1
|
|
|
<?php |
2
|
|
|
/** |
3
|
|
|
* Generate sitemap files in base XML as well as popular namespace extensions. |
4
|
|
|
* |
5
|
|
|
* @author Automattic |
6
|
|
|
* @link http://sitemaps.org/protocol.php Base sitemaps protocol. |
7
|
|
|
* @link http://www.google.com/support/webmasters/bin/answer.py?answer=74288 Google news sitemaps. |
8
|
|
|
*/ |
9
|
|
|
|
10
|
|
|
|
11
|
|
|
/** |
12
|
|
|
* Convert a MySQL datetime string to an ISO 8601 string. |
13
|
|
|
* |
14
|
|
|
* @module sitemaps |
15
|
|
|
* |
16
|
|
|
* @link http://www.w3.org/TR/NOTE-datetime W3C date and time formats document. |
17
|
|
|
* |
18
|
|
|
* @param string $mysql_date UTC datetime in MySQL syntax of YYYY-MM-DD HH:MM:SS. |
19
|
|
|
* |
20
|
|
|
* @return string ISO 8601 UTC datetime string formatted as YYYY-MM-DDThh:mm:ssTZD where timezone offset is always +00:00. |
21
|
|
|
*/ |
22
|
|
|
function jetpack_w3cdate_from_mysql( $mysql_date ) { |
23
|
|
|
return str_replace( ' ', 'T', $mysql_date ) . '+00:00'; |
24
|
|
|
} |
25
|
|
|
|
26
|
|
|
/** |
27
|
|
|
* Get the maximum comment_date_gmt value for approved comments for the given post_id. |
28
|
|
|
* |
29
|
|
|
* @module sitemaps |
30
|
|
|
* |
31
|
|
|
* @param int $post_id Post identifier. |
32
|
|
|
* |
33
|
|
|
* @return string datetime MySQL value or null if no comment found. |
34
|
|
|
*/ |
35
|
|
|
function jetpack_get_approved_comments_max_datetime( $post_id ) { |
36
|
|
|
global $wpdb; |
37
|
|
|
|
38
|
|
|
return $wpdb->get_var( $wpdb->prepare( "SELECT MAX(comment_date_gmt) FROM $wpdb->comments WHERE comment_post_ID = %d AND comment_approved = '1' AND comment_type=''", $post_id ) ); |
39
|
|
|
} |
40
|
|
|
|
41
|
|
|
/** |
42
|
|
|
* Return the content type used to serve a Sitemap XML file. |
43
|
|
|
* Uses text/xml by default, possibly overridden by jetpack_sitemap_content_type filter. |
44
|
|
|
* |
45
|
|
|
* @module sitemaps |
46
|
|
|
* |
47
|
|
|
* @return string Internet media type for the sitemap XML. |
48
|
|
|
*/ |
49
|
|
|
function jetpack_sitemap_content_type() { |
50
|
|
|
/** |
51
|
|
|
* Filter the content type used to serve the XML sitemap file. |
52
|
|
|
* |
53
|
|
|
* @module sitemaps |
54
|
|
|
* |
55
|
|
|
* @since 3.9.0 |
56
|
|
|
* |
57
|
|
|
* @param string $content_type By default, it's 'text/xml'. |
58
|
|
|
*/ |
59
|
|
|
return apply_filters( 'jetpack_sitemap_content_type', 'text/xml' ); |
60
|
|
|
} |
61
|
|
|
|
62
|
|
|
/** |
63
|
|
|
* Write an XML tag. |
64
|
|
|
* |
65
|
|
|
* @module sitemaps |
66
|
|
|
* |
67
|
|
|
* @param array $data Information to write an XML tag. |
68
|
|
|
*/ |
69
|
|
|
function jetpack_print_sitemap_item( $data ) { |
70
|
|
|
jetpack_print_xml_tag( array( 'url' => $data ) ); |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
/** |
74
|
|
|
* Write an opening tag and its matching closing tag. |
75
|
|
|
* |
76
|
|
|
* @module sitemaps |
77
|
|
|
* |
78
|
|
|
* @param array $array Information to write a tag, opening and closing it. |
79
|
|
|
*/ |
80
|
|
|
function jetpack_print_xml_tag( $array ) { |
81
|
|
|
foreach ( $array as $key => $value ) { |
82
|
|
|
if ( is_array( $value ) ) { |
83
|
|
|
echo "<$key>"; |
84
|
|
|
jetpack_print_xml_tag( $value ); |
85
|
|
|
echo "</$key>"; |
86
|
|
|
} else { |
87
|
|
|
echo "<$key>" . esc_html( $value ) . "</$key>"; |
88
|
|
|
} |
89
|
|
|
} |
90
|
|
|
} |
91
|
|
|
|
92
|
|
|
/** |
93
|
|
|
* Convert an array to a SimpleXML child of the passed tree. |
94
|
|
|
* |
95
|
|
|
* @module sitemaps |
96
|
|
|
* |
97
|
|
|
* @param array $data array containing element value pairs, including other arrays, for XML contruction. |
98
|
|
|
* @param SimpleXMLElement $tree A SimpleXMLElement class object used to attach new children. |
99
|
|
|
* |
100
|
|
|
* @return SimpleXMLElement full tree with new children mapped from array. |
101
|
|
|
*/ |
102
|
|
|
function jetpack_sitemap_array_to_simplexml( $data, &$tree ) { |
103
|
|
|
$doc_namespaces = $tree->getDocNamespaces(); |
104
|
|
|
|
105
|
|
|
foreach ( $data as $key => $value ) { |
106
|
|
|
