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 . $file_name,
  35                                                                  'changefreq' => 'daily'
  36                                                                  )
  37                                                    );
  38         }
  39
  40         write_file($output_paths['index_file'], urlset($index_urls));
  41
  42 }
  43
  44 # Generate sitemap of standard site elements.
  45 function standard_map() {
  46         global $output_paths;
  47
  48         $standard_map_urls .= url(
  49                                                           array(
  50                                                                         'url' => common_local_url('public'),
  51                                                                         'changefreq' => 'daily',
  52                                                                         'priority' => '1',
  53                                                                         )
  54                                                           );
  55
  56         $standard_map_urls .= url(
  57                                                           array(
  58                                                                         'url' => common_local_url('publicrss'),
  59                                                                         'changefreq' => 'daily',
  60                                                                         'priority' => '0.3',
  61                                                                         )
  62                                                           );
  63
  64         $docs = array('about', 'faq', 'contact', 'im', 'openid', 'openmublog', 'privacy', 'source');
  65
  66         foreach($docs as $title) {
  67                 $standard_map_urls .= url(
  68                                                                   array(
  69                                                                                 'url' => common_local_url('doc', array('title' => $title)),
  70                                                                                 'changefreq' => 'monthly',
  71                                                                                 'priority'   => '0.2',
  72                                                                                 )
  73                                                                   );
  74         }
  75
  76         $urlset_path = $output_paths['output_dir'] . 'standard.xml';
  77
  78         write_file($urlset_path, urlset($standard_map_urls));
  79 }
  80
  81 # Generate sitemaps of all notices.
  82 function notices_map() {
  83         global $output_paths;
  84
  85         $notices = DB_DataObject::factory('notice');
  86
  87         $notices->query('SELECT uri, modified FROM notice');
  88
  89         $notice_count = 0;
  90         $map_count = 1;
  91
  92         while ($notices->fetch()) {
  93
  94                 # Maximum 50,000 URLs per sitemap file.
  95                 if ($notice_count == 50000) {
  96                         $notice_count = 0;
  97                         $map_count++;
  98                 }
  99
 100                 $notice = array(
 101                                                 'url'        => $notices->uri,
 102                                                 'lastmod'    => common_date_w3dtf($notices->modified),
 103                                                 'changefreq' => 'daily',
 104                                                 'priority'   => '1',
 105                                                 );
 106
 107                 $notice_list[$map_count] .= url($notice);
 108
 109                 $notice_count++;
 110         }
 111
 112         # Make full sitemaps from the lists and save them.
 113         array_to_map($notice_list, 'notice');
 114 }
 115
 116 # Generate sitemaps of all users.
 117 function user_map() {
 118         global $output_paths;
 119
 120         $users = DB_DataObject::factory('user');
 121
 122         $users->query('SELECT id, nickname FROM user');
 123
 124         $user_count = 0;
 125         $map_count = 1;
 126
 127         while ($users->fetch()) {
 128
 129                 # Maximum 50,000 URLs per sitemap file.
 130                 if ($user_count == 50000) {
 131                         $user_count = 0;
 132                         $map_count++;
 133                 }
 134
 135                 $user_args = array('nickname' => $users->nickname);
 136
 137                 # Define parameters for generating <url></url> elements.
