text: fix alignment issues with box/block chars (#389)

Author: jedib0t

text/string.go

@@ -76,17 +76,20 @@ func LongestLineLen(str string) int {
 	return maxLength
 }
 
-// OverrideRuneWidthEastAsianWidth can *probably* help with alignment, and
-// length calculation issues when dealing with Unicode character-set and a
-// non-English language set in the LANG variable.
+// OverrideRuneWidthEastAsianWidth overrides the East Asian width detection in
+// the runewidth library. This is primarily for advanced use cases.
 //
-// Set this to 'false' to force the "runewidth" library to pretend to deal with
-// English character-set. Be warned that if the text/content you are dealing
-// with contains East Asian character-set, this may result in unexpected
-// behavior.
+// Box drawing (U+2500-U+257F) and block element (U+2580-U+259F) characters
+// are automatically handled and always reported as width 1, regardless of
+// this setting, fixing alignment issues that previously required setting this
+// to false.
 //
-// References:
-// * https://github.com/mattn/go-runewidth/issues/64#issuecomment-1221642154
+// Setting this to false forces runewidth to treat all characters as if in an
+// English locale. Warning: this may cause East Asian characters (Chinese,
+// Japanese, Korean) to be incorrectly reported as width 1 instead of 2.
+//
+// See:
+// * https://github.com/mattn/go-runewidth/issues/64
 // * https://github.com/jedib0t/go-pretty/issues/220
 // * https://github.com/jedib0t/go-pretty/issues/204
 func OverrideRuneWidthEastAsianWidth(val bool) {
@@ -184,16 +187,28 @@ func RuneCount(str string) int {
 	return StringWidthWithoutEscSequences(str)
 }
 
-// RuneWidth returns the mostly accurate character-width of the rune. This is
-// not 100% accurate as the character width is usually dependent on the
-// typeface (font) used in the console/terminal. For ex.:
+// RuneWidth returns the display width of a rune. Width accuracy depends on
+// the terminal font, as character width is font-dependent. Examples:
 //
 //	RuneWidth('A') == 1
 //	RuneWidth('ツ') == 2
 //	RuneWidth('⊙') == 1
 //	RuneWidth('︿') == 2
 //	RuneWidth(0x27) == 0
+//
+// Box drawing (U+2500-U+257F) and block element (U+2580-U+259F) characters
+// are always treated as width 1, regardless of locale, to ensure proper
+// alignment in tables and progress indicators. This fixes incorrect width 2
+// reporting in East Asian locales (e.g., LANG=zh_CN.UTF-8).
+//
+// See:
+// * https://github.com/mattn/go-runewidth/issues/64
+// * https://github.com/jedib0t/go-pretty/issues/220
+// * https://github.com/jedib0t/go-pretty/issues/204
 func RuneWidth(r rune) int {
+	if (r >= 0x2500 && r <= 0x257F) || (r >= 0x2580 && r <= 0x259F) {
+		return 1
+	}
 	return rwCondition.RuneWidth(r)
 }
 

text/string_test.go

@@ -94,19 +94,33 @@ func TestOverrideRuneWidthEastAsianWidth(t *testing.T) {
 		rwCondition.EastAsianWidth = originalValue
 	}()
 
+	// Box drawing characters (U+2500-U+257F) are now always reported as width 1,
+	// regardless of the EastAsianWidth setting. This fixes alignment issues
+	// that previously occurred when LANG was set to something like 'zh_CN.UTF-8'.
+	// Previously, '╋' would be reported as width 2 when EastAsianWidth was true.
 	OverrideRuneWidthEastAsianWidth(true)
-	assert.Equal(t, 2, StringWidthWithoutEscSequences("╋"))
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("╋"), "Box drawing character should always be width 1, even when EastAsianWidth is true")
 	OverrideRuneWidthEastAsianWidth(false)
-	assert.Equal(t, 1, StringWidthWithoutEscSequences("╋"))
-
-	// Note for posterity. We want the length of the box drawing character to
-	// be reported as 1. However, with an environment where LANG is set to
-	// something like 'zh_CN.UTF-8', the value being returned is 2, which breaks
-	// text alignment/padding logic in this library.
-	//
-	// If a future version of runewidth is able to address this internally and
-	// return 1 for the above, the function being tested can be marked for
-	// deprecation.
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("╋"), "Box drawing character should always be width 1, even when EastAsianWidth is false")
+
+	// Verify that block elements (U+2580-U+259F) are also handled correctly.
+	// These are used in progress indicators and should always be width 1.
+	OverrideRuneWidthEastAsianWidth(true)
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("█"), "Block element should always be width 1, even when EastAsianWidth is true")
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("▒"), "Block element should always be width 1, even when EastAsianWidth is true")
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("▓"), "Block element should always be width 1, even when EastAsianWidth is true")
+	OverrideRuneWidthEastAsianWidth(false)
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("█"), "Block element should always be width 1, even when EastAsianWidth is false")
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("▒"), "Block element should always be width 1, even when EastAsianWidth is false")
+	assert.Equal(t, 1, StringWidthWithoutEscSequences("▓"), "Block element should always be width 1, even when EastAsianWidth is false")
+
+	// Verify that actual East Asian characters are still handled correctly.
+	// Note: The runewidth library reports 'ツ' as width 2 regardless of the
+	// EastAsianWidth setting, as it's inherently a full-width character.
+	OverrideRuneWidthEastAsianWidth(true)
+	assert.Equal(t, 2, StringWidthWithoutEscSequences("ツ"), "East Asian character should be width 2")
+	OverrideRuneWidthEastAsianWidth(false)
+	assert.Equal(t, 2, StringWidthWithoutEscSequences("ツ"), "East Asian character should be width 2")
 }
 
 func ExamplePad() {