1
|
<?php
|
2
|
|
3
|
/**
|
4
|
* @file
|
5
|
* The Node export DSV format handler.
|
6
|
*
|
7
|
* Adds configurable DSV format to Node export.
|
8
|
*/
|
9
|
|
10
|
/**
|
11
|
* Settings callback.
|
12
|
*/
|
13
|
function node_export_dsv_settings($form, $form_state) {
|
14
|
$settings['dsv'] = array(
|
15
|
'#type' => 'fieldset',
|
16
|
'#title' => t('DSV format settings'),
|
17
|
'#description' => t(
|
18
|
'Select how your DSV output will be formatted - this must be configured the
|
19
|
same on both sites. By default this is configured to RFC4180 CSV format
|
20
|
where the delimiter is a comma (,), the enclosure is a double-quote ("),
|
21
|
and the seperator is CRLF (\r\n). Not all configurations may be possible,
|
22
|
use wisely. Enclosure will only be used to escape values that contain any
|
23
|
of the configured strings. Additionally single-quotes will be used to
|
24
|
escape values that are equivalent to reserved words (NULL, TRUE, FALSE).'
|
25
|
),
|
26
|
'#collapsible' => TRUE,
|
27
|
'#collapsed' => TRUE,
|
28
|
);
|
29
|
|
30
|
$settings['dsv']['node_export_dsv_delimiter'] = array(
|
31
|
'#type' => 'textfield',
|
32
|
'#title' => t('Value delimiter'),
|
33
|
'#size' => 5,
|
34
|
'#maxlength' => 255,
|
35
|
'#default_value' => variable_get('node_export_dsv_delimiter', ','),
|
36
|
'#required' => TRUE,
|
37
|
);
|
38
|
|
39
|
$settings['dsv']['node_export_dsv_enclosure'] = array(
|
40
|
'#type' => 'textfield',
|
41
|
'#title' => t('Escape enclosure'),
|
42
|
'#size' => 5,
|
43
|
'#maxlength' => 255,
|
44
|
'#default_value' => variable_get('node_export_dsv_enclosure', '"'),
|
45
|
'#required' => TRUE,
|
46
|
);
|
47
|
|
48
|
$settings['dsv']['node_export_dsv_seperator'] = array(
|
49
|
'#type' => 'textfield',
|
50
|
'#title' => t('Record seperator'),
|
51
|
'#size' => 5,
|
52
|
'#maxlength' => 255,
|
53
|
'#default_value' => variable_get('node_export_dsv_seperator', '\r\n'),
|
54
|
'#required' => TRUE,
|
55
|
);
|
56
|
|
57
|
$settings['dsv']['node_export_dsv_escape_eol'] = array(
|
58
|
'#type' => 'checkbox',
|
59
|
'#title' => t('Always escape values containing line breaks'),
|
60
|
'#default_value' => variable_get('node_export_dsv_escape_eol', 1),
|
61
|
'#description' => t('This is to overcome problems where Windows injects CRLF line breaks.'),
|
62
|
);
|
63
|
|
64
|
return $settings;
|
65
|
}
|
66
|
|
67
|
function node_export_dsv_string($string) {
|
68
|
$replace = array(
|
69
|
'\n' => "\n",
|
70
|
'\r' => "\r",
|
71
|
'\t' => "\t",
|
72
|
'\v' => "\v",
|
73
|
'\e' => "\e",
|
74
|
'\f' => "\f",
|
75
|
);
|
76
|
return str_replace(array_keys($replace), array_values($replace), $string);
|
77
|
}
|
78
|
|
79
|
/**
|
80
|
* Export callback.
|
81
|
*/
|
82
|
function node_export_dsv_export($nodes, $format) {
|
83
|
$delimiter = node_export_dsv_string(variable_get('node_export_dsv_delimiter', ','));
|
84
|
$enclosure = node_export_dsv_string(variable_get('node_export_dsv_enclosure', '"'));
|
85
|
$seperator = node_export_dsv_string(variable_get('node_export_dsv_seperator', '\r\n'));
|
86
|
$escape_eol = variable_get('node_export_dsv_escape_eol', 1);
|
87
|
return node_export_dsv_encode($nodes, $delimiter, $enclosure, $seperator, $escape_eol);
|
88
|
}
|
89
|
|
90
|
/**
|
91
|
* Build DSV string.
|
92
|
*/
|
93
|
function node_export_dsv_encode($nodes, $delimiter, $enclosure, $seperator, $escape_eol) {
|
94
|
$encoded_nodes = array();
|
95
|
$dsv_lines = array();
|
96
|
|
97
|
$node_keys = array();
|
98
|
foreach (array_keys($nodes) as $node_key) {
|
99
|
$new_node_key = 'node_' . $node_key;
|
100
|
$node_keys[] = $new_node_key;
|
101
|
node_export_dsv_encode_node($encoded_nodes, $new_node_key, $nodes[$node_key]);
|
102
|
}
|
103
|
|
104
|
$dsv_lines['node_export_dsv_header'] = array_keys($encoded_nodes);
|
105
|
|
106
|
foreach (array_keys($encoded_nodes) as $header_value) {
|
107
|
$encoded_nodes[$header_value] = array_merge(array_fill_keys($node_keys, NULL), $encoded_nodes[$header_value]);
|
108
|
foreach (array_keys($encoded_nodes[$header_value]) as $encoded_node_key) {
|
109
|
$dsv_lines[$encoded_node_key][$header_value] = $encoded_nodes[$header_value][$encoded_node_key];
|
110
|
}
|
111
|
}
|
112
|
|
113
|
return node_export_dsv_array_to_dsv($dsv_lines, $delimiter, $enclosure, $seperator, $escape_eol);
|
114
|
}
|
115
|
|
116
|
/**
|
117
|
* Process a node and update $header and $encoded_nodes accordingly.
