XHTMLCompiler/Page.php

   1 <?php
   2
   3 /**
   4  * Represents a page in our content management system. This is loosely
   5  * bound to the filesystem, although it doesn't actually refer to a
   6  * specific file, just a class of files.
   7  */
   8 class XHTMLCompiler_Page
   9 {
  10
  11     /**
  12      * Filename identifier of this page without extension
  13      */
  14     protected $pathStem;
  15
  16     /** File extension of source files (no period) */
  17     protected $sourceExt = 'xhtml';
  18     /** File extension of cache/served files */
  19     protected $cacheExt = 'html';
  20     /** File extension of dependency files */
  21     protected $depsExt = 'xc-deps';
  22
  23     /** Instance of XHTMLCompiler_File for source file */
  24     protected $source;
  25     /** Instance of XHTMLCompiler_File for cache file */
  26     protected $cache;
  27     /** Instance of XHTMLCompiler_File for dependency file */
  28     protected $deps;
  29
  30     /** Instance of XHTMLCompiler_Directory for all of the above files*/
  31     protected $dir;
  32
  33     /** Array of attributes about this file. Currently used by News/NewsLinker */
  34     public $attr = array();
  35
  36     /**
  37      * Constructs a page object, validates filename for correctness
  38      * @param $path String path filename, can be from untrusted source
  39      * @param $mute Whether or not to stop the class from complaining when
  40      *              the source file doesn't exist. This is a stopgap measure,
  41      *              please replace with better exception handling.
  42      * @todo Cleanup into subroutines
  43      * @todo Factor out allowed_directories realpath'ing to config class
  44      */
  45     public function __construct($path, $mute = false) {
  46
  47         $xc  = XHTMLCompiler::getInstance();
  48         $php = XHTMLCompiler::getPHPWrapper();
  49
  50         // test file extension
  51         $info = pathinfo($path);
  52         if (
  53             empty($info['extension']) || (
  54                 $info['extension'] !== $this->sourceExt &&
  55                 $info['extension'] !== $this->cacheExt
  56             )
  57         ) {
  58             throw new XHTMLCompiler_Exception(403, 'Forbidden extension',
  59               'File extension cannot be processed by XHTML Compiler, check
  60               for faulty <code>.htaccess</code> rules.');
  61         }
  62
  63         // test for directory's existence and resolve to real path
  64         $dir = $info['dirname'];
  65         if ($dir == '.') $dir .= '/';
  66         $dir = $php->realpath($dir);
  67         if ($dir === false) {
  68             throw new XHTMLCompiler_Exception(404, 'Missing directory',
  69                 'Requested directory cannot be found; check your file
  70                 path and try again.' );
  71         }
  72         if ($dir[strlen($dir)-1] == '/') $dir = substr($dir, 0, -1);
  73
  74         $allowed_dirs = $xc->getConf('allowed_dirs');
  75         $ok = false;
  76
  77         foreach ($allowed_dirs as $allowed_dir => $recursive) {
  78             $allowed_dir = $php->realpath($allowed_dir); // factor out!
  79             if (!is_string($allowed_dir)) continue;
  80             if ($dir === $allowed_dir) {
  81                 $ok = true;
  82                 break;
  83             // slash is required to prevent $allowed_dir = 'subdir' from
  84             // matching $dir = 'subdirectory', thanks Mordred!
  85             } elseif (strpos($dir, $allowed_dir . '/') === 0 && $recursive) {
  86                 $ok = true;
  87                 break;
  88             }
  89         }
  90
  91         if (!$ok) throw new XHTMLCompiler_Exception(403, 'Forbidden directory',
  92           'Requested directory is forbidden to XHTML Compiler; try
  93           accessing it directly or check for faulty <code>.htaccess</code> rules.');
  94
  95         // cannot use pathinfo, since PATHINFO_FILENAME is PHP 5.2.0
  96         $this->pathStem = substr($path, 0, strrpos($path, '.'));
  97
  98         // setup the files
  99         $this->source = new XHTMLCompiler_File($this->pathStem . '.' . $this->sourceExt);
 100         $this->cache  = new XHTMLCompiler_File($this->pathStem . '.' . $this->cacheExt);
 101         $this->deps   = new XHTMLCompiler_File($this->pathStem . '.' . $this->depsExt);
 102
 103         $this->dir    = new XHTMLCompiler_Directory(dirname($this->pathStem));
 104
 105         if (!$mute && !$this->source->exists()) {
 106             // Apache may have redirected to an ErrorDocument which got directed
 107             // via mod_rewrite to us, in that case, output the corresponding
 108             // status code.  Otherwise, we can give the regular 404.
