Unicode characters

A grid of 1795 characters one pastes rather than types, from extensions/unicode/data.json: arrows, math, Greek, currency, quotes and dashes, punctuation, typographic and zero-width spaces, superscripts and fractions, accented Latin letters (the Turkish, German, French, Nordic and Spanish sets), letterlike symbols, the Mac keyboard glyphs (⌘ ⌥ ⇧ ⌃ ⎋ ⏎ ⌫ ⇥ ⏏ and the control pictures), check marks, box drawing and block elements, geometric shapes, miscellaneous symbols, dingbats, enclosed numbers and letters.

The tile is the glyph (a space shows as the open box ␣); the name is the Unicode name in lower case; the code point is the subtitle; the section is the block. The search matches the name, the code point (2192 or u+2192), the HTML entity names (rarr, nbsp), the LaTeX command where there is an obvious one (\rightarrow, \alpha), the character's Unicode 1.0 name (command key) and plain words (cmd, turkish, tick, eur).

The last twelve characters copied lead the grid in a Recent section (kept in the extension's storage) and leave their block while they are there. The section is rebuilt when the palette lists again (at start, on cmd+r); between listings the core's frecency already moves a picked tile up.

The detail pane (cmd+i) shows the glyph large, the block, the code point, both HTML forms, the UTF-8 bytes (E2 86 92), the LaTeX command and the aliases.

Keyboard

In the grid, arrows move across and down the tiles and Enter runs the first action of the tile under the cursor.

Setup

Nothing to install. Pasting is a synthesised Cmd+V, which macOS only delivers from a process on the Accessibility list: without the permission pal shows the system prompt once per run and a toast saying what to grant (System Settings > Privacy & Security > Accessibility). Copying needs nothing. On Linux the paste is Ctrl+V through wtype, else ydotool (which needs ydotoold running).

The table is generated: bun run extensions/unicode/build.ts fetches UnicodeData.txt (names) and the WHATWG entities.json (entity names), applies the sections, LaTeX and keyword tables in the script, and writes data.json (160 KB), which is committed. The whole UCD is not shipped; add a range or a code point to the script's SECTIONS to widen the table.

No extension settings. Per palette, [palettes.unicode.settings]:

What it does not do

• Every Unicode character: 1795 curated ones, by block. A code point outside the table finds nothing; widen SECTIONS in build.ts.
• Emoji: they have their own palette (Emoji), with shortcodes and skin tones.
• Compose characters (a base letter plus a combining mark); the accented letters are the precomposed ones.
• Type a code point to get its character as an input palette would; the grid is searched, so 2192 finds the arrow's tile.

Platforms

macOS and Linux. The glyphs are drawn by the system font, so a character the font lacks shows as its fallback box; paste needs Accessibility on macOS and wtype or ydotool on Linux.