@turntrout/subfont 1.0.4 → 1.1.0

package/README.md

@@ -36,20 +36,26 @@ subfont path/to/index.html -i --cache
 
 ## Options
 
-|              Flag | Default | Description                                                  |
-| ----------------: | :-----: | :----------------------------------------------------------- |
-|  `-i, --in-place` |   off   | Modify files in-place                                        |
-|    `-o, --output` |         | Output directory                                             |
-| `-r, --recursive` |   off   | Crawl linked pages                                           |
-|       `--dynamic` |   off   | Trace with headless browser                                  |
-|       `--dry-run` |   off   | Preview without writing                                      |
-|     `--fallbacks` |   on    | Load the full original font for characters not in the subset |
-|  `--font-display` | `swap`  | `auto`/`block`/`swap`/`fallback`/`optional`                  |
-|          `--text` |         | Extra characters for every subset                            |
-|   `--cache [dir]` |   off   | Cache subset results to disk between runs                    |
-| `--concurrency N` |         | Max worker threads for parallel font tracing                 |
-|  `--chrome-flags` |         | Custom Chrome flags for `--dynamic`                          |
-|   `--source-maps` |   off   | Preserve CSS source maps (slower)                            |
+|               Flag | Default | Description                                                  |
+| -----------------: | :-----: | :----------------------------------------------------------- |
+|   `-i, --in-place` |   off   | Modify files in-place                                        |
+|     `-o, --output` |         | Output directory                                             |
+|           `--root` |         | Path to web root (deduced from input files if not specified) |
+| `--canonical-root` |         | URI root where the site will be deployed                     |
+|  `-r, --recursive` |   off   | Crawl linked pages                                           |
+|        `--dynamic` |   off   | Trace with headless browser                                  |
+|        `--dry-run` |   off   | Preview without writing                                      |
+|      `--fallbacks` |   on    | Load the full original font for characters not in the subset |
+|   `--font-display` | `swap`  | `auto`/`block`/`swap`/`fallback`/`optional`                  |
+|           `--text` |         | Extra characters for every subset                            |
+|    `--cache [dir]` |   off   | Cache subset results to disk between runs                    |
+|  `--concurrency N` |         | Max worker threads (capped by available memory, ~50 MB each) |
+|   `--chrome-flags` |         | Custom Chrome flags for `--dynamic`                          |
+|    `--source-maps` |   off   | Preserve CSS source maps (slower)                            |
+|     `-s, --silent` |   off   | Suppress all console output                                  |
+|      `-d, --debug` |   off   | Verbose timing and font glyph detection info                 |
+|  `--relative-urls` |   off   | Emit relative URLs instead of root-relative                  |
+|     `--inline-css` |   off   | Inline the subset @font-face CSS into HTML                   |
 
 Run `subfont --help` for the full list.
 

package/lib/FontTracerPool.js

@@ -159,6 +159,17 @@ class FontTracerPool {
     }
     this._pendingTasks = [];
 
+    // Reject any in-flight tasks still assigned to workers.
+    // Clear _taskByWorker before terminate() so _onWorkerExit won't double-reject.
+    for (const [, taskId] of this._taskByWorker) {
+      const cb = this._taskCallbacks.get(taskId);
+      if (cb) {
+        this._taskCallbacks.delete(taskId);
+        cb.reject(new Error('Worker pool destroyed'));
+      }
+    }
+    this._taskByWorker.clear();
+
     // Terminate workers with a 5-second timeout to prevent hanging
     const TERMINATE_TIMEOUT_MS = 5000;
     await Promise.all(

package/lib/HeadlessBrowser.js

@@ -180,7 +180,8 @@ class HeadlessBrowser {
         urlTools.resolveUrl(
           baseUrl,
           urlTools.buildRelativeUrl(assetGraph.root, htmlAsset.url)
-        )
+        ),
+        { timeout: 30000 }
       );
 
       await page.addScriptTag({
@@ -202,7 +203,13 @@ class HeadlessBrowser {
     const launchPromise = this._launchPromise;
     if (launchPromise) {
       this._launchPromise = undefined;
-      const browser = await launchPromise;
+      let browser;
+      try {
+        browser = await launchPromise;
+      } catch {
+        // Launch failed — nothing to close
+        return;
+      }
       await browser.close();
     }
   }

package/lib/concurrencyLimit.js

@@ -0,0 +1,10 @@
+const os = require('os');
+
+// Approximate memory footprint of each worker thread (jsdom + font-tracer).
+const WORKER_MEMORY_BYTES = 50 * 1024 * 1024; // 50 MB
+
+function getMaxConcurrency() {
+  return Math.max(1, Math.floor(os.totalmem() / WORKER_MEMORY_BYTES));
+}
+
+module.exports = { WORKER_MEMORY_BYTES, getMaxConcurrency };

package/lib/extractVisibleText.js

@@ -10,6 +10,7 @@ const INVISIBLE_ELEMENTS = new Set([
   'iframe',
   'object',
   'embed',
+  'datalist',
 ]);
 const TEXT_ATTRIBUTES = new Set([
   'alt',
@@ -72,3 +73,4 @@ function extractVisibleText(html) {
 }
 
 module.exports = extractVisibleText;
+module.exports.INVISIBLE_ELEMENTS = INVISIBLE_ELEMENTS;

package/lib/fontTracerWorker.js

@@ -28,11 +28,11 @@ parentPort.on('message', (msg) => {
   }
 
   if (msg.type === 'trace') {
+    const { taskId, htmlText, stylesheetsWithPredicates: serialized } = msg;
+    let dom;
     try {
-      const { taskId, htmlText, stylesheetsWithPredicates: serialized } = msg;
-
       // Re-parse HTML with jsdom to get a DOM document
-      const dom = new JSDOM(htmlText);
+      dom = new JSDOM(htmlText);
       const document = dom.window.document;
 
