anomalyco/opentui v0.6.2

Release v0.6.2

on GitHub

In 0.6.1, editors started to show control characters as notation. As a result,
an editor offset is no longer the same as a JavaScript string width. The string
"\f" has a width of 0, but the editor shows it as ^L in two cells.
EditBuffer.measureText() gives the correct offset, but only for a buffer you
already have. To get the offset of text that was not in an editor yet, code had
to create an EditBuffer only to measure that text.

In 0.6.2, measureEditText() measures text without a buffer. It returns the
same value as measureText() on an edit buffer with the same width method and
tab width. Pass the settings of the editor that will hold the text. Textarea
and Input use the renderer's width method and the default tab width, which is
2:

import { createCliRenderer, measureEditText } from "@opentui/core"

const renderer = await createCliRenderer()
measureEditText("\f@file", { widthMethod: renderer.widthMethod }) // 7: ^L takes two cells
measureEditText("a\tb", { widthMethod: renderer.widthMethod, tabWidth: 4 }) // 6

C programs get the same measurement from ot_edit_measure_text(). It takes a
context, but no edit buffer handle.

Added

• core: measureEditText(text, { widthMethod, tabWidth }) returns the
  editor offset at the end of text without an EditBuffer. tabWidth
  defaults to the tab width of a new edit buffer. See
  Editing buffers and views.
  (3ab0b954a)
• native: ot_edit_measure_text() gives the same measurement through the C
  interface. A tab_width over 255 fails. See
  Editing buffers and views.
  (3ab0b954a)

Fixed

• native: A bold, dim, italic, underlined, blinking, inverse, hidden, or
  struck-through cell in the last column of a row no longer gives its
  attributes to the next cells that the renderer draws. A row that ends in such
  a cell writes 4 more bytes. See Frames and output.
  (#1652)
• native: If you type an emoji with a skin-tone modifier, a ZWJ sequence,
  Hangul jamo, or U+FE0F in an editor, the editor now stores its correct width.
  Before, 👍🏽 took 4 cells instead of 2, so the cursor, wrapping, selection, and
  Right key were wrong. See Textarea.
  (#1650)
• native: A combining mark that starts a new text chunk now draws on the
  character before it, for example U+0301 typed after abc in an editor or
  added with TextBuffer.append. The renderer drops a mark that would make a
  1-cell character 2 cells wide, such as U+FE0F. See
  Text and terminal cells.
  (#1651)
• native: An embedded terminal redraws all rows when the child changes the
  default foreground or background color (OSC 10, 11, 110, or 111) or sets
  reverse video on or off (CSI ? 5 h or CSI ? 5 l). Before, rows on the
  screen kept the old colors. See
  Embedded terminal.
  (#1649)
• core: If the host terminal uses the Kitty keyboard protocol, an embedded
  terminal now sends keypad Enter, keypad navigation, lock, Print Screen, Pause,
  and left and right modifier keys to the child. Keypad digits reach a Kitty
  child as keypad keys, and Ctrl+keypad 0 no longer arrives as a bare 0. See
  Embedded terminal.
  (#1648)
• core: parseKeypress() no longer changes the caller's Buffer when it
  decodes a single byte above 127 as a meta key. The input parser of the
  renderer did not have this problem. See
  Keyboard input.
  (#1653)

API changes and documentation: https://opentui.com/docs/releases/0.6.2

What's Changed

• bench: time only drawBox in the box benchmark by @simonklee in #1643
• native: draw a combining mark that starts a chunk by @simonklee in #1651
• native: redraw embedded terminal when default colors change by @simonklee in #1649
• core: encode embedded terminal Kitty keypad, lock, and modifier keys by @simonklee in #1648
• native: measure merged editor chunks instead of adding their widths by @simonklee in #1650
• core: keep parseKeypress from changing the caller's Buffer by @simonklee in #1653
• native: close an attributed run at the end of its row by @simonklee in #1652

Full Changelog: v0.6.1...v0.6.2