fix(str): add invariant to avoid unexpected errors when parsing an in…

…valid UTF8 string (#410)
azjezz · Jul 18, 2023 · 1d015ad · 1d015ad
1 parent b0e98cd
commit 1d015ad
Show file tree

Hide file tree

Showing 13 changed files with 185 additions and 13 deletions.
diff --git a/docs/component/str.md b/docs/component/str.md
@@ -53,7 +53,7 @@
 - [range](./../../src/Psl/Str/range.php#L41)
 - [repeat](./../../src/Psl/Str/repeat.php#L26)
 - [replace](./../../src/Psl/Str/replace.php#L15)
-- [replace_ci](./../../src/Psl/Str/replace_ci.php#L16)
+- [replace_ci](./../../src/Psl/Str/replace_ci.php#L20)
 - [replace_every](./../../src/Psl/Str/replace_every.php#L15)
 - [replace_every_ci](./../../src/Psl/Str/replace_every_ci.php#L15)
 - [reverse](./../../src/Psl/Str/reverse.php#L14)
@@ -69,9 +69,9 @@
 - [strip_prefix](./../../src/Psl/Str/strip_prefix.php#L13)
 - [strip_suffix](./../../src/Psl/Str/strip_suffix.php#L13)
 - [to_int](./../../src/Psl/Str/to_int.php#L12)
-- [trim](./../../src/Psl/Str/trim.php#L18)
-- [trim_left](./../../src/Psl/Str/trim_left.php#L18)
-- [trim_right](./../../src/Psl/Str/trim_right.php#L18)
+- [trim](./../../src/Psl/Str/trim.php#L21)
+- [trim_left](./../../src/Psl/Str/trim_left.php#L21)
+- [trim_right](./../../src/Psl/Str/trim_right.php#L21)
 - [truncate](./../../src/Psl/Str/truncate.php#L25)
 - [uppercase](./../../src/Psl/Str/uppercase.php#L14)
 - [width](./../../src/Psl/Str/width.php#L14)

diff --git a/src/Psl/Encoding/Base64/Internal/Base64.php b/src/Psl/Encoding/Base64/Internal/Base64.php
@@ -102,6 +102,8 @@ public static function decode(string $base64, bool $explicit_padding = true): st
         if ($explicit_padding && $base64_length % 4 !== 0) {
             throw new Exception\IncorrectPaddingException('The given base64 string has incorrect padding.');
         }
+
+        /** @psalm-suppress MissingThrowsDocblock */
         $base64 = Str\trim_right($base64, '=');
         $base64_length = Str\length($base64, encoding: Str\Encoding::ASCII_8BIT);
 

diff --git a/src/Psl/Str/replace_ci.php b/src/Psl/Str/replace_ci.php
@@ -4,6 +4,8 @@
 
 namespace Psl\Str;
 
+use Psl\Regex;
+
 use function preg_quote;
 use function preg_split;
 
@@ -12,12 +14,24 @@
  * `$replacement` (case-insensitive).
  *
  * @pure
+ *
+ * @throws Exception\InvalidArgumentException if $needle is not a valid UTF-8 string.
  */
 function replace_ci(string $haystack, string $needle, string $replacement, Encoding $encoding = Encoding::UTF_8): string
 {
     if ('' === $needle || null === search_ci($haystack, $needle, 0, $encoding)) {
         return $haystack;
     }
 
-    return join(preg_split('{' . preg_quote($needle, '/') . '}iu', $haystack), $replacement);
+    try {
+        /** @var list<string> */
+        $pieces = Regex\Internal\call_preg(
+            'preg_split',
+            static fn() => preg_split('{' . preg_quote($needle, '/') . '}iu', $haystack, -1),
+        );
+    } catch (Regex\Exception\RuntimeException | Regex\Exception\InvalidPatternException $error) {
+        throw new Exception\InvalidArgumentException($error->getMessage(), previous: $error);
+    }
+
+    return join($pieces, $replacement);
 }
diff --git a/src/Psl/Str/trim.php b/src/Psl/Str/trim.php
@@ -4,8 +4,9 @@
 
 namespace Psl\Str;
 
+use Psl\Regex;
+
 use function preg_quote;
-use function preg_replace;
 
