AbstractCsv.php
2 weeks ago
ByteSequence.php
2 weeks ago
CannotInsertRecord.php
2 weeks ago
CharsetConverter.php
2 weeks ago
ColumnConsistency.php
2 weeks ago
EncloseField.php
2 weeks ago
EscapeFormula.php
2 weeks ago
Exception.php
9 months ago
HTMLConverter.php
2 weeks ago
Info.php
2 weeks ago
InvalidArgument.php
9 months ago
MapIterator.php
2 weeks ago
RFC4180Field.php
2 weeks ago
Reader.php
2 weeks ago
ResultSet.php
2 weeks ago
Statement.php
2 weeks ago
Stream.php
2 weeks ago
SyntaxError.php
2 weeks ago
TabularDataReader.php
2 weeks ago
UnableToProcessCsv.php
9 months ago
UnavailableFeature.php
9 months ago
UnavailableStream.php
9 months ago
Writer.php
2 weeks ago
XMLConverter.php
2 weeks ago
functions.php
9 months ago
functions_include.php
9 months ago
CharsetConverter.php
186 lines
| 1 | <?php |
| 2 | |
| 3 | /** |
| 4 | * League.Csv (https://csv.thephpleague.com) |
| 5 | * |
| 6 | * (c) Ignace Nyamagana Butera <nyamsprod@gmail.com> |
| 7 | * |
| 8 | * For the full copyright and license information, please view the LICENSE |
| 9 | * file that was distributed with this source code. |
| 10 | */ |
| 11 | declare (strict_types=1); |
| 12 | namespace IAWPSCOPED\League\Csv; |
| 13 | |
| 14 | use OutOfRangeException; |
| 15 | use php_user_filter; |
| 16 | use Traversable; |
| 17 | use function array_combine; |
| 18 | use function array_map; |
| 19 | use function in_array; |
| 20 | use function is_numeric; |
| 21 | use function mb_convert_encoding; |
| 22 | use function mb_list_encodings; |
| 23 | use function preg_match; |
| 24 | use function sprintf; |
| 25 | use function stream_bucket_append; |
| 26 | use function stream_bucket_make_writeable; |
| 27 | use function stream_filter_register; |
| 28 | use function stream_get_filters; |
| 29 | use function strpos; |
| 30 | use function strtolower; |
| 31 | use function substr; |
| 32 | /** |
| 33 | * Converts resource stream or tabular data content charset. |
| 34 | * @internal |
| 35 | */ |
| 36 | class CharsetConverter extends php_user_filter |
| 37 | { |
| 38 | const FILTERNAME = 'convert.league.csv'; |
| 39 | protected string $input_encoding = 'UTF-8'; |
| 40 | protected string $output_encoding = 'UTF-8'; |
| 41 | /** |
| 42 | * Static method to add the stream filter to a {@link AbstractCsv} object. |
| 43 | */ |
| 44 | public static function addTo(AbstractCsv $csv, string $input_encoding, string $output_encoding) : AbstractCsv |
| 45 | { |
| 46 | self::register(); |
| 47 | return $csv->addStreamFilter(self::getFiltername($input_encoding, $output_encoding)); |
| 48 | } |
| 49 | /** |
| 50 | * Static method to register the class as a stream filter. |
| 51 | */ |
| 52 | public static function register() : void |
| 53 | { |
| 54 | $filter_name = self::FILTERNAME . '.*'; |
| 55 | if (!in_array($filter_name, stream_get_filters(), \true)) { |
| 56 | stream_filter_register($filter_name, self::class); |
| 57 | } |
| 58 | } |
| 59 | /** |
| 60 | * Static method to return the stream filter filtername. |
| 61 | */ |
| 62 | public static function getFiltername(string $input_encoding, string $output_encoding) : string |
| 63 | { |
| 64 | return sprintf('%s.%s/%s', self::FILTERNAME, self::filterEncoding($input_encoding), self::filterEncoding($output_encoding)); |
| 65 | } |
| 66 | /** |
| 67 | * Filter encoding charset. |
| 68 | * |
| 69 | * @throws OutOfRangeException if the charset is malformed or unsupported |
| 70 | */ |
| 71 | protected static function filterEncoding(string $encoding) : string |
| 72 | { |
| 73 | static $encoding_list; |
| 74 | if (null === $encoding_list) { |
| 75 | $list = mb_list_encodings(); |
| 76 | $encoding_list = array_combine(array_map('strtolower', $list), $list); |
| 77 | } |
| 78 | $key = strtolower($encoding); |
| 79 | if (isset($encoding_list[$key])) { |
| 80 | return $encoding_list[$key]; |
