|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Oc\FieldNotes\Import; |
|
4
|
|
|
|
|
5
|
|
|
use League\Csv\Reader; |
|
6
|
|
|
use Oc\FieldNotes\Exception\FileFormatException; |
|
7
|
|
|
use Symfony\Component\HttpFoundation\File\File; |
|
8
|
|
|
|
|
9
|
|
|
/** |
|
10
|
|
|
* Dear brave soul |
|
11
|
|
|
* Once you are done trying to "optimize" this class, |
|
12
|
|
|
* and break everything but didn't recognize in first place, |
|
13
|
|
|
* please increment the following counter as a warning |
|
14
|
|
|
* for the next brave soul |
|
15
|
|
|
* total_hours_wasted = 42 |
|
16
|
|
|
*/ |
|
17
|
|
|
class FileParser |
|
18
|
|
|
{ |
|
19
|
|
|
/** |
|
20
|
|
|
* @var StructMapper |
|
21
|
|
|
*/ |
|
22
|
|
|
private $structMapper; |
|
23
|
|
|
|
|
24
|
|
|
public function __construct(StructMapper $structMapper) |
|
25
|
|
|
{ |
|
26
|
|
|
$this->structMapper = $structMapper; |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @throws FileFormatException |
|
31
|
|
|
*/ |
|
32
|
|
|
public function parseFile(File $file): array |
|
33
|
|
|
{ |
|
34
|
|
|
$csv = Reader::createFromPath($file->getRealPath()); |
|
35
|
|
|
$csv->setDelimiter(','); |
|
36
|
|
|
$csv->setEnclosure('"'); |
|
37
|
|
|
|
|
38
|
|
|
$rows = $this->getRowsFromCsv($csv); |
|
39
|
|
|
|
|
40
|
|
|
return $this->structMapper->map($rows); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
private function getRowsFromCsv(Reader $csv): array |
|
44
|
|
|
{ |
|
45
|
|
|
$lines = []; |
|
46
|
|
|
$rows = []; |
|
47
|
|
|
$content = $csv->getContent(); |
|
48
|
|
|
$content = mb_convert_encoding($content, 'UTF-8', 'UTF-16LE'); |
|
49
|
|
|
|
|
50
|
|
|
$rawLines = array_filter(explode("\n", $content)); |
|
51
|
|
|
$tmpString = ''; |
|
52
|
|
|
// is used to support line breaks in the comment section |
|
53
|
|
|
foreach ($rawLines as $key => $rawLine) { |
|
54
|
|
|
$columnCheck = count(str_getcsv($rawLines[$key + 1])); |
|
55
|
|
|
$tmpString .= $rawLine; |
|
56
|
|
|
if ($columnCheck === 4 || $rawLines[$key + 1] === null) { |
|
57
|
|
|
$lines[] = $tmpString; |
|
58
|
|
|
$tmpString = ''; |
|
59
|
|
|
} |
|
60
|
|
|
} |
|
61
|
|
|
|
|
62
|
|
|
foreach ($lines as $line) { |
|
63
|
|
|
$row = str_getcsv($line, ',', '"'); |
|
64
|
|
|
$row = array_map('trim', $row); |
|
65
|
|
|
|
|
66
|
|
|
if (count($row) < 4) { |
|
67
|
|
|
throw new FileFormatException('A row contains more or less than 4 columns'); |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
$rows[] = $row; |
|
71
|
|
|
} |
|
72
|
|
|
|
|
73
|
|
|
return $rows; |
|
74
|
|
|
} |
|
75
|
|
|
} |
|
76
|
|
|
|