fix: correctly handle message formatting for long truncated UTF8 strings

.github/workflows/tests.yml

@@ -42,3 +42,12 @@ jobs:
 
       - name: Running integration tests
         run: php vendor/bin/phpunit --testsuite=integration
+
+      - name: Running unit tests (no mb_strcut)
+        run: php -d disable_functions=mb_strcut vendor/bin/phpunit --testsuite=unit
+
+      - name: Running functional tests (no mb_strcut)
+        run: php -d disable_functions=mb_strcut vendor/bin/phpunit --testsuite=functional
+
+      - name: Running integration tests (no mb_strcut)
+        run: php -d disable_functions=mb_strcut vendor/bin/phpunit --testsuite=integration

src/Utility/String/StringCutter.php

@@ -0,0 +1,51 @@
+<?php
+
+declare(strict_types=1);
+
+namespace CuyZ\Valinor\Utility\String;
+
+use function substr;
+
+/** @internal */
+final class StringCutter
+{
+    public static function cut(string $s, int $length): string
+    {
+        if (function_exists('mb_strcut')) {
+            return mb_strcut($s, 0, $length);
+        }
+
+        return self::cutPolyfill($s, $length);
+    }
+
+    public static function cutPolyfill(string $s, int $length): string
+    {
+        $s = substr($s, 0, $length);
+        $cur = strlen($s) - 1;
+        // U+0000 - U+007F
+        if ((ord($s[$cur]) & 0b1000_0000) === 0) {
+            return $s;
+        }
+        $cnt = 0;
+        while ((ord($s[$cur]) & 0b1100_0000) === 0b1000_0000) {
+            ++$cnt;
+            if ($cur === 0) {
+                // @infection-ignore-all // Causes infinite loop
+                break;
+            }
+            --$cur;
+        }
+
+        assert($cur >= 0);
+
+        return match (true) {
+            default => substr($s, 0, $cur),
+            // U+0080 - U+07FF
+            $cnt === 1 && (ord($s[$cur]) & 0b1110_0000) === 0b1100_0000,
+            // U+0800 - U+FFFF
+            $cnt === 2 && (ord($s[$cur]) & 0b1111_0000) === 0b1110_0000,
+            // U+10000 - U+10FFFF
+            $cnt === 3 && (ord($s[$cur]) & 0b1111_1000) === 0b1111_0000 => $s
+        };
+    }
+}

src/Utility/String/StringFormatter.php

@@ -38,7 +38,7 @@ public static function for(HasParameters $message): string
     private static function formatWithIntl(string $locale, string $body, array $parameters): string
     {
         return MessageFormatter::formatMessage($locale, $body, $parameters)
-            ?: throw new StringFormatterError($body);
+            ?: throw new StringFormatterError($body, intl_get_error_message());
     }
 
     /**

src/Utility/String/StringFormatterError.php

@@ -9,8 +9,11 @@
 /** @internal */
 final class StringFormatterError extends RuntimeException
 {
-    public function __construct(string $body)
+    public function __construct(string $body, string $message = '')
     {
-        parent::__construct("Message formatter error using `$body`.", 1652901203);
+        if ($message !== '') {
+            $message = ": $message";
+        }
+        parent::__construct("Message formatter error using `$body`$message.", 1652901203);
     }
 }

src/Utility/ValueDumper.php

@@ -5,6 +5,7 @@
 namespace CuyZ\Valinor\Utility;
 
 use BackedEnum;
+use CuyZ\Valinor\Utility\String\StringCutter;
 use DateTimeInterface;
 use UnitEnum;
 
@@ -112,11 +113,11 @@ private static function crop(string $string): string
             return $string;
         }
 
-        $string = substr($string, 0, self::MAX_STRING_LENGTH + 1);
+        $string = StringCutter::cut($string, self::MAX_STRING_LENGTH + 1);
 
         for ($i = strlen($string) - 1; $i > 10; $i--) {
             if ($string[$i] === ' ') {
-                return substr($string, 0, $i) . '…';
+                return StringCutter::cut($string, $i) . '…';
             }
         }
 

tests/Integration/Mapping/Other/StrictMappingTest.php

@@ -192,6 +192,18 @@ public function test_invalid_union_value_throws_exception(): void
         }
     }
 
+    public function test_invalid_utf8_union_value_throws_exception(): void
+    {
+        try {
+            (new MapperBuilder())->mapper()->map('bool|int|float', '🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄');
+        } catch (MappingError $exception) {
+            $error = $exception->node()->messages()[0];
+
+            self::assertSame('1607027306', $error->code());
+            self::assertSame("Value '🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄…' does not match any of `bool`, `int`, `float`.", (string)$error);
+        }
+    }
+
     public function test_null_in_union_value_throws_exception(): void
     {
         try {