       // Re-parse CSS from serialized text — asset objects with PostCSS
@@ -50,9 +50,6 @@ parentPort.on('message', (msg) => {
         getCssRulesByProperty: memoizedGetCssRulesByProperty,
       });
 
-      // Clean up jsdom to free memory
-      dom.window.close();
-
       // Strip any non-serializable data from results
       const serializableResults = textByProps.map((entry) => ({
         text: entry.text,
@@ -71,6 +68,9 @@ parentPort.on('message', (msg) => {
         error: err.message,
         stack: err.stack,
       });
+    } finally {
+      // Clean up jsdom to free memory — must run even if fontTracer throws
+      if (dom) dom.window.close();
     }
   }
 });

package/lib/parseCommandLineOptions.js

@@ -1,8 +1,13 @@
 module.exports = function parseCommandLineOptions(argv) {
+  const os = require('os');
+  const { getMaxConcurrency } = require('./concurrencyLimit');
   let yargs = require('yargs');
   if (argv) {
     yargs = yargs(argv);
   }
+
+  const maxConcurrency = getMaxConcurrency();
+
   yargs
     .usage(
       'Create optimal font subsets from your actual font usage.\n$0 [options] <htmlFile(s) | url(s)>'
@@ -106,10 +111,22 @@ module.exports = function parseCommandLineOptions(argv) {
       },
     })
     .options('concurrency', {
-      describe:
-        'Maximum number of worker threads for parallel font tracing. Defaults to the number of CPU cores (max 8)',
+      describe: `Maximum number of worker threads for parallel font tracing. Defaults to the number of CPU cores (max 8). Upper bound: ${maxConcurrency} (based on available memory)`,
       type: 'number',
     })
+    .check((argv) => {
+      if (argv.concurrency !== undefined) {
+        if (!Number.isInteger(argv.concurrency) || argv.concurrency < 1) {
+          throw new Error('--concurrency must be a positive integer');
+        }
+        if (argv.concurrency > maxConcurrency) {
+          throw new Error(
+            `--concurrency must not exceed ${maxConcurrency} (each worker uses ~50 MB; system has ${Math.round(os.totalmem() / (1024 * 1024 * 1024))} GB total)`
+          );
+        }
+      }
+      return true;
+    })
     .options('source-maps', {
       describe: 'Preserve CSS source maps through subfont processing',
       type: 'boolean',

package/lib/subfont.js

@@ -1,5 +1,7 @@
 const fsPromises = require('fs/promises');
+const os = require('os');
 const pathModule = require('path');
+const { getMaxConcurrency } = require('./concurrencyLimit');
 const AssetGraph = require('assetgraph');
 const prettyBytes = require('pretty-bytes');
 const urlTools = require('urltools');
@@ -43,6 +45,12 @@ module.exports = async function subfont(
   ) {
     throw new UsageError('--concurrency must be a positive integer');
   }
+  const maxConcurrency = getMaxConcurrency();
+  if (concurrency !== undefined && concurrency > maxConcurrency) {
+    throw new UsageError(
+      `--concurrency must not exceed ${maxConcurrency} (each worker uses ~50 MB; system has ${Math.round(os.totalmem() / (1024 * 1024 * 1024))} GB total)`
+    );
+  }
 
   const formats = ['woff2'];
 
@@ -216,14 +224,15 @@ module.exports = async function subfont(
     }
   }
 
-  let sumSizesBefore = 0;
-  for (const asset of assetGraph.findAssets({
+  const sizeableAssetQuery = {
     isInline: false,
     isLoaded: true,
     type: {
       $in: ['Html', 'Svg', 'Css', 'JavaScript'],
     },
-  })) {
+  };
+  let sumSizesBefore = 0;
+  for (const asset of assetGraph.findAssets(sizeableAssetQuery)) {
     sumSizesBefore += asset.rawSrc.length;
   }
 
@@ -269,13 +278,7 @@ module.exports = async function subfont(
 
   phaseStart = Date.now();
   let sumSizesAfter = 0;
-  for (const asset of assetGraph.findAssets({
-    isInline: false,
-    isLoaded: true,
-    type: {
-      $in: ['Html', 'Svg', 'Css', 'JavaScript'],
-    },
-  })) {
+  for (const asset of assetGraph.findAssets(sizeableAssetQuery)) {
     sumSizesAfter += asset.rawSrc.length;
   }
 

package/lib/unicodeRange.js

@@ -14,7 +14,7 @@ const getUnicodeRanges = (codePoints) => {
   const ranges = [];
   let start, end;
 
-  codePoints.sort((a, b) => a - b);
+  codePoints = [...codePoints].sort((a, b) => a - b);
 
   for (let i = 0; i < codePoints.length; i++) {
     start = codePoints[i];

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@turntrout/subfont",
-  "version": "1.0.4",
+  "version": "1.1.0",
   "description": "Automatically subset web fonts to only the characters used on your pages. Fork of Munter/subfont with modern defaults.",
   "engines": {
     "node": ">=18.0.0"