|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Kasifi\PdfParserBundle\Util; |
|
4
|
|
|
|
|
5
|
|
|
class ParseHelper { |
|
6
|
|
|
/** |
|
7
|
|
|
* @param $data |
|
8
|
|
|
* @param $startConditions |
|
9
|
|
|
* |
|
10
|
|
|
* @return bool|int|null |
|
11
|
|
|
*/ |
|
12
|
|
|
public static function findPosition($data, $startConditions) |
|
13
|
|
|
{ |
|
14
|
|
|
$firstResult = null; |
|
15
|
|
|
foreach ($startConditions as $startCondition) { |
|
16
|
|
|
preg_match($startCondition, $data, $matches); |
|
17
|
|
|
if (count($matches)) { |
|
18
|
|
|
$pos = strpos($data, $matches[0]); |
|
19
|
|
|
if (is_null($firstResult) || $pos < $firstResult) { |
|
20
|
|
|
$firstResult = $pos; |
|
21
|
|
|
} |
|
22
|
|
|
} |
|
23
|
|
|
unset($matches); |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
return $firstResult; |
|
27
|
|
|
} |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* @param $rawRow |
|
31
|
|
|
* |
|
32
|
|
|
* @return array |
|
33
|
|
|
*/ |
|
34
|
|
|
public static function getSpacePositions($rawRow) |
|
35
|
|
|
{ |
|
36
|
|
|
$spacePositions = []; |
|
37
|
|
|
foreach (str_split($rawRow) as $key => $char) { |
|
38
|
|
|
if ($char == ' ') { |
|
39
|
|
|
$spacePositions[] = $key; |
|
40
|
|
|
} |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
return $spacePositions; |
|
44
|
|
|
} |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @param $row |
|
48
|
|
|
* @param $newRow |
|
49
|
|
|
* |
|
50
|
|
|
* @return mixed |
|
51
|
|
|
*/ |
|
52
|
|
|
public static function mergeRows($row, $newRow) |
|
53
|
|
|
{ |
|
54
|
|
|
foreach ($newRow as $newRowColumnIndex => $newRowColumnValue) { |
|
55
|
|
|
if (strlen($newRowColumnValue)) { |
|
56
|
|
|
$row[$newRowColumnIndex] = trim($row[$newRowColumnIndex] . "\n" . $newRowColumnValue); |
|
57
|
|
|
} |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
return $row; |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
/** |
|
64
|
|
|
* @param $colWidths |
|
65
|
|
|
* @param $rawRow |
|
66
|
|
|
* |
|
67
|
|
|
* @return array |
|
68
|
|
|
*/ |
|
69
|
|
|
public static function parseRow($colWidths, $rawRow) |
|
70
|
|
|
{ |
|
71
|
|
|
$colValues = []; |
|
72
|
|
|
foreach ($colWidths as $item) { |
|
73
|
|
|
$colValues[] = trim(substr($rawRow, $item['start'], $item['length'])); |
|
74
|
|
|
} |
|
75
|
|
|
|
|
76
|
|
|
return $colValues; |
|
77
|
|
|
} |
|
78
|
|
|
|
|
79
|
|
|
/** |
|
80
|
|
|
* @param $rows |
|
81
|
|
|
* |
|
82
|
|
|
* @return mixed |
|
83
|
|
|
*/ |
|
84
|
|
|
public static function inlineDates($rows) |
|
85
|
|
|
{ |
|
86
|
|
|
foreach ($rows as &$row) { |
|
87
|
|
|
foreach ($row as &$col) { |
|
88
|
|
|
$col = $col instanceof \DateTime ? $col->format(\DateTime::ISO8601) : $col; |
|
89
|
|
|
} |
|
90
|
|
|
} |
|
91
|
|
|
|
|
92
|
|
|
return $rows; |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* @param $rawRows |
|
97
|
|
|
* @param $skipKeys |
|
98
|
|
|
* @param $rowSkipConditions |
|
99
|
|
|
* |
|
100
|
|
|
* @return array |
|
101
|
|
|
*/ |
|
102
|
|
|
public static function prepareRows($rawRows, $skipKeys, $rowSkipConditions) |
|
103
|
|
|
{ |
|
104
|
|
|
$rows = []; |
|
105
|
|
|
$maxWidth = 0; |
|
106
|
|
|
foreach ($rawRows as $key => $rawRow) { |
|
107
|
|
|
// BYPASS EMPTY ROWS OR SPECIFIED ONES TO BE BYPASSED |
|
108
|
|
|
if (!strlen(trim($rawRow)) || in_array($key, $skipKeys)) { |
|
109
|
|
|
continue; |
|
110
|
|
|
} |
|
111
|
|
|
// SKIP ROWS TO SKIP |
|
112
|
|
|
foreach ($rowSkipConditions as $rowSkipCondition) { |
|
113
|
|
|
if (strpos($rawRow, $rowSkipCondition) !== false) { |
|
114
|
|
|
continue 2; |
|
115
|
|
|
} |
|
116
|
|
|
} |
|
117
|
|
|
|
|
118
|
|
|
if (strlen($rawRow) > $maxWidth) { |
|
119
|
|
|
$maxWidth = strlen($rawRow); |
|
120
|
|
|
} |
|
121
|
|
|
$rows[] = $rawRow; |
|
122
|
|
|
} |
|
123
|
|
|
// SET SAME PADDING FOR EACH ROWS |
|
124
|
|
|
foreach ($rows as &$row) { |
|
125
|
|
|
$row = str_pad($row, $maxWidth, ' '); |
|
126
|
|
|
} |
|
127
|
|
|
unset($row); |
|
128
|
|
|
|
|
129
|
|
|
return $rows; |
|
130
|
|
|
} |
|
131
|
|
|
|
|
132
|
|
|
} |
|
133
|
|
|
|
lucascherifi /
PdfParserBundle
