1 files changed, 50 insertions, 0 deletions
diff --git a/skate/verify.go b/skate/verify.go
index cd40279..e6eb8b8 100644
--- a/skate/verify.go
+++ b/skate/verify.go
@@ -247,6 +247,51 @@ func ZipUnverified(releases, refs io.Reader, mr MatchResult, provenance string,
 	return zipper.Run()
 }
 
+// ZipWikiUnverified takes a release and wiki reader (tsv, with ident, key, doc)
+// and assigns a fixed match result.
+func ZipWikiUnverified(releases, wiki io.Reader, mr MatchResult, provenance string, 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.
+	enc := json.NewEncoder(w)
+	keyer := 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
+		}
+	}
+	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()
+}
+
 // ZipVerifyRefs takes a release and refs reader (tsv, with ident, key, doc)
 // and will execute gf for each group found.
 func ZipVerifyRefs(releases, refs io.Reader, w io.Writer) error {
@@ -313,6 +358,11 @@ func stringToRef(s string) (r *Ref, err error) {
 	return
 }
 
+func stringToWiki(s string) (r *MinimalCitations, err error) {
+	err = json.Unmarshal([]byte(s), &r)
+	return
+}
+
 // Verify follows the fuzzycat (Python) implementation of this function: it
 // compares two release entities. The Go version can be used for large batch
 // processing (where the Python version might take two or more days).