reduce/schema: add csl

3 files changed, 70 insertions, 5 deletions

diff --git a/skate/reduce.go b/skate/reduce.go
index b28d976..3bb2466 100644
--- a/skate/reduce.go
+++ b/skate/reduce.go
@@ -511,7 +511,13 @@ func matchedRefsExtend(matched []*BiblioRef, refs []*Ref, stats *statsAugment) [
 		bref.SourceReleaseStage = r.ReleaseStage
 		bref.SourceWorkIdent = r.WorkIdent
 		bref.SourceYear = fmt.Sprintf("%d", r.ReleaseYear)
-		bref.TargetUnstructured = r.Biblio.Unstructured
+		if r.Biblio.Unstructured != "" {
+			bref.TargetUnstructured = r.Biblio.Unstructured
+		} else {
+			// TODO: create CSL subdoc
+			log.Printf("to-csl: %v", r)
+			bref.TargetCSL = nil
+		}
 		// Reuse fields for debugging, for now.
 		bref.MatchStatus = StatusUnmatched.Short()
 		bref.MatchReason = ReasonUnknown.Short()
diff --git a/skate/schema.go b/skate/schema.go
index dd088c2..afd9034 100644
--- a/skate/schema.go
+++ b/skate/schema.go
@@ -424,7 +424,66 @@ type BiblioRef struct {
 	MatchStatus            string `json:"match_status,omitempty"`
 	MatchReason            string `json:"match_reason,omitempty"`
 	TargetUnstructured     string `json:"target_unstructured,omitempty"`
-	TargetCSL              string `json:"target_csl,omitempty"`
+	TargetCSL              *CSL   `json:"target_csl,omitempty"`
+}
+
+// CSL is a subset of citation style language schema.
+// https://github.com/citation-style-language/schema,
+// https://navneethg.github.io/jsonschemaviewer/. This is a subset only.
+type CSL struct {
+	Accessed            CSLDate     `json:"accessed,omitempty"`
+	Author              []CSLAuthor `json:"author,omitempty"`
+	CollectionTitle     string      `json:"collection-title"`
+	ContainerTitle      string      `json:"container-title,omitempty"`
+	ContainerTitleShort string      `json:"container-title-short,omitempty"`
+	DOI                 string      `json:"DOI,omitempty"`
+	ID                  string      `json:"id,omitempty"`
+	ISBN                string      `json:"ISBN,omitempty"`
+	ISSN                string      `json:"ISSN,omitempty"`
+	Issue               string      `json:"issue,omitempty"`
+	Issued              CSLDate     `json:"issued,omitempty"`
+	JounralAbbreviation string      `json:"journalAbbreviation"`
+	Language            string      `json:"language"`
+	NumberOfPages       string      `json:"number-of-pages,omitempty"`
+	OriginalTitle       string      `json:"original-title,omitempty"`
+	PMCID               string      `json:"PMCID,omitempty"`
+	PMID                string      `json:"PMID,omitempty"`
+	Page                string      `json:"page,omitempty"`
+	PageFirst           string      `json:"page-first,omitempty"`
+	Publisher           string      `json:"publisher,omitempty"`
+	Source              string      `json:"source,omitempty"`
+	Status              string      `json:"status,omitempty"`
+	Title               string      `json:"title,omitempty"`
+	TitleShort          string      `json:"title-short,omitempty"`
+	Type                string      `json:"type,omitempty"`
+	URL                 string      `json:"URL,omitempty"`
+	Volume              string      `json:"volume,omitempty"`
+	VolumeTitle         string      `json:"volume-title,omitempty"`
+	VolumeTitleShort    string      `json:"volume-title-short,omitempty"`
+	YearSuffix          string      `json:"year-suffix,omitempty"`
+}
+
+// A date field is a complex field that expresses a date or a range of dates.
+// An example date field in CSL is issued, which identifies the date an item
+// was issued or published. Date fields can be expressed in two different
+// formats. The first format is an array format (note the double-nesting of the
+// array). To express a date range in this format, the ending date would be set
+// as a second array. The second date format is a raw string. The recommended
+// encoding is a string that represents the date in a numberic year-month-day
+// format.
+type CSLDate struct {
+	Raw   string  `json:"raw,omitempty"`
+	Parts [][]int `json:"date-parts,omitempty"`
+}
+
+// Some personal names are represented by a single field (e.g. mononyms such as
+// “Prince” or “Plato”). In such cases, the name can be delivered as a lone
+// family element. Institutional names may be delivered in the same way, but it
+// is preferred to set them instead as a literal element.
+type CSLAuthor struct {
+	Family  string `json:"family,omitempty"`
+	Given   string `json:"given,omitempty"`
+	Literal string `json:"literal,omitempty"`
 }
 
 func (b *BiblioRef) Reset() {
@@ -447,7 +506,7 @@ func (b *BiblioRef) Reset() {
 	b.MatchStatus = ""
 	b.MatchReason = ""
 	b.TargetUnstructured = ""
-	b.TargetCSL = ""
+	b.TargetCSL = nil
 }
 
 // Hash returns a string that will be the same, if source and target are
diff --git a/skate/schema_test.go b/skate/schema_test.go
index cb1cf83..d70c001 100644
--- a/skate/schema_test.go
+++ b/skate/schema_test.go
@@ -220,11 +220,11 @@ func TestLinkHash(t *testing.T) {
 	}{
 		{
 			bref:     BiblioRef{},
-			linkHash: "e39b4dd927dbfb19c8aa78b10f684b0a5c35fcca",
+			linkHash: "042db9d756488070ce841d1369f8d877b3982c40",
 		},
 		{
 			bref:     BiblioRef{SourceReleaseIdent: "123"},
-			linkHash: "6d244189489b57c8f70ac7a835f1bacfee00c42a",
+			linkHash: "1a31cc64bd5f6080dd81edc91079bdc41a95bcfe",
 		},
 		{
 			bref: BiblioRef{