|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/* For licensing terms, see /license.txt */ |
|
4
|
|
|
|
|
5
|
|
|
use League\Csv\Reader; |
|
6
|
|
|
use PhpOffice\PhpSpreadsheet\Reader\Xls; |
|
7
|
|
|
|
|
8
|
|
|
/** |
|
9
|
|
|
* Class Import |
|
10
|
|
|
* This class provides some functions which can be used when importing data from |
|
11
|
|
|
* external files into Chamilo. |
|
12
|
|
|
*/ |
|
13
|
|
|
class Import |
|
14
|
|
|
{ |
|
15
|
|
|
/** |
|
16
|
|
|
* @param string $path |
|
17
|
|
|
* @param bool $setFirstRowAsHeader |
|
18
|
|
|
* |
|
19
|
|
|
* @return array |
|
20
|
|
|
*/ |
|
21
|
|
|
public static function csv_reader($path, $setFirstRowAsHeader = true) |
|
22
|
|
|
{ |
|
23
|
|
|
return self::csvToArray($path); |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
/** |
|
27
|
|
|
* Guess the CSV delimiter by scanning the first non-empty line. |
|
28
|
|
|
*/ |
|
29
|
|
|
public static function detectCsvSeparator(string $filePath): ?string |
|
30
|
|
|
{ |
|
31
|
|
|
if (!is_readable($filePath)) { |
|
32
|
|
|
return null; |
|
33
|
|
|
} |
|
34
|
|
|
$h = @fopen($filePath, 'rb'); |
|
35
|
|
|
if (!$h) { |
|
|
|
|
|
|
36
|
|
|
return null; |
|
37
|
|
|
} |
|
38
|
|
|
$line = ''; |
|
39
|
|
|
for ($i = 0; $i < 5 && !feof($h); $i++) { |
|
40
|
|
|
$line = fgets($h, 1024 * 1024); |
|
41
|
|
|
if ($line !== false && trim($line) !== '') { |
|
42
|
|
|
break; |
|
43
|
|
|
} |
|
44
|
|
|
} |
|
45
|
|
|
fclose($h); |
|
46
|
|
|
if ($line === false || $line === '') { |
|
47
|
|
|
return null; |
|
48
|
|
|
} |
|
49
|
|
|
|
|
50
|
|
|
$cands = [',', ';', "\t", '|']; |
|
51
|
|
|
$scores = array_fill_keys($cands, 0); |
|
52
|
|
|
$inQuotes = false; |
|
53
|
|
|
$len = strlen($line); |
|
54
|
|
|
for ($i = 0; $i < $len; $i++) { |
|
55
|
|
|
$ch = $line[$i]; |
|
56
|
|
|
if ($ch === '"') { |
|
57
|
|
|
$prev = $i > 0 ? $line[$i - 1] : null; |
|
58
|
|
|
if ($prev !== '\\') { |
|
59
|
|
|
$inQuotes = !$inQuotes; |
|
60
|
|
|
} |
|
61
|
|
|
} elseif (!$inQuotes && isset($scores[$ch])) { |
|
62
|
|
|
$scores[$ch]++; |
|
63
|
|
|
} |
|
64
|
|
|
} |
|
65
|
|
|
arsort($scores); |
|
66
|
|
|
$top = array_key_first($scores); |
|
67
|
|
|
return ($scores[$top] > 0) ? $top : null; |
|
68
|
|
|
} |
|
69
|
|
|
|
|
70
|
|
|
/** |
|
71
|
|
|
* Check if the CSV is comma-separated. |
|
72
|
|
|
* - Default (returnMessage=false): returns bool (true = OK, false = not comma). |
|
73
|
|
|
* - With returnMessage=true: returns true (OK) OR a string with the error message. |
|
74
|
|
|
*/ |
|
75
|
|
|
public static function assertCommaSeparated(string $filePath, bool $returnMessage = false): bool|string |
|
76
|
|
|
{ |
|
77
|
|
|
$det = self::detectCsvSeparator($filePath); |
|
|
|
|
|
|
78
|
|
|
|
|
79
|
|
|
if ($det === null || $det === ',') { |
|
80
|
|
|
return true; |
|
81
|
|
|
} |
|
82
|
|
|
|
|
83
|
|
|
$msg = 'Semicolon (;) delimiter detected. This version of Chamilo requires comma (,) as the CSV separator. Please export your file again as CSV (comma-separated).'; |
|
84
|
|
|
|
|
85
|
|
|
return $returnMessage ? get_lang($msg) : false; |
|
86
|
|
|
} |
|
87
|
|
|
|
|
88
|
|
|
/** |
|
89
|
|
|
* Reads a CSV-file into an array. The first line of the CSV-file should contain the array-keys. |
|
90
|
|
|
* The encoding of the input file is tried to be detected. |
|
91
|
|
|
