mirror of
https://github.com/pbakaus/impeccable.git
synced 2026-09-19 17:46:36 +03:00
Paul BakausandClaude Fable 5.1
1d214f1e48
An author who wants a rule off for one component has had two choices: put `data-impeccable-ignore` on every instance, or silence the rule (or the file) for the whole project. On impeccable-site #34 that meant eleven attributes on eleven copies of the same 10px label for `undersized-ui-text`, 25 opt-out attributes in all. The count is the problem: the markup carries noise, and the reviewer never sees how much was waived. `detector.ignoreSelectors` is the declared twin of that attribute. One entry, `{ rule, selector }`, waives the rule for every element the selector matches and for that element's subtree, the same waiver the attribute grants the element carrying it: impeccable ignores add-selector undersized-ui-text ".ks-tag" \ --reason "10px mono index labels, confirmed" The waiver is never silent. The engines stamp a waived finding with `ignoredBy: "<selector>"` instead of dropping it; the config layer drops and counts, and every scan prints one line per entry on stderr, in `--json` runs too, so stdout stays the findings array: 3 undersized-ui-text hits ignored by detector.ignoreSelectors on .ks-tag. Where it applies: the browser engine (`BrowserConfig.ignoreSelectors`, also readable from `window.__IMPECCABLE_CONFIG__`), the static HTML engine (`DetectHtmlOptions.ignore_selectors`, and the `ignoreSelectors` option of the wasm `detect_html_source_json` export), the detect CLI, and the design hook. The text engine has no DOM and ignores the key. Entries can be scoped with `files` globs like `ignoreValues`; `--no-config` disables them; `doctor` validates their rule ids alongside `ignoreRules`. Nothing changes for a project without the key: the engines stamp nothing, the CLI prints nothing, the config writer does not add an empty `ignoreSelectors`, and the per-instance attribute keeps working exactly as before. Coverage: `crates/html/tests/selector_ignores.rs` (component, subtree, wrong-rule, `*`, attribute parity), driver tests over the fake DOM, `crates/detect` config tests (normalize, merge, per-target narrowing, the tally), and oracle cases `detect-selector-ignore-*` / `ignores-selector-*` over a new workspace. The eight re-recorded context/doctor goldens differ only in the recognized-detector-keys sentence, which now lists the new key. Assisted-by: Claude Code Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LQBUunp8QttxZqihybNmtL
8 lines
3.4 KiB
JSON
8 lines
3.4 KiB
JSON
{
|
|
"stdout": "Usage: impeccable detect [options] [file-or-dir-or-url...]\n\nScan files or URLs for UI anti-patterns and design quality issues.\n\nOptions:\n --json Output results as JSON\n --quiet In text mode, only print the final findings count\n --scope <name> Only report rules in the given design domain\n (type, layout). Comma-separated.\n --viewport <WxH> Browser viewport for URL scans (default 1280x800),\n e.g. --viewport 390x844 for a mobile-width pass\n --no-config Do not apply project config, detector ignores, inline\n ignore comments, or DESIGN.md\n --no-inline-ignores Do not honor in-file impeccable-disable* ignore comments\n --no-design-system Do not load local DESIGN.md / .impeccable/design.json context\n --no-advisory Suppress advisory findings entirely (e.g. em-dash overuse)\n --help Show this help message\n\nAdvisory findings:\n Some rules are advisory: detected and listed in a separate section, but never\n counted as failures and never changing the exit code. They stay out of the\n failure count so they never block automation. --no-advisory hides them.\n\nOutput streams:\n Human-readable findings go to stderr so stdout stays available for structured\n output. Use --json for JSON on stdout, or redirect text with 2> findings.txt.\n\nExit status:\n 0 Scan completed with no primary findings (advisories may still be listed)\n 1 At least one requested target could not be scanned\n 2 Scan completed with primary findings\n Operational failure takes precedence when a multi-target scan is partial.\n\nProject config:\n Respects .impeccable/config.json and .impeccable/config.local.json detector\n settings: detector.ignoreRules, detector.ignoreFiles, detector.ignoreValues,\n detector.ignoreSelectors, and detector.designSystem.enabled.\n\nComponent ignores:\n detector.ignoreSelectors waives one rule for every element a CSS selector\n matches, and for that element's subtree: one entry for a component instead\n of a data-impeccable-ignore attribute on each of its instances. Write one\n with `impeccable ignores add-selector <rule> \"<selector>\"`. Every scan\n prints what it suppressed on stderr, in --json runs too, so the exception\n stays visible:\n 3 undersized-ui-text hits ignored by detector.ignoreSelectors on .ks-tag.\n\nInline ignores:\n In-file comments waive a finding where it lives and travel with the file:\n <!-- impeccable-disable overused-font -- exported brand doc -->\n .brand { font-family: Inter } /* impeccable-disable-line overused-font */\n // impeccable-disable-next-line bounce-easing: intentional bounce\n impeccable-disable applies to the whole file; -line / -next-line are scoped.\n List one or more rule ids (comma-separated), or omit them / use * for all.\n\nDetection modes:\n HTML files Static HTML/CSS analysis (default, catches linked CSS)\n Non-HTML files Regex pattern matching (CSS, JSX, TSX, etc.)\n URLs Puppeteer full browser rendering (auto-detected;\n http(s):// and file:// URLs; accessible linked CSS included)\n\nExamples:\n impeccable detect src/\n impeccable detect index.html\n impeccable detect https://example.com\n impeccable detect --json .\n impeccable detect --no-config src/\n",
|
|
"stderr": "",
|
|
"exit": 0,
|
|
"signal": null,
|
|
"files": {}
|
|
}
|