1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* php-gedcom. |
4
|
|
|
* |
5
|
|
|
* php-gedcom is a library for parsing, manipulating, importing and exporting |
6
|
|
|
* GEDCOM 5.5 files in PHP 5.3+. |
7
|
|
|
* |
8
|
|
|
* @author Kristopher Wilson <[email protected]> |
9
|
|
|
* @copyright Copyright (c) 2010-2013, Kristopher Wilson |
10
|
|
|
* @license MIT |
11
|
|
|
* |
12
|
|
|
* @link http://github.com/mrkrstphr/php-gedcom |
13
|
|
|
*/ |
14
|
|
|
|
15
|
|
|
namespace Gedcom; |
16
|
|
|
|
17
|
|
|
class Parser |
18
|
|
|
{ |
19
|
|
|
protected $_file; |
20
|
|
|
|
21
|
|
|
protected $_gedcom; |
22
|
|
|
|
23
|
|
|
protected $_errorLog = []; |
24
|
|
|
|
25
|
|
|
protected $_linesParsed = 0; |
26
|
|
|
|
27
|
|
|
protected $_line = ''; |
28
|
|
|
|
29
|
|
|
protected $_lineRecord; |
30
|
|
|
|
31
|
|
|
protected $_linePieces = 0; |
32
|
|
|
|
33
|
|
|
protected $_returnedLine = ''; |
34
|
|
|
|
35
|
|
|
public function __construct(Gedcom $gedcom = null) |
36
|
|
|
{ |
37
|
|
|
if (!is_null($gedcom)) { |
38
|
|
|
$this->_gedcom = $gedcom; |
39
|
|
|
} else { |
40
|
|
|
$this->_gedcom = new Gedcom(); |
41
|
|
|
} |
42
|
|
|
} |
43
|
|
|
|
44
|
|
|
public function forward() |
45
|
|
|
{ |
46
|
|
|
// if there was a returned line by back(), set that as our current |
47
|
|
|
// line and blank out the returnedLine variable, otherwise grab |
48
|
|
|
// the next line from the file |
49
|
|
|
|
50
|
|
|
if (!empty($this->_returnedLine)) { |
51
|
|
|
$this->_line = $this->_returnedLine; |
52
|
|
|
$this->_returnedLine = ''; |
53
|
|
|
} else { |
54
|
|
|
$this->_line = fgets($this->_file); |
55
|
|
|
$this->_lineRecord = null; |
56
|
|
|
$this->_linesParsed++; |
57
|
|
|
} |
58
|
|
|
|
59
|
|
|
return $this; |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
public function back() |
63
|
|
|
{ |
64
|
|
|
// our parser object encountered a line it wasn't meant to parse |
65
|
|
|
// store this line for the previous parser to analyze |
66
|
|
|
|
67
|
|
|
$this->_returnedLine = $this->_line; |
68
|
|
|
|
69
|
|
|
return $this; |
70
|
|
|
} |
71
|
|
|
|
72
|
|
|
/** |
73
|
|
|
* Jump to the next level in the GEDCOM that is <= $level. This will leave the parser at the line above |
74
|
|
|
* this level, such that calling $parser->forward() will result in landing at the correct level. |
75
|
|
|
* |
76
|
|
|
* @param int $level |
77
|
|
|
*/ |
78
|
|
|
public function skipToNextLevel($level) |
79
|
|
|
{ |
80
|
|
|
$currentDepth = 999; |
81
|
|
|
|
82
|
|
|
while ($currentDepth > $level) { |
83
|
|
|
$this->forward(); |
84
|
|
|
$record = $this->getCurrentLineRecord(); |
85
|
|
|
$currentDepth = (int) $record[0]; |
86
|
|
|
} |
87
|
|
|
|
88
|
|
|
$this->back(); |
89
|
|
|
} |
90
|
|
|
|
91
|
|
|
public function getGedcom() |
92
|
|
|
{ |
93
|
|
|
return $this->_gedcom; |
94
|
|
|
} |
95
|
|
|
|
96
|
|
|
public function eof() |
