feat!: preserve the type of booleans, floats, scientific notations and integers when transforming back and forth between PostgreSQL and PHP

src/MartinGeorgiev/Doctrine/DBAL/Types/BaseIntegerArray.php

@@ -47,9 +47,10 @@ public function transformArrayItemForPHP($item): ?int
             return null;
         }
 
-        $isInvalidPHPInt = !(bool) \preg_match('/^-?\d+$/', (string) $item)
-            || (string) $item < $this->getMinValue()
-            || (string) $item > $this->getMaxValue();
+        $stringValue = (string) $item;
+        $isInvalidPHPInt = !(bool) \preg_match('/^-?\d+$/', $stringValue)
+            || $stringValue < $this->getMinValue()
+            || $stringValue > $this->getMaxValue();
         if ($isInvalidPHPInt) {
             throw new ConversionException(\sprintf('Given value of %s content cannot be transformed to valid PHP integer.', \var_export($item, true)));
         }

src/MartinGeorgiev/Utils/DataStructure.php

@@ -13,78 +13,131 @@
  */
 class DataStructure
 {
+    private const POSTGRESQL_EMPTY_ARRAY = '{}';
+
+    private const POSTGRESQL_NULL_VALUE = 'null';
+
     /**
      * This method supports only single-dimensioned text arrays and
      * relays on the default escaping strategy in PostgreSQL (double quotes).
      */
     public static function transformPostgresTextArrayToPHPArray(string $postgresArray): array
     {
-        $transform = static function (string $textArrayToTransform): array {
-            $indicatesMultipleDimensions = \mb_strpos($textArrayToTransform, '},{') !== false
-                || \mb_strpos($textArrayToTransform, '{{') === 0;
-            if ($indicatesMultipleDimensions) {
-                throw new \InvalidArgumentException('Only single-dimensioned arrays are supported');
-            }
-
-            $phpArray = \str_getcsv(\trim($textArrayToTransform, '{}'), escape: '\\');
-            foreach ($phpArray as $i => $text) {
-                if ($text === null) {
-                    unset($phpArray[$i]);
+        $trimmed = \trim($postgresArray);
 
-                    break;
-                }
+        if ($trimmed === '' || \strtolower($trimmed) === self::POSTGRESQL_NULL_VALUE) {
+            return [];
+        }
 
-                $isInteger = \is_numeric($text) && ''.(int) $text === $text;
-                if ($isInteger) {
-                    $phpArray[$i] = (int) $text;
+        if (\str_contains($trimmed, '},{') || \str_starts_with($trimmed, '{{')) {
+            throw new \InvalidArgumentException('Only single-dimensioned arrays are supported');
+        }
 
-                    continue;
-                }
+        if ($trimmed === self::POSTGRESQL_EMPTY_ARRAY) {
+            return [];
+        }
 
-                $isFloat = \is_numeric($text) && ''.(float) $text === $text;
-                if ($isFloat) {
-                    $phpArray[$i] = (float) $text;
+        $jsonArray = '['.\trim($trimmed, '{}').']';
 
-                    continue;
-                }
+        $decoded = \json_decode($jsonArray, true, 512, JSON_BIGINT_AS_STRING);
+        if ($decoded === null && \json_last_error() !== JSON_ERROR_NONE) {
+            throw new \InvalidArgumentException('Invalid array format: '.\json_last_error_msg());
+        }
 
-                $phpArray[$i] = \stripslashes(\str_replace('\"', '"', $text));
-            }
-
-            return $phpArray;
-        };
-
-        return $transform($postgresArray);
+        return \array_map(
+            static fn ($value): mixed => \is_string($value) ? self::unescapeString($value) : $value,
+            $decoded
+        );
     }
 
