wordpress-theme-aksrt/tools/generate-fa5-icons.php

<?php
/**
 * One-shot generator for src/Icons/fa5-icons.php.
 *
 * Sources (shipped with the theme):
 *  - assets/fontawesome5/css/all.min.css            → glyph class names + codepoints
 *  - assets/fontawesome5/webfonts/fa-*.woff         → per-style codepoints (cmap table)
 *
 * Usage: php tools/generate-fa5-icons.php
 *
 * The style split comes from each font's own cmap subtable rather than from the
 * CSS: FA5 assigns the same codepoint to different glyphs per style (e.g. far
 * "address-book" and fas "address-book" are different drawings), so membership
 * in fas / far / fab can only be answered by the font itself.
 */

defined('ABSPATH') || define('ABSPATH', __DIR__ . '/');

$root       = dirname(__DIR__);
$css_path   = $root . '/assets/fontawesome5/css/all.min.css';
$fonts_path = $root . '/assets/fontawesome5/webfonts';
$out_path   = $root . '/src/Icons/fa5-icons.php';

/** Font file per style group, matching the FA5 family naming. */
const FA5_FONTS = [
	'fas' => 'fa-solid-900.woff',
	'far' => 'fa-regular-400.woff',
	'fab' => 'fa-brands-400.woff',
];

/**
 * Extract one table's uncompressed bytes from a WOFF container.
 *
 * WOFF stores each table zlib-deflated when compLen differs from origLen.
 */
function fa5_woff_table(string $raw, string $tag): ?string {
	$num_tables = (int) (unpack('n', substr($raw, 12, 2))[1] ?? 0);
	if ($num_tables < 1 || strlen($raw) < 44 + $num_tables * 20) {
		return null;
	}
	for ($i = 0; $i < $num_tables; $i++) {
		$entry = substr($raw, 44 + $i * 20, 20);
		if (substr($entry, 0, 4) !== $tag) {
			continue;
		}
		$p = unpack('Noffset/NcompLen/NorigLen', substr($entry, 4, 12));
		$data = substr($raw, $p['offset'], $p['compLen']);
		if ($p['compLen'] !== $p['origLen']) {
			$data = gzuncompress($data);
			if ($data === false) {
				return null;
			}
		}
		return $data;
	}
	return null;
}

/**
 * Codepoints mapped by a cmap table, keyed for O(1) membership tests.
 *
 * Only format-4 (Unicode BMP) subtables are decoded — the (3,1) record is
 * preferred, falling back to the first (3,x) then (0,x) record so a font that
 * ships a different platform order still resolves.
 */
function fa5_cmap_codepoints(string $cmap): array {
	$records = (int) (unpack('n', substr($cmap, 2, 2))[1] ?? 0);
	$best = null;
	$fallback = null;
	for ($i = 0; $i < $records; $i++) {
		$rec = unpack('nplatform/nencoding/Noffset', substr($cmap, 4 + $i * 8, 8));
		if ($rec['platform'] === 3 && $rec['encoding'] === 1) {
			$best = $rec['offset'];
			break;
		}
		if ($fallback === null && ($rec['platform'] === 3 || $rec['platform'] === 0)) {
			$fallback = $rec['offset'];
		}
	}
	$offset = $best ?? $fallback;
	if ($offset === null || $offset + 14 > strlen($cmap)) {
		return [];
	}

	$table = substr($cmap, $offset);
	$format = (int) (unpack('n', substr($table, 0, 2))[1] ?? 0);
	if ($format !== 4) {
		fwrite(STDERR, "Unsupported cmap format {$format}; only format 4 is decoded\n");
		exit(1);
	}

	$seg_x2    = (int) (unpack('n', substr($table, 6, 2))[1] ?? 0);
	$seg_count = intdiv($seg_x2, 2);
	if ($seg_count < 1) {
		return [];
	}
	$end_base   = 14;
	$start_base = $end_base + $seg_x2 + 2;

	$codes = [];
	for ($i = 0; $i < $seg_count; $i++) {
		$start = (int) (unpack('n', substr($table, $start_base + $i * 2, 2))[1] ?? 0);
		$end   = (int) (unpack('n', substr($table, $end_base + $i * 2, 2))[1] ?? 0);
		if ($start > $end || $end === 0xFFFF) {
			continue;
		}
		for ($code = $start; $code <= $end; $code++) {
			$codes[$code] = true;
		}
	}
	return $codes;
}

$css = (string) file_get_contents($css_path);

// class name → codepoint (e.g. 'home' => 0xf015)
if (!preg_match_all('/\.fa-([a-z0-9-]+):before\{content:"\\\\([0-9a-f]{4})"\}/i', $css, $m, PREG_SET_ORDER)) {
	fwrite(STDERR, "No glyph rules matched in all.min.css\n");
	exit(1);
}
$name_to_code = [];
foreach ($m as $hit) {
	$name_to_code[$hit[1]] = hexdec($hit[2]);
}

// style → set of codepoints, decoded from the shipped woff cmaps
$code_to_style = [];
foreach (FA5_FONTS as $style => $file) {
	$path = $fonts_path . '/' . $file;
	if (!is_readable($path)) {
		fwrite(STDERR, "Missing font: {$path}\n");
		exit(1);
	}
	$cmap = fa5_woff_table((string) file_get_contents($path), 'cmap');
	if ($cmap === null) {
		fwrite(STDERR, "No readable cmap table in {$path}\n");
		exit(1);
	}
	$code_to_style[$style] = fa5_cmap_codepoints($cmap);
}

$groups = ['fas' => [], 'far' => [], 'fab' => []];
foreach ($name_to_code as $name => $code) {
	foreach ($groups as $style => $unused) {
		if (!empty($code_to_style[$style][$code])) {
			$groups[$style][] = $name;
		}
	}
}
foreach ($groups as $style => $names) {
	sort($groups[$style], SORT_NATURAL | SORT_FLAG_CASE);
}

$export = var_export($groups, true);

$php = <<<PHP
<?php
/**
 * Font Awesome 5 Free (5.15.4) icon catalog, grouped by style.
 *
 * Generated by tools/generate-fa5-icons.php — do not edit by hand.
 * Bare glyph names; the picker composes "fa {fas|far|fab} fa-{name}".
 *
 * @package AKSRT
 */

defined('ABSPATH') || exit;

return {$export}
PHP;

// .gitattributes declares *.php eol=crlf for the working tree; emit CRLF so the
// regenerated file matches its neighbours instead of tripping the normalize warning.
file_put_contents($out_path, str_replace("\n", "\r\n", $php . ";\n"));

printf(
	"fas=%d far=%d fab=%d (css glyph classes=%d)\nWrote %s\n",
	count($groups['fas']),
	count($groups['far']),
	count($groups['fab']),
	count($name_to_code),
	$out_path
);