scripts/sitemap.php

   1 <?php
   2
   3 define('INSTALLDIR', realpath(dirname(__FILE__) . '/..'));
   4 define('LACONICA', true);
   5
   6 require_once(INSTALLDIR . '/lib/common.php');
   7 require_once(INSTALLDIR . '/lib/util.php');
   8
   9 $output_paths = parse_args();
  10
  11 standard_map();
  12 notices_map();
  13 user_map();
  14 index_map();
  15
  16 # ------------------------------------------------------------------------------
  17 # Main functions: get data out and turn them into sitemaps
  18 # ------------------------------------------------------------------------------
  19
  20 # Generate index sitemap of all other sitemaps.
  21 function index_map() {
  22         global $output_paths;
  23         $output_dir = $output_paths['output_dir'];
  24         $output_url = $output_paths['output_url'];
  25
  26         foreach (glob("$output_dir*.xml") as $file_name) {
  27
  28                 # Just the file name please.
  29                 $file_name = preg_replace("|$output_dir|", '', $file_name);
  30
  31                 $index_urls .= sitemap(
  32                                                    array(
  33                                                                  'url' => $output_url . $file_name,
  34                                                                  'changefreq' => 'daily'
  35                                                                  )
  36                                                    );
  37         }
  38
  39         write_file($output_paths['index_file'], sitemapindex($index_urls));
  40 }
  41
  42 # Generate sitemap of standard site elements.
  43 function standard_map() {
  44         global $output_paths;
  45
  46         $standard_map_urls .= url(
  47                                                           array(
  48                                                                         'url' => common_local_url('public'),
  49                                                                         'changefreq' => 'daily',
  50                                                                         'priority' => '1',
  51                                                                         )
  52                                                           );
  53
  54         $standard_map_urls .= url(
  55                                                           array(
  56                                                                         'url' => common_local_url('publicrss'),
  57                                                                         'changefreq' => 'daily',
  58                                                                         'priority' => '0.3',
  59                                                                         )
  60                                                           );
  61
  62         $docs = array('about', 'faq', 'contact', 'im', 'openid', 'openmublog', 'privacy', 'source');
  63
  64         foreach($docs as $title) {
  65                 $standard_map_urls .= url(
  66                                                                   array(
  67                                                                                 'url' => common_local_url('doc', array('title' => $title)),
  68                                                                                 'changefreq' => 'monthly',
  69                                                                                 'priority'   => '0.2',
  70                                                                                 )
  71                                                                   );
  72         }
  73
  74         $urlset_path = $output_paths['output_dir'] . 'standard.xml';
  75
  76         write_file($urlset_path, urlset($standard_map_urls));
  77 }
  78
  79 # Generate sitemaps of all notices.
  80 function notices_map() {
  81         global $output_paths;
  82
  83         $notices = DB_DataObject::factory('notice');
  84
  85         $notices->query('SELECT id, uri, url, modified FROM notice');
  86
  87         $notice_count = 0;
  88         $map_count = 1;
  89
  90         while ($notices->fetch()) {
  91
  92                 # Maximum 50,000 URLs per sitemap file.
  93                 if ($notice_count == 50000) {
  94                         $notice_count = 0;
  95                         $map_count++;
  96                 }
  97
  98                 # remote notices have an URL
  99
 100                 if (!$notices->url && $notices->uri) {
 101                         $notice = array(
 102                                                 'url'        => ($notices->uri) ? $notices->uri : common_local_url('shownotice', array('notice' => $notices->id)),
 103                                                 'lastmod'    => common_date_w3dtf($notices->modified),
 104                                                 'changefreq' => 'never',
 105                                                 'priority'   => '1',
 106                                                 );
 107
 108                         $notice_list[$map_count] .= url($notice);
 109                         $notice_count++;
 110                 }
 111         }
 112
 113         # Make full sitemaps from the lists and save them.
 114         array_to_map($notice_list, 'notice');
 115 }
 116
 117 # Generate sitemaps of all users.
 118 function user_map() {
 119         global $output_paths;
 120
 121         $users = DB_DataObject::factory('user');
 122
 123         $users->query('SELECT id, nickname FROM user');
 124
 125         $user_count = 0;
 126         $map_count = 1;
 127
 128         while ($users->fetch()) {
 129
 130                 # Maximum 50,000 URLs per sitemap file.
