Reduce production bundle size (#11)

* Reduce production bundle size

* fix fc seed

.changeset/silver-coins-serve.md

@@ -0,0 +1,9 @@
+---
+"unicode-segmenter": patch
+---
+
+Reduce the production bundle size
+
+Previously I did unescape `"\u{1F680}"` to `"🚀"` in the Unicode table. Since extra characters are required to escape, it reduces the output size.
+
+However, escape sequences compress better. So leaving the build output as is makes more sense for production.

README.md

@@ -62,7 +62,7 @@ Tested on:
 
 | Name                         | ESM? | Size    | Size (min)       | Size (min+gzip)  | Size (min+br)    |
 |------------------------------|------|--------:|-----------------:|-----------------:|-----------------:|
-| `unicode-segmenter/grapheme` |    ✔️ |  44,669 |           30,698 |            9,696 |            5,917 |
+| `unicode-segmenter/grapheme` |    ✔️ |  54,383 |           40,412 |            9,586 |            5,321 |
 | `graphemer`                  |    ✖️ ️| 410,424 |           95,104 |           15,752 |           10,660 |
 | `grapheme-splitter`          |    ✖️ | 122,241 |           23,680 |            7,852 |            4,841 |
 

scripts/build-exports.js

@@ -11,25 +11,15 @@ let bundleDir = path.join(distDir, 'bundle');
 await fs.mkdir(distDir, { recursive: true });
 
 let src = name => path.join(srcDir, name);
-let dist = name => path.join(distDir, name);
-let bundle = name => path.join(bundleDir, name);
-
 let modules = await fs.readdir(srcDir);
 
-async function unescapeCodePoints(file) {
-  let content = await fs.readFile(file, 'utf8');
-  content = content.replace(/(\\u\{?[0-9a-zA-Z]+\}?)/g, (_match, group) => {
-    return eval(`"${group}"`);
-  });
-  await fs.writeFile(file, content, 'utf8');
-}
-
-async function rewriteCjsEntries(file) {
-  let content = await fs.readFile(file, 'utf8');
-  content = content.replace(/(require\("(?<id>.*)\.js"\))/g, (_match, _group1, id) => {
+/**
+ * @param {string} content
+ */
+function rewriteCjsEntries(content) {
+  return content.replace(/(require\("(?<id>.*)\.js"\))/g, (_match, _group1, id) => {
     return `require("${id}.cjs")`;
   });
-  await fs.writeFile(file, content, 'utf8');
 }
 
 {
@@ -42,21 +32,16 @@ async function rewriteCjsEntries(file) {
     treeShaking: true,
     write: true,
   });
-  await build({
+  let { outputFiles: cjsOutputFiles = [] } = await build({
     entryPoints,
     outdir: distDir,
     outExtension: { '.js': '.cjs' },
     format: 'cjs',
     treeShaking: true,
-    write: true,
+    write: false,
   });
-  for (let file of await fs.readdir(distDir)) {
-    if (file.startsWith('_')) {
-      await unescapeCodePoints(dist(file));
-    }
-    if (file.endsWith('.cjs')) {
-      await rewriteCjsEntries(dist(file));
-    }
+  for (let file of cjsOutputFiles) {
+    await fs.writeFile(file.path, rewriteCjsEntries(file.text), 'utf8');
   }
 }
 
@@ -86,7 +71,4 @@ async function rewriteCjsEntries(file) {
     treeShaking: true,
     write: true,
   });
-  for (let file of await fs.readdir(bundleDir)) {
-    await unescapeCodePoints(bundle(file));
-  }
 }

test/property.js

@@ -5,6 +5,8 @@ import fc from 'fast-check';
 import { graphemeSegments } from 'unicode-segmenter/grapheme';
 
 fc.configureGlobal({
+  // Fix seed here for stable coverage report
+  seed: 1713140942000,
   numRuns: 100_000,
 });