From ec6a3cee5818d8f6d9f11dbb1702a37df4c6d5e0 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C3=89ric=20NICOLAS=20=28ccjmne=29?= <ccjmne@gmail.com>
Date: Tue, 15 Oct 2024 00:59:46 +0200
Subject: [PATCH]

---
 src/utils/easy-htmlelement.ts | 12 ++++++++++--
 webpack.config.ts             |  8 +++++---
 2 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/src/utils/easy-htmlelement.ts b/src/utils/easy-htmlelement.ts
index 10344f8a..fd0d3d13 100644
--- a/src/utils/easy-htmlelement.ts
+++ b/src/utils/easy-htmlelement.ts
@@ -4,6 +4,7 @@ import externalLink from 'src/assets/external-link.svg?template'
 import { type RegExpGroups } from 'src/types'
 
 const SVGNS = 'http://www.w3.org/2000/svg'
+const HYPHENATE = /^y/i.test(process.env.HYPHENATE ?? 'no')
 
 export default class EasyHTMLElement {
 
@@ -49,20 +50,27 @@ export default class EasyHTMLElement {
   /**
    * Parses contents and:
    * - discard empty strings
-   * - automatically mark for hyphenation (for `en-gb`) with `\u00AD` (soft hyphen),
    * - replace linefeeds (literal `\n`) with `<br />` elements
    * - replace markdown-style hyperlinks with `<a href="...">...</a>` elements
    * - replace `&nbsp;` with `\u00A0` (non-breaking space)
+   * - replace `&shy;`  with `\u00AD` (soft hyphen)
+   *
+   * Additionally, while in `development` mode, automatically mark for
+   * hyphenation (for `en-gb`) with `\u00AD` (soft hyphen).
+   *
+   * This isn't done in `production` builds so as to avoid needlessly confusing
+   * crawlers and possible ATSs (Applicant Tracking System).
    */
   private static prepare(elements: ReadonlyArray<string | EasyHTMLElement>): ReadonlyArray<string | HTMLElement | SVGElement> {
     return elements
       .filter(content => content !== '')
       .flatMap(content => (typeof content !== 'string' ? content.elem : content
         .replace(/&nbsp;/g, '\u00A0')
+        .replace(/&shy;/g,  '\u00AD')
         .split(/(?<=\[[^\]]+\]\([^)]+\))|(?=\[[^\]]+\]\([^)]+\))/) // split around markdown-style hyperlinks
         .map(fragment => fragment.match(/^\[(?<text>[^\]]+)\]\((?<href>[^)]+)\)$/)?.groups as RegExpGroups<'text' | 'href'> | undefined ?? fragment)
         .flatMap(fragment => (typeof fragment === 'string'
-          ? fragment.split(/\n/g).flatMap(t => [new EasyHTMLElement('br').elem, hyphenate(t)]).slice(1)
+          ? fragment.split(/\n/g).flatMap(t => [new EasyHTMLElement('br').elem, HYPHENATE ? hyphenate(t) : t]).slice(1)
           : EasyHTMLElement.anchor(fragment).elem))
       ))
   }
diff --git a/webpack.config.ts b/webpack.config.ts
index 93de9eff..c1c1056a 100644
--- a/webpack.config.ts
+++ b/webpack.config.ts
@@ -10,6 +10,7 @@ import { Compiler } from 'webpack'
 import 'webpack-dev-server' // Augment "Configuration" type
 import { author, description, homepage, keywords, name, repository, title } from './package.json'
 import { PDFPrinter } from './tooling/pdf-printer-plugin'
+import { DefinePlugin } from 'webpack'
 
 const src = resolve(__dirname, 'src')
 const dist = resolve(__dirname, 'dist')
@@ -32,7 +33,7 @@ const pages = readdirSync(src, { withFileTypes: true })
   .reduce((acc, { name, path }) => ({ ...acc, [name]: path }), {})
 
 export default (
-  _env: string,
+  env: NodeJS.ProcessEnv,
   { mode = 'production', port = '8042' }: { mode?: 'production' | 'development', port?: string } = {},
 ): Configuration => ({
   entry: pages,
@@ -89,9 +90,10 @@ export default (
     path: dist,
   },
   plugins: [
+    new DefinePlugin({ 'process.env': JSON.stringify(env) }),
     ...mode === 'production' ? [new CleanWebpackPlugin()] : [],
     new TypedScssModulesPlugin({ watch: mode === 'development' }),
-    ...Object.entries(pages).map(([name, path]) => new HtmlWebpackPlugin({
+    ...Object.keys(pages).map(name => new HtmlWebpackPlugin({
       title: `Page ${name}`,
       meta: { author, description, repository, keywords: keywords.join(', ') },
       chunks: [name],
@@ -102,7 +104,7 @@ export default (
       ...mode === 'production'
         ? { scheme: 'file', paths: Object.keys(pages).map(name => resolve(dist, `${name}.html`)) }
         : { port, paths: Object.keys(pages).map(name => `${name}.html`) },
-      output: resolve(dist, `${out}.pdf`),
+      output: resolve(dist, env.OUTPUT ?? `${out}.pdf`),
       properties: { title, author, subject: description, keywords: keywords.join(', '), creator: `${name} (${homepage})` },
       blocking: mode === 'production',
     }),