From 25c10e412df5775d101c0c07752cf826e0850934 Mon Sep 17 00:00:00 2001
From: Martin Czygan <martin.czygan@gmail.com>
Date: Tue, 8 Jun 2021 22:27:03 +0200
Subject: reduce: add rere mode

---
 skate/cmd/skate-reduce/main.go | 17 ++++++++++++++
 skate/zippy.go                 | 50 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 67 insertions(+)

(limited to 'skate')

diff --git a/skate/cmd/skate-reduce/main.go b/skate/cmd/skate-reduce/main.go
index c0ef9fc..098240a 100644
--- a/skate/cmd/skate-reduce/main.go
+++ b/skate/cmd/skate-reduce/main.go
@@ -44,6 +44,11 @@
 // * oled  | zippy mode for releases and OL inputs, emit biblioref.
 //         |
 //         | $ skate-reduce -m oled -F a.ndj -O b.ndj
+//         |
+//         |
+// * rere  | zippy mode for OL release and ref (as release) inputs, emit biblioref.
+//         |
+//         | $ skate-reduce -m rere -O a.ndj -F b.ndj
 //
 package main
 
@@ -159,6 +164,18 @@ func main() {
 		if err := skate.ZippyVerifyRefsOpenLibrary(o, f, bw); err != nil {
 			log.Fatal(err)
 		}
+	case "rere":
+		o, f, err := xio.OpenTwo(*releases, *refs)
+		if err != nil {
+			log.Fatal(err)
+		}
+		r, ok := reasonMap[*reason]
+		if !ok {
+			log.Fatalf("unknown reason: %v", *reason)
+		}
+		if err := skate.ZippyExactReleases(o, f, r, bw); err != nil {
+			log.Fatal(err)
+		}
 	default:
 		log.Fatalf("invalid mode")
 	}
diff --git a/skate/zippy.go b/skate/zippy.go
index 7ee12cf..793fe4a 100644
--- a/skate/zippy.go
+++ b/skate/zippy.go
@@ -73,6 +73,56 @@ func ZippyExact(releases, refs io.Reader, matchResult MatchResult, w io.Writer)
 	return zipper.Run()
 }
 
+// ZippyExactReleases takes two release readers (key, doc) and assigns a fixed
+// match result.
+func ZippyExactReleases(olReader, reReader io.Reader, matchResult MatchResult, w io.Writer) error {
+	var (
+		enc     = json.NewEncoder(w)
+		keyer   = makeKeyFunc("\t", 1)
+		i       = 0
+		grouper = func(g *zipkey.Group) error {
+			i++
+			if i%10000 == 0 {
+				log.Printf("processed %v groups", i)
+			}
+			var (
+				target, re *Release
+				err        error
+			)
+			if len(g.G0) == 0 || len(g.G1) == 0 {
+				return nil
+			}
+			if target, err = stringToRelease(Cut(g.G0[0], 2)); err != nil {
+				groupLogf(g, "[skip] failed to parse release: %v", err)
+				return nil
+			}
+			for _, line := range g.G1 {
+				if re, err = stringToRelease(Cut(line, 2)); err != nil {
+					groupLogf(g, "[skip] failed to parse release: %v", err)
+					continue
+				}
+				var bref BiblioRef
+				bref.SourceReleaseIdent = re.Ident
+				bref.SourceWorkIdent = re.WorkID
+				bref.SourceReleaseStage = re.ReleaseStage
+				bref.SourceYear = fmt.Sprintf("%d", re.ReleaseYear())
+				bref.RefIndex = re.Extra.Skate.Ref.Index + 1 // we want 1-index (also helps with omitempty)
+				bref.RefKey = re.Extra.Skate.Ref.Key
+				bref.TargetOpenLibraryWork = target.WorkID
+				bref.MatchProvenance = re.Extra.Skate.Ref.Source
+				bref.MatchStatus = matchResult.Status.Short()
+				bref.MatchReason = matchResult.Reason.Short()
+				if err := enc.Encode(bref); err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+	)
+	zipper := zipkey.New(olReader, reReader, keyer, grouper)
+	return zipper.Run()
+}
+
 // ZippyExactWiki takes a release and wiki reader (key, doc) and assigns a
 // fixed match result.
 func ZippyExactWiki(releases, wiki io.Reader, mr MatchResult, w io.Writer) error {
-- 
cgit v1.2.3