 131                 if ($user_count == 50000) {
 132                         $user_count = 0;
 133                         $map_count++;
 134                 }
 135
 136                 $user_args = array('nickname' => $users->nickname);
 137
 138                 # Define parameters for generating <url></url> elements.
 139                 $user = array(
 140                                           'url'        => common_local_url('showstream', $user_args),
 141                                           'changefreq' => 'daily',
 142                                           'priority'   => '1',
 143                                           );
 144
 145                 $user_rss = array(
 146                                                   'url'        => common_local_url('userrss', $user_args),
 147                                                   'changefreq' => 'daily',
 148                                                   'priority'   => '0.3',
 149                                                   );
 150
 151                 $all = array(
 152                                          'url'        => common_local_url('all', $user_args),
 153                                          'changefreq' => 'daily',
 154                                          'priority'   => '1',
 155                                          );
 156
 157                 $all_rss = array(
 158                                                  'url'        => common_local_url('allrss', $user_args),
 159                                                  'changefreq' => 'daily',
 160                                                  'priority'   => '0.3',
 161                                                  );
 162
 163                 $replies = array(
 164                                                  'url'        => common_local_url('replies', $user_args),
 165                                                  'changefreq' => 'daily',
 166                                                  'priority'   => '1',
 167                                                  );
 168
 169                 $replies_rss = array(
 170                                                          'url'        => common_local_url('repliesrss', $user_args),
 171                                                          'changefreq' => 'daily',
 172                                                          'priority'   => '0.3',
 173                                                          );
 174
 175                 $foaf = array(
 176                                           'url'        => common_local_url('foaf', $user_args),
 177                                           'changefreq' => 'weekly',
 178                                           'priority'   => '0.5',
 179                                           );
 180
 181                 # Construct a <url></url> element for each user facet and add it
 182                 # to our existing list of those.
 183                 $user_list[$map_count]        .= url($user);
 184                 $user_rss_list[$map_count]    .= url($user_rss);
 185                 $all_list[$map_count]         .= url($all);
 186                 $all_rss_list[$map_count]     .= url($all_rss);
 187                 $replies_list[$map_count]     .= url($replies);
 188                 $replies_rss_list[$map_count] .= url($replies_rss);
 189                 $foaf_list[$map_count]        .= url($foaf);
 190
 191                 $user_count++;
 192         }
 193
 194         # Make full sitemaps from the lists and save them.
 195         # Possible factoring: put all the lists into a master array, thus allowing
 196         # calling with single argument (i.e., array_to_map('user')).
 197         array_to_map($user_list, 'user');
 198         array_to_map($user_rss_list, 'user_rss');
 199         array_to_map($all_list, 'all');
 200         array_to_map($all_rss_list, 'all_rss');
 201         array_to_map($replies_list, 'replies');
 202         array_to_map($replies_rss_list, 'replies_rss');
 203         array_to_map($foaf_list, 'foaf');
 204 }
 205
 206 # ------------------------------------------------------------------------------
 207 # XML generation functions
 208 # ------------------------------------------------------------------------------
 209
 210 # Generate a <url></url> element.
 211 function url($url_args) {
 212         $url        = preg_replace('/&/', '&amp;', $url_args['url']); # escape ampersands for XML
 213         $lastmod    = $url_args['lastmod'];
 214         $changefreq = $url_args['changefreq'];
 215         $priority   = $url_args['priority'];
 216
 217         if (is_null($url)) {
 218                 error("url() arguments require 'url' value.");
 219         }
 220
 221         $url_out = "\t<url>\n";
 222         $url_out .= "\t\t<loc>$url</loc>\n";
 223
 224         if ($changefreq) {
 225                 $url_out .= "\t\t<changefreq>$changefreq</changefreq>\n";
 226         }
 227
 228         if ($lastmod) {
 229                 $url_out .= "\t\t<lastmod>$lastmod</lastmod>\n";
 230         }
 231
 232         if ($priority) {
 233                 $url_out .= "\t\t<priority>$priority</priority>\n";
 234         }
 235
 236         $url_out .= "\t</url>\n";
 237
 238         return $url_out;
 239 }
 240
 241 function sitemap($sitemap_args) {
 242         $url        = preg_replace('/&/', '&amp;', $sitemap_args['url']); # escape ampersands for XML
 243         $lastmod    = $sitemap_args['lastmod'];
 244
 245         if (is_null($url)) {
 246                 error("url() arguments require 'url' value.");
 247         }
 248
 249         $sitemap_out = "\t<sitemap>\n";
 250         $sitemap_out .= "\t\t<loc>$url</loc>\n";
 251
 252         if ($lastmod) {
 253                 $sitemap_out .= "\t\t<lastmod>$lastmod</lastmod>\n";
 254         }
 255
 256         $sitemap_out .= "\t</sitemap>\n";
 257
 258         return $sitemap_out;
 259 }
 260
 261 # Generate a <urlset></urlset> element.