     /**
      * This method supports only single-dimensioned PHP arrays.
      * This method relays on the default escaping strategy in PostgreSQL (double quotes).
-     *
-     * @see https://stackoverflow.com/a/5632171/3425372 Kudos to jmz for the inspiration
      */
     public static function transformPHPArrayToPostgresTextArray(array $phpArray): string
     {
-        $transform = static function (array $phpArrayToTransform): string {
-            $result = [];
-            foreach ($phpArrayToTransform as $text) {
-                if (\is_array($text)) {
-                    throw new \InvalidArgumentException('Only single-dimensioned arrays are supported');
-                }
-
-                if (\is_numeric($text) || \ctype_digit($text)) {
-                    $escapedText = $text;
-                } else {
-                    \assert(\is_string($text));
-                    $escapedText = \sprintf('"%s"', \addcslashes($text, '"\\'));
-                }
-
-                $result[] = $escapedText;
-            }
-
-            return '{'.\implode(',', $result).'}';
-        };
-
-        return $transform($phpArray);
+        if ($phpArray === []) {
+            return self::POSTGRESQL_EMPTY_ARRAY;
+        }
+
+        if (\array_filter($phpArray, 'is_array')) {
+            throw new \InvalidArgumentException('Only single-dimensioned arrays are supported');
+        }
+
+        $processed = \array_map(static fn ($value): string => self::formatValue($value), $phpArray);
+
+        return '{'.\implode(',', $processed).'}';
+    }
+
+    /**
+     * Formats a single value for PostgreSQL array.
+     */
+    private static function formatValue(mixed $value): string
+    {
+        // Handle null
+        if ($value === null) {
+            return 'NULL';
+        }
+
+        // Handle actual numbers
+        if (\is_int($value) || \is_float($value)) {
+            return (string) $value;
+        }
+
+        // Convert to string if not already
+        $stringValue = (string) $value;
+
+        // Handle empty string
+        if ($stringValue === '') {
+            return '""';
+        }
+
+        if (self::isNumericSimple($stringValue)) {
+            return '"'.$stringValue.'"';
+        }
+
+        // Double the backslashes and escape quotes
+        $escaped = \str_replace(
+            ['\\', '"'],
+            ['\\\\', '\"'],
+            $stringValue
+        );
+
+        return '"'.$escaped.'"';
+    }
+
+    private static function isNumericSimple(string $value): bool
+    {
+        // Fast path for obvious numeric strings
+        if ($value === '' || $value[0] === '"') {
+            return false;
+        }
+
+        // Handle scientific notation
+        $lower = \strtolower($value);
+        if (\str_contains($lower, 'e')) {
+            $value = \str_replace('e', '', $lower);
+        }
+
+        // Use built-in numeric check
+        return \is_numeric($value);
+    }
+
+    private static function unescapeString(string $value): string
+    {
+        // First handle escaped quotes
+        $value = \str_replace('\"', '___QUOTE___', $value);
+
+        // Handle double backslashes
+        $value = \str_replace('\\\\', '___DBLBACK___', $value);
+
+        // Handle remaining single backslashes
+        $value = \str_replace('\\', '\\', $value);
+
+        // Restore double backslashes
+        $value = \str_replace('___DBLBACK___', '\\\\', $value);
+
+        // Finally restore quotes
+        return \str_replace('___QUOTE___', '"', $value);
     }
 }

tests/MartinGeorgiev/Doctrine/DBAL/Types/TextArrayTest.php

@@ -74,6 +74,8 @@ public static function provideValidTransformations(): array
                 'phpValue' => [
                     1,
                     '2',
+                    3.4,
+                    '5.6',
                     'text',
                     'some text here',
                     'and some here',
@@ -84,7 +86,7 @@ public static function provideValidTransformations(): array
                     'and "double-quotes"',
                 ],
                 'postgresValue' => <<<'END'
-{1,2,"text","some text here","and some here","''\"quotes\"'' ain't no \"\"\"worry\"\"\", '''right''' Alexander O'Vechkin?","back-slashing\\double-slashing\\\\hooking though","and \"double-quotes\""}
+{1,"2",3.4,"5.6","text","some text here","and some here","''\"quotes\"'' ain't no \"\"\"worry\"\"\", '''right''' Alexander O'Vechkin?","back-slashing\\double-slashing\\\\hooking though","and \"double-quotes\""}
 END,
             ],
         ];

