fix(cli): ASCII glyph fallback for Windows console mojibake (#168) (#178)

The shimmer progress renderer writes from a worker thread via
`fs.writeSync(1, ...)` to keep the animation smooth while the main
thread is busy in SQLite. That path bypasses Node's TTY-aware
UTF-8->codepage conversion on Windows, so glyphs like `|`/`<>`/`-`
were emitted as raw UTF-8 bytes and reinterpreted by the console's
OEM codepage (CP437, CP936, ...), producing strings like
`鋍?[0m 鉒?[0m Scanning files 鈥?N found`.

Add `src/ui/glyphs.ts` with `supportsUnicode()` detection plus
matched Unicode + ASCII glyph sets, and route all CLI/shimmer
output through `getGlyphs()`. Defaults: ASCII on Windows and on
Linux kernel consoles (`TERM=linux`), Unicode everywhere else.
`CODEGRAPH_UNICODE=1` and `CODEGRAPH_ASCII=1` are escape hatches.

Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Colby Mchenry
2026-05-19 10:45:20 -05:00
committed by GitHub
co-authored by Claude Opus 4.7
parent 36c8dbc404
commit e176062c56
7 changed files with 322 additions and 34 deletions
+91
View File
@@ -0,0 +1,91 @@
/**
* Glyph selection for CLI output.
*
* On Windows, console output is interpreted via the active output
* codepage. PowerShell 5.1 and cmd.exe default to OEM codepages
* (CP437, CP936, ...), so UTF-8 bytes written to the console render
* as mojibake (see #168). The shimmer worker is hit hardest because
* it uses `fs.writeSync(1, ...)` (raw bytes, no TTY-aware encoding
* conversion) to keep animation smooth while the main thread is
* blocked in SQLite. To stay readable everywhere, we fall back to
* ASCII glyphs whenever the terminal is not known to handle UTF-8.
*
* Detection is intentionally simple:
* - `CODEGRAPH_ASCII=1` -> ASCII (escape hatch for any terminal)
* - `CODEGRAPH_UNICODE=1` -> Unicode (opt-in on Windows)
* - Windows -> ASCII by default
* - Linux kernel console (`TERM=linux`) -> ASCII
* - Everything else -> Unicode
*/
export function supportsUnicode(): boolean {
if (process.env.CODEGRAPH_ASCII === '1') return false;
if (process.env.CODEGRAPH_UNICODE === '1') return true;
if (process.platform === 'win32') return false;
return process.env.TERM !== 'linux';
}
export interface Glyphs {
ok: string;
err: string;
info: string;
warn: string;
spinner: string[];
barFilled: string;
barEmpty: string;
rail: string;
phaseDone: string;
dash: string;
hLine: string;
treeBranch: string;
treeLast: string;
treePipe: string;
}
export const UNICODE_GLYPHS: Glyphs = {
ok: '✓',
err: '✗',
info: '',
warn: '⚠',
spinner: ['·', '✢', '✳', '✶', '✻', '✽'],
barFilled: '█',
barEmpty: '░',
rail: '│',
phaseDone: '◆',
dash: '—',
hLine: '─',
treeBranch: '├── ',
treeLast: '└── ',
treePipe: '│ ',
};
export const ASCII_GLYPHS: Glyphs = {
ok: '[OK]',
err: '[ERR]',
info: '[i]',
warn: '[!]',
spinner: ['.', '*', '+', 'x', 'o', 'O'],
barFilled: '#',
barEmpty: '-',
rail: '|',
phaseDone: '*',
dash: '-',
hLine: '-',
treeBranch: '|-- ',
treeLast: '`-- ',
treePipe: '| ',
};
let cached: Glyphs | null = null;
export function getGlyphs(): Glyphs {
if (cached === null) {
cached = supportsUnicode() ? UNICODE_GLYPHS : ASCII_GLYPHS;
}
return cached;
}
/** Reset the cached glyph set. Test-only; production code should call `getGlyphs()`. */
export function _resetGlyphsCache(): void {
cached = null;
}
+17 -11
View File
@@ -1,5 +1,6 @@
import { parentPort, workerData } from 'worker_threads';
import { writeSync } from 'fs';
import { getGlyphs } from './glyphs';
