1 files changed, 58 insertions, 1 deletions
diff --git a/skate/zippy.go b/skate/zippy.go
index 5767b34..bd7bf70 100644
--- a/skate/zippy.go
+++ b/skate/zippy.go
@@ -14,8 +14,10 @@ import (
 	"fmt"
 	"io"
 	"log"
+	"sort"
 	"strings"
 
+	"git.archive.org/martin/cgraph/skate/set"
 	"git.archive.org/martin/cgraph/skate/zipkey"
 	json "github.com/segmentio/encoding/json"
 )
@@ -316,7 +318,57 @@ func ZippyVerifyRefsOpenLibrary(olr, refs io.Reader, w io.Writer) error {
 //
 // We can identify, which docs have been matched by checking the ref key and index.
 func ZippyBrefAugment(bref, raw io.Reader, w io.Writer) error {
-	return nil
+	var (
+		enc     = json.NewEncoder(w)
+		keyer   = makeKeyFunc("\t", 1)
+		grouper = func(g *zipkey.Group) error {
+			// g.G0 contains a matched docs for a given work id, g.G1 all raw
+			// refs, with the same work id.
+
+			// First, iterate over all matches and sort out duplicates, e.g.
+			// docs that have the same source and target id.
+			uniqueBref, err := uniqueMatches(g.G0)
+			if err != nil {
+				return err
+			}
+			return nil
+		}
+	)
+	zipper := zipkey.New(bref, raw, keyer, grouper)
+	return zipper.Run()
+}
+
+// uniqueMatches takes a list of bref docs (unserialized) and will return a
+// list of serialized bref docs, containing unique matches.
+func uniqueMatches(docs []string) (result []*BiblioRef, err error) {
+	var (
+		brefs []*BiblioRef
+		bref  BiblioRef
+	)
+	for _, doc := range docs {
+		if err := json.Unmarshal([]byte(doc), &bref); err != nil {
+			return nil, err
+		}
+		brefs = append(brefs, &bref)
+	}
+	// Make sure we exact matches come first.
+	sort.Slice(brefs, func(i, j int) bool {
+		return brefs[i].MatchStatus != StatusExact.Short()
+	})
+	// We consider a match unique, if source and target match.
+	hash := func(bref *BiblioRef) string {
+		return bref.SourceReleaseIdent + bref.TargetReleaseIdent
+	}
+	seen := set.New()
+	for _, doc := range brefs {
+		v := hash(doc)
+		if seen.Contains(v) {
+			continue
+		}
+		seen.Add(v)
+		result = append(result, doc)
+	}
+	return result, nil
 }
 
 // Cut returns a specific column (1-indexed, like CutSep) from a tabular
@@ -371,3 +423,8 @@ func stringToWiki(s string) (r *MinimalCitations, err error) {
 	err = json.Unmarshal([]byte(s), &r)
 	return
 }
+
+func stringToBiblioref(s string) (r *BiblioRef, err error) {
+	err = json.Unmarshal([]byte(s), &r)
+	return
+}