Repository navigation
feat: terminal coverage (theme detection, 16 colours, ASCII mode, clipboard, doctor) - #20
Merged
Merged
Conversation
With no --theme and no config theme, ask the terminal for its background colour (OSC 11, bounded by a trailing DA1 request and a 100 ms timeout, 400 ms over SSH) before the TUI starts, in raw mode with the mode restored on every path. Parses rgb:R/G/B (1-4 hex digits) and #RGB forms; light when perceived luminance > 0.5. Precedence: flag/config > OSC 11 > COLORFGBG > dark. Skipped for TERM=dumb/linux, non-TTY stdin/stdout, and Windows outside Windows Terminal. COLORFGBG now treats index 7 as light; the dead TERM_PROGRAM branch (with its Ghostty casing bug) is gone.
… under screen clip.exe decodes its input with the console code page, so accented, CJK and emoji text arrived as mojibake on Windows and WSL. Feed it UTF-16LE with a byte-order mark instead. On WSL without clip.exe on PATH, fall back to PowerShell's Set-Clipboard reading UTF-8. Use termux-clipboard-set when TERMUX_VERSION is set. Helper selection and the bytes written are now a pure, tested function. Under GNU screen (STY set, TERM=screen*, no TMUX) the OSC 52 sequence is sent in short DCS chunks so screen forwards it to the outer terminal.
The file browser always enabled mouse capture, ignoring [behavior] mouse_capture = false (it has no mouse actions, so capture only took the terminal's own selection away). It now follows the same setting as the reader. New --no-mouse flag forces capture off for the session in both; the flag beats the config.
The ratatui-image probe waits up to ~2 s for terminals that never reply, so every start on them stalled. Skip it (half-blocks) when stdout is not a terminal, TERM is linux or dumb, or TERM_PROGRAM=Apple_Terminal. An explicit --image-protocol kitty|iterm2|sixel still applies, with an assumed cell size. Remove the catch_unwind around the probe: release builds abort on panic, so it never caught anything.
ink doctor now has [environment] (WSL, SSH, multiplexers, terminal and locale variables, UTF-8 locale, display presence), [colour] (depth and the signal that decided it, whether colour is on and why), [theme] (resolved theme and source, including the OSC 11 colour), [clipboard] (mode, native helper, OSC 52 wrapping), [input] (mouse capture and why) and [config] (path, existence, warnings), next to the existing graphics and decoder sections. Under tmux it shows allow-passthrough. Nothing identifying is printed: display and session variables are presence-only, SSH is yes/no, and the home directory is replaced with ~. Piped output skips terminal queries and says 'not a terminal'.
…r / NO_COLOR Colour depth is now truecolour, 256 or ANSI 16. TERM=linux, xterm, vt100, ansi, screen, cygwin (and any TERM without 256color and no other colour signal) get the 16 ANSI colours, as named colours so the terminal's own scheme applies; *-256color terminals keep exactly the old behaviour, and tmux, a non-truecolor COLORTERM, Terminal.app or an unset TERM stay at 256. Pastel theme colours keep their hue instead of collapsing to grey. Wired through the reader and --plain (30-37/90-97 SGR codes). The reader used to ignore --color=never and NO_COLOR. It now draws with the terminal's default colours and attributes only, with selections and code-block hint labels in reverse video. The unused adapt() helper is gone.
… readers Box borders, bullets, task boxes, quote bars, heading markers, rules, ellipses, image markers, popup borders, status-bar separators, the progress bar and mermaid diagrams now come from one Glyphs set (Unicode or ASCII), chosen once at startup: --ascii, [behavior] ascii = true|false, or automatically for TERM=linux and non-UTF-8 locales (Unix). ASCII mode also turns off smart punctuation and emoji shortcodes and names images instead of drawing them in half-blocks, so an ASCII document prints as pure 7-bit ASCII. ink stats/outline use the same set; ink doctor reports it. Default (Unicode) output is byte-identical.
Merged
1 task
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Makes ink behave correctly on more terminals and operating systems.
COLORFGBG, dark.clip.exenow receives UTF-16LE with a BOM, so non-ASCII text survives on Windows and WSL. Adds a PowerShell fallback on WSL,termux-clipboard-seton Termux, and OSC 52 wrapping under GNU screen.--no-mouseflag; the file browser now honoursmouse_capture.--color=neverorNO_COLORthe reader uses no colour, with reverse video for selection and hints.--asciimode for the Linux console, legacy Windows fonts and screen readers: body, tables, code boxes, popups, status bar, browser and mermaid. Automatic forTERM=linuxand non-UTF-8 locales; also[behavior] ascii = true.TERM=linuxordumb, and in Apple Terminal.ink doctornow reports environment, colour, theme, clipboard, input and config, with home paths and session values redacted.Testing
main, including a new pty suite (tests/pty.rs, Unix only) that plays the terminal side: background replies, mouse sequences, startup time, ASCII-only reader output.Known issue, not fixed here
If a terminal answers none of the image-protocol queries, a leftover reader thread in ratatui-image can swallow the next keystroke. Real terminals answer DA1, so this is rare.
🤖 Generated with Claude Code
https://claude.ai/code/session_01UVcvmQrSXd2eyt32g4ko19