|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace OfxParser; |
|
4
|
|
|
|
|
5
|
|
|
use SimpleXMLElement; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* An OFX parser library |
|
9
|
|
|
* |
|
10
|
|
|
* Heavily refactored from Guillaume Bailleul's grimfor/ofxparser |
|
11
|
|
|
* |
|
12
|
|
|
* @author Guillaume BAILLEUL <[email protected]> |
|
13
|
|
|
* @author James Titcumb <[email protected]> |
|
14
|
|
|
* @author Oliver Lowe <[email protected]> |
|
15
|
|
|
*/ |
|
16
|
|
|
class Parser |
|
17
|
|
|
{ |
|
18
|
|
|
/** |
|
19
|
|
|
* Factory to extend support for OFX document structures. |
|
20
|
|
|
* @param SimpleXMLElement $xml |
|
21
|
|
|
* @return Ofx |
|
22
|
|
|
*/ |
|
23
|
15 |
|
protected function createOfx(SimpleXMLElement $xml) |
|
24
|
|
|
{ |
|
25
|
15 |
|
return new Ofx($xml); |
|
26
|
|
|
} |
|
27
|
|
|
|
|
28
|
|
|
/** |
|
29
|
|
|
* Load an OFX file into this parser by way of a filename |
|
30
|
|
|
* |
|
31
|
|
|
* @param string $ofxFile A path that can be loaded with file_get_contents |
|
32
|
|
|
* @return Ofx |
|
33
|
|
|
* @throws \InvalidArgumentException |
|
34
|
|
|
*/ |
|
35
|
16 |
|
public function loadFromFile($ofxFile) |
|
36
|
|
|
{ |
|
37
|
16 |
|
if (!file_exists($ofxFile)) { |
|
38
|
1 |
|
throw new \InvalidArgumentException("File '{$ofxFile}' could not be found"); |
|
39
|
|
|
} |
|
40
|
|
|
|
|
41
|
15 |
|
return $this->loadFromString(file_get_contents($ofxFile)); |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
/** |
|
45
|
|
|
* Load an OFX by directly using the text content |
|
46
|
|
|
* |
|
47
|
|
|
* @param string $ofxContent |
|
48
|
|
|
* @return Ofx |
|
49
|
|
|
*/ |
|
50
|
15 |
|
public function loadFromString($ofxContent) |
|
51
|
|
|
{ |
|
52
|
15 |
|
$ofxContent = str_replace(["\r\n", "\r"], "\n", $ofxContent); |
|
53
|
15 |
|
$ofxContent = utf8_encode($ofxContent); |
|
54
|
|
|
|
|
55
|
15 |
|
$sgmlStart = stripos($ofxContent, '<OFX>'); |
|
56
|
15 |
|
$ofxHeader = trim(substr($ofxContent, 0, $sgmlStart)); |
|
57
|
15 |
|
$header = $this->parseHeader($ofxHeader); |
|
58
|
|
|
|
|
59
|
15 |
|
$ofxSgml = trim(substr($ofxContent, $sgmlStart)); |
|
60
|
15 |
|
if (stripos($ofxHeader, '<?xml') === 0) { |
|
61
|
2 |
|
$ofxXml = $ofxSgml; |
|
62
|
|
|
} else { |
|
63
|
13 |
|
$ofxSgml = $this->conditionallyAddNewlines($ofxSgml); |
|
64
|
13 |
|
$ofxXml = $this->convertSgmlToXml($ofxSgml); |
|
65
|
|
|
} |
|
66
|
|
|
|
|
67
|
15 |
|
$xml = $this->xmlLoadString($ofxXml); |
|
68
|
|
|
|
|
69
|
15 |
|
$ofx = $this->createOfx($xml); |
|
70
|
15 |
|
$ofx->buildHeader($header); |
|
71
|
|
|
|
|
72
|
15 |
|
return $ofx; |
|
73
|
|
|
} |