| 81 | } |
| 82 | throw new OutOfRangeException('The submitted charset ' . $encoding . ' is not supported by the mbstring extension.'); |
| 83 | } |
| 84 | public function onCreate() : bool |
| 85 | { |
| 86 | $prefix = self::FILTERNAME . '.'; |
| 87 | if (0 !== strpos($this->filtername, $prefix)) { |
| 88 | return \false; |
| 89 | } |
| 90 | $encodings = substr($this->filtername, \strlen($prefix)); |
| 91 | if (1 !== preg_match(',^(?<input>[-\\w]+)\\/(?<output>[-\\w]+)$,', $encodings, $matches)) { |
| 92 | return \false; |
| 93 | } |
| 94 | try { |
| 95 | $this->input_encoding = self::filterEncoding($matches['input']); |
| 96 | $this->output_encoding = self::filterEncoding($matches['output']); |
| 97 | } catch (OutOfRangeException $e) { |
| 98 | return \false; |
| 99 | } |
| 100 | return \true; |
| 101 | } |
| 102 | /** |
| 103 | * @param resource $in |
| 104 | * @param resource $out |
| 105 | * @param int $consumed |
| 106 | * @param bool $closing |
| 107 | */ |
| 108 | public function filter($in, $out, &$consumed, $closing) : int |
| 109 | { |
| 110 | while (null !== ($bucket = stream_bucket_make_writeable($in))) { |
| 111 | $bucket->data = @mb_convert_encoding($bucket->data, $this->output_encoding, $this->input_encoding); |
| 112 | $consumed += $bucket->datalen; |
| 113 | stream_bucket_append($out, $bucket); |
| 114 | } |
| 115 | return \PSFS_PASS_ON; |
| 116 | } |
| 117 | /** |
| 118 | * Convert Csv records collection into UTF-8. |
| 119 | */ |
| 120 | public function convert(iterable $records) : iterable |
| 121 | { |
| 122 | if ($this->output_encoding === $this->input_encoding) { |
| 123 | return $records; |
| 124 | } |
| 125 | if (\is_array($records)) { |
| 126 | return array_map($this, $records); |
| 127 | } |
| 128 | /* @var Traversable $records */ |
| 129 | return new MapIterator($records, $this); |
| 130 | } |
| 131 | /** |
| 132 | * Enable using the class as a formatter for the {@link Writer}. |
| 133 | */ |
| 134 | public function __invoke(array $record) : array |
| 135 | { |
| 136 | $outputRecord = []; |
| 137 | foreach ($record as $offset => $value) { |
| 138 | [$newOffset, $newValue] = $this->encodeField($value, $offset); |
| 139 | $outputRecord[$newOffset] = $newValue; |
| 140 | } |
| 141 | return $outputRecord; |
| 142 | } |
| 143 | /** |
| 144 | * Walker method to convert the offset and the value of a CSV record field. |
| 145 | * |
| 146 | * @param int|float|string|null $value can be a scalar type or null |
| 147 | * @param int|string $offset can be a string or an int |
| 148 | */ |
| 149 | protected function encodeField($value, $offset) : array |
| 150 | { |
| 151 | if (null !== $value && !is_numeric($value)) { |
| 152 | $value = mb_convert_encoding($value, $this->output_encoding, $this->input_encoding); |
| 153 | } |
| 154 | if (!is_numeric($offset)) { |
| 155 | $offset = mb_convert_encoding($offset, $this->output_encoding, $this->input_encoding); |
| 156 | } |
| 157 | return [$offset, $value]; |
| 158 | } |
| 159 | /** |
| 160 | * Sets the records input encoding charset. |
| 161 | */ |
| 162 | public function inputEncoding(string $encoding) : self |
| 163 | { |
| 164 | $encoding = self::filterEncoding($encoding); |
| 165 | if ($encoding === $this->input_encoding) { |
| 166 | return $this; |
| 167 | } |
| 168 | $clone = clone $this; |
| 169 | $clone->input_encoding = $encoding; |
| 170 | return $clone; |
| 171 | } |
| 172 | /** |
| 173 | * Sets the records output encoding charset. |
| 174 | */ |
| 175 | public function outputEncoding(string $encoding) : self |
| 176 | { |
| 177 | $encoding = self::filterEncoding($encoding); |
| 178 | if ($encoding === $this->output_encoding) { |
| 179 | return $this; |
| 180 | } |
| 181 | $clone = clone $this; |
| 182 | $clone->output_encoding = $encoding; |
| 183 | return $clone; |
| 184 | } |
| 185 | } |
| 186 |