97
|
|
|
{ |
98
|
|
|
return feof($this->_file); |
99
|
|
|
} |
100
|
|
|
|
101
|
|
|
/** |
102
|
|
|
* @return string |
103
|
|
|
*/ |
104
|
|
|
public function parseMultiLineRecord() |
105
|
|
|
{ |
106
|
|
|
$record = $this->getCurrentLineRecord(); |
107
|
|
|
|
108
|
|
|
$depth = (int) $record[0]; |
109
|
|
|
$data = isset($record[2]) ? trim($record[2]) : ''; |
110
|
|
|
|
111
|
|
|
$this->forward(); |
112
|
|
|
|
113
|
|
|
while (!$this->eof()) { |
114
|
|
|
$record = $this->getCurrentLineRecord(); |
115
|
|
|
$recordType = strtoupper(trim($record[1])); |
116
|
|
|
$currentDepth = (int) $record[0]; |
117
|
|
|
|
118
|
|
|
if ($currentDepth <= $depth) { |
119
|
|
|
$this->back(); |
120
|
|
|
break; |
121
|
|
|
} |
122
|
|
|
|
123
|
|
|
switch ($recordType) { |
124
|
|
|
case 'CONT': |
125
|
|
|
$data .= "\n"; |
126
|
|
|
|
127
|
|
|
if (isset($record[2])) { |
128
|
|
|
$data .= trim($record[2]); |
129
|
|
|
} |
130
|
|
|
break; |
131
|
|
|
case 'CONC': |
132
|
|
|
if (isset($record[2])) { |
133
|
|
|
$data .= ' '.trim($record[2]); |
134
|
|
|
} |
135
|
|
|
break; |
136
|
|
|
default: |
137
|
|
|
$this->back(); |
138
|
|
|
break 2; |
139
|
|
|
} |
140
|
|
|
|
141
|
|
|
$this->forward(); |
142
|
|
|
} |
143
|
|
|
|
144
|
|
|
return $data; |
145
|
|
|
} |
146
|
|
|
|
147
|
|
|
/** |
148
|
|
|
* @return string The current line |
149
|
|
|
*/ |
150
|
|
|
public function getCurrentLine() |
151
|
|
|
{ |
152
|
|
|
return $this->_line; |
153
|
|
|
} |
154
|
|
|
|
155
|
|
|
public function getCurrentLineRecord($pieces = 3) |
156
|
|
|
{ |
157
|
|
|
if (!is_null($this->_lineRecord)) { |
158
|
|
|
if ($this->_linePieces == $pieces) { |
159
|
|
|
return $this->_lineRecord; |
160
|
|
|
} |
161
|
|
|
} |
162
|
|
|
|
163
|
|
|
if (empty($this->_line)) { |
164
|
|
|
return false; |
165
|
|
|
} |
166
|
|
|
|
167
|
|
|
$line = trim($this->_line); |
168
|
|
|
|
169
|
|
|
$this->_lineRecord = explode(' ', $line, $pieces); |
170
|
|
|
$this->_linePieces = $pieces; |
171
|
|
|
|
172
|
|
|
return $this->_lineRecord; |
173
|
|
|
} |
174
|
|
|
|
175
|
|
|
protected function logError($error) |
176
|
|
|
{ |
177
|
|
|
$this->_errorLog[] = $error; |
178
|
|
|
} |
179
|
|
|
|
180
|
|
|
public function logUnhandledRecord($additionalInfo = '') |
181
|
|
|
{ |
182
|
|
|
$this->logError( |
183
|
|
|
$this->_linesParsed.': (Unhandled) '.trim(implode('|', $this->getCurrentLineRecord())). |
|
|
|
|
184
|
|
|
(!empty($additionalInfo) ? ' - '.$additionalInfo : '') |
185
|
|
|
); |
186
|
|
|
} |
187
|
|
|
|
188
|
|
|
public function logSkippedRecord($additionalInfo = '') |
189
|
|
|
{ |
190
|
|
|
$this->logError( |
191
|
|
|
$this->_linesParsed.': (Skipping) '.trim(implode('|', $this->getCurrentLineRecord())). |
|
|
|
|
192
|
|
|
(!empty($additionalInfo) ? ' - '.$additionalInfo : '') |
193
|
|
|
); |
194
|
|
|
} |
195
|
|
|
|
196
|
|
|