 109             $code = $php->getRedirectStatus();
 110             if (!$code || $code == 200) $code = 404;
 111             throw new XHTMLCompiler_Exception($code, 'Page not found', 'Requested page not found; check the URL in your address bar.');
 112         }
 113     }
 114
 115     // Note: Do not use this functions internally inside the class
 116
 117     /** Returns path stem, full filename without file extension */
 118     public function getPathStem() { return $this->pathStem; }
 119     /** Returns relative path to cache */
 120     public function getCachePath() { return $this->cache->getName(); }
 121     /** Returns relative path to source */
 122     public function getSourcePath() { return $this->source->getName(); }
 123     /** Returns XHTMLCompiler_Directory representation of directory */
 124     public function getDir() { return $this->dir; }
 125     /** Returns directory of the files without trailing slash */
 126     public function getDirName() { return $this->dir->getName(); }
 127     /** Returns directory of the files with trailing slash (unless there is none) */
 128     public function getDirSName() { return $this->dir->getSName(); }
 129     /** Returns how deep from the root the file is */
 130     public function getDepth() { return substr_count($this->getSourcePath(), '/'); }
 131
 132     /** Normalizes a relative path as if it were from this page's directory */
 133     public function normalizePath($path) {
 134         return $this->getDirName() . '/' . $path;
 135     }
 136
 137     /**
 138      * Returns a fully formed web path with web domain to the file. This path
 139      * is valid anywhere on the web.
 140      */
 141     public function getWebPath() {
 142         $xc = XHTMLCompiler::getInstance();
 143         $domain = $xc->getConf('web_domain');
 144         if (!$domain) {
 145             throw new Exception('Configuration value web_domain must be set for command line');
 146         }
 147         return 'http://' . $domain . $this->getAbsolutePath();
 148     }
 149
 150     /**
 151      * Returns a fully formed absolute web path valid anywhere on the
 152      * current domain to the cached file.
 153      */
 154     public function getAbsolutePath() {
 155         $xc = XHTMLCompiler::getInstance();
 156         $name = $this->cache->getName();
 157         // a little icky
 158         if ($name[0] !== '/') $name = "/$name";
 159         if (strncmp($name, './', 2) === 0) $name = substr($name, 1);
 160         return $xc->getConf('web_path') . $name;
 161     }
 162
 163     /** Returns contents of the cache/served file */
 164     public function getCache() { return $this->cache->get(); }
 165     /** Returns contents of the source file */
 166     public function getSource() { return $this->source->get(); }
 167
 168     /** Reports whether or not cache file exists and is a file */
 169     public function isCacheExistent() { return $this->cache->exists(); }
 170     /** Reports whether or not source file exists and is a file */
 171     public function isSourceExistent() { return $this->source->exists(); }
 172
 173     /**
 174      * Reports whether or not the cache is stale by comparing the file
 175      * modification times between the source file and the cache file.
 176      * @warning You must not call this function until you've also called
 177      *          isCacheExistent().
 178      */
 179     public function isCacheStale() {
 180         if (!$this->cache->exists()) {
 181             throw new Exception('Cannot check for stale cache when cache
 182                 does not exist, please call isCacheExistent and take
 183                 appropriate action with the result');
 184         }
 185         if ($this->source->getMTime() > $this->cache->getMTime()) return true;
 186         // check dependencies
 187         if (!$this->deps->exists()) return true; // we need a dependency file!
 188         $deps = unserialize($this->deps->get());
 189         foreach ($deps as $filename => $time) {
 190             if ($time < filemtime($filename)) return true;
 191         }
 192         return false;
 193     }
 194
 195     /**
 196      * Writes text to the cache file, overwriting any previous contents
 197      * and creating the cache file if it doesn't exist.
 198      * @param $contents String contents to write to cache
 199      */
 200     public function writeCache($contents) {$this->cache->write($contents);}
 201
 202     /**
 203      * Attempts to display contents from the cache, otherwise returns false
 204      * @return True if successful, false if not.
