paperclip/scripts/codemod-extract-type.mjs

240 lines
11 KiB
JavaScript

#!/usr/bin/env node
/**
* codemod-extract-type.mjs
*
* Phase 2 (extraction), Batch 2/4 of the design-token audit
* (branch design/token-extraction). Replaces hardcoded TYPE values —
* arbitrary Tailwind font-size (`text-[11px]`), letter-spacing
* (`tracking-[0.18em]`), line-height (`leading-[...]`), and raw inline
* `fontSize` style literals — in `ui/src/components/**` and
* `ui/src/pages/**` (including their *.test.tsx companions) with
* references to CSS custom-property tokens defined in `ui/src/index.css`.
*
* Unlike Batch 1's color codemod (which used a hand-audited site table to
* avoid false-positiving on non-color hex-like strings such as issue
* references), this batch's patterns are unambiguous: `text-[Npx]`,
* `text-[N.Nrem]`, `tracking-[N em]`, and `leading-[...]` inside Tailwind
* class strings, and `fontSize: "Npx"` / `fontSize: "N.Nrem"` inline-style
* string literals, cannot mean anything other than a type-size/spacing
* value. A blanket regex sweep is therefore safe and is used here, scoped
* to `ui/src/components/**` and `ui/src/pages/**` only. Numeric or
* computed `fontSize` forms (e.g. `fontSize: 12`, `fontSize: Math.round(...)`)
* are functional (third-party config objects / runtime-computed values)
* and are left untouched — see ALLOWLIST_NOTES below.
*
* Token naming (verbatim value, no normalizing):
* --fs-<N> font-size, px values, e.g. --fs-11: 11px;
* --fs-0_<N>rem font-size, rem values, e.g. --fs-0_7rem: 0.7rem;
* --ls-0_<N> letter-spacing, em values, e.g. --ls-0_18: 0.18em;
* --lh-<N> line-height (px or unitless — none found this batch)
*
* Tailwind v4 paren-shorthand rewrite forms used:
* text-[Npx] -> text-(length:--fs-N) (length hint REQUIRED —
* bare text-(--x) means color)
* tracking-[N em] -> tracking-(--ls-0_N) (unambiguous, no hint)
* leading-[...] -> leading-(--lh-N) (unambiguous, no hint)
* All variant/modifier prefixes (`sm:`, `dark:`, `group-hover:`,
* `[&>x]:`, trailing `!important` marker, etc.) are preserved verbatim —
* the regex only rewrites the bracket portion itself.
*
* Idempotent: the FIND regex only matches the ORIGINAL bracket-literal
* form (`text-[11px]` etc.); once rewritten to `text-(length:--fs-11)` the
* pattern no longer matches, so re-running is a no-op. The inline-style
* FIND is likewise the literal `fontSize: "11px"` string form.
*
* Usage: node scripts/codemod-extract-type.mjs [--check]
* --check Report what WOULD change without writing files (dry run).
*/
import { readFileSync, writeFileSync, readdirSync, statSync } from "node:fs";
import { resolve, dirname, join, relative } from "node:path";
import { fileURLToPath } from "node:url";
const __dirname = dirname(fileURLToPath(import.meta.url));
const REPO_ROOT = resolve(__dirname, "..");
const UI_SRC = resolve(REPO_ROOT, "ui/src");
const SCAN_DIRS = ["components", "pages"];
const DRY_RUN = process.argv.includes("--check");
// ── Helpers ────────────────────────────────────────────────────────────
function walk(dir, out) {
for (const entry of readdirSync(dir, { withFileTypes: true })) {
const p = join(dir, entry.name);
if (entry.isDirectory()) walk(p, out);
else if (/\.(tsx?|jsx?)$/.test(entry.name)) out.push(p);
}
}
function tokenSuffixForPx(value) {
// "11" -> "11", "0.65" -> "0_65"
return value.replace(".", "_");
}
function tokenSuffixForEm(value) {
// "0.18" -> "0_18"
return value.replace(".", "_");
}
// ── Token registries (populated as sites are discovered) ───────────────
// Map of token name (without --) -> { value, comment, kind }
const fsTokens = new Map(); // font-size
const lsTokens = new Map(); // letter-spacing
const lhTokens = new Map(); // line-height
function registerFsToken(rawValue, unit, sourceNote) {
const name = unit === "px" ? `fs-${tokenSuffixForPx(rawValue)}` : `fs-${tokenSuffixForPx(rawValue)}rem`;
if (!fsTokens.has(name)) {
fsTokens.set(name, { value: `${rawValue}${unit}`, comment: sourceNote });
}
return name;
}
function registerLsToken(rawValue, sourceNote) {
const name = `ls-${tokenSuffixForEm(rawValue)}`;
if (!lsTokens.has(name)) {
lsTokens.set(name, { value: `${rawValue}em`, comment: sourceNote });
}
return name;
}
function registerLhToken(rawValue, sourceNote) {
// rawValue includes unit already stripped by caller; store as given
const safeName = rawValue.replace(/[^a-zA-Z0-9]/g, "_");
const name = `lh-${safeName}`;
if (!lhTokens.has(name)) {
lhTokens.set(name, { value: rawValue, comment: sourceNote });
}
return name;
}
// ── Regexes ──────────────────────────────────────────────────────────
// text-[11px], text-[0.65rem], with optional /[Npx] line-height suffix
// (none found in this codebase, but handled for completeness/future-proofing).
