schema: tweaks

add String() to CSLDate; we only cover a few typical cases

2 files changed, 112 insertions, 5 deletions

diff --git a/skate/schema.go b/skate/schema.go
index 7717f9f..18d0e83 100644
--- a/skate/schema.go
+++ b/skate/schema.go
@@ -21,6 +21,10 @@ var (
 	isbn10Regex = regexp.MustCompile(`[O0-9xX -]{10,18}`)
 	isbn13Regex = regexp.MustCompile(`9[O0-9xX -]{12,20}`)
 
+	// Related to CSL processing.
+	yearMonthPat    = regexp.MustCompile(`[12][0-9]{3,3}-(0?[1-9]|1[12])`)
+	yearMonthDayPat = regexp.MustCompile(`[12][0-9]{3,3}-(0?[1-9]|1[12])-([0]?[1-9]|[12][0-9]|[3][01])`)
+
 	// openLibraryDateLayouts, e.g. as found in Open Library Editions,
 	// .publish_date.
 	openLibraryDateLayouts = []string{
@@ -73,14 +77,16 @@ type Biblio struct {
 	Unstructured    string   `json:"unstructured,omitempty"`
 	Url             string   `json:"url,omitempty"`
 	Volume          string   `json:"volume,omitempty"`
-	Year            int64    `json:"year,omitempty"`
-	// More non-standard fields go into extra.
+	Year            int      `json:"year,omitempty"`
+	// Any field we may require as part of our processing should go into an
+	// extra section.
 	Extra struct {
 		ISBN []string `json:"isbn"`
 	} `json:"extra"`
 }
 
-// RefToRelease converts a ref to a release.
+// RefToRelease converts a ref to a release. We want this e.g. for a release to
+// release fuzzy verification, when one of the docs is a ref.
 func RefToRelease(ref *Ref) (*Release, error) {
 	var (
 		release  Release
@@ -103,6 +109,7 @@ func RefToRelease(ref *Ref) (*Release, error) {
 	release.Volume = b.Volume
 	release.Issue = b.Issue
 	release.Pages = b.Pages
+	// Skip some accidental zero or bogus years.
 	if ref.ReleaseYear > 1000 {
 		release.ReleaseYearValue = fmt.Sprintf("%d", ref.ReleaseYear)
 	}
@@ -124,7 +131,7 @@ func RefToRelease(ref *Ref) (*Release, error) {
 
 // ReleaseToUnstructured tries to render a sensible string, e.g. for frontend
 // display of unmatched and other relations. Some examples:
-// https://guides.lib.uw.edu/c.php?g=341448&p=4076094 No specific style, just
+// https://guides.lib.uw.edu/c.php?g=341448&p=4076094 - no specific style, just
 // try to be readable.
 func ReleaseToUnstructured(r *Release) string {
 	var (
@@ -199,9 +206,10 @@ func ReleaseToUnstructured(r *Release) string {
 	return sb.String()
 }
 
-// ParseIsbn tries to find and validate ISBN from unstructured data. Returns a
+// ParseIsbn tries to find and validate ISBN from a string. Returns a
 // list of unique, unsorted and validated ISBN13, e.g. 9780123838520.
 func ParseIsbn(s string) []string {
+	// Some example inputs:
 	// ISBN: 10: 0137822693, pp: 373
 	// Robotec, E. (1996). Scorbot ER VII, User's Manual, Eshed Robotec,
 	// ISBN9652910333. Shannon, C. (1948). A Mathematical Theory of
@@ -317,6 +325,8 @@ type Release struct {
 			MetadataVersion int                `json:"metadataVersion,omitempty"`
 			Relations       []DataCiteRelation `json:"relations,omitempty"`
 		} `json:"datacite,omitempty"`
+		// Anything in the "Skate" substruct should be considered private to
+		// skate, nothing to depend upon outside this pipeline.
 		Skate struct {
 			// Mark as converted from "ref", "rg" or other schemas.
 			Status string `json:"status,omitempty"`
@@ -330,6 +340,9 @@ type Release struct {
 			ResearchGate struct {
 				URL string `json:"url,omitempty"`
 			} `json:"rg,omitempty"`
+			// At a point, where we have a release that was a ref and we only
+			// have a partial container name, we can include any discovered
+			// container name here.
 			ResolvedContainerName string `json:"resolved_container_name"`
 		} `json:"skate,omitempty"`
 		OpenLibrary struct {
@@ -471,11 +484,74 @@ type CSL struct {
 // 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.
+// https://docs.citationstyles.org/en/stable/specification.html#date-part, in a
+// json schema: https://git.io/J411z
+//
+// Raw might be removed in the future: https://discourse.citationstyles.org/t/raw-dates-vs-date-parts/1533/12
 type CSLDate struct {
 	Raw   string  `json:"raw,omitempty"`
 	Parts [][]int `json:"date-parts,omitempty"`
 }
 
