|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
declare(strict_types=1); |
|
4
|
|
|
|
|
5
|
|
|
namespace PhpCfdi\SatCatalogosPopulate\Converters; |
|
6
|
|
|
|
|
7
|
|
|
use Iterator; |
|
8
|
|
|
use SplFileObject; |
|
9
|
|
|
|
|
10
|
|
|
use function PhpCfdi\SatCatalogosPopulate\Utils\array_rtrim; |
|
11
|
|
|
|
|
12
|
|
|
class CsvFolderJoinFiles |
|
13
|
|
|
{ |
|
14
|
4 |
|
public function joinFilesInFolder(string $csvFolder): void |
|
15
|
|
|
{ |
|
16
|
4 |
|
$destinations = $this->obtainFilesThatAreSplitted($csvFolder); |
|
17
|
|
|
|
|
18
|
4 |
|
foreach ($destinations as $destination => $files) { |
|
19
|
4 |
|
$this->joinFilesToDestination(array_values($files), $destination); |
|
20
|
|
|
} |
|
21
|
|
|
} |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* @param array<int, string> $files |
|
25
|
|
|
*/ |
|
26
|
5 |
|
public function joinFilesToDestination(array $files, string $destination): void |
|
27
|
|
|
{ |
|
28
|
5 |
|
$skipFirstLines = 0; |
|
29
|
5 |
|
$firstSource = $files[0]; |
|
30
|
5 |
|
file_put_contents($destination, ''); // clear the file contents |
|
31
|
5 |
|
foreach ($files as $i => $source) { |
|
32
|
5 |
|
if ($i > 0 && 0 === $skipFirstLines) { |
|
33
|
5 |
|
$skipFirstLines = $this->findLinesToSkip($firstSource, $source); |
|
34
|
|
|
} |
|
35
|
5 |
|
$skipLastLines = 0; |
|
36
|
5 |
|
if ($this->lastLineContains($source, ['Continúa en'])) { |
|
37
|
5 |
|
$skipLastLines = 1; |
|
38
|
|
|
} |
|
39
|
5 |
|
$this->writeLines($source, $destination, $skipFirstLines, $skipLastLines); |
|
40
|
|
|
} |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
/** |
|
44
|
|
|
* @return array<string, array<int, string>> |
|
45
|
|
|
*/ |
|
46
|
6 |
|
public function obtainFilesThatAreSplitted(string $csvFolder): array |
|
47
|
|
|
{ |
|
48
|
6 |
|
$files = array_filter( |
|
49
|
6 |
|
array_map( |
|
50
|
6 |
|
function ($path): array { |
|
51
|
6 |
|
$file = basename($path); |
|
52
|
6 |
|
$matches = []; |
|
53
|
|
|
if ( |
|
54
|
6 |
|
! preg_match('/^ *(.+)_Parte_([0-9]+) *\.csv$/', $file, $matches) |
|
55
|
6 |
|
&& ! preg_match('/^ *(.+) \(Parte ([0-9]+)\) *\.csv$/', $file, $matches) |
|
56
|
6 |
|
&& ! preg_match('/^ *(.+)_([0-9]+) *\.csv$/', $file, $matches) |
|
57
|
|
|
) { |
|
58
|
4 |
|
return []; |
|
59
|
|
|
} |
|
60
|
|
|
return [ |
|
61
|
6 |
|
'destination' => dirname($path) . '/' . trim($matches[1]) . '.csv', |
|
62
|
6 |
|
'index' => (int) $matches[2], |
|
63
|
|
|
'source' => $path, |
|
64
|
|
|
]; |
|
65
|
|
|
}, |
|
66
|
6 |
|
glob($csvFolder . '/*.csv') ?: [] |
|
67
|
|
|
) |
|
68
|
|
|
); |
|
69
|
|
|
|
|
70
|
6 |
|
uasort($files, $this->compareFiles(...)); |
|
|
|
|
|
|
71
|
|
|
|
|
72
|
6 |
|
$destinations = []; |
|
73
|
6 |
|
foreach ($files as $file) { |
|
74
|
6 |
|
$destinations[strval($file['destination'])][] = strval($file['source']); |
|
75
|
|
|
} |
|
76
|
|
|
|
|
77
|
6 |
|
return $destinations; |
|
78
|
|
|
} |
|
79
|
|
|
|
|
80
|
|
|
/** |
|
81
|
|
|
* @param array{destination: string, index: int} $first |
|
82
|
|
|
* @param array{destination: string, index: int} $second |
|
83
|
|
|
*/ |
|
84
|
6 |
|
private function compareFiles(array $first, array $second): int |
|
85
|
|
|
{ |
|
86
|
6 |
|
return ($first['destination'] <=> $second['destination']) ?: $first['index'] <=> $second['index']; |
|
87
|
|
|
} |
|
88
|
|
|
|
|
89
|
5 |
|
public function writeLines(string $source, string $destination, int $skipFirstLines, int $skipLastLines): void |