 138                 $user = array(
 139                                           'url'        => common_local_url('showstream', $user_args),
 140                                           'changefreq' => 'daily',
 141                                           'priority'   => '1',
 142                                           );
 143
 144                 $user_rss = array(
 145                                                   'url'        => common_local_url('userrss', $user_args),
 146                                                   'changefreq' => 'daily',
 147                                                   'priority'   => '0.3',
 148                                                   );
 149
 150                 $all = array(
 151                                          'url'        => common_local_url('all', $user_args),
 152                                          'changefreq' => 'daily',
 153                                          'priority'   => '1',
 154                                          );
 155
 156                 $all_rss = array(
 157                                                  'url'        => common_local_url('allrss', $user_args),
 158                                                  'changefreq' => 'daily',
 159                                                  'priority'   => '0.3',
 160                                                  );
 161
 162                 $replies = array(
 163                                                  'url'        => common_local_url('replies', $user_args),
 164                                                  'changefreq' => 'daily',
 165                                                  'priority'   => '1',
 166                                                  );
 167
 168                 $replies_rss = array(
 169                                                          'url'        => common_local_url('repliesrss', $user_args),
 170                                                          'changefreq' => 'daily',
 171                                                          'priority'   => '0.3',
 172                                                          );
 173
 174                 $foaf = array(
 175                                           'url'        => common_local_url('foaf', $user_args),
 176                                           'changefreq' => 'weekly',
 177                                           'priority'   => '0.5',
 178                                           );
 179
 180                 # Construct a <url></url> element for each user facet and add it
 181                 # to our existing list of those.
 182                 $user_list[$map_count]        .= url($user);
 183                 $user_rss_list[$map_count]    .= url($user_rss);
 184                 $all_list[$map_count]         .= url($all);
 185                 $all_rss_list[$map_count]     .= url($all_rss);
 186                 $replies_list[$map_count]     .= url($replies);
 187                 $replies_rss_list[$map_count] .= url($replies_rss);
 188                 $foaf_list[$map_count]        .= url($foaf);
 189
 190                 $user_count++;
 191         }
 192
 193         # Make full sitemaps from the lists and save them.
 194         # Possible factoring: put all the lists into a master array, thus allowing
 195         # calling with single argument (i.e., array_to_map('user')).
 196         array_to_map($user_list, 'user');
 197         array_to_map($user_rss_list, 'user_rss');
 198         array_to_map($all_list, 'all');
 199         array_to_map($all_rss_list, 'all_rss');
 200         array_to_map($replies_list, 'replies');
 201         array_to_map($replies_rss_list, 'replies_rss');
 202         array_to_map($foaf_list, 'foaf');
 203 }
 204
 205 # Generate sitemaps of all avatars.
 206 function avatar_map() {
 207         global $output_paths;
 208
 209         $avatars = DB_DataObject::factory('avatar');
 210
 211         $avatars->query('SELECT url, modified FROM avatar');
 212
 213         $avatar_count = 0;
 214         $map_count = 1;
 215
 216         while ($avatars->fetch()) {
 217
 218                 # We only want the original size and 24px thumbnail version - skip 96px.
 219                 if (preg_match('/-96-/', $avatars->url)) {
 220                         continue;
 221                 }
 222
 223                 # Maximum 50,000 URLs per sitemap file.
 224                 if ($avatar_count == 50000) {
 225                         $avatar_count = 0;
 226                         $map_count++;
 227                 }
 228 w3cdate($avatars->modified);
 229                 $image = array(
 230                                            'url'        => common_avatar_display_url($avatars),
 231                                            'lastmod'    => common_date_w3dtf($avatars->modified),
 232                                            'changefreq' => 'monthly',
 233                                            'priority'   => '0.2',
 234                                            );
 235
 236                 # Construct a <url></url> element for each avatar and add it
 237                 # to our existing list of those.
 238                 $avatar_list[$map_count] .= url($image);
 239         }
 240
 241         array_to_map($avatar_list, 'avatars');
 242 }
 243
 244 # ------------------------------------------------------------------------------
 245 # XML generation functions
 246 # ------------------------------------------------------------------------------
 247
 248 # Generate a <url></url> element.