// Allow namespaced keys by use of colon in $key, namespaces must be part of the document |
107
|
|
|
$namespace = null; |
108
|
|
|
if ( false !== strpos( $key, ':' ) && 'image' != $key ) { |
109
|
|
|
list( $namespace_prefix, $key ) = explode( ':', $key ); |
110
|
|
|
if ( isset( $doc_namespaces[ $namespace_prefix ] ) ) { |
111
|
|
|
$namespace = $doc_namespaces[ $namespace_prefix ]; |
112
|
|
|
} |
113
|
|
|
} |
114
|
|
|
|
115
|
|
|
if ( 'image' != $key ) { |
116
|
|
|
if ( is_array( $value ) ) { |
117
|
|
|
$child = $tree->addChild( $key, null, $namespace ); |
118
|
|
|
jetpack_sitemap_array_to_simplexml( $value, $child ); |
119
|
|
|
} else { |
120
|
|
|
$tree->addChild( $key, esc_html( $value ), $namespace ); |
121
|
|
|
} |
122
|
|
|
} elseif ( is_array( $value ) ) { |
123
|
|
|
foreach ( $value as $image ) { |
124
|
|
|
$child = $tree->addChild( $key, null, $namespace ); |
125
|
|
|
jetpack_sitemap_array_to_simplexml( $image, $child ); |
126
|
|
|
} |
127
|
|
|
} |
128
|
|
|
} |
129
|
|
|
|
130
|
|
|
return $tree; |
131
|
|
|
} |
132
|
|
|
|
133
|
|
|
/** |
134
|
|
|
* Define an array of attribute value pairs for use inside the root element of an XML document. |
135
|
|
|
* Intended for mapping namespace and namespace URI values. |
136
|
|
|
* Passes array through jetpack_sitemap_ns for other functions to add their own namespaces. |
137
|
|
|
* |
138
|
|
|
* @module sitemaps |
139
|
|
|
* |
140
|
|
|
* @return array array of attribute value pairs passed through the jetpack_sitemap_ns filter |
141
|
|
|
*/ |
142
|
|
|
function jetpack_sitemap_namespaces() { |
143
|
|
|
/** |
144
|
|
|
* Filter the attribute value pairs used for namespace and namespace URI mappings. |
145
|
|
|
* |
146
|
|
|
* @module sitemaps |
147
|
|
|
* |
148
|
|
|
* @since 3.9.0 |
149
|
|
|
* |
150
|
|
|
* @param array $namespaces Associative array with namespaces and namespace URIs. |
151
|
|
|
*/ |
152
|
|
|
return apply_filters( 'jetpack_sitemap_ns', array( |
153
|
|
|
'xmlns:xsi' => 'http://www.w3.org/2001/XMLSchema-instance', |
154
|
|
|
'xsi:schemaLocation' => 'http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd', |
155
|
|
|
'xmlns' => 'http://www.sitemaps.org/schemas/sitemap/0.9', |
156
|
|
|
// Mobile namespace from http://support.google.com/webmasters/bin/answer.py?hl=en&answer=34648 |
157
|
|
|
'xmlns:mobile' => 'http://www.google.com/schemas/sitemap-mobile/1.0', |
158
|
|
|
'xmlns:image' => 'http://www.google.com/schemas/sitemap-image/1.1', |
159
|
|
|
) ); |
160
|
|
|
} |
161
|
|
|
|
162
|
|
|
/** |
163
|
|
|
* Start sitemap XML document, writing its heading and <urlset> tag with namespaces. |
164
|
|
|
* |
165
|
|
|
* @module sitemaps |
166
|
|
|
* |
167
|
|
|
* @param $charset string Charset for current XML document. |
168
|
|
|
* |
169
|
|
|
* @return string |
170
|
|
|
*/ |
171
|
|
|
function jetpack_sitemap_initstr( $charset ) { |
172
|
|
|
global $wp_rewrite; |
173
|
|
|
// URL to XSLT |
174
|
|
|
if ( $wp_rewrite->using_index_permalinks() ) { |
175
|
|
|
$xsl = home_url( '/index.php/sitemap.xsl' ); |
176
|
|
|
} else if ( $wp_rewrite->using_permalinks() ) { |
177
|
|
|
$xsl = home_url( '/sitemap.xsl' ); |
178
|
|
|
} else { |
179
|
|
|
$xsl = home_url( '/?jetpack-sitemap-xsl=true' ); |
180
|
|
|
} |
181
|
|
|
|
182
|
|
|
$initstr = '<?xml version="1.0" encoding="' . $charset . '"?>' . "\n"; |
183
|
|
|
$initstr .= '<?xml-stylesheet type="text/xsl" href="' . esc_url( $xsl ) . '"?>' . "\n"; |
184
|
|
|
$initstr .= '<!-- generator="jetpack-' . JETPACK__VERSION . '" -->' . "\n"; |
185
|
|
|
$initstr .= '<urlset'; |
186
|
|
|
foreach ( jetpack_sitemap_namespaces() as $attribute => $value ) { |
187
|
|
|
$initstr .= ' ' . esc_html( $attribute ) . '="' . esc_attr( $value ) . '"'; |
188
|
|
|
} |
189
|
|
|
$initstr .= ' />'; |
190
|
|
|
|
191
|
|
|
return $initstr; |
192
|
|
|
} |
193
|
|
|
|
194
|
|
|
/** |
195
|
|
|
* Load XSLT for sitemap. |
196
|
|
|
* |
197
|
|
|
* @module sitemaps |
198
|
|
|
* |
199
|
|
|
* @param string $type XSLT to load. |
200
|
|
|
*/ |
201
|
|
|