tests/Unit/Utility/String/StringCutterTest.php

@@ -0,0 +1,105 @@
+<?php
+
+declare(strict_types=1);
+
+namespace CuyZ\Valinor\Tests\Unit\Utility\String;
+
+use CuyZ\Valinor\Utility\String\StringCutter;
+use PHPUnit\Framework\TestCase;
+
+final class StringCutterTest extends TestCase
+{
+    /**
+     * @dataProvider mb_strcut_polyfill_data_provider
+     */
+    public function test_mb_strcut_polyfill(string $base, int $length, string $expected): void
+    {
+        $cut = StringCutter::cutPolyfill($base, $length);
+
+        self::assertSame($expected, $cut);
+    }
+
+    public function mb_strcut_polyfill_data_provider(): iterable
+    {
+        yield '1 byte' => [
+            'base' => 'foobar',
+            'length' => 3,
+            'expected' => 'foo',
+        ];
+
+        yield '2 bytes not cut' => [
+            'base' => "foo\u{07FF}bar",
+            'length' => 5,
+            'expected' => "foo\u{07FF}",
+        ];
+
+        yield '2 bytes cut' => [
+            'base' => "foo\u{07FF}",
+            'length' => 4,
+            'expected' => 'foo',
+        ];
+
+        yield '3 bytes not cut' => [
+            'base' => "foo\u{FFFF}bar",
+            'length' => 6,
+            'expected' => "foo\u{FFFF}",
+        ];
+
+        yield '3 bytes cut' => [
+            'base' => "foo\u{FFFF}bar",
+            'length' => 5,
+            'expected' => 'foo',
+        ];
+
+        yield '4 bytes not cut #1' => [
+            'base' => "foo\u{10FFFD}bar",
+            'length' => 7,
+            'expected' => "foo\u{10FFFD}",
+        ];
+
+        yield '4 bytes cut #1' => [
+            'base' => "foo\u{10FFFD}bar",
+            'length' => 6,
+            'expected' => 'foo',
+        ];
+
+        yield '4 bytes not cut #2' => [
+            'base' => "foo\u{90000}bar",
+            'length' => 7,
+            'expected' => "foo\u{90000}",
+        ];
+
+        yield '4 bytes not cut #3' => [
+            'base' => "foo\u{40000}bar",
+            'length' => 7,
+            'expected' => "foo\u{40000}",
+        ];
+
+        yield '4 bytes #4' => [
+            'base' => "foo🦄bar",
+            'length' => 7,
+            'expected' => "foo🦄",
+        ];
+
+        yield '4 bytes cut #4' => [
+            'base' => "foo🦄bar",
+            'length' => 6,
+            'expected' => 'foo',
+        ];
+    }
+
+    public function test_invalid_utf8(): void
+    {
+        // Invalid utf8 values are trimmed, if present at the end of the string
+        // (really just an edge case we shouldn't care about)
+
+        $base = "\u{07FF}";
+        $trailer = substr($base, 1);
+        self::assertSame('', StringCutter::cutPolyfill($trailer, 10));
+        self::assertSame('', StringCutter::cutPolyfill($base . $trailer, 10));
+        self::assertSame('', StringCutter::cutPolyfill($base . $trailer . $trailer, 10));
+        self::assertSame('', StringCutter::cutPolyfill($base . $trailer . $trailer . $trailer, 10));
+
+        self::assertSame('', StringCutter::cutPolyfill(substr($base, 0, 1), 10));
+    }
+}

tests/Unit/Utility/String/StringFormatterTest.php

@@ -17,7 +17,7 @@ final class StringFormatterTest extends TestCase
     public function test_wrong_intl_format_throws_exception(): void
     {
         $this->expectException(StringFormatterError::class);
-        $this->expectExceptionMessage('Message formatter error using `some {wrong.format}`.');
+        $this->expectExceptionMessage('Message formatter error using `some {wrong.format}`: pattern syntax error (parse error at offset 6, after "some {", before or at "wrong.format}"): U_PATTERN_SYNTAX_ERROR.');
         $this->expectExceptionCode(1652901203);
 
         StringFormatter::format('en', 'some {wrong.format}', []);
@@ -26,7 +26,7 @@ public function test_wrong_intl_format_throws_exception(): void
     public function test_wrong_message_body_format_throws_exception(): void
     {
         $this->expectException(StringFormatterError::class);
-        $this->expectExceptionMessage('Message formatter error using `some message with {invalid format}`.');
+        $this->expectExceptionMessage('Message formatter error using `some message with {invalid format}`: pattern syntax error (parse error at offset 19, after " message with {", before or at "invalid format}"): U_PATTERN_SYNTAX_ERROR.');
         $this->expectExceptionCode(1652901203);
 
         StringFormatter::format('en', 'some message with {invalid format}');

tests/Unit/Utility/ValueDumperTest.php

@@ -35,6 +35,7 @@ public function dump_value_returns_correct_signature_data_provider(): array
             'string with both quotes' => ['"foo\'bar"', '\'"foo\\\'bar"\''],
             'string with exact max length' => ['Lorem ipsum dolor sit amet, consectetur adipiscing', "'Lorem ipsum dolor sit amet, consectetur adipiscing'"],
             'string cropped' => ['Lorem ipsum dolor sit amet, consectetur adipiscing elit.', "'Lorem ipsum dolor sit amet, consectetur adipiscing…'"],
+            'utf8 string cropped' => ['🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄', "'🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄🦄…'"],
             'string cropped only after threshold' => ['Lorem12345 ipsumdolorsitamet,consecteturadipiscingelit.Curabitur', "'Lorem12345 ipsumdolorsitamet,consecteturadipiscinge…'"],
             'string without space cropped' => ['Loremipsumdolorsitamet,consecteturadipiscingelit.Curabitur',"'Loremipsumdolorsitamet,consecteturadipiscingelit.Cu…'"],
             'date' => [new DateTimeImmutable('@1648733888'), '2022/03/31 13:38:08'],