import type { ShimmerWorkerMessage } from './types';
// Write directly to fd 1 (stdout) instead of writeStdout().
@@ -7,11 +8,16 @@ import type { ShimmerWorkerMessage } from './types';
// thread's event loop — so if the main thread is blocked (e.g. SQLite),
// stdout writes from the worker queue up and the animation freezes.
// fs.writeSync(1, ...) is a direct kernel syscall that bypasses this.
//
// Side effect: bypasses Node's TTY-aware encoding conversion on Windows,
// so UTF-8 bytes hit the console raw and mojibake on OEM codepages.
// `getGlyphs()` returns ASCII fallbacks on Windows to avoid this (#168).
function writeStdout(s: string): void {
writeSync(1, s);
}
const SPINNER_GLYPHS = ['·', '✢', '✳', '✶', '✻', '✽'];
const G = getGlyphs();
const SPINNER_GLYPHS = G.spinner;
const ANIM_INTERVAL = 150;
const FRAMES_PER_GLYPH = 3;
@@ -43,7 +49,7 @@ function formatNumber(n: number): string {
}
function renderBar(frame: number, filled: number, empty: number): string {
if (filled === 0) return `${DM}${'░'.repeat(empty)}${RST}`;
if (filled === 0) return `${DM}${G.barEmpty.repeat(empty)}${RST}`;
const cycleFrames = 24;
const shimmerPos = ((frame % cycleFrames) / cycleFrames) * (filled + 6) - 3;
const shimmerWidth = 3;
@@ -54,9 +60,9 @@ function renderBar(frame: number, filled: number, empty: number): string {
const r = lerp(160, 251, t);
const g = lerp(100, 191, t);
const b = lerp(9, 36, t);
bar += `\x1b[38;2;${r};${g};${b}m${BOLD}`;
bar += `\x1b[38;2;${r};${g};${b}m${BOLD}${G.barFilled}`;
}
bar += `${RST}${DM}${'░'.repeat(empty)}${RST}`;
bar += `${RST}${DM}${G.barEmpty.repeat(empty)}${RST}`;
return bar;
}
@@ -69,7 +75,7 @@ function render(): void {
if (!currentMessage) return;
const frame = animFrame();
const glyphIdx = Math.floor(frame / FRAMES_PER_GLYPH) % SPINNER_GLYPHS.length;
const glyph = SPINNER_GLYPHS[glyphIdx] ?? '·';
const glyph = SPINNER_GLYPHS[glyphIdx] ?? SPINNER_GLYPHS[0] ?? '.';
const color = shimmerColor(frame);
let line: string;
@@ -77,11 +83,11 @@ function render(): void {
const barWidth = 25;
const filled = Math.round(barWidth * currentPercent / 100);
const empty = barWidth - filled;
line = `${DM}${RST} ${color}${glyph}${RST} ${currentMessage} ${renderBar(frame, filled, empty)} ${currentPercent}%`;
line = `${DM}${G.rail}${RST} ${color}${glyph}${RST} ${currentMessage} ${renderBar(frame, filled, empty)} ${currentPercent}%`;
} else if (currentCount > 0) {
line = `${DM}${RST} ${color}${glyph}${RST} ${currentMessage}... ${formatNumber(currentCount)} found`;
line = `${DM}${G.rail}${RST} ${color}${glyph}${RST} ${currentMessage}... ${formatNumber(currentCount)} found`;
} else {
line = `${DM}${RST} ${color}${glyph}${RST} ${currentMessage}...`;
line = `${DM}${G.rail}${RST} ${color}${glyph}${RST} ${currentMessage}...`;
}
writeStdout(`\r\x1b[K${line}`);
@@ -91,9 +97,9 @@ function finishPhase(): void {
if (!currentMessage) return;
writeStdout(`\r\x1b[K`);
let detail = '';
if (currentPercent >= 0) detail = ' — done';
else if (currentCount > 0) detail = ` ${formatNumber(currentCount)} found`;
writeStdout(`${DM}${RST} ${GRN}${RST} ${currentMessage}${detail}\n`);
if (currentPercent >= 0) detail = ` ${G.dash} done`;
else if (currentCount > 0) detail = ` ${G.dash} ${formatNumber(currentCount)} found`;
writeStdout(`${DM}${G.rail}${RST} ${GRN}${G.phaseDone}${RST} ${currentMessage}${detail}\n`);
currentMessage = '';
currentPercent = -1;
currentCount = 0;