function jetpack_load_xsl( $type = '' ) { |
202
|
|
|
|
203
|
|
|
$transient_xsl = empty( $type ) ? 'jetpack_sitemap_xsl' : "jetpack_{$type}_sitemap_xsl"; |
204
|
|
|
|
205
|
|
|
$xsl = get_transient( $transient_xsl ); |
206
|
|
|
|
207
|
|
|
if ( $xsl ) { |
208
|
|
|
header( 'Content-Type: ' . jetpack_sitemap_content_type(), true ); |
209
|
|
|
echo $xsl; |
210
|
|
|
die(); |
|
|
|
|
211
|
|
|
} |
212
|
|
|
|
213
|
|
|
// Populate $xsl. Use $type. |
214
|
|
|
include_once JETPACK__PLUGIN_DIR . 'modules/sitemaps/sitemap-xsl.php'; |
215
|
|
|
|
216
|
|
|
if ( ! empty( $xsl ) ) { |
217
|
|
|
set_transient( $transient_xsl, $xsl, DAY_IN_SECONDS ); |
218
|
|
|
echo $xsl; |
219
|
|
|
} |
220
|
|
|
|
221
|
|
|
die(); |
|
|
|
|
222
|
|
|
} |
223
|
|
|
|
224
|
|
|
/** |
225
|
|
|
* Responds with an XSLT to stylize sitemap. |
226
|
|
|
* |
227
|
|
|
* @module sitemaps |
228
|
|
|
*/ |
229
|
|
|
function jetpack_print_sitemap_xsl() { |
230
|
|
|
jetpack_load_xsl(); |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* Responds with an XSLT to stylize news sitemap. |
235
|
|
|
* |
236
|
|
|
* @module sitemaps |
237
|
|
|
*/ |
238
|
|
|
function jetpack_print_news_sitemap_xsl() { |
239
|
|
|
jetpack_load_xsl( 'news' ); |
240
|
|
|
} |
241
|
|
|
|
242
|
|
|
/** |
243
|
|
|
* Print an XML sitemap conforming to the Sitemaps.org protocol. |
244
|
|
|
* Outputs an XML list of up to the latest 1000 posts. |
245
|
|
|
* |
246
|
|
|
* @module sitemaps |
247
|
|
|
* |
248
|
|
|
* @link http://sitemaps.org/protocol.php Sitemaps.org protocol. |
249
|
|
|
*/ |
250
|
|
|
function jetpack_print_sitemap() { |
251
|
|
|
global $wpdb, $post; |
252
|
|
|
|
253
|
|
|
$xml = get_transient( 'jetpack_sitemap' ); |
254
|
|
|
|
255
|
|
|
if ( $xml ) { |
256
|
|
|
header( 'Content-Type: ' . jetpack_sitemap_content_type(), true ); |
257
|
|
|
echo $xml; |
258
|
|
|
die(); |
|
|
|
|
259
|
|
|
} |
260
|
|
|
|
261
|
|
|
// Compatibility with PHP 5.3 and older |
262
|
|
|
if ( ! defined( 'ENT_XML1' ) ) { |
263
|
|
|
define( 'ENT_XML1', 16 ); |
264
|
|
|
} |
265
|
|
|
|
266
|
|
|
/** |
267
|
|
|
* Filter the post types that will be included in sitemap. |
268
|
|
|
* |
269
|
|
|
* @module sitemaps |
270
|
|
|
* |
271
|
|
|
* @since 3.9.0 |
272
|
|
|
* |
273
|
|
|
* @param array $post_types Array of post types. |
274
|
|
|
*/ |
275
|
|
|
$post_types = apply_filters( 'jetpack_sitemap_post_types', array( 'post', 'page' ) ); |
|
|
|
|
276
|
|
|
|
277
|
|
|
$post_types_in = array(); |
278
|
|
|
foreach ( (array) $post_types as $post_type ) { |
279
|
|
|
$post_types_in[] = $wpdb->prepare( '%s', $post_type ); |
280
|
|
|
} |
281
|
|
|
$post_types_in = join( ",", $post_types_in ); |
282
|
|
|
|
283
|
|
|
// use direct query instead because get_posts was acting too heavy for our needs |
284
|
|
|
//$posts = get_posts( array( 'numberposts'=>1000, 'post_type'=>$post_types, 'post_status'=>'published' ) ); |
285
|
|
|
$posts = $wpdb->get_results( "SELECT ID, post_type, post_modified_gmt, comment_count FROM $wpdb->posts WHERE post_status='publish' AND post_type IN ({$post_types_in}) ORDER BY post_modified_gmt DESC LIMIT 1000" ); |
286
|
|
|
if ( empty( $posts ) ) { |
287
|
|
|
status_header( 404 ); |
288
|
|
|
} |
289
|
|
|
header( 'Content-Type: ' . jetpack_sitemap_content_type() ); |
290
|
|
|
$initstr = jetpack_sitemap_initstr( get_bloginfo( 'charset' ) ); |
291
|
|
|
$tree = simplexml_load_string( $initstr ); |
292
|
|
|
// If we did not get a valid string, force UTF-8 and try again. |
293
|
|
|
if ( false === $tree ) { |
294
|
|
|
$initstr = jetpack_sitemap_initstr( 'UTF-8' ); |
295
|
|
|
$tree = simplexml_load_string( $initstr ); |
296
|
|
|
} |
297
|
|
|
|
298
|
|
|
unset( $initstr ); |
299
|
|
|
$latest_mod = ''; |
300
|
|
|
foreach ( $posts as $post ) { |
301
|
|
|
setup_postdata( $post ); |
302
|
|
|
|
303
|
|
|
/** |
304
|
|
|
* Filter condition to allow skipping specific posts in sitemap. |
305
|
|
|
* |
306
|
|
|
* @module sitemaps |
307
|
|
|
* |
308
|
|
|
* @since 3.9.0 |
309
|
|
|
* |
310
|
|
|
