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 . $file_name,
  35                                                                  'changefreq' => 'daily',
  36                                                                  'priority' => '1',
  37                                                                  )
  38                                                    );
  39         }
  40
  41         write_file($output_paths['index_file'], urlset($index_urls));
  42
  43 }
  44
  45 # Generate sitemap of standard site elements.
  46 function standard_map() {
  47         global $output_paths;
  48
  49         $standard_map_urls .= url(
  50                                                           array(
  51                                                                         'url' => common_local_url('public'),
  52                                                                         'changefreq' => 'daily',
  53                                                                         'priority' => '1',
  54                                                                         )
  55                                                           );
  56
  57         $standard_map_urls .= url(
  58                                                           array(
  59                                                                         'url' => common_local_url('publicrss'),
  60                                                                         'changefreq' => 'daily',
  61                                                                         'priority' => '0.3',
  62                                                                         )
  63                                                           );
  64
  65         $docs = array('about', 'faq', 'contact', 'im', 'openid', 'openmublog', 'privacy', 'source');
  66
  67         foreach($docs as $title) {
  68                 $standard_map_urls .= url(
  69                                                                   array(
  70                                                                                 'url' => common_local_url('doc', array('title' => $title)),
  71                                                                                 'changefreq' => 'monthly',
  72                                                                                 'priority'   => '0.2',
  73                                                                                 )
  74                                                                   );
  75         }
  76
  77         $urlset_path = $output_paths['output_dir'] . 'standard.xml';
  78
  79         write_file($urlset_path, urlset($standard_map_urls));
  80 }
  81
  82 # Generate sitemaps of all notices.
  83 function notices_map() {
  84         global $output_paths;
  85
  86         $notices = DB_DataObject::factory('notice');
  87
  88         $notices->query('SELECT uri, modified FROM notice');
  89
  90         $notice_count = 0;
  91         $map_count = 1;
  92
  93         while ($notices->fetch()) {
  94
  95                 # Maximum 50,000 URLs per sitemap file.
  96                 if ($notice_count == 50000) {
  97                         $notice_count = 0;
  98                         $map_count++;
  99                 }
 100
 101                 $notice = array(
 102                                                 'url'        => $notices->uri,
 103                                                 'lastmod'    => w3cdate($notices->modified),
 104                                                 'changefreq' => 'daily',
 105                                                 'priority'   => '1',
 106                                                 );
 107
 108                 $notice_list[$map_count] .= url($notice);
 109
 110                 $notice_count++;
 111         }
 112
 113         # Make full sitemaps from the lists and save them.
 114         array_to_map($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 # Generate sitemaps of all avatars.
 207 function avatar_map() {
 208         global $output_paths;
 209
 210         $avatars = DB_DataObject::factory('avatar');
 211
 212         $avatars->query('SELECT url, modified FROM avatar');
 213
 214         $avatar_count = 0;
 215         $map_count = 1;
 216
 217         while ($avatars->fetch()) {
 218
 219                 # We only want the original size and 24px thumbnail version - skip 96px.
 220                 if (preg_match('/-96-/', $avatars->url)) {
 221                         continue;
 222                 }
 223
 224                 # Maximum 50,000 URLs per sitemap file.
 225                 if ($avatar_count == 50000) {
 226                         $avatar_count = 0;
 227                         $map_count++;
 228                 }
 229 w3cdate($avatars->modified);
 230                 $image = array(
 231                                            'url'        => $avatars->url,
 232                                            'lastmod'    => w3cdate($avatars->modified),
 233                                            'changefreq' => 'monthly',
 234                                            'priority'   => '0.2',
 235                                            );
 236
 237                 # Construct a <url></url> element for each avatar and add it
 238                 # to our existing list of those.
 239                 $avatar_list[$map_count] .= url($image);
 240         }
 241
 242         array_to_map($avatar_list, 'avatars');
 243 }
 244
 245 # ------------------------------------------------------------------------------
 246 # XML generation functions
 247 # ------------------------------------------------------------------------------
 248
 249 # Generate a <url></url> element.
 250 function url($url_args) {
 251         $url        = preg_replace('/&/', '&amp;', $url_args['url']); # escape ampersands for XML
 252         $lastmod    = $url_args['lastmod'];
 253         $changefreq = $url_args['changefreq'];
 254         $priority   = $url_args['priority'];
 255
 256         if (is_null($url)) {
 257                 error("url() arguments require 'url' value.");
 258         }
 259
 260         $url_out = "\t<url>\n";
 261         $url_out .= "\t\t<loc>$url</loc>\n";
 262
 263         if ($changefreq) {
 264                 $url_out .= "\t\t<changefreq>$changefreq</changefreq>\n";
 265         }
 266
 267         if ($lastmod) {
 268                 $url_out .= "\t\t<lastmod>$lastmod</lastmod>\n";
 269         }
 270
 271         if ($priority) {
 272                 $url_out .= "\t\t<priority>$priority</priority>\n";
 273         }
 274
 275         $url_out .= "\t</url>\n";
 276
 277         return $url_out;
 278 }
 279
 280 # Generate a <urlset></urlset> element.
 281 function urlset($urlset_text) {
 282         $urlset = '<?xml version="1.0" encoding="UTF-8"?>' . "\n" .
 283           '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n" .
 284           $urlset_text .
 285           '</urlset>';
 286
 287         return $urlset;
 288 }
 289
 290 # Generate a sitemap from an array containing <url></url> elements and write it to a file.
 291 function array_to_map($url_list, $filename_prefix) {
 292         global $output_paths;
 293
 294         if ($url_list) {
 295                 # $map_urls is a long string containing concatenated <url></url> elements.
 296                 while (list($map_idx, $map_urls) = each($url_list)) {
 297                         $urlset_path = $output_paths['output_dir'] . "$filename_prefix-$map_idx.xml";
 298
 299                         write_file($urlset_path, urlset($map_urls));
 300                 }
 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 path> -d <output directory path> -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 $paths;
 347 }
 348
 349 # Format database timestamps as W3C DTF.
 350 function w3cdate ($timestamp) {
 351         preg_match('/(\d{4})-(\d\d)-(\d\d) (\d\d):(\d\d):(\d\d)/', $timestamp, $date);
 352
 353         return date(DATE_W3C, mktime($date[4], $date[5], $date[6], $date[2], $date[3], $date[1]));
 354 }
 355
 356 # Ensure paths end with a "/".
 357 function trailing_slash($path) {
 358         if (preg_match('/\/$/', $path) == 0) {
 359                 $path .= '/';
 360         }
 361
 362         return $path;
 363 }
 364
 365 # Write data to disk.
 366 function write_file($path, $data) {
 367         if (is_null($path)) {
 368                 error('No path specified for writing to.');
 369         }        elseif (is_null($data)) {
 370                 error('No data specified for writing.');
 371         }
 372
 373         if (($fh_out = fopen($path,'w')) === FALSE) {
 374                 error("couldn't open $path for writing.");
 375         }
 376
 377         if (fwrite($fh_out, $data) === FALSE) {
 378                 error("couldn't write to $path.");
 379         }
 380 }
 381
 382 # Display an error message and exit.
 383 function error ($error_msg) {
 384         if (is_null($error_msg)) {
 385                 $error_msg = 'error() was called without any explanation!';
 386         }
 387
 388         echo "Error: $error_msg\n";
 389         exit(1);
 390 }
 391
 392 ?>