const FS_RE = /text-\[([0-9.]+)(px|rem)\](?:\/\[([0-9.]+)(px|rem)\])?/g;
const LS_RE = /tracking-\[([0-9.]+)em\]/g;
const LEADING_RE = /leading-\[([^\]]+)\]/g;
const FONTSIZE_STYLE_RE = /fontSize:\s*"([0-9.]+)(px|rem)"/g;
function rewriteFile(filePath, relPath) {
const original = readFileSync(filePath, "utf8");
let content = original;
let siteCount = 0;
// -- font-size Tailwind class utilities --
content = content.replace(FS_RE, (match, num, unit, lhNum, lhUnit) => {
const fsName = registerFsToken(num, unit, `Extracted from ${relPath} (text-[${num}${unit}]).`);
let replacement = `text-(length:--${fsName})`;
if (lhNum) {
const lhName = registerLhToken(`${lhNum}${lhUnit}`, `Extracted from ${relPath} (text-[...]/[${lhNum}${lhUnit}] line-height suffix).`);
replacement += `/(--${lhName})`;
}
siteCount++;
return replacement;
});
// -- letter-spacing Tailwind class utilities --
content = content.replace(LS_RE, (match, num) => {
const lsName = registerLsToken(num, `Extracted from ${relPath} (tracking-[${num}em]).`);
siteCount++;
return `tracking-(--${lsName})`;
});
// -- line-height Tailwind class utilities (standalone leading-[...]) --
content = content.replace(LEADING_RE, (match, raw) => {
// Only rewrite numeric/unit literals (px, rem, unitless number). Skip
// keyword forms like leading-[inherit] or var()-based (already tokenized).
if (!/^[0-9.]+(px|rem)?$/.test(raw)) return match;
const lhName = registerLhToken(raw, `Extracted from ${relPath} (leading-[${raw}]).`);
siteCount++;
return `leading-(--${lhName})`;
});
// -- inline style fontSize string literals --
content = content.replace(FONTSIZE_STYLE_RE, (match, num, unit) => {
const fsName = registerFsToken(num, unit, `Extracted from ${relPath} (inline style fontSize: "${num}${unit}").`);
siteCount++;
return `fontSize: "var(--${fsName})"`;
});
if (content !== original && !DRY_RUN) {
writeFileSync(filePath, content, "utf8");
}
return { changed: content !== original, siteCount };
}
function main() {
const files = [];
for (const dir of SCAN_DIRS) walk(resolve(UI_SRC, dir), files);
let totalSites = 0;
let filesChanged = 0;
const changedFiles = [];
for (const filePath of files) {
const relPath = "ui/src/" + relative(UI_SRC, filePath);
const { changed, siteCount } = rewriteFile(filePath, relPath);
if (changed) {
filesChanged++;
changedFiles.push(relPath);
}
totalSites += siteCount;
}
// ── index.css token block ──────────────────────────────────────────
const cssPath = resolve(UI_SRC, "index.css");
const cssOriginal = readFileSync(cssPath, "utf8");
const marker = "/* ── Extracted verbatim TYPE tokens (Phase 2 Batch 2, design/token-extraction) ── */";
let cssNext = cssOriginal;
let cssChanged = false;
if (!cssOriginal.includes(marker) && (fsTokens.size || lsTokens.size || lhTokens.size)) {
const lines = [];
lines.push(marker);
lines.push("/* Batch 2/4: font-size + letter-spacing + line-height literals, verbatim");
lines.push(" (no normalizing — 9/10/11/12/13/14/15px and 0.08-0.24em all stay distinct;");
lines.push(" the human scale-collapse decision comes later per DESIGN.md/TOKEN-AUDIT.md).");
lines.push("");
lines.push(" Allowlist (sites intentionally left as hardcoded / functional literals,");
lines.push(" NOT converted to tokens — each also carries an inline");
lines.push(" `token-extraction: allowlisted` comment at the site):");
lines.push(" - pages/CompanyEnvironments.tsx (fontSize: 12) — xterm.js terminal theme");
lines.push(" config; functional third-party numeric option, not a rendered CSS value.");
lines.push(" Same allowlisted object as Batch 1's color entry for this file.");
lines.push(" - pages/CompanySkills.tsx (fontSize: Math.round(size * 0.42)) — computed at");
lines.push(" runtime from a prop; not a static literal, nothing to extract.");
lines.push("*/");
lines.push(":root {");
for (const [name, { value, comment }] of fsTokens) {
lines.push(` --${name}: ${value}; /* ${comment} */`);
}
for (const [name, { value, comment }] of lsTokens) {
lines.push(` --${name}: ${value}; /* ${comment} */`);
}
for (const [name, { value, comment }] of lhTokens) {
lines.push(` --${name}: ${value}; /* ${comment} */`);
}
lines.push("}");
const block = "\n" + lines.join("\n") + "\n";
cssNext = cssOriginal + block;
cssChanged = true;
}
if (cssChanged && !DRY_RUN) writeFileSync(cssPath, cssNext, "utf8");
// ── Summary ─────────────────────────────────────────────────────────
console.log(`\n${DRY_RUN ? "[DRY RUN] " : ""}codemod-extract-type summary`);
console.log(` Sites rewritten: ${totalSites}`);
console.log(` Files changed: ${filesChanged}`);
console.log(` New --fs-* tokens: ${fsTokens.size}`);
console.log(` New --ls-* tokens: ${lsTokens.size}`);
console.log(` New --lh-* tokens: ${lhTokens.size}`);
console.log(` index.css token block: ${cssChanged ? "added" : "already present or nothing to add (idempotent no-op)"}`);
if (changedFiles.length) {
console.log(`\n Changed files:`);
for (const f of changedFiles) console.log(` - ${f}`);
}
}
main();