 249 function url($url_args) {
 250         $url        = preg_replace('/&/', '&amp;', $url_args['url']); # escape ampersands for XML
 251         $lastmod    = $url_args['lastmod'];
 252         $changefreq = $url_args['changefreq'];
 253         $priority   = $url_args['priority'];
 254
 255         if (is_null($url)) {
 256                 error("url() arguments require 'url' value.");
 257         }
 258
 259         $url_out = "\t<url>\n";
 260         $url_out .= "\t\t<loc>$url</loc>\n";
 261
 262         if ($changefreq) {
 263                 $url_out .= "\t\t<changefreq>$changefreq</changefreq>\n";
 264         }
 265
 266         if ($lastmod) {
 267                 $url_out .= "\t\t<lastmod>$lastmod</lastmod>\n";
 268         }
 269
 270         if ($priority) {
 271                 $url_out .= "\t\t<priority>$priority</priority>\n";
 272         }
 273
 274         $url_out .= "\t</url>\n";
 275
 276         return $url_out;
 277 }
 278
 279 # Generate a <urlset></urlset> element.
 280 function urlset($urlset_text) {
 281         $urlset = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" .
 282           '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n" .
 283           $urlset_text .
 284           '</urlset>';
 285
 286         return $urlset;
 287 }
 288
 289 # Generate a sitemap from an array containing <url></url> elements and write it to a file.
 290 function array_to_map($url_list, $filename_prefix) {
 291         global $output_paths;
 292
 293         if ($url_list) {
 294                 # $map_urls is a long string containing concatenated <url></url> elements.
 295                 while (list($map_idx, $map_urls) = each($url_list)) {
 296                         $urlset_path = $output_paths['output_dir'] . "$filename_prefix-$map_idx.xml";
 297
 298                         write_file($urlset_path, urlset($map_urls));
 299                 }
 300         }
 301 }
 302
 303 # ------------------------------------------------------------------------------
 304 # Internal functions
 305 # ------------------------------------------------------------------------------
 306
 307 # Parse command line arguments.
 308 function parse_args() {
 309         $args = getopt('f:d:u:');
 310
 311         if (is_null($args[f]) && is_null($args[d]) && is_null($args[u])) {
 312                 error('Mandatory arguments: -f <index file path> -d <output directory path> -u <URL of sitemaps directory>');
 313         }
 314
 315         if (is_null($args[f])) {
 316                 error('You must specify an index file name with the -f option.');
 317         }
 318
 319         if (is_null($args[d])) {
 320                 error('You must specify a directory for the output file with the -d option.');
 321         }
 322
 323         if (is_null($args[u])) {
 324                 error('You must specify a URL for the directory where the sitemaps will be kept with the -u option.');
 325         }
 326
 327         $index_file = $args[f];
 328         $output_dir = $args[d];
 329         $output_url = $args[u];
 330
 331         if (file_exists($output_dir)) {
 332                 if (is_writable($output_dir) === FALSE) {
 333                         error("$output_dir is not writable.");
 334                 }
 335         }        else {
 336                 error("output directory $output_dir does not exist.");
 337         }
 338
 339         $paths = array(
 340                                    'index_file' => $index_file,
 341                                    'output_dir' => trailing_slash($output_dir),
 342                                    'output_url' => trailing_slash($output_url),
 343                                    );
 344
 345         return $paths;
 346 }
 347
 348 # Ensure paths end with a "/".
 349 function trailing_slash($path) {
 350         if (preg_match('/\/$/', $path) == 0) {
 351                 $path .= '/';
 352         }
 353
 354         return $path;
 355 }
 356
 357 # Write data to disk.
 358 function write_file($path, $data) {
 359         if (is_null($path)) {
 360                 error('No path specified for writing to.');
 361         }        elseif (is_null($data)) {
 362                 error('No data specified for writing.');
 363         }
 364
 365         if (($fh_out = fopen($path,'w')) === FALSE) {
 366                 error("couldn't open $path for writing.");
 367         }
 368
 369         if (fwrite($fh_out, $data) === FALSE) {
 370                 error("couldn't write to $path.");
 371         }
 372 }
 373
 374 # Display an error message and exit.
 375 function error ($error_msg) {
 376         if (is_null($error_msg)) {
 377                 $error_msg = 'error() was called without any explanation!';
 378         }
 379
 380         echo "Error: $error_msg\n";
 381         exit(1);
 382 }
 383
 384 ?>