|
118
|
*/
|
119
|
function node_export_dsv_encode_node(&$encoded_nodes, $node_key, $var, $parent = NULL) {
|
120
|
|
121
|
foreach ($var as $k => &$v) {
|
122
|
|
123
|
// Get the new header value.
|
124
|
$header_value = node_export_dsv_encode_header_value($parent, $var, $k);
|
125
|
|
126
|
if (is_object($v) || is_array($v)) {
|
127
|
// Recurse through the structure.
|
128
|
node_export_dsv_encode_node($encoded_nodes, $node_key, $v, $header_value);
|
129
|
}
|
130
|
else {
|
131
|
// Create a safe text version of this value and store it against the header using a safe key.
|
132
|
$encoded_nodes[$header_value][$node_key] = node_export_dsv_encode_sanitize_value($v);
|
133
|
}
|
134
|
}
|
135
|
}
|
136
|
|
137
|
/**
|
138
|
* Encode a value.
|
139
|
*/
|
140
|
function node_export_dsv_encode_sanitize_value($var) {
|
141
|
if (is_numeric($var)) {
|
142
|
return $var;
|
143
|
}
|
144
|
elseif (is_bool($var)) {
|
145
|
return ($var ? 'TRUE' : 'FALSE');
|
146
|
}
|
147
|
elseif (is_null($var)) {
|
148
|
return 'NULL';
|
149
|
}
|
150
|
elseif (is_string($var) && !empty($var)) {
|
151
|
// Single-quote strings that could be confused for null or boolean.
|
152
|
if (in_array(strtoupper($var), array('TRUE', 'FALSE', 'NULL'))) {
|
153
|
$var = "'" . $var . "'";
|
154
|
}
|
155
|
|
156
|
return $var;
|
157
|
}
|
158
|
else {
|
159
|
return '';
|
160
|
}
|
161
|
}
|
162
|
|
163
|
/**
|
164
|
* Decode a value.
|
165
|
*/
|
166
|
function node_export_dsv_decode_sanitize_value($var) {
|
167
|
// Allow numeric, bool, and null values to pass right back as is.
|
168
|
if (is_numeric($var) || is_bool($var) || is_null($var)) {
|
169
|
return $var;
|
170
|
}
|
171
|
// Allow the special case strings back as is.
|
172
|
elseif (in_array(strtoupper($var), array("'TRUE'", "'FALSE'", "'NULL'"))) {
|
173
|
return $var;
|
174
|
}
|
175
|
// Assume this is a string.
|
176
|
return "'" . str_replace("'", "\'", $var) . "'";
|
177
|
}
|
178
|
|
179
|
/**
|
180
|
* Create header value from $parents, $var, and $k.
|
181
|
*/
|
182
|
function node_export_dsv_encode_header_value($parents, $var, $k) {
|
183
|
if (is_null($parents)) {
|
184
|
// Special case; on the first level do not prefix the key.
|
185
|
$header_value = $k;
|
186
|
}
|
187
|
elseif (is_object($var)) {
|
188
|
$header_value = $parents . "->" . $k;
|
189
|
}
|
190
|
elseif (is_array($var)) {
|
191
|
$header_value = $parents . "['" . $k . "']";
|
192
|
}
|
193
|
return $header_value;
|
194
|
}
|
195
|
|
196
|
/**
|
197
|
* Import callback.
|
198
|
*/
|
199
|
function node_export_dsv_import($code_string) {
|
200
|
$delimiter = node_export_dsv_string(variable_get('node_export_dsv_delimiter', ','));
|
201
|
$enclosure = node_export_dsv_string(variable_get('node_export_dsv_enclosure', '"'));
|
202
|
$seperator = node_export_dsv_string(variable_get('node_export_dsv_seperator', '\r\n'));
|
203
|
return node_export_dsv_decode($code_string, $delimiter, $enclosure, $seperator);
|
204
|
}
|
205
|
|
206
|
/**
|
207
|
* Interpret a DSV string.
|
208
|
*/
|
209
|
function node_export_dsv_decode($code_string, $delimiter, $enclosure, $seperator) {
|
210
|
// Get array data from DSV.
|
211
|
$array = @node_export_dsv_dsv_to_array($code_string, $delimiter, $enclosure, $seperator);
|
212
|
|
213
|
// If the first two rows are of equal length, we can assume this is a DSV.
|
214
|
// Also checks there are a decent number of fields.