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* Detect if the OFX file is on one line. If it is, add newlines automatically. |
|
77
|
|
|
* |
|
78
|
|
|
* @param string $ofxContent |
|
79
|
|
|
* @return string |
|
80
|
|
|
*/ |
|
81
|
13 |
|
private function conditionallyAddNewlines($ofxContent) |
|
82
|
|
|
{ |
|
83
|
13 |
|
if (preg_match('/<OFX>.*<\/OFX>/', $ofxContent) === 1) { |
|
84
|
1 |
|
return str_replace('<', "\n<", $ofxContent); // add line breaks to allow XML to parse |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
12 |
|
return $ofxContent; |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
/** |
|
91
|
|
|
* Load an XML string without PHP errors - throws exception instead |
|
92
|
|
|
* |
|
93
|
|
|
* @param string $xmlString |
|
94
|
|
|
* @throws \RuntimeException |
|
95
|
|
|
* @return \SimpleXMLElement |
|
96
|
|
|
*/ |
|
97
|
17 |
|
private function xmlLoadString($xmlString) |
|
98
|
|
|
{ |
|
99
|
17 |
|
libxml_clear_errors(); |
|
100
|
17 |
|
libxml_use_internal_errors(true); |
|
101
|
17 |
|
$xml = simplexml_load_string($xmlString); |
|
102
|
|
|
|
|
103
|
17 |
|
if ($errors = libxml_get_errors()) { |
|
104
|
1 |
|
throw new \RuntimeException('Failed to parse OFX: ' . var_export($errors, true)); |
|
105
|
|
|
} |
|
106
|
|
|
|
|
107
|
16 |
|
return $xml; |
|
108
|
|
|
} |
|
109
|
|
|
|
|
110
|
|
|
/** |
|
111
|
|
|
* Detect any unclosed XML tags - if they exist, close them |
|
112
|
|
|
* |
|
113
|
|
|
* @param string $line |
|
114
|
|
|
* @return string |
|
115
|
|
|
*/ |
|
116
|
24 |
|
private function closeUnclosedXmlTags($line) |
|
117
|
|
|
{ |
|
118
|
|
|
// Special case discovered where empty content tag wasn't closed |
|
119
|
24 |
|
$line = trim($line); |
|
120
|
24 |
|
if (preg_match('/<MEMO>$/', $line) === 1) { |
|
121
|
1 |
|
return '<MEMO></MEMO>'; |
|
122
|
|
|
} |
|
123
|
|
|
|
|
124
|
|
|
// Matches: <SOMETHING>blah |
|
125
|
|
|
// Does not match: <SOMETHING> |
|
126
|
|
|
// Does not match: <SOMETHING>blah</SOMETHING> |
|
127
|
23 |
|
if (preg_match( |
|
128
|
23 |
|
"/<([A-Za-z0-9.]+)>([\wà-úÀ-Ú0-9\.\-\_\+\, ;:\[\]\'\&\/\\\*\(\)\+\{\|\}\!\£\$\?=@€£#%±§~`\"]+)$/", |
|
129
|
23 |
|
$line, |
|
130
|
23 |
|
$matches |
|
131
|
|
|
)) { |
|
132
|
17 |
|
return "<{$matches[1]}>{$matches[2]}</{$matches[1]}>"; |
|
133
|
|
|
} |
|
134
|
21 |
|
return $line; |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
/** |
|
138
|
|
|
* Parse the SGML Header to an Array |
|
139
|
|
|
* |
|
140
|
|
|
* @param string $ofxHeader |
|
141
|
|
|
* @param int $sgmlStart |
|
142
|
|
|
* @return array |
|
143
|
|
|
*/ |
|
144
|
15 |
|
private function parseHeader($ofxHeader) |
|
