sitemap.php

   1 <?php
   2
   3 define('INSTALLDIR', 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 avatar_map();
  15 index_map();
  16
  17 # ------------------------------------------------------------------------------
  18 # Main functions: get data out and turn them into sitemaps
  19 # ------------------------------------------------------------------------------
  20
  21 # Generate index sitemap of all other sitemaps.
  22 function index_map() {
  23         global $output_paths;
  24         $output_dir = $output_paths['output_dir'];
  25         $output_url = $output_paths['output_url'];
  26
  27         foreach (glob("$output_dir*.xml") as $file_name) {
  28
  29                 # Just the file name please.
  30                 $file_name = preg_replace("|$output_dir|", '', $file_name);
  31
  32                 $index_urls .= url(
  33                                                    array(
  34                                                                  'url' => $output_url . $file_name,
  35                                                                  'changefreq' => 'hourly',
  36                                                                  'priority' => '1',
  37                                                                  )
  38                                                    );
  39         }
  40
  41         $index_path = $output_dir . $output_paths['index_file'];
  42
  43         write_file($index_path, urlset($index_urls));
  44
  45 }
  46
  47 # Generate sitemap of standard site elements.
  48 function standard_map() {
  49         global $output_paths;
  50
  51         $standard_map_urls .= url(
  52                                                           array(
  53                                                                         'url' => common_local_url('public'),
  54                                                                         'changefreq' => 'hourly',
  55                                                                         'priority' => '1',
  56                                                                         )
  57                                                           );
  58
  59         $standard_map_urls .= url(
  60                                                           array(
  61                                                                         'url' => common_local_url('publicrss'),
  62                                                                         'changefreq' => 'hourly',
  63                                                                         'priority' => '0.3',
  64                                                                         )
  65                                                           );
  66
  67         $docs = array('about', 'faq', 'contact', 'im', 'openid', 'openmublog', 'privacy', 'source');
  68
  69         foreach($docs as $title) {
  70                 $standard_map_urls .= url(
  71                                                                   array(
  72                                                                                 'url' => common_local_url('doc', array('title' => $title)),
  73                                                                                 'changefreq' => 'monthly',
  74                                                                                 'priority'   => '0.2',
  75                                                                                 )
  76                                                                   );
  77         }
  78
  79         $urlset_path = $output_paths['output_dir'] . 'standard.xml';
  80
  81         write_file($urlset_path, urlset($standard_map_urls));
  82 }
  83
  84 # Generate sitemaps of all notices.
  85 function notices_map() {
  86         global $output_paths;
  87
  88         $notices = DB_DataObject::factory('notice');
  89
  90         $notices->query('SELECT uri, modified FROM notice');
  91
  92         $notice_count = 0;
  93         $map_count = 1;
  94
  95         while ($notices->fetch()) {
  96
  97                 # Maximum 50,000 URLs per sitemap file.
  98                 if ($notice_count == 50000) {
  99                         $notice_count = 0;
 100                         $map_count++;
 101                 }
 102
 103                 $notice = array(
 104                                                 'url'        => $notices->uri,
 105                                                 'lastmod'    => preg_replace('/ /', 'T', $notices->modified), # W3C DTF requires "T" separator
 106                                                 'changefreq' => 'hourly',
 107                                                 'priority'   => '1',
 108                                                 );
 109
 110                 $notice_list[$map_count] .= url($notice);
 111
 112                 $notice_count++;
 113         }
 114
 115         # Make full sitemaps from the lists and save them.
 116         array_to_map($notice_list, 'notice');
 117 }
 118
 119 # Generate sitemaps of all users.
 120 function user_map() {
 121         global $output_paths;
 122
 123         $users = DB_DataObject::factory('user');
 124
 125         $users->query('SELECT id, nickname FROM user');
 126
 127         $user_count = 0;
 128         $map_count = 1;
 129
 130         while ($users->fetch()) {
 131
 132                 # Maximum 50,000 URLs per sitemap file.
 133                 if ($user_count == 50000) {
 134                         $user_count = 0;
 135                         $map_count++;
 136                 }
 137
 138                 $user_args = array('nickname' => $users->nickname);
 139
 140                 # Define parameters for generating <url></url> elements.
