From 92e080465e07f3f1f9b25792c58cd0785fd36aff Mon Sep 17 00:00:00 2001
From: Jun Pataleta <jun@moodle.com>
Date: Fri, 27 May 2022 14:24:29 +0800
Subject: [PATCH] MDL-72885 core: New weblib function
 get_html_lang_attribute_value()

Converts a language code to hyphen-separated format in accordance to the
BCP47 syntax appropriate for the HTML lang attribute.

See
https://datatracker.ietf.org/doc/html/rfc5646#section-2.1
https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/lang
---
 lib/tests/weblib_test.php | 26 ++++++++++++++++++++++++++
 lib/weblib.php            | 18 ++++++++++++++++++
 2 files changed, 44 insertions(+)

diff --git a/lib/tests/weblib_test.php b/lib/tests/weblib_test.php
index 8fa7d6a9924..a6c6aa9b3c3 100644
--- a/lib/tests/weblib_test.php
+++ b/lib/tests/weblib_test.php
@@ -941,4 +941,30 @@ EXPECTED;
 
         $this->assertNotEquals($policydisabled, print_password_policy());
     }
+
+    /**
+     * Data provider for the testing get_html_lang_attribute_value().
+     *
+     * @return string[][]
+     */
+    public function get_html_lang_attribute_value_provider() {
+        return [
+            'Empty lang code' => ['    ', 'unknown'],
+            'English' => ['en', 'en'],
+            'English, US' => ['en_us', 'en-us'],
+        ];
+    }
+
+    /**
+     * Test for get_html_lang_attribute_value().
+     *
+     * @covers ::get_html_lang_attribute_value()
+     * @dataProvider get_html_lang_attribute_value_provider
+     * @param string $langcode The language code to convert.
+     * @param string $expected The expected converted value.
+     * @return void
+     */
+    public function test_get_html_lang_attribute_value(string $langcode, string $expected): void {
+        $this->assertEquals($expected, get_html_lang_attribute_value($langcode));
+    }
 }
diff --git a/lib/weblib.php b/lib/weblib.php
index 7b142723eed..ed2b73f163f 100644
--- a/lib/weblib.php
+++ b/lib/weblib.php
@@ -2206,6 +2206,24 @@ function highlightfast($needle, $haystack) {
     return str_replace('<span class="highlight"></span>', '', join('', $parts));
 }
 
+/**
+ * Converts a language code to hyphen-separated format in accordance to the
+ * {@link https://datatracker.ietf.org/doc/html/rfc5646#section-2.1 BCP47 syntax}.
+ *
+ * For additional information, check out
+ * {@link https://developer.mozilla.org/en-US/docs/Web/HTML/Global_attributes/lang MDN web docs - lang}.
+ *
+ * @param string $langcode The language code to convert.
+ * @return string
+ */
+function get_html_lang_attribute_value(string $langcode): string {
+    if (empty(trim($langcode))) {
+        // If the language code passed is an empty string, return 'unknown'.
+        return 'unknown';
+    }
+    return str_replace('_', '-', $langcode);
+}
+
 /**
  * Return a string containing 'lang', xml:lang and optionally 'dir' HTML attributes.
  *