145
|
|
|
{ |
|
146
|
15 |
|
$header = []; |
|
147
|
|
|
|
|
148
|
|
|
|
|
149
|
15 |
|
$ofxHeader = trim($ofxHeader); |
|
150
|
|
|
// Remove empty new lines. |
|
151
|
15 |
|
$ofxHeader = preg_replace('/^\n+/m', '', $ofxHeader); |
|
152
|
|
|
|
|
153
|
|
|
// Check if it's an XML file (OFXv2) |
|
154
|
15 |
|
if(preg_match('/^<\?xml/', $ofxHeader) === 1) { |
|
155
|
|
|
// Only parse OFX headers and not XML headers. |
|
156
|
2 |
|
$ofxHeader = preg_replace('/<\?xml .*?\?>\n?/', '', $ofxHeader); |
|
157
|
2 |
|
$ofxHeader = preg_replace(['/"/', '/\?>/', '/<\?OFX/i'], '', $ofxHeader); |
|
158
|
2 |
|
$ofxHeaderLine = explode(' ', trim($ofxHeader)); |
|
159
|
|
|
|
|
160
|
2 |
|
foreach ($ofxHeaderLine as $value) { |
|
161
|
2 |
|
$tag = explode('=', $value); |
|
162
|
2 |
|
$header[$tag[0]] = $tag[1]; |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
2 |
|
return $header; |
|
166
|
|
|
} |
|
167
|
|
|
|
|
168
|
13 |
|
$ofxHeaderLines = explode("\n", $ofxHeader); |
|
169
|
13 |
|
foreach ($ofxHeaderLines as $value) { |
|
170
|
13 |
|
$tag = explode(':', $value); |
|
171
|
13 |
|
$header[$tag[0]] = $tag[1]; |
|
172
|
|
|
} |
|
173
|
|
|
|
|
174
|
13 |
|
return $header; |
|
175
|
|
|
} |
|
176
|
|
|
|
|
177
|
|
|
/** |
|
178
|
|
|
* Convert an SGML to an XML string |
|
179
|
|
|
* |
|
180
|
|
|
* @param string $sgml |
|
181
|
|
|
* @return string |
|
182
|
|
|
*/ |
|
183
|
16 |
|
private function convertSgmlToXml($sgml) |
|
184
|
|
|
{ |
|
185
|
16 |
|
$sgml = preg_replace('/&(?!#?[a-z0-9]+;)/', '&', $sgml); |
|
186
|
|
|
|
|
187
|
16 |
|
$lines = explode("\n", $sgml); |
|
188
|
16 |
|
$tags = []; |
|
189
|
|
|
|
|
190
|
16 |
|
foreach ($lines as $i => &$line) { |
|
191
|
16 |
|
$line = trim($this->closeUnclosedXmlTags($line)) . "\n"; |
|
192
|
|
|
|
|
193
|
|
|
// Matches tags like <SOMETHING> or </SOMETHING> |
|
194
|
16 |
|
if (!preg_match("/^<(\/?[A-Za-z0-9.]+)>$/", trim($line), $matches)) { |
|
195
|
16 |
|
continue; |
|
196
|
|
|
} |
|
197
|
|
|
|
|
198
|
|
|
// If matches </SOMETHING>, looks back and replaces all tags like |
|
199
|
|
|
// <OTHERTHING> to <OTHERTHING/> until finds the opening tag <SOMETHING> |
|
200
|
16 |
|
if ($matches[1][0] == '/') { |
|
201
|
16 |
|
$tag = substr($matches[1], 1); |
|
202
|
|
|
|
|
203
|
16 |
|
while (($last = array_pop($tags)) && $last[1] != $tag) { |
|
204
|
1 |
|
$lines[$last[0]] = "<{$last[1]}/>"; |
|
205
|
|
|
} |
|
206
|
|
|
} else { |
|
207
|
16 |
|
$tags[] = [$i, $matches[1]]; |
|
208
|
|
|
} |
|
209
|
|
|
} |
|
210
|
|
|
|
|
211
|
16 |
|
return implode("\n", array_map('trim', $lines)); |
|
212
|
|
|
} |
|
213
|
|
|
} |
|
214
|
|
|
|
jacques /
ofxparser
