulltext: false,
+					},
+				},
+			},
+			err: nil,
+		},
+	}
+	for _, c := range cases {
+		r, err := OpenLibraryToRelease(&c.work)
+		if err != nil {
+			t.Fatalf("got %v, want %v", err, c.err)
+		}
+		if !reflect.DeepEqual(r, &c.release) {
+			t.Fatalf(prettyStructDiff(r, &c.release))
+		}
+	}
+}
+
+func prettyStructDiff(s, t interface{}) string {
+	b, err := json.MarshalIndent(s, "", "    ")
+	if err != nil {
+		return fmt.Sprintf("diff failed: %v", err)
+	}
+	c, err := json.MarshalIndent(t, "", "    ")
+	if err != nil {
+		return fmt.Sprintf("diff failed: %v", err)
+	}
+	opts := jsondiff.DefaultConsoleOptions()
+	_, d := jsondiff.Compare(b, c, &opts)
+	return d
+}