[perf] Optimize AWK-style tokenizer for --nth

Approx. 50% less memory footprint and 40% improvement in query time

src/util/chars.go

@@ -111,3 +111,10 @@ func (chars *Chars) ToRunes() []rune {
 	}
 	return runes
 }
+
+func (chars *Chars) Slice(b int, e int) Chars {
+	if chars.runes != nil {
+		return Chars{runes: chars.runes[b:e]}
+	}
+	return Chars{bytes: chars.bytes[b:e]}
+}

src/util/chars_test.go

@@ -34,3 +34,24 @@ func TestCharsToString(t *testing.T) {
 		t.Error()
 	}
 }
+
+func TestTrimLength(t *testing.T) {
+	check := func(str string, exp int) {
+		chars := ToChars([]byte(str))
+		trimmed := chars.TrimLength()
+		if trimmed != exp {
+			t.Errorf("Invalid TrimLength result for '%s': %d (expected %d)",
+				str, trimmed, exp)
+		}
+	}
+	check("hello", 5)
+	check("hello ", 5)
+	check("hello  ", 5)
+	check(" hello", 5)
+	check("  hello", 5)
+	check(" hello ", 5)
+	check("  hello  ", 5)
+	check("h   o", 5)
+	check("  h   o  ", 5)
+	check("         ", 0)
+}

src/util/util.go

@@ -83,30 +83,6 @@ func IsTty() bool {
 	return int(C.isatty(C.int(os.Stdin.Fd()))) != 0
 }
 
-// TrimLen returns the length of trimmed rune array
-func TrimLen(runes []rune) int {
-	var i int
-	for i = len(runes) - 1; i >= 0; i-- {
-		char := runes[i]
-		if char != ' ' && char != '\t' {
-			break
-		}
-	}
-	// Completely empty
-	if i < 0 {
-		return 0
-	}
-
-	var j int
-	for j = 0; j < len(runes); j++ {
-		char := runes[j]
-		if char != ' ' && char != '\t' {
-			break
-		}
-	}
-	return i - j + 1
-}
-
 // ExecCommand executes the given command with $SHELL
 func ExecCommand(command string) *exec.Cmd {
 	shell := os.Getenv("SHELL")

src/util/util_test.go

@@ -20,23 +20,3 @@ func TestContrain(t *testing.T) {
 		t.Error("Expected", 3)
 	}
 }
-
-func TestTrimLen(t *testing.T) {
-	check := func(str string, exp int) {
-		trimmed := TrimLen([]rune(str))
-		if trimmed != exp {
-			t.Errorf("Invalid TrimLen result for '%s': %d (expected %d)",
-				str, trimmed, exp)
-		}
-	}
-	check("hello", 5)
-	check("hello ", 5)
-	check("hello  ", 5)
-	check(" hello", 5)
-	check("  hello", 5)
-	check(" hello ", 5)
-	check("  hello  ", 5)
-	check("h   o", 5)
-	check("  h   o  ", 5)
-	check("         ", 0)
-}