 /**
  * Returns the given string with whitespace stripped from the beginning and end.
@@ -14,11 +15,17 @@
  * be stripped: space, tab, newline, carriage return, NUL byte, vertical tab.
  *
  * @pure
+ *
+ * @throws Exception\InvalidArgumentException if $string is not a valid UTF-8 string.
  */
 function trim(string $string, ?string $char_mask = null): string
 {
     $char_mask ??= " \t\n\r\0\x0B\x0C\u{A0}\u{FEFF}";
     $char_mask   = preg_quote($char_mask, null);
 
-    return preg_replace("{^[{$char_mask}]++|[{$char_mask}]++$}uD", '', $string);
+    try {
+        return Regex\replace($string, "{^[{$char_mask}]++|[{$char_mask}]++$}uD", '');
+    } catch (Regex\Exception\RuntimeException | Regex\Exception\InvalidPatternException $error) {
+        throw new Exception\InvalidArgumentException($error->getMessage(), previous: $error);
+    }
 }
diff --git a/src/Psl/Str/trim_left.php b/src/Psl/Str/trim_left.php
@@ -4,8 +4,9 @@
 
 namespace Psl\Str;
 
+use Psl\Regex;
+
 use function preg_quote;
-use function preg_replace;
 
 /**
  * Returns the given string with whitespace stripped from the left.
@@ -14,11 +15,17 @@
  * be stripped: space, tab, newline, carriage return, NUL byte, vertical tab.
  *
  * @pure
+ *
+ * @throws Exception\InvalidArgumentException if $string is not a valid UTF-8 string.
  */
 function trim_left(string $string, ?string $char_mask = null): string
 {
     $char_mask ??= " \t\n\r\0\x0B\x0C\u{A0}\u{FEFF}";
     $char_mask   = preg_quote($char_mask, null);
 
-    return preg_replace("{^[{$char_mask}]++}uD", '', $string);
+    try {
+        return Regex\replace($string, "{^[{$char_mask}]++}uD", '');
+    } catch (Regex\Exception\RuntimeException | Regex\Exception\InvalidPatternException $error) {
+        throw new Exception\InvalidArgumentException($error->getMessage(), previous: $error);
+    }
 }
diff --git a/src/Psl/Str/trim_right.php b/src/Psl/Str/trim_right.php
@@ -4,8 +4,9 @@
 
 namespace Psl\Str;
 
+use Psl\Regex;
+
 use function preg_quote;
-use function preg_replace;
 
 /**
  * Returns the given string with whitespace stripped from the right.
@@ -14,11 +15,17 @@
  * be stripped: space, tab, newline, carriage return, NUL byte, vertical tab.
  *
  * @pure
+ *
+ * @throws Exception\InvalidArgumentException if $string is not a valid UTF-8 string.
  */
 function trim_right(string $string, ?string $char_mask = null): string
 {
     $char_mask ??= " \t\n\r\0\x0B\x0C\u{A0}\u{FEFF}";
     $char_mask   = preg_quote($char_mask, null);
 
-    return preg_replace("{[{$char_mask}]++$}uD", '', $string);
+    try {
+        return Regex\replace($string, "{[{$char_mask}]++$}uD", '');
+    } catch (Regex\Exception\RuntimeException | Regex\Exception\InvalidPatternException $error) {
+        throw new Exception\InvalidArgumentException($error->getMessage(), previous: $error);
+    }
 }
diff --git a/src/Psl/Type/Internal/LiteralScalarType.php b/src/Psl/Type/Internal/LiteralScalarType.php
@@ -122,6 +122,7 @@ public function toString(): string
         }
 
         if (Type\float()->matches($value)) {
+            /** @psalm-suppress MissingThrowsDocblock */
             $string_representation = Str\trim_right(Str\format('%.14F', $value), '0');
             /** @psalm-suppress MissingThrowsDocblock */
             if (Str\ends_with($string_representation, '.')) {

diff --git a/src/Psl/Type/Internal/PositiveIntType.php b/src/Psl/Type/Internal/PositiveIntType.php
@@ -47,8 +47,13 @@ public function coerce(mixed $value): int
                 return $int;
             }
 
