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