tests/MartinGeorgiev/Utils/DataStructureTest.php

@@ -43,26 +43,44 @@
      */
     public static function provideValidTransformations(): array
     {
         return [
-            [
+            'simple integer strings as strings are preserved as strings' => [
                 'phpValue' => [
                     0 => '1',
                     1 => '2',
                     2 => '3',
                     3 => '4',
                 ],
+                'postgresValue' => '{"1","2","3","4"}',
+            ],
+            'simple integer strings' => [
+                'phpValue' => [
+                    0 => 1,
+                    1 => 2,
+                    2 => 3,
+                    3 => 4,
+                ],
                 'postgresValue' => '{1,2,3,4}',
             ],
-            [
+            'decimal numbers represented as strings are preserved as strings' => [
                 'phpValue' => [
                     0 => '1.23',
                     1 => '2.34',
                     2 => '3.45',
                     3 => '4.56',
                 ],
+                'postgresValue' => '{"1.23","2.34","3.45","4.56"}',
+            ],
+            'decimal numbers' => [
+                'phpValue' => [
+                    0 => 1.23,
+                    1 => 2.34,
+                    2 => 3.45,
+                    3 => 4.56,
+                ],
                 'postgresValue' => '{1.23,2.34,3.45,4.56}',
             ],
-            [
+            'mixed content with special characters' => [
                 'phpValue' => [
                     0 => 'dfasdf',
                     1 => 'qw,,e{q"we',
@@ -72,17 +90,49 @@
                 ],
                 'postgresValue' => '{"dfasdf","qw,,e{q\"we","\'qrer\'",604,"\"aaa\",\"b\"\"bb\",\"ccc\""}',
             ],
-            [
+            'empty strings' => [
                 'phpValue' => [
                     0 => '',
                     1 => '',
                 ],
                 'postgresValue' => '{"",""}',
             ],
-            [
+            'empty array' => [
                 'phpValue' => [],
                 'postgresValue' => '{}',
             ],
+            'scientific notation as strings' => [
+                'phpValue' => ['1.23e4', '2.34e5', '3.45e6'],
+                'postgresValue' => '{"1.23e4","2.34e5","3.45e6"}',
+            ],
+            'scientific notation with negative exponents' => [
+                'phpValue' => ['1.23e-4', '2.34e-5', '3.45e-6'],
+                'postgresValue' => '{"1.23e-4","2.34e-5","3.45e-6"}',
+            ],
+            'whole floats that look like integers' => [
+                'phpValue' => ['1.0', '2.00', '3.000', '4.0000'],
+                'postgresValue' => '{"1.0","2.00","3.000","4.0000"}',
+            ],
+            'large integers beyond PHP_INT_MAX' => [
+                'phpValue' => [
+                    '9223372036854775808',  // PHP_INT_MAX + 1
+                    '9999999999999999999',
+                    '-9223372036854775809', // PHP_INT_MIN - 1
+                ],
+                'postgresValue' => '{"9223372036854775808","9999999999999999999","-9223372036854775809"}',
+            ],
+            'mixed numeric formats' => [
+                'phpValue' => [
+                    '1.23',           // regular float string
+                    1.23,             // regular float
+                    '1.230',          // float with trailing zeros
+                    '1.23e4',         // scientific notation
+                    '1.0',            // whole float as string
+                    1.0,              // whole float
+                    '9999999999999999999', // large integer
+                ],
+                'postgresValue' => '{"1.23",1.23,"1.230","1.23e4","1.0",1,"9999999999999999999"}',
+            ],
         ];
     }