 205      * @todo Purge check needs to be factored into XHTMLCompiler
 206      */
 207     public function tryCache() {
 208         if (
 209             !isset($_GET['purge']) &&
 210             $this->cache->exists() &&
 211             !$this->isCacheStale()
 212         ) {
 213             // cached version is fresh, serve it. This shouldn't happen normally
 214             set_response_code(200); // if we used ErrorDocument, override
 215             readfile($this->getCachePath());
 216             return true;
 217         }
 218         return false;
 219     }
 220
 221     /**
 222      * Generates the final version of a page from the source file and writes
 223      * it to the cache.
 224      * @note This function needs to be extended greatly
 225      * @return Generated contents from source
 226      */
 227     public function generate() {
 228         $source = $this->source->get();
 229         $xc = XHTMLCompiler::getInstance();
 230         $filters = $xc->getFilterManager();
 231         $contents = $filters->process($source, $this);
 232         $deps = $filters->getDeps();
 233         if (empty($contents)) return ''; // don't write, probably an error
 234         $contents .= '<!-- generated by XHTML Compiler -->';
 235         $this->cache->write($contents);
 236         $this->cache->chmod(0664);
 237         $this->deps->write(serialize($deps));
 238         return $contents;
 239     }
 240
 241     /**
 242      * Displays the page, either from cache or fresh regeneration.
 243      */
 244     public function display() {
 245         if($this->tryCache()) return;
 246         $ret = $this->generate();
 247         if ($ret) {
 248             if (stripos($_SERVER["HTTP_ACCEPT"], 'application/xhtml+xml') !== false) {
 249                 header("Content-type: application/xhtml+xml");
 250             } else {
 251                 header("Content-type: text/html");
 252             }
 253         }
 254         echo $ret;
 255     }
 256
 257     /**
 258      * Retrieves the Git_Repo that represents this page.
 259      */
 260     public function getRepo() {
 261         return new Git_Repo($this->source->getDirectory());
 262     }
 263
 264     /**
 265      * Retrieves the filename relative to the Git repository root.
 266      */
 267     public function getGitPath() {
 268         $repo = $this->getRepo();
 269         // This won't work with bare repositories
 270         return $name = str_replace(
 271             '\\', '/', // account for Windows
 272             substr(
 273                 realpath($this->source->getName()), // $repo->path is full
 274                 strlen(dirname($repo->path))+1 // chop off "repo" path (w/o .git) + leading slash
 275             )
 276         );
 277     }
 278
 279     /**
 280      * Retrieves the log that represents this page.
 281      */
 282     public function getLog() {
 283         // This doesn't account for sub-repositories
 284         $repo = $this->getRepo();
 285         return $repo->log('master', array($this->getGitPath()), array('follow' => true));
 286     }
 287
 288     // this is metadata stuff that needs to be moved and cached
 289
 290     /**
 291      * Retrieves the DateTime this page was created, according to Git's logs.
 292      */
 293     public function getCreatedTime() {
 294         // As a backwards-compatibility measure, we allow the first meta tag
 295         // with the specific signature:
 296         //    <meta name="Date" contents="..."
 297         // to specify an ISO 8601 formatted date (or date compatible with
 298         // GNU strtotime; Metadata will convert it into ISO 8601 as per
 299         // the Dublin core specification).
 300         $source = $this->source->get();
 301         if (($p = strpos($source, '<meta name="Date" content="')) !== false) {
 302             $p += 27; // cursor is now after the quote
 303             // Grab the time
 304             $time = substr($source, $p, strpos($source, '"', $p) - $p);
 305             return new DateTime($time);
 306         }
 307
 308         $repo = $this->getRepo();
 309         // This is extremely memory inefficient, but I can't figure out
 310         // how to get Git to limit the commits (-n) without undoing
 311         // --reverse.
 312         $log = $repo->log('master', array($this->getGitPath()), array(
 313             'reverse' => true,
 314         ));
 315         if (empty($log)) return;
 316         return $log[0]->authoredDate;
 317     }
 318
 319     /**
 320      * Retrieves the DateTime this page was last updated, according to Git's logs.
 321      */
 322     public function getLastModifiedTime() {
 323         $repo = $this->getRepo();
 324         $log = $repo->log('master', array($this->getGitPath()), array(
 325             'n' => 1,
 326         ));
 327         if (empty($log)) return;
 328         // or committedDate?
 329         return $log[0]->authoredDate;
 330     }
 331
 332 }