1 files changed, 171 insertions, 0 deletions
diff --git a/skate/zippy.go b/skate/zippy.go
new file mode 100644
index 0000000..76f576d
--- /dev/null
+++ b/skate/zippy.go
@@ -0,0 +1,171 @@
+package skate
+
+import (
+	"fmt"
+	"io"
+	"strings"
+
+	"git.archive.org/martin/cgraph/skate/zipkey"
+	json "github.com/segmentio/encoding/json"
+)
+
+// This file contains the two-stream (zippy) matchers.
+
+// ZippyFixed takes a release and refs reader (tsv, with ident, key, doc)
+// and assigns a fixed match result.
+func ZippyFixed(releases, refs io.Reader, mr MatchResult, provenance string, w io.Writer) error {
+	var (
+		enc     = json.NewEncoder(w)
+		keyer   = makeKeyFunc("\t", 2)
+		grouper = func(g *zipkey.Group) error {
+			if len(g.G0) == 0 || len(g.G1) == 0 {
+				return nil
+			}
+			target, err := stringToRelease(lineColumn(g.G0[0], "\t", 3))
+			if err != nil {
+				return err
+			}
+			for _, line := range g.G1 {
+				ref, err := stringToRef(lineColumn(line, "\t", 3))
+				if err != nil {
+					return err
+				}
+				var bref BiblioRef
+				bref.SourceReleaseIdent = ref.ReleaseIdent
+				bref.SourceWorkIdent = ref.WorkIdent
+				bref.SourceReleaseStage = ref.ReleaseStage
+				bref.SourceYear = fmt.Sprintf("%d", ref.ReleaseYear)
+				bref.RefIndex = ref.Index + 1 // we want 1-index (also helps with omitempty)
+				bref.RefKey = ref.Key
+				bref.TargetReleaseIdent = target.Ident
+				bref.TargetWorkIdent = target.WorkID
+				bref.MatchProvenance = provenance
+				bref.MatchStatus = mr.Status.Short()
+				bref.MatchReason = mr.Reason.Short()
+				if err := enc.Encode(bref); err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+	)
+	zipper := zipkey.New(releases, refs, keyer, grouper)
+	return zipper.Run()
+}
+
+// ZippyFixedWiki takes a release and wiki reader (tsv, with ident, key, doc)
+// and assigns a fixed match result.
+func ZippyFixedWiki(releases, wiki io.Reader, mr MatchResult, provenance string, w io.Writer) error {
+	var (
+		enc     = json.NewEncoder(w)
+		keyer   = makeKeyFunc("\t", 2)
+		grouper = func(g *zipkey.Group) error {
+			if len(g.G0) == 0 || len(g.G1) == 0 {
+				return nil
+			}
+			target, err := stringToRelease(lineColumn(g.G0[0], "\t", 3))
+			if err != nil {
+				return err
+			}
+			for _, line := range g.G1 {
+				wiki, err := stringToWiki(lineColumn(line, "\t", 3))
+				if err != nil {
+					return err
+				}
+				var bref BiblioRef
+				bref.Key = fmt.Sprintf("%s_%s", slugifyString(wiki.PageTitle), target.Ident) // XXX: what should we use?
+				bref.SourceWikipediaArticle = wiki.PageTitle
+				bref.TargetReleaseIdent = target.Ident
+				bref.TargetWorkIdent = target.WorkID
+				bref.MatchProvenance = provenance
+				bref.MatchStatus = mr.Status.Short()
+				bref.MatchReason = mr.Reason.Short()
+				if err := enc.Encode(bref); err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+	)
+	zipper := zipkey.New(releases, wiki, keyer, grouper)
+	return zipper.Run()
+}
+
+// ZippyVerifyRefs takes a release and refs reader (tsv, with ident, key, doc)
+// and will execute gf for each group found.
+func ZippyVerifyRefs(releases, refs io.Reader, w io.Writer) error {
+	// Define a grouper, working on one set of refs and releases with the same
+	// key at a time. Here, we do verification and write out the generated
+	// biblioref.
+	var (
+		enc     = json.NewEncoder(w)
+		keyer   = makeKeyFunc("\t", 2)
+		grouper = func(g *zipkey.Group) error {
+			if len(g.G0) == 0 || len(g.G1) == 0 {
+				return nil
+			}
+			pivot, err := stringToRelease(lineColumn(g.G0[0], "\t", 3))
+			if err != nil {
+				return err
+			}
+			for _, line := range g.G1 {
+				re, err := stringToRelease(lineColumn(line, "\t", 3))
+				if err != nil {
+					return err
+				}
+				result := Verify(pivot, re)
+				switch result.Status {
+				case StatusExact, StatusStrong:
+					if result.Reason == ReasonDOI {
+						continue
+					}
+					br := generateBiblioRef(re, pivot, result, "fuzzy")
+					if err := enc.Encode(br); err != nil {
+						return err
+					}
+				}
+			}
+			return nil
+		}
+	)
+	zipper := zipkey.New(releases, refs, keyer, grouper)
+	return zipper.Run()
+}
+
+// makeKeyFunc creates a function that can be used as keyFunc, selecting a
+// column from sep.
+func makeKeyFunc(sep string, column int) func(string) (string, error) {
+	return func(s string) (string, error) {
+		if k := lineColumn(s, "\t", 2); k == "" {
+			return k, fmt.Errorf("cannot get key: %s", s)
+		} else {
+			return k, nil
+		}
+	}
+}
+
+// lineColumn returns a specific column (1-indexed, like cut) from a tabular
+// file, returns empty string if column is invalid.
+func lineColumn(line, sep string, column int) string {
+	parts := strings.Split(strings.TrimSpace(line), sep)
+	if len(parts) < column {
+		return ""
+	} else {
+		return parts[column-1]
+	}
+}
+
+func stringToRelease(s string) (r *Release, err error) {
+	err = json.Unmarshal([]byte(s), &r)
+	return
+}
+
+func stringToRef(s string) (r *Ref, err error) {
+	err = json.Unmarshal([]byte(s), &r)
+	return
+}
+
+func stringToWiki(s string) (r *MinimalCitations, err error) {
+	err = json.Unmarshal([]byte(s), &r)
+	return
+}