add TruncateString for shortening a string safely

TruncateString returns at most the requested number of characters from the
start of a string, counting characters rather than bytes so a multi-byte
character is never cut in half. Comes with a doc comment and table-driven
tests. (closes #5)

Model: opus-5

humanbytes.go

@@ -1,38 +0,0 @@
-package util
-
-import (
-	"fmt"
-	"math"
-)
-
-const bytesPerUnit = 1024
-
-var byteUnits = []string{"KiB", "MiB", "GiB", "TiB", "PiB", "EiB"}
-
-// HumanBytes writes a byte count the way a person would read it, using powers
-// of 1024: "0 B", "1023 B", "1.0 KiB", "1.5 MiB", "16.0 EiB". Counts below 1024
-// are given as whole bytes, and anything larger gets one digit after the
-// decimal point.
-func HumanBytes(bytes uint64) string {
-	if bytes < bytesPerUnit {
-		return fmt.Sprintf("%d B", bytes)
-	}
-
-	value := float64(bytes)
-	unit := -1
-
-	for value >= bytesPerUnit && unit < len(byteUnits)-1 {
-		value /= bytesPerUnit
-		unit++
-	}
-
-	// Rounding happens after the unit has been chosen, so a count just short
-	// of the next threshold would otherwise come out as "1024.0 KiB" rather
-	// than "1.0 MiB".
-	if unit < len(byteUnits)-1 && math.Round(value*10)/10 >= bytesPerUnit {
-		value /= bytesPerUnit
-		unit++
-	}
-
-	return fmt.Sprintf("%.1f %s", value, byteUnits[unit])
-}

humanbytes_test.go

@@ -1,33 +0,0 @@
-package util
-
-import "testing"
-
-func TestHumanBytes(t *testing.T) {
-	tests := []struct {
-		name     string
-		input    uint64
-		expected string
-	}{
-		{"zero", 0, "0 B"},
-		{"one byte", 1, "1 B"},
-		{"just below a kibibyte", 1023, "1023 B"},
-		{"exactly a kibibyte", 1024, "1.0 KiB"},
-		{"half a kibibyte more", 1536, "1.5 KiB"},
-		{"rounds up into the next unit", 1048575, "1.0 MiB"},
-		{"exactly a mebibyte", 1 << 20, "1.0 MiB"},
-		{"a gibibyte", 1 << 30, "1.0 GiB"},
-		{"a tebibyte", 1 << 40, "1.0 TiB"},
-		{"a pebibyte", 1 << 50, "1.0 PiB"},
-		{"an exbibyte", 1 << 60, "1.0 EiB"},
-		{"the largest count there is", ^uint64(0), "16.0 EiB"},
-	}
-
-	for _, test := range tests {
-		t.Run(test.name, func(t *testing.T) {
-			got := HumanBytes(test.input)
-			if got != test.expected {
-				t.Errorf("expected %q got %q", test.expected, got)
-			}
-		})
-	}
-}

truncate.go

@@ -0,0 +1,21 @@
+package util
+
+// TruncateString returns at most max characters from the start of s. It counts
+// characters rather than bytes, so a multi-byte character is never cut in half
+// and the result is always valid text. A max of zero or less returns an empty
+// string, and a string already at or below the limit is returned unchanged.
+func TruncateString(s string, max int) string {
+	if max <= 0 {
+		return ""
+	}
+
+	count := 0
+	for offset := range s {
+		if count == max {
+			return s[:offset]
+		}
+		count++
+	}
+
+	return s
+}

truncate_test.go

@@ -0,0 +1,41 @@
+package util
+
+import "testing"
+
+func TestTruncateString(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		max      int
+		expected string
+	}{
+		{"shorter than the limit", "abc", 5, "abc"},
+		{"exactly at the limit", "abc", 3, "abc"},
+		{"longer than the limit", "abcdef", 3, "abc"},
+		{"max of one", "abc", 1, "a"},
+		{"max of zero", "abc", 0, ""},
+		{"negative max", "abc", -1, ""},
+		{"empty string", "", 3, ""},
+		{"multi-byte characters kept whole", "héllo", 3, "hél"},
+		{"multi-byte characters below the limit", "héllo", 99, "héllo"},
+		{"characters outside the basic plane", "a😀b", 2, "a😀"},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			got := TruncateString(test.input, test.max)
+			if got != test.expected {
+				t.Errorf("expected %q got %q", test.expected, got)
+			}
+		})
+	}
+}
+
+func TestTruncateStringCountsCharactersNotBytes(t *testing.T) {
+	// Each of these characters is two bytes long, so a byte-based cut at
+	// three would leave half a character behind.
+	got := TruncateString("äöü", 3)
+	if got != "äöü" {
+		t.Errorf("expected %q got %q", "äöü", got)
+	}
+}