public function getErrors() |
197
|
|
|
{ |
198
|
|
|
return $this->_errorLog; |
199
|
|
|
} |
200
|
|
|
|
201
|
|
|
public function normalizeIdentifier($identifier) |
202
|
|
|
{ |
203
|
|
|
$identifier = trim($identifier); |
204
|
|
|
|
205
|
|
|
return trim($identifier, '@'); |
206
|
|
|
} |
207
|
|
|
|
208
|
|
|
/** |
209
|
|
|
* @param string $fileName |
210
|
|
|
* |
211
|
|
|
* @return Gedcom |
212
|
|
|
*/ |
213
|
|
|
public function parse($fileName) |
214
|
|
|
{ |
215
|
|
|
$this->_file = fopen($fileName, 'r'); //explode("\n", mb_convert_encoding($contents, 'UTF-8')); |
216
|
|
|
|
217
|
|
|
if (!$this->_file) { |
218
|
|
|
return null; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
$this->forward(); |
222
|
|
|
|
223
|
|
|
while (!$this->eof()) { |
224
|
|
|
$record = $this->getCurrentLineRecord(); |
225
|
|
|
|
226
|
|
|
if ($record === false) { |
227
|
|
|
continue; |
228
|
|
|
} |
229
|
|
|
|
230
|
|
|
$depth = (int) $record[0]; |
231
|
|
|
|
232
|
|
|
// We only process 0 level records here. Sub levels are processed |
233
|
|
|
// in methods for those data types (individuals, sources, etc) |
234
|
|
|
|
235
|
|
|
if ($depth == 0) { |
236
|
|
|
// Although not always an identifier (HEAD,TRLR): |
237
|
|
|
if (isset($record[1])) { |
238
|
|
|
$this->normalizeIdentifier($record[1]); |
239
|
|
|
} |
240
|
|
|
|
241
|
|
|
if (isset($record[1]) && trim($record[1]) == 'HEAD') { |
242
|
|
|
\Gedcom\Parser\Head::parse($this); |
243
|
|
|
} elseif (isset($record[2]) && trim($record[2]) == 'SUBN') { |
244
|
|
|
\Gedcom\Parser\Subn::parse($this); |
245
|
|
|
} elseif (isset($record[2]) && trim($record[2]) == 'SUBM') { |
246
|
|
|
\Gedcom\Parser\Subm::parse($this); |
247
|
|
|
} elseif (isset($record[2]) && $record[2] == 'SOUR') { |
248
|
|
|
\Gedcom\Parser\Sour::parse($this); |
249
|
|
|
} elseif (isset($record[2]) && $record[2] == 'INDI') { |
250
|
|
|
\Gedcom\Parser\Indi::parse($this); |
251
|
|
|
} elseif (isset($record[2]) && $record[2] == 'FAM') { |
252
|
|
|
\Gedcom\Parser\Fam::parse($this); |
253
|
|
|
} elseif (isset($record[2]) && substr(trim($record[2]), 0, 4) == 'NOTE') { |
254
|
|
|
\Gedcom\Parser\Note::parse($this); |
255
|
|
|
} elseif (isset($record[2]) && $record[2] == 'REPO') { |
256
|
|
|
\Gedcom\Parser\Repo::parse($this); |
257
|
|
|
} elseif (isset($record[2]) && $record[2] == 'OBJE') { |
258
|
|
|
\Gedcom\Parser\Obje::parse($this); |
259
|
|
|
} elseif (isset($record[1]) && trim($record[1]) == 'TRLR') { |
260
|
|
|
// EOF |
261
|
|
|
break; |
262
|
|
|
} else { |
263
|
|
|
$this->logUnhandledRecord(self::class.' @ '.__LINE__); |
264
|
|
|
} |
265
|
|
|
} else { |
266
|
|
|
$this->logUnhandledRecord(self::class.' @ '.__LINE__); |
267
|
|
|
} |
268
|
|
|
|
269
|
|
|
$this->forward(); |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
return $this->getGedcom(); |
273
|
|
|
} |
274
|
|
|
} |
275
|
|
|
|