Merge pull request #16 from mennodekker/master

Partial fix #15 - labels no longer trimmed for multibyte chars

Author: tiamo

src/Sav/Record/Variable.php

@@ -140,9 +140,17 @@ public function write(Buffer $buffer)
         $buffer->writeString($this->name, 8);
 
         if ($hasLabel) {
+            // Maxlength is 255 bytes, since we write utf8 a char can be multiple bytes
             $labelLength = min(mb_strlen($this->label), 255);
-            $buffer->writeInt($labelLength);
-            $buffer->writeString($this->label, Utils::roundUp($labelLength, 4));
+            $label = mb_substr($this->label, 0, $labelLength);
+            $labelLengthBytes = mb_strlen($label, '8bit');
+            while ($labelLengthBytes > 255) {
+                // Strip one char, can be multiple bytes
+                $label = mb_substr($this->label, 0, -1);
+                $labelLengthBytes = mb_strlen($label, '8bit');
+            }
+            $buffer->writeInt($labelLengthBytes);
+            $buffer->writeString($label, Utils::roundUp($labelLengthBytes, 4));
         }
 
         // TODO: test
@@ -156,7 +164,8 @@ public function write(Buffer $buffer)
             }
         }
 
-        $this->writeBlank($buffer, $seg0width);
+        // I think we don't need an empty record
+        //$this->writeBlank($buffer, $seg0width);
 
         // Write additional segments for very long string variables.
         if (self::isVeryLong($this->width)) {

tests/SavRandomReadWriteTest.php

@@ -126,7 +126,8 @@ public function testWriteRead($data)
             //         )
             //     );
             // }
-            $index += $var['width'] > 0 ? Utils::widthToOcts($var['width']) : 1;
+            //$index += $var['width'] > 0 ? Utils::widthToOcts($var['width']) : 1;
+            $index++;
         }
 
         // TODO: valueLabels

tests/WriteMultibyteTest.php

@@ -0,0 +1,51 @@
+<?php
+
+namespace SPSS\Tests;
+
+use SPSS\Sav\Reader;
+use SPSS\Sav\Writer;
+
+class WriteMultibyteTest extends TestCase
+{
+
+    public function testMultiByteLabel()
+    {
+        $data   = [
+            'header'    => [
+                'prodName'     => '@(#) IBM SPSS STATISTICS',
+                'layoutCode'   => 2,
+                'creationDate' => date('d M y'),
+                'creationTime' => date('H:i:s'),
+            ],
+            'variables' => [
+                [
+                    'name'   => 'longname_longerthanexpected',
+                    'label'  => 'Data zákończenia',
+                    'width'  => 16,
+                    'format' => 1,
+                ],
+                [
+                    'name'   => 'ccc',
+                    'label'  => '12345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901233á',
+                    'format' => 5,
+                    'values' => [
+                        1 => 'Panel',
+                    ],
+                ],
+            ],
+        ];
+        $writer = new Writer($data);
+
+        $buffer = $writer->getBuffer();
+        $buffer->rewind();
+
+        $reader = Reader::fromString($buffer->getStream())->read();
+
+        // Sort name
+        $this->assertEquals($data['variables'][0]['label'], $reader->variables[0]->label);
+
+        // Long name
+        $this->assertEquals(mb_substr($data['variables'][1]['label'], 0, -1), $reader->variables[1]->label);
+    }
+
+}