chinese is mojibake in the generated file

[zhengyuhe123]

When I set valuelabel, variablelable and data in Chinese, the value of the field in the sav file is Mojibake.

[zhengyuhe123]

$writer = new \SPSS\Sav\Writer([ 'header' => [ 'prodName' => '@(#) IBM SPSS STATISTICS 64-bit Macintosh 23.0.0.0', 'creationDate' => '05 Oct 18', 'creationTime' => '01:36:53', 'weightIndex' => 0, ], 'variables' => [ [ 'name' => 'test1', 'format' => Variable::FORMAT_TYPE_F, 'width' => 4, 'decimals' => 2, 'label' => 'test', 'values' => [ 1 => '1测试中文标签1', 2 => '2测试中文标签2', ], // 'missing' => [], 'columns' => 5, 'alignment' => Variable::ALIGN_RIGHT, 'measure' => Variable::MEASURE_SCALE, 'attributes' => [ '$@Role' => Variable::ROLE_PARTITION, ], 'data' => [1, 1, 1], ], [ 'name' => 'test2', 'format' => Variable::FORMAT_TYPE_A, 'width' => 100, 'label' => 'test', 'columns' => 100, 'alignment' => Variable::ALIGN_LEFT, 'measure' => Variable::MEASURE_NOMINAL, 'attributes' => [ '$@Role' => Variable::ROLE_SPLIT, ], 'data' => ['测试中文数据1', '测试中文数据2', '测试中文数据3'], ], ], ]);

[tiamo]

I'll check it as soon as possible.

[zhengyuhe123]

I think the problem is on line 249.

spss/src/Sav/Record/Data.php

Lines 247 to 256 in a17eb3a

	for ($i = $segWidth; $i > 0; $i -= 8, $offset += 8) {
	// $chunkSize = min($i, 8);
	$val = mb_substr($value, $offset, 8);
	if ($val == "") {
	$this->writeOpcode($buffer, $dataBuffer, self::OPCODE_WHITESPACES);
	} else {
	$this->writeOpcode($buffer, $dataBuffer, self::OPCODE_RAW_DATA);
	$dataBuffer->writeString($val, 8);
	}
	}

[mennodekker]

Line 249:
$val = substr($value, $offset, 8);

That seems to do the trick for me. At least in spss the data looks the same. This probably goes for other parts of the code as well. The catch is that we do not chop chars that can be 1, 2 or 3 bytes but really segments of 8 bytes. I also tried mb_strcut but ended up having spaces between the segments.

[mennodekker]

[mennodekker]

I will push a fix with unit test later