zippy: simplify cut

2 files changed, 34 insertions, 29 deletions

diff --git a/skate/zippy.go b/skate/zippy.go
index 146d2e7..43fc911 100644
--- a/skate/zippy.go
+++ b/skate/zippy.go
@@ -1,3 +1,5 @@
+// This file contains various "reducers", e.g. functions, that take two data
+// streams and will apply a computation on matched groups.
 package skate
 
 import (
@@ -10,17 +12,15 @@ import (
 	json "github.com/segmentio/encoding/json"
 )
 
-// This file contains the two-stream (zippy) matchers.
-
-// groupLogf logs a message and a serialized group.
+// groupLogf logs a message alongsize a serialized group.
 func groupLogf(g *zipkey.Group, s string, vs ...interface{}) {
 	log.Printf(s, vs...)
 	b, _ := json.Marshal(g)
 	log.Println(string(b))
 }
 
-// ZippyExact takes a release and refs reader (tsv, with ident, key, doc)
-// and assigns a fixed match result. XXX: allow empty keys
+// ZippyExact takes a release and refs reader (key, doc) and assigns a fixed
+// match result. XXX: allow empty keys
 func ZippyExact(releases, refs io.Reader, matchResult MatchResult, w io.Writer) error {
 	var (
 		enc     = json.NewEncoder(w)
@@ -39,12 +39,12 @@ func ZippyExact(releases, refs io.Reader, matchResult MatchResult, w io.Writer)
 			if len(g.G0) == 0 || len(g.G1) == 0 {
 				return nil
 			}
-			if target, err = stringToRelease(cut(g.G0[0], "\t", 2)); err != 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 ref, err = stringToRef(cut(line, "\t", 2)); err != nil {
+				if ref, err = stringToRef(Cut(line, 2)); err != nil {
 					groupLogf(g, "[skip] failed to parse ref: %v", err)
 					continue
 				}
@@ -71,8 +71,8 @@ func ZippyExact(releases, refs io.Reader, matchResult MatchResult, w io.Writer)
 	return zipper.Run()
 }
 
-// ZippyExactWiki takes a release and wiki reader (tsv, with key, doc)
-// and assigns a fixed match result.
+// 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 {
 	var (
 		enc     = json.NewEncoder(w)
@@ -86,11 +86,11 @@ func ZippyExactWiki(releases, wiki io.Reader, mr MatchResult, w io.Writer) error
 			if len(g.G0) == 0 || len(g.G1) == 0 {
 				return nil
 			}
-			if target, err = stringToRelease(cut(g.G0[0], "\t", 2)); err != nil {
+			if target, err = stringToRelease(Cut(g.G0[0], 2)); err != nil {
 				return err
 			}
 			for _, line := range g.G1 {
-				if wiki, err = stringToWiki(cut(line, "\t", 2)); err != nil {
+				if wiki, err = stringToWiki(Cut(line, 2)); err != nil {
 					return err
 				}
 				var bref BiblioRef
@@ -129,11 +129,11 @@ func ZippyVerifyRefs(releases, refs io.Reader, w io.Writer) error {
 			if len(g.G0) == 0 || len(g.G1) == 0 {
 				return nil
 			}
-			if pivot, err = stringToRelease(cut(g.G0[0], "\t", 2)); err != nil {
+			if pivot, err = stringToRelease(Cut(g.G0[0], 2)); err != nil {
 				return err
 			}
 			for _, line := range g.G1 {
-				if re, err = stringToRelease(cut(line, "\t", 2)); err != nil {
+				if re, err = stringToRelease(Cut(line, 2)); err != nil {
 					return err
 				}
 				result := Verify(pivot, re)
@@ -175,11 +175,11 @@ func ZippyVerifyRefsOpenLibrary(olr, refs io.Reader, w io.Writer) error {
 				return nil
 			}
 			// We take a single edition from OL; XXX: might want to link to work.
-			if pivot, err = stringToRelease(cut(g.G0[0], "\t", 2)); err != nil {
+			if pivot, err = stringToRelease(Cut(g.G0[0], 2)); err != nil {
 				return err
 			}
 			for _, line := range g.G1 {
-				if re, err = stringToRelease(cut(line, "\t", 2)); err != nil {
+				if re, err = stringToRelease(Cut(line, 2)); err != nil {
 					return err
 				}
 				// The refs have a container name, but not a title, but here we
@@ -189,7 +189,7 @@ func ZippyVerifyRefsOpenLibrary(olr, refs io.Reader, w io.Writer) error {
 				fmt.Printf("%s\t%s\t%s\t%s\t%s\n",
 					result.Status.Short(),
 					result.Reason.Short(),
-					cut(g.G0[0], "\t", 1),
+					CutSep(g.G0[0], "\t", 1),
 					pivot.Title,
 					re.Title)
 			}
@@ -200,11 +200,27 @@ func ZippyVerifyRefsOpenLibrary(olr, refs io.Reader, w io.Writer) error {
 	return zipper.Run()
 }
 
+// Cut returns a specific column (1-indexed, like CutSep) from a tabular
+// file, returns empty string if column is invalid.
+func Cut(line string, column int) string {
+	return CutSep(line, "\t", column)
+}
+
+// CutSep allows to specify a separator.
+func CutSep(line, sep string, column int) string {
+	parts := strings.Split(strings.TrimSpace(line), sep)
+	if len(parts) < column {
+		return ""
+	} else {
+		return parts[column-1]
+	}
+}
+
 // makeKeyFunc creates a function that can be used as keyFunc, selecting a
 // column from fields separated by sep; column is 1-indexed.
 func makeKeyFunc(sep string, column int) func(string) (string, error) {
 	return func(s string) (string, error) {
-		if k := cut(s, sep, column); k == "" {
+		if k := CutSep(s, sep, column); k == "" {
 			return k, fmt.Errorf("cannot get key: %s", s)
 		} else {
 			return k, nil
@@ -212,17 +228,6 @@ func makeKeyFunc(sep string, column int) func(string) (string, error) {
 	}
 }
 
-// cut returns a specific column (1-indexed, like cut) from a tabular
-// file, returns empty string if column is invalid.
-func cut(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
diff --git a/skate/zippy_test.go b/skate/zippy_test.go
index c966d67..1448472 100644
--- a/skate/zippy_test.go
+++ b/skate/zippy_test.go
@@ -17,7 +17,7 @@ func TestLineColumn(t *testing.T) {
 		{"1 2 3", "\t", 1, "1 2 3"},
 	}
 	for _, c := range cases {
-		result := cut(c.line, c.sep, c.column)
+		result := CutSep(c.line, c.sep, c.column)
 		if result != c.result {
 			t.Fatalf("got %v, want %v", result, c.result)
 		}