* @param bool $skip Current boolean. False by default, so no post is skipped. |
311
|
|
|
* @param WP_POST $post Current post object. |
312
|
|
|
*/ |
313
|
|
|
if ( apply_filters( 'jetpack_sitemap_skip_post', false, $post ) ) { |
314
|
|
|
continue; |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
$post_latest_mod = null; |
318
|
|
|
$url = array( 'loc' => esc_url( get_permalink( $post->ID ) ) ); |
319
|
|
|
|
320
|
|
|
// If this post is configured to be the site home, skip since it's added separately later |
321
|
|
|
if ( untrailingslashit( get_permalink( $post->ID ) ) == untrailingslashit( get_option( 'home' ) ) ) { |
322
|
|
|
continue; |
323
|
|
|
} |
324
|
|
|
|
325
|
|
|
// Mobile node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=34648 |
326
|
|
|
$url['mobile:mobile'] = ''; |
327
|
|
|
|
328
|
|
|
// Image node specified in http://support.google.com/webmasters/bin/answer.py?hl=en&answer=178636 |
329
|
|
|
// These attachments were produced with batch SQL earlier in the script |
330
|
|
|
if ( ! post_password_required( $post->ID ) ) { |
331
|
|
|
|
332
|
|
|
$media = array(); |
333
|
|
|
$methods = array( |
334
|
|
|
'from_thumbnail' => false, |
335
|
|
|
'from_slideshow' => false, |
336
|
|
|
'from_gallery' => false, |
337
|
|
|
'from_attachment' => false, |
338
|
|
|
'from_html' => false, |
339
|
|
|
); |
340
|
|
|
foreach ( $methods as $method => $value ) { |
341
|
|
|
$methods[ $method ] = true; |
342
|
|
|
$images_collected = Jetpack_PostImages::get_images( $post->ID, $methods ); |
343
|
|
|
if ( is_array( $images_collected ) ) { |
344
|
|
|
$media = array_merge( $media, $images_collected ); |
345
|
|
|
} |
346
|
|
|
$methods[ $method ] = false; |
347
|
|
|
} |
348
|
|
|
|
349
|
|
|
$images = array(); |
350
|
|
|
|
351
|
|
|
foreach ( $media as $item ) { |
352
|
|
|
if ( ! isset( $item['type'] ) || 'image' != $item['type'] ) { |
353
|
|
|
continue; |
354
|
|
|
} |
355
|
|
|
$one_image = array(); |
356
|
|
|
|
357
|
|
|
if ( isset( $item['src'] ) ) { |
358
|
|
|
$one_image['image:loc'] = esc_url( $item['src'] ); |
359
|
|
|
$one_image['image:title'] = sanitize_title_with_dashes( $name = pathinfo( $item['src'], PATHINFO_FILENAME ) ); |
360
|
|
|
} |
361
|
|
|
|
362
|
|
|
$images[] = $one_image; |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
if ( ! empty( $images ) ) { |
366
|
|
|
$url['image:image'] = $images; |
367
|
|
|
} |
368
|
|
|
} |
369
|
|
|
|
370
|
|
|
if ( $post->post_modified_gmt && $post->post_modified_gmt != '0000-00-00 00:00:00' ) { |
371
|
|
|
$post_latest_mod = $post->post_modified_gmt; |
372
|
|
|
} |
373
|
|
|
if ( $post->comment_count > 0 ) { |
374
|
|
|
// last modified based on last comment |
375
|
|
|
$latest_comment_datetime = jetpack_get_approved_comments_max_datetime( $post->ID ); |
376
|
|
|
if ( ! empty( $latest_comment_datetime ) ) { |
377
|
|
|
if ( is_null( $post_latest_mod ) || $latest_comment_datetime > $post_latest_mod ) { |
378
|
|
|
$post_latest_mod = $latest_comment_datetime; |
379
|
|
|
} |
380
|
|
|
} |
381
|
|
|
unset( $latest_comment_datetime ); |
382
|
|
|
} |
383
|
|
|
if ( ! empty( $post_latest_mod ) ) { |
384
|
|
|
$latest_mod = max( $latest_mod, $post_latest_mod ); |
385
|
|
|
$url['lastmod'] = jetpack_w3cdate_from_mysql( $post_latest_mod ); |
386
|
|
|
} |
387
|
|
|
unset( $post_latest_mod ); |
388
|
|
|
if ( $post->post_type == 'page' ) { |
389
|
|
|
$url['changefreq'] = 'weekly'; |
390
|
|
|
$url['priority'] = '0.6'; // set page priority above default priority of 0.5 |
391
|
|
|
} else { |
392
|
|
|
$url['changefreq'] = 'monthly'; |
393
|
|
|
} |
394
|
|
|
/** |
395
|
|
|
* Filter associative array with data to build <url> node and its descendants for current post. |
396
|
|
|
* |
397
|
|
|
* @module sitemaps |
398
|
|
|
* |
399
|
|
|
* @since 3.9.0 |
400
|
|
|
* |
401
|
|
|
* @param array $url Data to build parent and children nodes for current post. |
402
|
|
|
* @param int $post_id Current post ID. |
403
|
|
|
*/ |
404
|
|
|
$url_node = apply_filters( 'jetpack_sitemap_url', $url, $post->ID ); |
405
|
|
|