* The elements of the returned array are encoded in the system encoding. |
|
92
|
|
|
* Example: |
|
93
|
|
|
* FirstName;LastName;Email |
|
94
|
|
|
* John;Doe;[email protected] |
|
95
|
|
|
* Adam;Adams;[email protected] |
|
96
|
|
|
* returns |
|
97
|
|
|
* $result [0]['FirstName'] = 'John'; |
|
98
|
|
|
* $result [0]['LastName'] = 'Doe'; |
|
99
|
|
|
* $result [0]['Email'] = 'john.doe@mail. com'; |
|
100
|
|
|
* $result [1]['FirstName'] = 'Adam'; |
|
101
|
|
|
* ... |
|
102
|
|
|
* |
|
103
|
|
|
* @param string $filename the path to the CSV-file which should be imported |
|
104
|
|
|
* |
|
105
|
|
|
* @return array returns an array (in the system encoding) that contains all data from the CSV-file |
|
106
|
|
|
*/ |
|
107
|
|
|
public static function csvToArray($filename, $delimiter = ','): array |
|
108
|
|
|
{ |
|
109
|
|
|
if (empty($filename)) { |
|
110
|
|
|
return []; |
|
111
|
|
|
} |
|
112
|
|
|
|
|
113
|
|
|
$reader = Reader::createFromPath($filename, 'r'); |
|
114
|
|
|
if ($reader) { |
|
115
|
|
|
$reader->setDelimiter($delimiter); |
|
116
|
|
|
$reader->setHeaderOffset(0); |
|
117
|
|
|
$iterator = $reader->getRecords(); |
|
118
|
|
|
|
|
119
|
|
|
return iterator_to_array($iterator); |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
return []; |
|
123
|
|
|
} |
|
124
|
|
|
|
|
125
|
|
|
/** |
|
126
|
|
|
* @param string $filename |
|
127
|
|
|
* |
|
128
|
|
|
* @return array |
|
129
|
|
|
*/ |
|
130
|
|
|
public static function xlsToArray($filename) |
|
131
|
|
|
{ |
|
132
|
|
|
if (empty($filename)) { |
|
133
|
|
|
return []; |
|
134
|
|
|
} |
|
135
|
|
|
|
|
136
|
|
|
$reader = new Xls(); |
|
137
|
|
|
$spreadsheet = $reader->load($filename); |
|
138
|
|
|
$sheet = $spreadsheet->getActiveSheet(); |
|
139
|
|
|
|
|
140
|
|
|
return $sheet->toArray(); |
|
141
|
|
|
} |
|
142
|
|
|
|
|
143
|
|
|
public static function csvColumnToArray($filename, $columnIndex = 0): array |
|
144
|
|
|
{ |
|
145
|
|
|
if (empty($filename) || !is_readable($filename)) { |
|
146
|
|
|
return []; |
|
147
|
|
|
} |
|
148
|
|
|
|
|
149
|
|
|
$reader = Reader::createFromPath($filename, 'r'); |
|
150
|
|
|
|
|
151
|
|
|
// Use semicolon as delimiter. |
|
152
|
|
|
$reader->setDelimiter(';'); |
|
153
|
|
|
|
|
154
|
|
|
// Ensure we do not include BOM in data (defensive: only if the method exists). |
|
155
|
|
|
// In League CSV v9 the BOM is excluded by default; this call is just a safe guard. |
|
156
|
|
|
if (method_exists($reader, 'includeInputBOM')) { |
|
157
|
|
|
// Do not include the BOM as part of the first value. |
|
158
|
|
|
$reader->includeInputBOM(false); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
// Skip empty records to avoid [null] rows. |
|
162
|
|
|
$reader->skipEmptyRecords(); |
|
163
|
|
|
|
|
164
|
|
|
// Read the requested column as an iterator and materialize it. |
|
165
|
|
|
$values = iterator_to_array($reader->fetchColumn($columnIndex), false); |
|
166
|
|
|
|
|
167
|
|
|
// Extra safety: remove UTF-8 BOM if it somehow slipped into the first value. |
|
168
|
|
|
if (isset($values[0]) && is_string($values[0])) { |
|
169
|
|
|
// Remove leading BOM from the first element only. |
|
170
|
|
|
$values[0] = preg_replace('/^\xEF\xBB\xBF/u', '', $values[0]); |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
return $values; |
|
174
|
|
|
} |
|
175
|
|
|
} |
|
176
|
|
|
|