|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Endeken\OFX; |
|
4
|
|
|
|
|
5
|
|
|
use RuntimeException; |
|
6
|
|
|
use SimpleXMLElement; |
|
7
|
|
|
|
|
8
|
|
|
class OFXUtils |
|
9
|
|
|
{ |
|
10
|
|
|
public static function normalizeOfx(string $ofxContent): string|false|SimpleXMLElement |
|
11
|
|
|
{ |
|
12
|
|
|
$ofxContent = str_replace(['\r\n'], '\n', $ofxContent); |
|
13
|
|
|
$ofxContent = mb_convert_encoding($ofxContent, 'UTF-8', 'ISO-8859-1'); |
|
14
|
|
|
$sgmlStart = stripos($ofxContent, '<OFX>'); |
|
|
|
|
|
|
15
|
|
|
$ofxHeader = trim(substr($ofxContent, 0, $sgmlStart)); |
|
|
|
|
|
|
16
|
|
|
$header = self::parseHeader($ofxHeader); |
|
|
|
|
|
|
17
|
|
|
$ofxSgml = trim(substr($ofxContent, $sgmlStart)); |
|
18
|
|
|
if (stripos($ofxHeader, '<?xml') === 0) { |
|
19
|
|
|
$ofxXml = $ofxSgml; |
|
20
|
|
|
} else { |
|
21
|
|
|
if (preg_match('/<OFX>.*<\/OFX>/', $ofxSgml) === 1) { |
|
22
|
|
|
$ofxSgml = str_replace('<', "\n<", $ofxSgml); // add line breaks to allow XML to parse |
|
23
|
|
|
} |
|
24
|
|
|
$ofxXml = self::convertSgmlToXml($ofxSgml); |
|
25
|
|
|
} |
|
26
|
|
|
libxml_clear_errors(); |
|
27
|
|
|
libxml_use_internal_errors(true); |
|
28
|
|
|
$xml = simplexml_load_string($ofxXml); |
|
29
|
|
|
|
|
30
|
|
|
if ($errors = libxml_get_errors()) { |
|
31
|
|
|
throw new RuntimeException('Failed to parse OFX: ' . var_export($errors, true)); |
|
32
|
|
|
} |
|
33
|
|
|
|
|
34
|
|
|
return $xml; |
|
35
|
|
|
} |
|
36
|
|
|
|
|
37
|
|
|
private static function parseHeader(string $ofxHeader): array |
|
38
|
|
|
{ |
|
39
|
|
|
$header = []; |
|
40
|
|
|
|
|
41
|
|
|
$ofxHeader = trim($ofxHeader); |
|
42
|
|
|
// Remove empty new lines. |
|
43
|
|
|
$ofxHeader = preg_replace('/^\n+/m', '', $ofxHeader); |
|
44
|
|
|
|
|
45
|
|
|
// Check if it's an XML file (OFXv2) |
|
46
|
|
|
if(preg_match('/^<\?xml/', $ofxHeader) === 1) { |
|
47
|
|
|
// Only parse OFX headers and not XML headers. |
|
48
|
|
|
$ofxHeader = preg_replace('/<\?xml .*?\?>\n?/', '', $ofxHeader); |
|
49
|
|
|
$ofxHeader = preg_replace(['/"/', '/\?>/', '/<\?OFX/i'], '', $ofxHeader); |
|
50
|
|
|
$ofxHeaderLine = explode(' ', trim($ofxHeader)); |
|
51
|
|
|
|
|
52
|
|
|
foreach ($ofxHeaderLine as $value) { |
|
53
|
|
|
$tag = explode('=', $value); |
|
54
|
|
|
$header[$tag[0]] = $tag[1]; |
|
55
|
|
|
} |
|
56
|
|
|
|
|
57
|
|
|
return $header; |
|
58
|
|
|
} |
|
59
|
|
|
|
|
60
|
|
|
$ofxHeaderLines = explode("\n", $ofxHeader); |
|
61
|
|
|
foreach ($ofxHeaderLines as $value) { |
|
62
|
|
|
$tag = explode(':', $value); |
|
63
|
|
|
$header[$tag[0]] = $tag[1]; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
return $header; |
|
67
|
|
|
} |
|
68
|
|
|
|
|
69
|
|
|
private static function convertSgmlToXml($sgml): string |
|
70
|
|
|
{ |
|
71
|
|
|
$sgml = preg_replace('/&(?!#?[a-z0-9]+;)/', '&', $sgml); |
|
72
|
|
|
|
|
73
|
|
|
$lines = explode("\n", $sgml); |
|
74
|
|
|
$tags = []; |
|
75
|
|
|
|
|
76
|
|
|
foreach ($lines as $i => &$line) { |
|
77
|
|
|
$line = trim(self::closeUnclosedXmlTags($line)) . "\n"; |
|
78
|
|
|
|
|
79
|
|
|
// Matches tags like <SOMETHING> or </SOMETHING> |
|
80
|
|
|
if (!preg_match("/^<(\/?[A-Za-z0-9.]+)>$/", trim($line), $matches)) { |
|
81
|
|
|
continue; |
|
82
|
|
|
} |
|
83
|
|
|
|
|
84
|
|
|
// If matches </SOMETHING>, looks back and replaces all tags like |
|
85
|
|
|
// <OTHERTHING> to <OTHERTHING/> until finds the opening tag <SOMETHING> |
|
86
|
|
|
if ($matches[1][0] == '/') { |
|
87
|
|
|
$tag = substr($matches[1], 1); |
|
88
|
|
|
|
|
89
|
|
|
while (($last = array_pop($tags)) && $last[1] != $tag) { |
|
90
|
|
|
$lines[$last[0]] = "<{$last[1]}/>"; |
|
91
|
|
|
} |
|
92
|
|
|
} else { |
|
93
|
|
|
$tags[] = [$i, $matches[1]]; |
|
94
|
|
|
} |
|
95
|
|
|
} |
|
96
|
|
|
|
|
97
|
|
|
return implode("\n", array_map('trim', $lines)); |
|
98
|
|
|
} |
|
99
|
|
|
|
|
100
|
|
|
private static function closeUnclosedXmlTags($line): string |
|
101
|
|
|
{ |
|
102
|
|
|
// Special case discovered where empty content tag wasn't closed |
|
103
|
|
|
$line = trim($line); |
|
104
|
|
|
if (preg_match('/<MEMO>$/', $line) === 1) { |
|
105
|
|
|
return '<MEMO></MEMO>'; |
|
106
|
|
|
} |
|
107
|
|
|
|
|
108
|
|
|
// Matches: <SOMETHING>blah |
|
109
|
|
|
// Does not match: <SOMETHING> |
|
110
|
|
|
// Does not match: <SOMETHING>blah</SOMETHING> |
|
111
|
|
|
if (preg_match( |
|
112
|
|
|
"/<([A-Za-z0-9.]+)>([\wà-úÀ-Ú0-9.\-_+, ;:\[\]'&\/\\\*()+{|}!£\$?=@€£#%±§~`\"]+)$/", |
|
113
|
|
|
$line, |
|
114
|
|
|
$matches |
|
115
|
|
|
)) { |
|
116
|
|
|
return "<$matches[1]>$matches[2]</$matches[1]>"; |
|
117
|
|
|
} |
|
118
|
|
|
return $line; |
|
119
|
|
|
} |
|
120
|
|
|
|
|
121
|
|
|
} |
|
122
|
|
|
|
endeken-com /
ofx-php-parser
Loading history...