jetpack_sitemap_array_to_simplexml( array( 'url' => $url_node ), $tree ); |
406
|
|
|
unset( $url ); |
407
|
|
|
} |
408
|
|
|
wp_reset_postdata(); |
409
|
|
|
$blog_home = array( |
410
|
|
|
'loc' => esc_url( get_option( 'home' ) ), |
411
|
|
|
'changefreq' => 'daily', |
412
|
|
|
'priority' => '1.0' |
413
|
|
|
); |
414
|
|
|
if ( ! empty( $latest_mod ) ) { |
415
|
|
|
$blog_home['lastmod'] = jetpack_w3cdate_from_mysql( $latest_mod ); |
416
|
|
|
header( 'Last-Modified:' . mysql2date( 'D, d M Y H:i:s', $latest_mod, 0 ) . ' GMT' ); |
417
|
|
|
} |
418
|
|
|
/** |
419
|
|
|
* Filter associative array with data to build <url> node and its descendants for site home. |
420
|
|
|
* |
421
|
|
|
* @module sitemaps |
422
|
|
|
* |
423
|
|
|
* @since 3.9.0 |
424
|
|
|
* |
425
|
|
|
* @param array $blog_home Data to build parent and children nodes for site home. |
426
|
|
|
*/ |
427
|
|
|
$url_node = apply_filters( 'jetpack_sitemap_url_home', $blog_home ); |
428
|
|
|
jetpack_sitemap_array_to_simplexml( array( 'url' => $url_node ), $tree ); |
429
|
|
|
unset( $blog_home ); |
430
|
|
|
|
431
|
|
|
/** |
432
|
|
|
* Filter data before rendering it as XML. |
433
|
|
|
* |
434
|
|
|
* @module sitemaps |
435
|
|
|
* |
436
|
|
|
* @since 3.9.0 |
437
|
|
|
* |
438
|
|
|
* @param SimpleXMLElement $tree Data tree for sitemap. |
439
|
|
|
* @param string $latest_mod Date of last modification. |
440
|
|
|
*/ |
441
|
|
|
$tree = apply_filters( 'jetpack_print_sitemap', $tree, $latest_mod ); |
442
|
|
|
|
443
|
|
|
$xml = $tree->asXML(); |
444
|
|
|
unset( $tree ); |
445
|
|
|
if ( ! empty( $xml ) ) { |
446
|
|
|
set_transient( 'jetpack_sitemap', $xml, DAY_IN_SECONDS ); |
447
|
|
|
echo $xml; |
448
|
|
|
} |
449
|
|
|
|
450
|
|
|
die(); |
|
|
|
|
451
|
|
|
} |
452
|
|
|
|
453
|
|
|
/** |
454
|
|
|
* Prints the news XML sitemap conforming to the Sitemaps.org protocol. |
455
|
|
|
* Outputs an XML list of up to 1000 posts published in the last 2 days. |
456
|
|
|
* |
457
|
|
|
* @module sitemaps |
458
|
|
|
* |
459
|
|
|
* @link http://sitemaps.org/protocol.php Sitemaps.org protocol. |
460
|
|
|
*/ |
461
|
|
|
function jetpack_print_news_sitemap() { |
462
|
|
|
|
463
|
|
|
$xml = get_transient( 'jetpack_news_sitemap' ); |
464
|
|
|
|
465
|
|
|
if ( $xml ) { |
466
|
|
|
header( 'Content-Type: application/xml' ); |
467
|
|
|
echo $xml; |
468
|
|
|
die(); |
|
|
|
|
469
|
|
|
} |
470
|
|
|
|
471
|
|
|
global $wpdb, $post; |
472
|
|
|
|
473
|
|
|
/** |
474
|
|
|
* Filter post types to be included in news sitemap. |
475
|
|
|
* |
476
|
|
|
* @module sitemaps |
477
|
|
|
* |
478
|
|
|
* @since 3.9.0 |
479
|
|
|
* |
480
|
|
|
* @param array $post_types Array with post types to include in news sitemap. |
481
|
|
|
*/ |
482
|
|
|
$post_types = apply_filters( 'jetpack_sitemap_news_sitemap_post_types', array( 'post' ) ); |
483
|
|
|
if ( empty( $post_types ) ) { |
484
|
|
|
return; |
485
|
|
|
} |
486
|
|
|
|
487
|
|
|
$post_types_in = array(); |
488
|
|
|
foreach ( $post_types as $post_type ) { |
489
|
|
|
$post_types_in[] = $wpdb->prepare( '%s', $post_type ); |
490
|
|
|
} |
491
|
|
|
$post_types_in_string = implode( ', ', $post_types_in ); |
492
|
|
|
|
493
|
|
|
/** |
494
|
|
|
* Filter limit of entries to include in news sitemap. |
495
|
|
|
* |
496
|
|
|
* @module sitemaps |
497
|
|
|
* |
498
|
|
|
* @since 3.9.0 |
499
|
|
|
* |
500
|
|
|
* @param int $count Number of entries to include in news sitemap. |
501
|
|
|
*/ |
502
|
|
|
$limit = apply_filters( 'jetpack_sitemap_news_sitemap_count', 1000 ); |
503
|
|
|
$cur_datetime = current_time( 'mysql', true ); |
504
|
|
|
|
505
|
|
|
$query = $wpdb->prepare( " |
506
|
|
|
SELECT p.ID, p.post_title, p.post_type, p.post_date, p.post_name, p.post_date_gmt, GROUP_CONCAT(t.name SEPARATOR ', ') AS keywords |
507
|
|
|
FROM |
508
|
|
|
$wpdb->posts AS p LEFT JOIN $wpdb->term_relationships AS r ON p.ID = r.object_id |
509
|
|
|
LEFT JOIN $wpdb->term_taxonomy AS tt ON r.term_taxonomy_id = tt.term_taxonomy_id AND tt.taxonomy = 'post_tag' |
510
|
|
|
LEFT JOIN $wpdb->terms AS t ON tt.term_id = t.term_id |