-            $trimmed = Str\trim_left($str, '0');
-            $int     = Str\to_int($trimmed);
+            try {
+                $trimmed = Str\trim_left($str, '0');
+            } catch (Str\Exception\InvalidArgumentException $e) {
+                throw CoercionException::withValue($value, $this->toString(), $this->getTrace());
+            }
+
+            $int = Str\to_int($trimmed);
             if (null !== $int && $int > 0) {
                 return $int;
             }

diff --git a/tests/unit/Str/ReplaceCiTest.php b/tests/unit/Str/ReplaceCiTest.php
@@ -26,4 +26,36 @@ public function provideData(): array
             ['foo', 'foo', 'bar', 'baz'],
         ];
     }
+
+    /**
+     * @dataProvider provideBadUtf8Data
+     */
+    public function testBadUtf8(string $string, string $expectedException, string $expectedExceptionMessage): void
+    {
+        $this->expectException($expectedException);
+        $this->expectExceptionMessage($expectedExceptionMessage);
+
+        Str\replace_ci($string, $string, $string);
+    }
+
+    public function provideBadUtf8Data(): iterable
+    {
+        yield [
+            "\xc1\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Compilation failed: UTF-8 error: overlong 2-byte sequence at offset 0',
+        ];
+
+        yield [
+            "\xe0\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Compilation failed: UTF-8 error: overlong 3-byte sequence at offset 0',
+        ];
+
+        yield [
+            "\xf0\x80\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Compilation failed: UTF-8 error: overlong 4-byte sequence at offset 0',
+        ];
+    }
 }
diff --git a/tests/unit/Str/TrimLeftTest.php b/tests/unit/Str/TrimLeftTest.php
@@ -47,4 +47,36 @@ public function provideData(): array
             ],
         ];
     }
+
+    /**
+     * @dataProvider provideBadUtf8Data
+     */
+    public function testBadUtf8(string $string, string $expectedException, string $expectedExceptionMessage): void
+    {
+        $this->expectException($expectedException);
+        $this->expectExceptionMessage($expectedExceptionMessage);
+
+        Str\trim_left($string);
+    }
+
+    public function provideBadUtf8Data(): iterable
+    {
+        yield [
+            "\xc1\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xe0\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xf0\x80\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+    }
 }
diff --git a/tests/unit/Str/TrimRightTest.php b/tests/unit/Str/TrimRightTest.php
@@ -57,4 +57,36 @@ public function provideData(): array
             ],
         ];
     }
+
+    /**
+     * @dataProvider provideBadUtf8Data
+     */
+    public function testBadUtf8(string $string, string $expectedException, string $expectedExceptionMessage): void
+    {
+        $this->expectException($expectedException);
+        $this->expectExceptionMessage($expectedExceptionMessage);
+
+        Str\trim_right($string);
+    }
+
+    public function provideBadUtf8Data(): iterable
+    {
+        yield [
+            "\xc1\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xe0\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xf0\x80\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+    }
 }
diff --git a/tests/unit/Str/TrimTest.php b/tests/unit/Str/TrimTest.php
@@ -52,4 +52,36 @@ public function provideData(): array
             ],
         ];
     }
+
+    /**
+     * @dataProvider provideBadUtf8Data
+     */
+    public function testBadUtf8(string $string, string $expectedException, string $expectedExceptionMessage): void
+    {
+        $this->expectException($expectedException);
+        $this->expectExceptionMessage($expectedExceptionMessage);
+
+        Str\trim($string);
+    }
+
+    public function provideBadUtf8Data(): iterable
+    {
+        yield [
+            "\xc1\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xe0\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+
+        yield [
+            "\xf0\x80\x81\xbf",
+            Str\Exception\InvalidArgumentException::class,
+            'Malformed UTF-8 characters, possibly incorrectly encoded',
+        ];
+    }
 }
diff --git a/tests/unit/Type/PositiveIntTypeTest.php b/tests/unit/Type/PositiveIntTypeTest.php
@@ -59,6 +59,7 @@ public function getInvalidCoercions(): iterable
         yield [$this->stringable('-9223372036854775809')];
         yield ['0xFF'];
         yield ['-0xFF'];
+        yield ["\xc1\xbf"];
         yield [''];
     }