starred/opentui

Fork 0

mirror of https://github.com/anomalyco/opentui.git synced 2026-04-25 13:06:00 +03:00

[GH-ISSUE #609] lineStarts returns column offset instead of byte offset when wrapping multi-byte text #165

New issue

Open

opened 2026-03-02 23:45:00 +03:00 by kerem · 1 comment

kerem commented

2026-03-02 23:45:00 +03:00

Owner

Originally created by @zenyr on GitHub (Feb 1, 2026).
Original GitHub issue: https://github.com/anomalyco/opentui/issues/609

Problem

After #255 fix, CJK/emoji text renders correctly in most cases. However, lineInfo.lineStarts still returns column offsets instead of byte offsets when text wraps. This causes character corruption at line boundaries.

Korean text wrap (representative case)

buffer.setStyledText(stringToStyledText("흐름도")) // 9 bytes
view.setWrapWidth(4) // wraps after "흐름"

	`lineStarts`	Line 0	Line 1
AS-IS	`[0, 4]`	`"흐�"`	`"��도"`
TO-BE	`[0, 6]`	`"흐름"`	`"도"`

Chinese text wrap

buffer.setStyledText(stringToStyledText("你好世界")) // 12 bytes
view.setWrapWidth(4)

	`lineStarts`	Line 0	Line 1
AS-IS	`[0, 4]`	`"你�"`	`"��世界"`
TO-BE	`[0, 6]`	`"你好"`	`"世界"`

Korean with leading ASCII

buffer.setStyledText(stringToStyledText(" 안녕a")) // 8 bytes
view.setWrapWidth(5)

	`lineStarts`	Line 0	Line 1
AS-IS	`[0, 5]`	`" 안�"`	`"��a"`
TO-BE	`[0, 7]`	`" 안녕"`	`"a"`

Flag emoji

buffer.setStyledText(stringToStyledText("🇰🇷🇯🇵🇨🇳")) // 24 bytes
view.setWrapWidth(4)

	`lineStarts`	Line 0	Line 1
AS-IS	`[0, 4]`	`"🇰"`	`"🇷🇯🇵🇨🇳"`
TO-BE	`[0, 16]`	`"🇰🇷🇯🇵"`	`"🇨🇳"`

Skin tone emoji

buffer.setStyledText(stringToStyledText("👋🏻👋🏿hi")) // 18 bytes
view.setWrapWidth(4)

	`lineStarts`	Line 0	Line 1	Line 2
AS-IS	`[0, 4, 8]`	`"👋"`	`"🏻"`	`"👋🏿hi"`
TO-BE	`[0, 8, 16]`	`"👋🏻"`	`"👋🏿"`	`"hi"`

AS-IS splits skin tone modifier from base emoji.

Mixed keycap + CJK

buffer.setStyledText(stringToStyledText("1️⃣한글")) // 13 bytes
view.setWrapWidth(4)

	`lineStarts`	Line 0	Line 1
AS-IS	`[0, 3]`	`"1�"`	`"�⃣한글"`
TO-BE	`[0, 10]`	`"1️⃣한"`	`"글"`

Long Korean (multiple wraps)

buffer.setStyledText(stringToStyledText("가나다라마바사")) // 21 bytes
view.setWrapWidth(6)

	`lineStarts`	Line 0	Line 1	Line 2
AS-IS	`[0, 6, 12]`	`"가나"`	`"다라"`	`"마바사"`
TO-BE	`[0, 9, 18]`	`"가나"`	`"다라"`	`"마바사"`

Happens to display correctly but byte offsets are still wrong (6≠9, 12≠18).

Root Cause

In text-buffer-view.zig, cached_line_starts was populated using char_offset (display column width) instead of tracking actual byte position through the wrap logic.