 141                 $user = array(
 142                                           'url'        => common_local_url('showstream', $user_args),
 143                                           'changefreq' => 'hourly',
 144                                           'priority'   => '1',
 145                                           );
 146
 147                 $user_rss = array(
 148                                                   'url'        => common_local_url('userrss', $user_args),
 149                                                   'changefreq' => 'hourly',
 150                                                   'priority'   => '0.3',
 151                                                   );
 152
 153                 $all = array(
 154                                          'url'        => common_local_url('all', $user_args),
 155                                          'changefreq' => 'hourly',
 156                                          'priority'   => '1',
 157                                          );
 158
 159                 $all_rss = array(
 160                                                  'url'        => common_local_url('allrss', $user_args),
 161                                                  'changefreq' => 'hourly',
 162                                                  'priority'   => '0.3',
 163                                                  );
 164
 165                 $replies = array(
 166                                                  'url'        => common_local_url('replies', $user_args),
 167                                                  'changefreq' => 'hourly',
 168                                                  'priority'   => '1',
 169                                                  );
 170
 171                 $replies_rss = array(
 172                                                          'url'        => common_local_url('repliesrss', $user_args),
 173                                                          'changefreq' => 'hourly',
 174                                                          'priority'   => '0.3',
 175                                                          );
 176
 177                 $foaf = array(
 178                                           'url'        => common_local_url('foaf', $user_args),
 179                                           'changefreq' => 'weekly',
 180                                           'priority'   => '0.5',
 181                                           );
 182
 183                 # Construct a <url></url> element for each user facet and add it
 184                 # to our existing list of those.
 185                 $user_list[$map_count]        .= url($user);
 186                 $user_rss_list[$map_count]    .= url($user_rss);
 187                 $all_list[$map_count]         .= url($all);
 188                 $all_rss_list[$map_count]     .= url($all_rss);
 189                 $replies_list[$map_count]     .= url($replies);
 190                 $replies_rss_list[$map_count] .= url($replies_rss);
 191                 $foaf_list[$map_count]        .= url($foaf);
 192
 193                 $user_count++;
 194         }
 195
 196         # Make full sitemaps from the lists and save them.
 197         # Possible factoring: put all the lists into a master array, thus allowing
 198         # calling with single argument (i.e., array_to_map('user')).
 199         array_to_map($user_list, 'user');
 200         array_to_map($user_rss_list, 'user_rss');
 201         array_to_map($all_list, 'all');
 202         array_to_map($all_rss_list, 'all_rss');
 203         array_to_map($replies_list, 'replies');
 204         array_to_map($replies_rss_list, 'replies_rss');
 205         array_to_map($foaf_list, 'foaf');
 206 }
 207
 208 # Generate sitemaps of all avatars.
 209 function avatar_map() {
 210         global $output_paths;
 211
 212         $avatars = DB_DataObject::factory('avatar');
 213
 214         $avatars->query('SELECT url, modified FROM avatar');
 215
 216         $avatar_count = 0;
 217         $map_count = 1;
 218
 219         while ($avatars->fetch()) {
 220
 221                 # We only want the original size and 24px thumbnail version - skip 96px.
 222                 if (preg_match('/-96-/', $avatars->url)) {
 223                         continue;
 224                 }
 225
 226                 # Maximum 50,000 URLs per sitemap file.
 227                 if ($avatar_count == 50000) {
 228                         $avatar_count = 0;
 229                         $map_count++;
 230                 }
 231
 232                 $image = array(
 233                                            'url'        => $avatars->url,
 234                                            'lastmod'    => preg_replace('/ /', 'T', $avatars->modified), # W3C DTF requires "T" separator
 235                                            'changefreq' => 'monthly',
 236                                            'priority'   => '0.2',
 237                                            );
 238
 239                 # Construct a <url></url> element for each avatar and add it
 240                 # to our existing list of those.
 241                 $avatar_list[$map_count] .= url($image);
 242         }
 243
 244         array_to_map($avatar_list, 'avatars');
 245 }
 246
 247 # ------------------------------------------------------------------------------
 248 # XML generation functions
 249 # ------------------------------------------------------------------------------
 250
 251 # Generate a <url></url> element.