511
|
|
|
WHERE |
512
|
|
|
post_status='publish' AND post_type IN ( {$post_types_in_string} ) AND post_date_gmt > (%s - INTERVAL 2 DAY) |
513
|
|
|
GROUP BY p.ID |
514
|
|
|
ORDER BY p.post_date_gmt DESC LIMIT %d", $cur_datetime, $limit ); |
515
|
|
|
|
516
|
|
|
// URL to XSLT |
517
|
|
|
$xsl = get_option( 'permalink_structure' ) ? home_url( 'news-sitemap.xsl' ) : home_url( '/?jetpack-news-sitemap-xsl=true' ); |
518
|
|
|
|
519
|
|
|
// Unless it's zh-cn for Simplified Chinese or zh-tw for Traditional Chinese, |
520
|
|
|
// trim national variety so an ISO 639 language code as required by Google. |
521
|
|
|
$language_code = strtolower( get_locale() ); |
522
|
|
|
if ( in_array( $language_code, array( 'zh_tw', 'zh_cn' ) ) ) { |
523
|
|
|
$language_code = str_replace( '_', '-', $language_code ); |
524
|
|
|
} else { |
525
|
|
|
$language_code = preg_replace( '/(_.*)$/i', '', $language_code ); |
526
|
|
|
} |
527
|
|
|
|
528
|
|
|
header( 'Content-Type: application/xml' ); |
529
|
|
|
ob_start(); |
530
|
|
|
echo '<?xml version="1.0" encoding="UTF-8"?>' . "\n"; |
531
|
|
|
echo '<?xml-stylesheet type="text/xsl" href="' . esc_url( $xsl ) . '"?>' . "\n"; |
532
|
|
|
echo '<!-- generator="jetpack-' . JETPACK__VERSION . '" -->' . "\n"; |
533
|
|
|
?> |
534
|
|
|
<!-- generator="jetpack" --> |
535
|
|
|
<urlset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" |
536
|
|
|
xsi:schemaLocation="http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd" |
537
|
|
|
xmlns="http://www.sitemaps.org/schemas/sitemap/0.9" |
538
|
|
|
xmlns:news="http://www.google.com/schemas/sitemap-news/0.9" |
539
|
|
|
xmlns:image="http://www.google.com/schemas/sitemap-image/1.1" |
540
|
|
|
> |
541
|
|
|
<?php |
542
|
|
|
$posts = $wpdb->get_results( $query ); |
543
|
|
|
foreach ( $posts as $post ): |
544
|
|
|
setup_postdata( $post ); |
545
|
|
|
|
546
|
|
|
/** |
547
|
|
|
* Filter condition to allow skipping specific posts in news sitemap. |
548
|
|
|
* |
549
|
|
|
* @module sitemaps |
550
|
|
|
* |
551
|
|
|
* @since 3.9.0 |
552
|
|
|
* |
553
|
|
|
* @param bool $skip Current boolean. False by default, so no post is skipped. |
554
|
|
|
* @param WP_POST $post Current post object. |
555
|
|
|
*/ |
556
|
|
|
if ( apply_filters( 'jetpack_sitemap_news_skip_post', false, $post ) ) { |
557
|
|
|
continue; |
558
|
|
|
} |
559
|
|
|
|
560
|
|
|
$GLOBALS['post'] = $post; |
561
|
|
|
$url = array(); |
562
|
|
|
$url['loc'] = get_permalink( $post->ID ); |
563
|
|
|
$news = array(); |
564
|
|
|
$news['news:publication']['news:name'] = get_bloginfo_rss( 'name' ); |
565
|
|
|
$news['news:publication']['news:language'] = $language_code; |
566
|
|
|
$news['news:publication_date'] = jetpack_w3cdate_from_mysql( $post->post_date_gmt ); |
567
|
|
|
$news['news:title'] = get_the_title_rss(); |
568
|
|
|
if ( $post->keywords ) { |
569
|
|
|
$news['news:keywords'] = html_entity_decode( ent2ncr( $post->keywords ), ENT_HTML5 ); |
570
|
|
|
} |
571
|
|
|
$url['news:news'] = $news; |
572
|
|
|
|
573
|
|
|
// Add image to sitemap |
574
|
|
|
$post_thumbnail = Jetpack_PostImages::get_image( $post->ID ); |
575
|
|
|
if ( isset( $post_thumbnail['src'] ) ) { |
576
|
|
|
$url['image:image'] = array( 'image:loc' => esc_url( $post_thumbnail['src'] ) ); |
577
|
|
|
} |
578
|
|
|
|
579
|
|
|
/** |
580
|
|
|
* Filter associative array with data to build <url> node and its descendants for current post in news sitemap. |
581
|
|
|
* |
582
|
|
|
* @module sitemaps |
583
|
|
|
* |
584
|
|
|
* @since 3.9.0 |
585
|
|
|
* |
586
|
|
|
* @param array $url Data to build parent and children nodes for current post. |
587
|
|
|
* @param int $post_id Current post ID. |
588
|
|
|
*/ |
589
|
|
|
$url = apply_filters( 'jetpack_sitemap_news_sitemap_item', $url, $post ); |
590
|
|
|
|
591
|
|
|
if ( empty( $url ) ) { |
592
|
|
|
continue; |
593
|
|
|
} |
594
|
|
|
|
595
|
|
|
jetpack_print_sitemap_item( $url ); |
596
|
|
|
endforeach; |
597
|
|
|
wp_reset_postdata(); |
598
|
|
|
?> |
599
|
|
|
</urlset> |
600
|
|
|
<?php |
601