|
215
|
if (!empty($array[0]) && !empty($array[1]) && count($array[0]) > 10 && count($array[0]) == count($array[1])) {
|
216
|
$nodes = array();
|
217
|
|
218
|
// Assume row 0 is the header, and the rest of the rows are the nodes.
|
219
|
$header = array_shift($array);
|
220
|
|
221
|
// Build the nodes.
|
222
|
foreach ($array as &$row) {
|
223
|
$node = (object)array();
|
224
|
foreach ($row as $key => $item) {
|
225
|
$item = node_export_dsv_decode_sanitize_value($item);
|
226
|
eval('$node->' . $header[$key] . ' = ' . $item . ';');
|
227
|
}
|
228
|
$nodes[] = $node;
|
229
|
}
|
230
|
|
231
|
return $nodes;
|
232
|
}
|
233
|
}
|
234
|
|
235
|
/**
|
236
|
* Encode DSV.
|
237
|
*/
|
238
|
function node_export_dsv_array_to_dsv($array, $delimiter, $enclosure, $seperator, $escape_eol) {
|
239
|
$lines = array();
|
240
|
foreach ($array as $line) {
|
241
|
$out_item = array();
|
242
|
foreach ($line as $item) {
|
243
|
if (stripos($item, $enclosure) !== FALSE) {
|
244
|
$item = str_replace($enclosure, $enclosure . $enclosure, $item);
|
245
|
}
|
246
|
if (
|
247
|
(stripos($item, $delimiter) !== FALSE)
|
248
|
|| (stripos($item, $enclosure) !== FALSE)
|
249
|
|| (stripos($item, $seperator) !== FALSE)
|
250
|
|| ($escape_eol && stripos($item, "\n") !== FALSE)
|
251
|
) {
|
252
|
$item = $enclosure . $item . $enclosure;
|
253
|
}
|
254
|
$out_item[] = $item;
|
255
|
}
|
256
|
$lines[] = implode($delimiter, $out_item);
|
257
|
}
|
258
|
return implode($seperator, $lines);
|
259
|
}
|
260
|
|
261
|
/**
|
262
|
* Decode DSV.
|
263
|
*/
|
264
|
function node_export_dsv_dsv_to_array($string, $delimiter, $enclosure, $seperator) {
|
265
|
$lines = array();
|
266
|
$out_item = array();
|
267
|
$count = strlen($string);
|
268
|
$escape = FALSE;
|
269
|
$double_escape = FALSE;
|
270
|
$position = 0;
|
271
|
$i = 0;
|
272
|
$seperators = str_split($seperator);
|
273
|
|
274
|
while ($i < $count) {
|
275
|
$c = $string[$i];
|
276
|
|
277
|
// Determine whether this is an EOL.
|
278
|
$is_eol = TRUE;
|
279
|
for ($j = 0; $j < count($seperators); $j++) {
|
280
|
if (!isset($string[$i + $j]) || $string[$i + $j] != $seperators[$j]) {
|
281
|
$is_eol = FALSE;
|
282
|
break;
|
283
|
}
|
284
|
}
|
285
|
|
286
|
if ($is_eol) {
|
287
|
if ($escape) {
|
288
|
$out_item[$position] .= $c;
|
289
|
}
|
290
|
else {
|
291
|
$i += count($seperators);
|
292
|
$lines[] = $out_item;
|
293
|
$out_item = array();
|
294
|
$position = 0;
|
295
|
continue;
|
296
|
}
|
297
|
}
|
298
|
elseif ($c == $delimiter) {
|
299
|
if ($escape) {
|
300
|
$out_item[$position] .= $c;
|
301
|
}
|
302
|
else {
|
303
|
if ($string[$i - 1] == $delimiter) {
|
304
|
$out_item[$position] .= '';
|
305
|
}
|
306
|
$position++;
|
307
|
$escape = FALSE;
|
308
|
$double_escape = FALSE;
|
309
|
}
|
310
|
}
|
311
|
elseif ($c == $enclosure) {
|
312
|
if ($double_escape) {
|
313
|
$out_item[$position] .= $enclosure;
|
314
|
$double_escape = FALSE;
|
315
|
}
|
316
|
if ($escape) {
|
317
|
$escape = FALSE;
|
318
|
$double_escape = TRUE;
|
319
|
}
|
320
|
else {
|
321
|
$escape = TRUE;
|
322
|
$double_escape = FALSE;
|
323
|
}
|
324
|
}
|
325
|
else {
|
326
|
if ($double_escape) {
|
327
|
$out_item[$position] .= $enclosure;
|
328
|
$double_escape = FALSE;
|
329
|
}
|
330
|
$out_item[$position] .= $c;
|
331
|
}
|
332
|
$i++;
|
333
|
}
|
334
|
if (!empty($out_item)) {
|
335
|
$lines[] = $out_item;
|
336
|
}
|
337
|
return $lines;
|
338
|
}
|
339
|
|
340
|
/**
|
341
|
* Callback for actions.
|
342
|
*/
|
343
|
function node_export_dsv_action_form($context, &$form_state) {
|
344
|
return node_export_action_form($context, $form_state, 'dsv');
|
345
|
}
|