reduce: temp bug fix for line cutter

we wanted to trim whitespace at one point, because values contained the
separator values; however this breaks with empty values; move back to
not trimming values except for the newline, when requesting the last
value; moving forward, we need to clean or reject dirty values or use a
different delimiter

2 files changed, 61 insertions, 32 deletions

diff --git a/skate/reduce.go b/skate/reduce.go
index 9986152..eac2e83 100644
--- a/skate/reduce.go
+++ b/skate/reduce.go
@@ -598,10 +598,14 @@ func Cut(line string, column int) string {
 
 // CutSep allows to specify a separator, column is 1-indexed.
 func CutSep(line, sep string, column int) string {
-	parts := strings.Split(strings.TrimSpace(line), sep)
+	// XXX: This will cut the tab separator, if there is no other value.
+	parts := strings.Split(line, sep)
 	if len(parts) < column {
 		return ""
 	} else {
+		if len(parts) == column {
+			return strings.TrimSuffix(parts[column-1], "\n")
+		}
 		return parts[column-1]
 	}
 }
diff --git a/skate/reduce_test.go b/skate/reduce_test.go
index 4db0687..99c0ed7 100644
--- a/skate/reduce_test.go
+++ b/skate/reduce_test.go
@@ -34,37 +34,6 @@ func TestLineColumn(t *testing.T) {
 	}
 }
 
-func TestCutBatch(t *testing.T) {
-	var cases = []struct {
-		lines  []string
-		column int
-		result []string
-	}{
-		{
-			[]string{},
-			1,
-			nil,
-		},
-		{
-			[]string{},
-			9,
-			nil,
-		},
-		{
-			[]string{"1\t2\n", "3\t4\n"},
-			2,
-			[]string{"2", "4"},
-		},
-	}
-	for _, c := range cases {
-		result := CutBatch(c.lines, c.column)
-		if !reflect.DeepEqual(result, c.result) {
-			t.Fatalf("got %v (%d), want %v (%d)",
-				result, len(result), c.result, len(c.result))
-		}
-	}
-}
-
 func TestUniqueMatches(t *testing.T) {
 	var cases = []struct {
 		about  string
@@ -453,3 +422,59 @@ func tempWriteFile(buf *bytes.Buffer) (string, error) {
 	}
 	return f.Name(), nil
 }
+
+func TestCutBatch(t *testing.T) {
+	var cases = []struct {
+		lines  []string
+		column int
+		result []string
+	}{
+		{
+			[]string{},
+			1,
+			nil,
+		},
+		{
+			[]string{},
+			9,
+			nil,
+		},
+		{
+			[]string{"1\t2\n", "3\t4\n"},
+			2,
+			[]string{"2", "4"},
+		},
+	}
+	for _, c := range cases {
+		result := CutBatch(c.lines, c.column)
+		if !reflect.DeepEqual(result, c.result) {
+			t.Fatalf("got %v (%d), want %v (%d)",
+				result, len(result), c.result, len(c.result))
+		}
+	}
+}
+
+func TestCutSep(t *testing.T) {
+	var cases = []struct {
+		line   string
+		sep    string
+		column int
+		result string
+	}{
+		{"", "\t", 1, ""},
+		{"", "\t", 2, ""},
+		{"a\tb", "\t", 1, "a"},
+		{"a\tb", "\t", 2, "b"},
+		{"a\tb", "\t", 3, ""},
+		{"a\t\tb", "\t", 1, "a"},
+		{"a\t\tb", "\t", 2, ""},
+		{"a\t\tb", "\t", 3, "b"},
+		{"\tb", "\t", 1, ""},
+	}
+	for _, c := range cases {
+		result := CutSep(c.line, c.sep, c.column)
+		if !reflect.DeepEqual(result, c.result) {
+			t.Fatalf("got %v, want %v", result, c.result)
+		}
+	}
+}