|
|
|
$xml = ob_get_contents(); |
602
|
|
|
ob_end_clean(); |
603
|
|
|
if ( ! empty( $xml ) ) { |
604
|
|
|
set_transient( 'jetpack_news_sitemap', $xml, DAY_IN_SECONDS ); |
605
|
|
|
echo $xml; |
606
|
|
|
} |
607
|
|
|
|
608
|
|
|
die(); |
|
|
|
|
609
|
|
|
} |
610
|
|
|
|
611
|
|
|
/** |
612
|
|
|
* Absolute URL of the current blog's sitemap. |
613
|
|
|
* |
614
|
|
|
* @module sitemaps |
615
|
|
|
* |
616
|
|
|
* @return string Sitemap URL. |
617
|
|
|
*/ |
618
|
|
View Code Duplication |
function jetpack_sitemap_uri() { |
|
|
|
|
619
|
|
|
global $wp_rewrite; |
620
|
|
|
|
621
|
|
|
if ( $wp_rewrite->using_index_permalinks() ) { |
622
|
|
|
$sitemap_url = home_url( '/index.php/sitemap.xml' ); |
623
|
|
|
} else if ( $wp_rewrite->using_permalinks() ) { |
624
|
|
|
$sitemap_url = home_url( '/sitemap.xml' ); |
625
|
|
|
} else { |
626
|
|
|
$sitemap_url = home_url( '/?jetpack-sitemap=true' ); |
627
|
|
|
} |
628
|
|
|
|
629
|
|
|
/** |
630
|
|
|
* Filter sitemap URL relative to home URL. |
631
|
|
|
* |
632
|
|
|
* @module sitemaps |
633
|
|
|
* |
634
|
|
|
* @since 3.9.0 |
635
|
|
|
* |
636
|
|
|
* @param string $sitemap_url Sitemap URL. |
637
|
|
|
*/ |
638
|
|
|
return apply_filters( 'jetpack_sitemap_location', $sitemap_url ); |
639
|
|
|
} |
640
|
|
|
|
641
|
|
|
/** |
642
|
|
|
* Absolute URL of the current blog's news sitemap. |
643
|
|
|
* |
644
|
|
|
* @module sitemaps |
645
|
|
|
*/ |
646
|
|
View Code Duplication |
function jetpack_news_sitemap_uri() { |
|
|
|
|
647
|
|
|
global $wp_rewrite; |
648
|
|
|
|
649
|
|
|
if ( $wp_rewrite->using_index_permalinks() ) { |
650
|
|
|
$news_sitemap_url = home_url( '/index.php/news-sitemap.xml' ); |
651
|
|
|
} else if ( $wp_rewrite->using_permalinks() ) { |
652
|
|
|
$news_sitemap_url = home_url( '/news-sitemap.xml' ); |
653
|
|
|
} else { |
654
|
|
|
$news_sitemap_url = home_url( '/?jetpack-news-sitemap=true' ); |
655
|
|
|
} |
656
|
|
|
|
657
|
|
|
/** |
658
|
|
|
* Filter news sitemap URL relative to home URL. |
659
|
|
|
* |
660
|
|
|
* @module sitemaps |
661
|
|
|
* |
662
|
|
|
* @since 3.9.0 |
663
|
|
|
* |
664
|
|
|
* @param string $news_sitemap_url News sitemap URL. |
665
|
|
|
*/ |
666
|
|
|
return apply_filters( 'jetpack_news_sitemap_location', $news_sitemap_url ); |
667
|
|
|
} |
668
|
|
|
|
669
|
|
|
/** |
670
|
|
|
* Output the default sitemap URL. |
671
|
|
|
* |
672
|
|
|
* @module sitemaps |
673
|
|
|
*/ |
674
|
|
|
function jetpack_sitemap_discovery() { |
675
|
|
|
echo 'Sitemap: ' . esc_url( jetpack_sitemap_uri() ) . PHP_EOL; |
676
|
|
|
} |
677
|
|
|
|
678
|
|
|
/** |
679
|
|
|
* Output the news sitemap URL. |
680
|
|
|
* |
681
|
|
|
* @module sitemaps |
682
|
|
|
*/ |
683
|
|
|
function jetpack_news_sitemap_discovery() { |
684
|
|
|
echo 'Sitemap: ' . esc_url( jetpack_news_sitemap_uri() ) . PHP_EOL . PHP_EOL; |
685
|
|
|
} |
686
|
|
|
|
687
|
|
|
/** |
688
|
|
|
* Clear the sitemap cache when a sitemap action has changed. |
689
|
|
|
* |
690
|
|
|
* @module sitemaps |
691
|
|
|
* |
692
|
|
|
* @param int $post_id unique post identifier. not used. |
693
|
|
|
*/ |
694
|
|
|
function jetpack_sitemap_handle_update( $post_id ) { |
|
|
|
|
695
|
|
|
delete_transient( 'jetpack_sitemap' ); |
696
|
|
|
delete_transient( 'jetpack_news_sitemap' ); |
697
|
|
|
} |
698
|
|
|
|
699
|
|
|
/** |
700
|
|
|
* Clear sitemap cache when an entry changes. Make sitemaps discoverable to robots. Render sitemaps. |
701
|
|
|
* |
702
|
|
|
* @module sitemaps |
703
|
|
|
*/ |
704
|
|
|
function jetpack_sitemap_initialize() { |
705
|
|
|
add_action( 'publish_post', 'jetpack_sitemap_handle_update', 12, 1 ); |
706
|
|
|
add_action( 'publish_page', 'jetpack_sitemap_handle_update', 12, 1 ); |
707
|
|
|
add_action( 'trash_post', 'jetpack_sitemap_handle_update', 12, 1 ); |
708
|
|
|
add_action( 'deleted_post', 'jetpack_sitemap_handle_update', 12, 1 ); |
709
|
|
|
|
710
|
|
|
/** |
711
|
|
|
* Filter whether to make the default sitemap discoverable to robots or not. |
712
|
|
|
* |
713
|
|
|
* @module sitemaps |
714
|
|
|
* |
715
|
|
|
