art.php: bitwise byte-string fast path and memoized color codes (105ms -> 21ms uncached)

inc/art.php

@@ -58,11 +58,89 @@ function ansi_html_cached(string $file, bool $mirror = false): string
 }
 
 function ansi_html(string $file, bool $mirror = false): string
+{
+    $raw = (string) file_get_contents($file);
+    if (!preg_match('/[\x80-\xff]/', $raw)) {
+        $html = ansi_html_bytes($raw, $mirror);
+        if ($html !== null) {
+            return $html;
+        }
+    }
+    return ansi_html_cells($raw, $mirror);
+}
+
+/**
+ * Fast path for plain-ASCII art (one byte per cell): each line becomes a
+ * string of characters plus a parallel string of color indices, and every
+ * color layer is cut out of the whole art with bitwise string ops, which run
+ * in C. Returns null if there are too many colors to index in a byte.
+ */
+function ansi_html_bytes(string $raw, bool $mirror): ?string
+{
+    static $from = null, $to = null;          // byte 0 -> 0xff, anything else -> 0
+    $from ??= implode(array_map('chr', range(0, 255)));
+    $to ??= "\xff" . str_repeat("\0", 255);
+
+    static $memo = [];
+    $index = ['' => 0];                       // color hex => byte (0 = uncolored)
+    $chars = [];
+    $masks = [];
+    foreach (explode("\n", rtrim($raw, "\n")) as $line) {
+        $c = '';
+        $m = '';
+        $cur = null;
+        foreach (preg_split('/(\e\[[0-9;]*m)/', $line, -1, PREG_SPLIT_DELIM_CAPTURE | PREG_SPLIT_NO_EMPTY) as $part) {
+            if ($part[0] === "\e") {
+                // a handful of distinct codes repeated thousands of times: decode each once
+                $cur = $memo[$part . '|' . $cur] ??= ansi_color(explode(';', substr($part, 2, -1)), $cur);
+                continue;
+            }
+            $k = $index[$cur ?? ''] ??= count($index);
+            $c .= $part;
+            $m .= str_repeat(chr($k), strlen($part));
+        }
+        $chars[] = $c;
+        $masks[] = $m;
+    }
+    if (count($index) > 250) {
+        return null;
+    }
+
+    $width = max(array_map('strlen', $chars) ?: [0]);
+    foreach ($chars as $i => $c) {
+        $chars[$i] = str_pad($c, $width);
+        $masks[$i] = str_pad($masks[$i], $width, "\0");
+        if ($mirror) {
+            $chars[$i] = strtr(strrev($chars[$i]), '/\\()<>[]{}', '\\/)(><][}{');
+            $masks[$i] = strrev($masks[$i]);
+        }
+    }
+    // the whole art as one string; 0xfe in the mask marks the line breaks
+    $all = implode("\n", $chars);
+    $mask = implode("\xfe", $masks);
+    $blank = implode("\n", array_fill(0, count($chars), str_repeat(' ', $width)));
+    $n = strlen($all);
+
+    $html = '';
+    foreach ($index as $color => $k) {
+        $sel = strtr($mask ^ str_repeat(chr($k), $n), $from, $to); // 0xff where this color, else 0
+        $layer = ($all & $sel) | ($blank & ~$sel);
+        if (trim($layer) === '') {
+            continue;                                             // only blanks in this color
+        }
+        $style = $color !== '' ? " style=\"color:#$color\"" : '';
+        $html .= "<span$style>" . htmlspecialchars($layer) . '</span>';
+    }
+    return $html;
+}
+
+/** General path: a [char, color] cell at a time, so multibyte characters work. */
+function ansi_html_cells(string $raw, bool $mirror): string
 {
     static $flip = ['/' => '\\', '\\' => '/', '(' => ')', ')' => '(', '<' => '>', '>' => '<',
                     '[' => ']', ']' => '[', '{' => '}', '}' => '{'];
     $lines = [];
-    foreach (explode("\n", rtrim((string) file_get_contents($file), "\n")) as $line) {
+    foreach (explode("\n", rtrim($raw, "\n")) as $line) {
         // split into [char, color] cells
         $cells = [];
         $cur = null;