add CollapseWhitespace for flattening runs of whitespace

CollapseWhitespace turns every run of whitespace into a single space and
trims the ends, so a block of text becomes one tidy line. Comes with a doc
comment and table-driven tests. (closes #7)

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)
-			}
-		})
-	}
-}

whitespace.go

@@ -0,0 +1,12 @@
+package util
+
+import "strings"
+
+// CollapseWhitespace replaces every run of whitespace in s with a single space
+// and removes any whitespace at the start and end. Tabs, newlines and spaces
+// outside ASCII such as the non-breaking space all count as whitespace, so a
+// block of text comes back as one line with single spaces between the words.
+// A string that is nothing but whitespace comes back empty.
+func CollapseWhitespace(s string) string {
+	return strings.Join(strings.Fields(s), " ")
+}

whitespace_test.go

@@ -0,0 +1,36 @@
+package util
+
+import "testing"
+
+// Written as a code point so that it is visible in the source rather than
+// looking like an ordinary space.
+const nonBreakingSpace = string(rune(0x00a0))
+
+func TestCollapseWhitespace(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected string
+	}{
+		{"nothing to change", "one two", "one two"},
+		{"leading and trailing spaces", "  one two  ", "one two"},
+		{"run in the middle", "one    two", "one two"},
+		{"tabs and newlines", "one\t\ttwo\nthree", "one two three"},
+		{"windows line endings", "one\r\ntwo", "one two"},
+		{"non-breaking space", "one" + nonBreakingSpace + "two", "one two"},
+		{"mixed whitespace run", "one \t" + nonBreakingSpace + " two", "one two"},
+		{"only whitespace", " \t\n ", ""},
+		{"empty string", "", ""},
+		{"single word with padding", "\tword\n", "word"},
+		{"paragraph becomes one line", "first line\n\nsecond line\n", "first line second line"},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			got := CollapseWhitespace(test.input)
+			if got != test.expected {
+				t.Errorf("expected %q got %q", test.expected, got)
+			}
+		})
+	}
+}