Stdlib: grapheme_str_split() JIT compile-time fold (#6246)

ext/intl/JitGrapheme.php

@@ -8,6 +8,8 @@
 use PHPCompiler\JIT\Builtin\StringUtf8Runtime;
 use PHPCompiler\JIT\Context;
 use PHPCompiler\JIT\Variable as JITVariable;
+use PHPCompiler\VM\HashTable;
+use PHPCompiler\VM\Variable;
 use PHPLLVM\Builder;
 use PHPLLVM\Value;
 
@@ -50,6 +52,33 @@ public static function tryStrlenFold(Context $context, array $args): ?Value
         return $context->constantFromInteger($result, 'int64');
     }
 
+    /**
+     * @param JITVariable[] $args
+     */
+    public static function tryStrSplitFold(Context $context, array $args): ?Value
+    {
+        $string = self::compileTimeString($args, 0);
+        if (null === $string) {
+            return null;
+        }
+        $length = 1;
+        if (isset($args[1])) {
+            $lengthCt = self::compileTimeInt($args, 1);
+            if (null === $lengthCt) {
+                return null;
+            }
+            $length = $lengthCt;
+        }
+        $parts = VmGrapheme::strSplit($string, $length);
+        if (false === $parts) {
+            return $context->getTypeFromString('bool')->constInt(0, false);
+        }
+        $ht = self::hashTableFromStringList($parts);
+        $cacheKey = 'grapheme_str_split_'.md5($string."\0".$length);
+
+        return $context->constantArrayFromVmHashTable($cacheKey, $ht);
+    }
+
     /**
      * @param JITVariable[] $args
      */
@@ -273,4 +302,19 @@ private static function compileTimeBool(array $args, int $index): ?bool
 
         return (bool) ($args[$index]->compileTimeBool ?? false);
     }
+
+    /**
+     * @param list<string> $parts
+     */
+    private static function hashTableFromStringList(array $parts): HashTable
+    {
+        $ht = new HashTable();
+        foreach ($parts as $part) {
+            $stored = new Variable();
+            $stored->string($part);
+            $ht->append($stored);
+        }
+
+        return $ht;
+    }
 }

ext/intl/grapheme_str_split.php

@@ -9,6 +9,7 @@
 use PHPCompiler\Frame;
 use PHPCompiler\Func\Internal;
 use PHPCompiler\JIT\Context;
+use PHPCompiler\JIT\JitStringBuiltinArg;
 use PHPCompiler\JIT\Variable as JITVariable;
 use PHPCompiler\VM\BuiltinExecute;
 use PHPCompiler\VM\HashTable;
@@ -18,7 +19,7 @@
 /**
  * grapheme_str_split() — split string into grapheme clusters (php-src ext/intl/grapheme; #5958).
  *
- * VM: {@see VmGrapheme}; JIT: compile-time fold via {@see JitGrapheme} (runtime defer #5958).
+ * VM: {@see VmGrapheme}; JIT: compile-time fold via {@see JitGrapheme::tryStrSplitFold} (#5958, #6246).
  */
 final class grapheme_str_split extends Internal
 {
@@ -76,9 +77,14 @@ public function call(Context $context, JITVariable ...$args): Value
         if ($argc < 1 || $argc > 2) {
             throw new \LogicException('grapheme_str_split() requires one or two arguments');
         }
+        $folded = JitGrapheme::tryStrSplitFold($context, $args);
+        if (null !== $folded) {
+            return $folded;
+        }
+        JitStringBuiltinArg::lower($context, $args[0], 'grapheme_str_split', 0, 'string');
 
         throw new \LogicException(
-            'grapheme_str_split() JIT runtime lowering is deferred; use VM (#5958)'
+            'grapheme_str_split() JIT runtime lowering is deferred; use VM or compile-time literals (#5958)'
         );
     }
 }

test/compliance/cases/stdlib/grapheme_str_split_jit.phpt

@@ -0,0 +1,32 @@
+--TEST--
+stdlib grapheme_str_split() JIT — compile-time fold (#5958, #6246)
+--FILE--
+<?php
+echo (int) function_exists('grapheme_str_split'), "\n";
+
+$one = grapheme_str_split("e\xCC\x81");
+var_export($one);
+echo "\n";
+
+$ascii = grapheme_str_split('abc');
+var_export($ascii);
+echo "\n";
+
+$chunked = grapheme_str_split('abcdef', 2);
+var_export($chunked);
+echo "\n";
+--EXPECT--
+0
+array (
+  0 => 'é',
+)
+array (
+  0 => 'a',
+  1 => 'b',
+  2 => 'c',
+)
+array (
+  0 => 'ab',
+  1 => 'cd',
+  2 => 'ef',
+)

test/fixtures/aot/compile-only/grapheme_str_split_literals.php

@@ -0,0 +1,9 @@
+<?php
+
+declare(strict_types=1);
+
+// Compile-only (#6246): grapheme_str_split compile-time fold for AOT lint.
+function probe_str_split(string $string, int $length = 1): array|false
+{
+    return grapheme_str_split($string, $length);
+}

test/unit/GraphemeStrSplitJitCompileTest.php

@@ -0,0 +1,37 @@
+<?php
+
+declare(strict_types=1);
+
+namespace PHPCompiler;
+
+use PHPUnit\Framework\TestCase;
+
+/** AOT lint verify for grapheme_str_split() JIT compile-time fold (#6246). */
+final class GraphemeStrSplitJitCompileTest extends TestCase
+{
+    public function testGraphemeStrSplitAotLint(): void
+    {
+        $root = dirname(__DIR__, 2);
+        $bin = realpath($root.'/bin/compile.php');
+        $this->assertNotFalse($bin);
+        $target = $root.'/test/fixtures/aot/compile-only/grapheme_str_split_literals.php';
+        $cmd = [PHP_BINARY, $bin, '-l', $target];
+        $descriptorSpec = [
+            0 => ['pipe', 'r'],
+            1 => ['pipe', 'w'],
+            2 => ['pipe', 'w'],
+        ];
+        $proc = proc_open($cmd, $descriptorSpec, $pipes, $root);
+        $this->assertIsResource($proc);
+        fclose($pipes[0]);
+        fclose($pipes[1]);
+        $stderr = stream_get_contents($pipes[2]);
+        fclose($pipes[2]);
+        $exit = proc_close($proc);
+        $this->assertSame(
+            0,
+            $exit,
+            trim($stderr !== false ? $stderr : '')."\n".'compile.php -l failed for grapheme_str_split_literals.php'
+        );
+    }
+}