 262 function urlset($urlset_text) {
 263         $urlset = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" .
 264           '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n" .
 265           $urlset_text .
 266           '</urlset>';
 267
 268         return $urlset;
 269 }
 270
 271 # Generate a <urlset></urlset> element.
 272 function sitemapindex($sitemapindex_text) {
 273         $sitemapindex = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" .
 274           '<sitemapindex xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n" .
 275           $sitemapindex_text .
 276           '</sitemapindex>';
 277
 278         return $sitemapindex;
 279 }
 280
 281 # Generate a sitemap from an array containing <url></url> elements and write it to a file.
 282 function array_to_map($url_list, $filename_prefix) {
 283         global $output_paths;
 284
 285         if ($url_list) {
 286                 # $map_urls is a long string containing concatenated <url></url> elements.
 287                 while (list($map_idx, $map_urls) = each($url_list)) {
 288                         $urlset_path = $output_paths['output_dir'] . "$filename_prefix-$map_idx.xml";
 289
 290                         write_file($urlset_path, urlset($map_urls));
 291                 }
 292         }
 293 }
 294
 295 # ------------------------------------------------------------------------------
 296 # Internal functions
 297 # ------------------------------------------------------------------------------
 298
 299 # Parse command line arguments.
 300 function parse_args() {
 301         $args = getopt('f:d:u:');
 302
 303         if (is_null($args[f]) && is_null($args[d]) && is_null($args[u])) {
 304                 error('Mandatory arguments: -f <index file path> -d <output directory path> -u <URL of sitemaps directory>');
 305         }
 306
 307         if (is_null($args[f])) {
 308                 error('You must specify an index file name with the -f option.');
 309         }
 310
 311         if (is_null($args[d])) {
 312                 error('You must specify a directory for the output file with the -d option.');
 313         }
 314
 315         if (is_null($args[u])) {
 316                 error('You must specify a URL for the directory where the sitemaps will be kept with the -u option.');
 317         }
 318
 319         $index_file = $args[f];
 320         $output_dir = $args[d];
 321         $output_url = $args[u];
 322
 323         if (file_exists($output_dir)) {
 324                 if (is_writable($output_dir) === FALSE) {
 325                         error("$output_dir is not writable.");
 326                 }
 327         }        else {
 328                 error("output directory $output_dir does not exist.");
 329         }
 330
 331         $paths = array(
 332                                    'index_file' => $index_file,
 333                                    'output_dir' => trailing_slash($output_dir),
 334                                    'output_url' => trailing_slash($output_url),
 335                                    );
 336
 337         return $paths;
 338 }
 339
 340 # Ensure paths end with a "/".
 341 function trailing_slash($path) {
 342         if (preg_match('/\/$/', $path) == 0) {
 343                 $path .= '/';
 344         }
 345
 346         return $path;
 347 }
 348
 349 # Write data to disk.
 350 function write_file($path, $data) {
 351         if (is_null($path)) {
 352                 error('No path specified for writing to.');
 353         }        elseif (is_null($data)) {
 354                 error('No data specified for writing.');
 355         }
 356
 357         if (($fh_out = fopen($path,'w')) === FALSE) {
 358                 error("couldn't open $path for writing.");
 359         }
 360
 361         if (fwrite($fh_out, $data) === FALSE) {
 362                 error("couldn't write to $path.");
 363         }
 364 }
 365
 366 # Display an error message and exit.
 367 function error ($error_msg) {
 368         if (is_null($error_msg)) {
 369                 $error_msg = 'error() was called without any explanation!';
 370         }
 371
 372         echo "Error: $error_msg\n";
 373         exit(1);
 374 }
 375
 376 ?>