|
90
|
|
|
{ |
|
91
|
5 |
|
$command = implode(' ', [ |
|
92
|
5 |
|
'cat ' . escapeshellarg($source), // send the file to the pipes |
|
93
|
5 |
|
sprintf('| tail -n +%d', $skipFirstLines + 1), // without firsts n lines |
|
94
|
5 |
|
sprintf('| head -n -%d', $skipLastLines), // without last n lines |
|
95
|
5 |
|
'>> ' . escapeshellarg($destination), // create/append to destination |
|
96
|
|
|
]); |
|
97
|
5 |
|
shell_exec($command); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
6 |
|
public function findLinesToSkip(string $firstPath, string $secondPath): int |
|
101
|
|
|
{ |
|
102
|
6 |
|
$first = new SplFileObject($firstPath, 'r'); |
|
103
|
6 |
|
$second = new SplFileObject($secondPath, 'r'); |
|
104
|
|
|
|
|
105
|
6 |
|
return $this->findLinesToSkipFromIterators($first, $second); |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
/** |
|
109
|
|
|
* @param Iterator<string> $first |
|
110
|
|
|
* @param Iterator<string> $second |
|
111
|
|
|
*/ |
|
112
|
10 |
|
public function findLinesToSkipFromIterators(Iterator $first, Iterator $second): int |
|
113
|
|
|
{ |
|
114
|
10 |
|
$firstTenLines = $this->splReadTenLines($first); |
|
115
|
10 |
|
$secondTenLines = $this->splReadTenLines($second); |
|
116
|
|
|
|
|
117
|
10 |
|
for ($i = 9; $i > 0; $i--) { |
|
118
|
10 |
|
$firstValue = $firstTenLines[$i] ?? null; |
|
119
|
10 |
|
$secondValue = $secondTenLines[$i] ?? null; |
|
120
|
10 |
|
if (null !== $firstValue && null !== $secondValue && $firstValue === $secondValue) { |
|
121
|
10 |
|
return $i + 1; |
|
122
|
|
|
} |
|
123
|
|
|
} |
|
124
|
1 |
|
return 0; |
|
125
|
|
|
} |
|
126
|
|
|
|
|
127
|
|
|
/** |
|
128
|
|
|
* @param Iterator<string> $iterator |
|
129
|
|
|
* @return array<int, string> |
|
130
|
|
|
*/ |
|
131
|
10 |
|
public function splReadTenLines(Iterator $iterator): array |
|
132
|
|
|
{ |
|
133
|
10 |
|
$result = []; |
|
134
|
10 |
|
$counter = 0; |
|
135
|
10 |
|
foreach ($iterator as $line) { |
|
136
|
10 |
|
$result[] = $this->splCurrentLinesNormalizeValue($line); |
|
137
|
10 |
|
$counter = $counter + 1; |
|
138
|
10 |
|
if (10 === $counter) { |
|
139
|
1 |
|
break; |
|
140
|
|
|
} |
|
141
|
|
|
} |
|
142
|
10 |
|
return $result; |
|
143
|
|
|
} |
|
144
|
|
|
|
|
145
|
10 |
|
private function splCurrentLinesNormalizeValue(string $current): string |
|
146
|
|
|
{ |
|
147
|
|
|
// explode values, trim values, remove empty values at end and implode values back |
|
148
|
10 |
|
return implode(',', array_rtrim( |
|
149
|
10 |
|
array_map(static fn (?string $value): string => trim($value ?? ''), str_getcsv($current)) |
|
150
|
|
|
)); |
|
151
|
|
|
} |
|
152
|
|
|
|
|
153
|
|
|
/** @param string[] $searchterms */ |
|
154
|
5 |
|
public function lastLineContains(string $filename, array $searchterms): bool |
|
155
|
|
|
{ |
|
156
|
5 |
|
$lastline = $this->obtainFileLastLine($filename); |
|
157
|
5 |
|
foreach ($searchterms as $search) { |
|
158
|
5 |
|
if (str_contains($lastline, $search)) { |
|
159
|
5 |
|
return true; |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
|
|
163
|
5 |
|
return false; |
|
164
|
|
|
} |
|
165
|
|
|
|
|
166
|
5 |
|
public function obtainFileLastLine(string $filename): string |
|
167
|
|
|
{ |
|
168
|
5 |
|
$command = sprintf("tail -n 5 %s | grep -v '/^$/' | tail -n 1", escapeshellarg($filename)); |
|
169
|
5 |
|
return shell_exec($command) ?: ''; |
|
170
|
|
|
} |
|
171
|
|
|
} |
|
172
|
|
|
|