#255

Originally created by @zenyr on GitHub (Feb 1, 2026). Original GitHub issue: https://github.com/anomalyco/opentui/issues/609 ### Problem After #255 fix, CJK/emoji text renders correctly in most cases. However, `lineInfo.lineStarts` still returns **column offsets** instead of **byte offsets** when text wraps. This causes character corruption at line boundaries. <details open> <summary>Korean text wrap (representative case)</summary> ```ts buffer.setStyledText(stringToStyledText("흐름도")) // 9 bytes view.setWrapWidth(4) // wraps after "흐름" ``` | | `lineStarts` | Line 0 | Line 1 | | ----- | ---------- | ------ | ------ | | **AS-IS** | `[0, 4]` | `"흐�"` | `"��도"` | | **TO-BE** | `[0, 6]` | `"흐름"` | `"도"` | </details> <details> <summary>Chinese text wrap</summary> ```ts buffer.setStyledText(stringToStyledText("你好世界")) // 12 bytes view.setWrapWidth(4) ``` | | `lineStarts` | Line 0 | Line 1 | | ----- | ---------- | ------ | -------- | | **AS-IS** | `[0, 4]` | `"你�"` | `"��世界"` | | **TO-BE** | `[0, 6]` | `"你好"` | `"世界"` | </details> <details> <summary>Korean with leading ASCII</summary> ```ts buffer.setStyledText(stringToStyledText(" 안녕a")) // 8 bytes view.setWrapWidth(5) ``` | | `lineStarts` | Line 0 | Line 1 | | ----- | ---------- | ------- | ------ | | **AS-IS** | `[0, 5]` | `" 안�"` | `"��a"` | | **TO-BE** | `[0, 7]` | `" 안녕"` | `"a"` | </details> <details> <summary>Flag emoji</summary> ```ts buffer.setStyledText(stringToStyledText("🇰🇷🇯🇵🇨🇳")) // 24 bytes view.setWrapWidth(4) ``` | | `lineStarts` | Line 0 | Line 1 | | ----- | ---------- | ------ | ------- | | **AS-IS** | `[0, 4]` | `"🇰"` | `"🇷🇯🇵🇨🇳"` | | **TO-BE** | `[0, 16]` | `"🇰🇷🇯🇵"` | `"🇨🇳"` | </details> <details> <summary>Skin tone emoji</summary> ```ts buffer.setStyledText(stringToStyledText("👋🏻👋🏿hi")) // 18 bytes view.setWrapWidth(4) ``` | | `lineStarts` | Line 0 | Line 1 | Line 2 | | ----- | ---------- | ------ | ------ | ------ | | **AS-IS** | `[0, 4, 8]` | `"👋"` | `"🏻"` | `"👋🏿hi"` | | **TO-BE** | `[0, 8, 16]` | `"👋🏻"` | `"👋🏿"` | `"hi"` | AS-IS splits skin tone modifier from base emoji. </details> <details> <summary>Mixed keycap + CJK</summary> ```ts buffer.setStyledText(stringToStyledText("1️⃣한글")) // 13 bytes view.setWrapWidth(4) ``` | | `lineStarts` | Line 0 | Line 1 | | ----- | ---------- | ------ | -------- | | **AS-IS** | `[0, 3]` | `"1�"` | `"�⃣한글"` | | **TO-BE** | `[0, 10]` | `"1️⃣한"` | `"글"` | </details> <details> <summary>Long Korean (multiple wraps)</summary> ```ts buffer.setStyledText(stringToStyledText("가나다라마바사")) // 21 bytes view.setWrapWidth(6) ``` | | `lineStarts` | Line 0 | Line 1 | Line 2 | | ----- | ---------- | ------ | ------ | -------- | | **AS-IS** | `[0, 6, 12]` | `"가나"` | `"다라"` | `"마바사"` | | **TO-BE** | `[0, 9, 18]` | `"가나"` | `"다라"` | `"마바사"` | Happens to display correctly but byte offsets are still wrong (6≠9, 12≠18). </details> ### Root Cause In `text-buffer-view.zig`, `cached_line_starts` was populated using `char_offset` (display column width) instead of tracking actual byte position through the wrap logic. ### Related - #255

kerem added the

core

bug

labels

2026-03-02 23:45:00 +03:00

kerem commented

2026-03-02 23:45:01 +03:00

Author

Owner

@simonklee commented on GitHub (Mar 2, 2026):

tests/fix-609-cases

@simonklee commented on GitHub (Mar 2, 2026): [tests/fix-609-cases](https://github.com/anomalyco/opentui/tree/tests/fix-609-cases)

kerem referenced this issue

2026-03-02 23:46:02 +03:00

[PR #165] [CLOSED] Custom solid renderer #340

kerem referenced this issue

2026-03-14 09:19:26 +03:00

[PR #165] [CLOSED] Custom solid renderer #1119