* @since 3.9.0 |
716
|
|
|
* |
717
|
|
|
* @param bool $discover_sitemap Make default sitemap discoverable to robots. |
718
|
|
|
*/ |
719
|
|
|
$discover_sitemap = apply_filters( 'jetpack_sitemap_generate', true ); |
720
|
|
View Code Duplication |
if ( $discover_sitemap ) { |
721
|
|
|
add_action( 'do_robotstxt', 'jetpack_sitemap_discovery', 5, 0 ); |
722
|
|
|
|
723
|
|
|
if ( get_option( 'permalink_structure' ) ) { |
724
|
|
|
/** This filter is documented in modules/sitemaps/sitemaps.php */ |
725
|
|
|
$sitemap = apply_filters( 'jetpack_sitemap_location', home_url( '/sitemap.xml' ) ); |
726
|
|
|
$sitemap = parse_url( $sitemap, PHP_URL_PATH ); |
727
|
|
|
} else { |
728
|
|
|
/** This filter is documented in modules/sitemaps/sitemaps.php */ |
729
|
|
|
$sitemap = apply_filters( 'jetpack_sitemap_location', home_url( '/?jetpack-sitemap=true' ) ); |
730
|
|
|
$sitemap = preg_replace( '/(=.*?)$/i', '', parse_url( $sitemap, PHP_URL_QUERY ) ); |
731
|
|
|
} |
732
|
|
|
|
733
|
|
|
// Sitemap XML |
734
|
|
|
if ( preg_match( '#(' . $sitemap . ')$#i', $_SERVER['REQUEST_URI'] ) || ( isset( $_GET[ $sitemap ] ) && 'true' == $_GET[ $sitemap ] ) ) { |
735
|
|
|
// run later so things like custom post types have been registered |
736
|
|
|
add_action( 'init', 'jetpack_print_sitemap', 999 ); |
737
|
|
|
} |
738
|
|
|
|
739
|
|
|
// XSLT for sitemap |
740
|
|
|
if ( preg_match( '#(/sitemap\.xsl)$#i', $_SERVER['REQUEST_URI'] ) || ( isset( $_GET['jetpack-sitemap-xsl'] ) && 'true' == $_GET['jetpack-sitemap-xsl'] ) ) { |
741
|
|
|
add_action( 'init', 'jetpack_print_sitemap_xsl' ); |
742
|
|
|
} |
743
|
|
|
} |
744
|
|
|
|
745
|
|
|
/** |
746
|
|
|
* Filter whether to make the news sitemap discoverable to robots or not. |
747
|
|
|
* |
748
|
|
|
* @module sitemaps |
749
|
|
|
* |
750
|
|
|
* @since 3.9.0 |
751
|
|
|
* |
752
|
|
|
* @param bool $discover_news_sitemap Make default news sitemap discoverable to robots. |
753
|
|
|
*/ |
754
|
|
|
$discover_news_sitemap = apply_filters( 'jetpack_news_sitemap_generate', true ); |
755
|
|
View Code Duplication |
if ( $discover_news_sitemap ) { |
756
|
|
|
add_action( 'do_robotstxt', 'jetpack_news_sitemap_discovery', 5, 0 ); |
757
|
|
|
|
758
|
|
|
if ( get_option( 'permalink_structure' ) ) { |
759
|
|
|
/** This filter is documented in modules/sitemaps/sitemaps.php */ |
760
|
|
|
$sitemap = apply_filters( 'jetpack_news_sitemap_location', home_url( '/news-sitemap.xml' ) ); |
761
|
|
|
$sitemap = parse_url( $sitemap, PHP_URL_PATH ); |
762
|
|
|
} else { |
763
|
|
|
/** This filter is documented in modules/sitemaps/sitemaps.php */ |
764
|
|
|
$sitemap = apply_filters( 'jetpack_news_sitemap_location', home_url( '/?jetpack-news-sitemap=true' ) ); |
765
|
|
|
$sitemap = preg_replace( '/(=.*?)$/i', '', parse_url( $sitemap, PHP_URL_QUERY ) ); |
766
|
|
|
} |
767
|
|
|
|
768
|
|
|
// News Sitemap XML |
769
|
|
|
if ( preg_match( '#(' . $sitemap . ')$#i', $_SERVER['REQUEST_URI'] ) || ( isset( $_GET[ $sitemap ] ) && 'true' == $_GET[ $sitemap ] ) ) { |
770
|
|
|
// run later so things like custom post types have been registered |
771
|
|
|
add_action( 'init', 'jetpack_print_news_sitemap', 999 ); |
772
|
|
|
} |
773
|
|
|
|
774
|
|
|
// XSLT for sitemap |
775
|
|
|
if ( preg_match( '#(/news-sitemap\.xsl)$#i', $_SERVER['REQUEST_URI'] ) || ( isset( $_GET['jetpack-news-sitemap-xsl'] ) && 'true' == $_GET['jetpack-news-sitemap-xsl'] ) ) { |
776
|
|
|
add_action( 'init', 'jetpack_print_news_sitemap_xsl' ); |
777
|
|
|
} |
778
|
|
|
} |
779
|
|
|
} |
780
|
|
|
|
781
|
|
|
// Initialize sitemaps once themes can filter the initialization. |
782
|
|
|
add_action( 'after_setup_theme', 'jetpack_sitemap_initialize' ); |
783
|
|
|
|
An exit expression should only be used in rare cases. For example, if you write a short command line script.
In most cases however, using an
exit
expression makes the code untestable and often causes incompatibilities with other libraries. Thus, unless you are absolutely sure it is required here, we recommend to refactor your code to avoid its usage.