+func (c *CSLDate) String() string {
+	if c.Raw != "" {
+		return c.Raw
+	}
+	// The "date-parts" field has quite some spec around them, also some open
+	// issues (e.g. schema allows for string and numbers, which might lead to
+	// issues like: https://github.com/zotero/zotero/issues/1603).
+	var renderField = func(v interface{}, padding int) string {
+		switch padding {
+		case 2:
+			switch w := v.(type) {
+			case string:
+				return w
+			case int:
+				return fmt.Sprintf("%02d", v)
+			default:
+				return fmt.Sprintf("%v", v)
+			}
+		default:
+			switch w := v.(type) {
+			case string:
+				return w
+			case int:
+				return fmt.Sprintf("%d", v)
+			default:
+				return fmt.Sprintf("%v", v)
+			}
+		}
+	}
+	switch {
+	case len(c.Parts) == 1:
+		switch len(c.Parts[0]) {
+		case 0:
+			return ""
+		case 1:
+			return renderField(c.Parts[0][0], 4)
+		case 2:
+			v := fmt.Sprintf("%s-%s",
+				renderField(c.Parts[0][0], 4),
+				renderField(c.Parts[0][1], 2),
+			)
+			// Assume year, month for the moment.
+			if yearMonthPat.MatchString(v) {
+				return v
+			}
+		case 3:
+			v := fmt.Sprintf("%s-%s-%s",
+				renderField(c.Parts[0][0], 4),
+				renderField(c.Parts[0][1], 2),
+				renderField(c.Parts[0][2], 2))
+			// Assume year, month, day for the moment.
+			if yearMonthDayPat.MatchString(v) {
+				return v
+			}
+		}
+	}
+	return ""
+}
+
 // 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
diff --git a/skate/schema_test.go b/skate/schema_test.go
index 59b1f58..3e783df 100644
--- a/skate/schema_test.go
+++ b/skate/schema_test.go
@@ -364,6 +364,37 @@ func TestReleaseToUnstructured(t *testing.T) {
 	}
 }
 
+func TestCSLDateString(t *testing.T) {
+	var cases = []struct {
+		date CSLDate
+		s    string
+	}{
+		{CSLDate{}, ""},
+		{CSLDate{Raw: "2012"}, "2012"},
+		{CSLDate{Raw: "about 2012"}, "about 2012"},
+		{CSLDate{Raw: "", Parts: nil}, ""},
+		{CSLDate{Raw: "", Parts: [][]int{}}, ""},
+		{CSLDate{Raw: "", Parts: [][]int{
+			[]int{2001},
+		}}, "2001"},
+		{CSLDate{Raw: "", Parts: [][]int{
+			[]int{2001, 1},
+		}}, "2001-01"},
+		{CSLDate{Raw: "", Parts: [][]int{
+			[]int{2001, 1, 1},
+		}}, "2001-01-01"},
+		{CSLDate{Raw: "", Parts: [][]int{
+			[]int{2001, 1, 12},
+		}}, "2001-01-12"},
+	}
+	for i, c := range cases {
+		result := c.date.String()
+		if result != c.s {
+			t.Fatalf("[%d] got %v, want %v", i, result, c.s)
+		}
+	}
+}
+
 func BenchmarkParseIsbn(b *testing.B) {
 	for n := 0; n < b.N; n++ {
 		ParseIsbn("House Pvt. Limited., (2006), ISBN 9788183561426. Date accessed: August 2015.")