 252 function url($url_args) {
 253         $url        = preg_replace('/&/', '&amp;', $url_args['url']); # escape ampersands for XML
 254         $lastmod    = $url_args['lastmod'];
 255         $changefreq = $url_args['changefreq'];
 256         $priority   = $url_args['priority'];
 257
 258         if (is_null($url)) {
 259                 error("url() arguments require 'url' value.");
 260         }
 261
 262         $url_out = "\t<url>\n";
 263         $url_out .= "\t\t<loc>$url</loc>\n";
 264
 265         if ($changefreq) {
 266                 $url_out .= "\t\t<changefreq>$changefreq</changefreq>\n";
 267         }
 268
 269         if ($lastmod) {
 270                 $url_out .= "\t\t<lastmod>$lastmod</lastmod>\n";
 271         }
 272
 273         if ($priority) {
 274                 $url_out .= "\t\t<priority>$priority</priority>\n";
 275         }
 276
 277         $url_out .= "\t</url>\n";
 278
 279         return $url_out;
 280 }
 281
 282 # Generate a <urlset></urlset> element.
 283 function urlset($urlset_text) {
 284         $urlset = '<?xml version="1.0" encoding="UTF-8"?>'.
 285           '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'.
 286           $urlset_text.
 287           '</urlset>';
 288
 289         return $urlset;
 290 }
 291
 292 # Generate a sitemap from an array containing <url></url> elements and write it to a file.
 293 function array_to_map($url_list, $filename_prefix) {
 294         global $output_paths;
 295
 296         # $map_urls is a long string containing concatenated <url></url> elements.
 297         while (list($map_idx, $map_urls) = each($url_list)) {
 298                 $urlset_path = $output_paths['output_dir'] . "$filename_prefix-$map_idx.xml";
 299
 300                 write_file($urlset_path, urlset($map_urls));
 301         }
 302 }
 303
 304 # ------------------------------------------------------------------------------
 305 # Internal functions
 306 # ------------------------------------------------------------------------------
 307
 308 # Parse command line arguments.
 309 function parse_args() {
 310         $args = getopt('f:d:u:');
 311
 312         if (is_null($args[f]) && is_null($args[d]) && is_null($args[u])) {
 313                 error('Mandatory arguments: -f <index file name> -d <output directory> -u <URL of sitemaps directory>');
 314         }
 315
 316         if (is_null($args[f])) {
 317                 error('You must specify an index file name with the -f option.');
 318         }
 319
 320         if (is_null($args[d])) {
 321                 error('You must specify a directory for the output file with the -d option.');
 322         }
 323
 324         if (is_null($args[u])) {
 325                 error('You must specify a URL for the directory where the sitemaps will be kept with the -u option.');
 326         }
 327
 328         $index_file = $args[f];
 329         $output_dir = $args[d];
 330         $output_url = $args[u];
 331
 332         if (file_exists($output_dir)) {
 333                 if (is_writable($output_dir) === FALSE) {
 334                         error("$output_dir is not writable.");
 335                 }
 336         }        else {
 337                 error("output directory $output_dir does not exist.");
 338         }
 339
 340         $paths = array(
 341                                    'index_file' => $index_file,
 342                                    'output_dir' => trailing_slash($output_dir),
 343                                    'output_url' => trailing_slash($output_url),
 344                                    );
 345
 346         return $paths;
 347 }
 348
 349 # Ensure paths end with a "/".
 350 function trailing_slash($path) {
 351         if (preg_match('/\/$/', $path) == 0) {
 352                 $path .= '/';
 353         }
 354
 355         return $path;
 356 }
 357
 358 # Write data to disk.
 359 function write_file($path, $data) {
 360         if (is_null($path)) {
 361                 error('No path specified for writing to.');
 362         }        elseif (is_null($data)) {
 363                 error('No data specified for writing.');
 364         }
 365
 366         if (($fh_out = fopen($path,'w')) === FALSE) {
 367                 error("couldn't open $path for writing.");
 368         }
 369
 370         if (fwrite($fh_out, $data) === FALSE) {
 371                 error("couldn't write to $path.");
 372         }
 373 }
 374
 375 # Display an error message and exit.
 376 function error ($error_msg) {
 377         if (is_null($error_msg)) {
 378                 $error_msg = 'error() was called without any explanation!';
 379         }
 380
 381         echo "Error: $error_msg\n";
 382         exit(1);
 383 }
 384
 385 ?>