application/LinkFilter.php

   1 <?php
   2
   3 /**
   4  * Class LinkFilter.
   5  *
   6  * Perform search and filter operation on link data list.
   7  */
   8 class LinkFilter
   9 {
  10     /**
  11      * @var string permalinks.
  12      */
  13     public static $FILTER_HASH   = 'permalink';
  14
  15     /**
  16      * @var string text search.
  17      */
  18     public static $FILTER_TEXT   = 'fulltext';
  19
  20     /**
  21      * @var string tag filter.
  22      */
  23     public static $FILTER_TAG    = 'tags';
  24
  25     /**
  26      * @var string filter by day.
  27      */
  28     public static $FILTER_DAY    = 'FILTER_DAY';
  29
  30     /**
  31      * @var string Allowed characters for hashtags (regex syntax).
  32      */
  33     public static $HASHTAG_CHARS = '\p{Pc}\p{N}\p{L}\p{Mn}';
  34
  35     /**
  36      * @var array all available links.
  37      */
  38     private $links;
  39
  40     /**
  41      * @param array $links initialization.
  42      */
  43     public function __construct($links)
  44     {
  45         $this->links = $links;
  46     }
  47
  48     /**
  49      * Filter links according to parameters.
  50      *
  51      * @param string $type          Type of filter (eg. tags, permalink, etc.).
  52      * @param mixed  $request       Filter content.
  53      * @param bool   $casesensitive Optional: Perform case sensitive filter if true.
  54      * @param bool   $privateonly   Optional: Only returns private links if true.
  55      *
  56      * @return array filtered link list.
  57      */
  58     public function filter($type, $request, $casesensitive = false, $privateonly = false)
  59     {
  60         switch($type) {
  61             case self::$FILTER_HASH:
  62                 return $this->filterSmallHash($request);
  63             case self::$FILTER_TAG | self::$FILTER_TEXT:
  64                 if (!empty($request)) {
  65                     $filtered = $this->links;
  66                     if (isset($request[0])) {
  67                         $filtered = $this->filterTags($request[0], $casesensitive, $privateonly);
  68                     }
  69                     if (isset($request[1])) {
  70                         $lf = new LinkFilter($filtered);
  71                         $filtered = $lf->filterFulltext($request[1], $privateonly);
  72                     }
  73                     return $filtered;
  74                 }
  75                 return $this->noFilter($privateonly);
  76             case self::$FILTER_TEXT:
  77                 return $this->filterFulltext($request, $privateonly);
  78             case self::$FILTER_TAG:
  79                 return $this->filterTags($request, $casesensitive, $privateonly);
  80             case self::$FILTER_DAY:
  81                 return $this->filterDay($request);
  82             default:
  83                 return $this->noFilter($privateonly);
  84         }
  85     }
  86
  87     /**
  88      * Unknown filter, but handle private only.
  89      *
  90      * @param bool $privateonly returns private link only if true.
  91      *
  92      * @return array filtered links.
  93      */
  94     private function noFilter($privateonly = false)
  95     {
  96         if (! $privateonly) {
  97             krsort($this->links);
  98             return $this->links;
  99         }
 100
 101         $out = array();
 102         foreach ($this->links as $value) {
 103             if ($value['private']) {
 104                 $out[$value['linkdate']] = $value;
 105             }
 106         }
 107
 108         krsort($out);
 109         return $out;
 110     }
 111
 112     /**
 113      * Returns the shaare corresponding to a smallHash.
 114      *
 115      * @param string $smallHash permalink hash.
 116      *
 117      * @return array $filtered array containing permalink data.
 118      *
 119      * @throws LinkNotFoundException if the smallhash doesn't match any link.
 120      */
 121     private function filterSmallHash($smallHash)
 122     {
 123         $filtered = array();
 124         foreach ($this->links as $l) {
 125             if ($smallHash == smallHash($l['linkdate'])) {
 126                 // Yes, this is ugly and slow
 127                 $filtered[$l['linkdate']] = $l;
 128                 return $filtered;
 129             }
 130         }
 131
 132         if (empty($filtered)) {
 133             throw new LinkNotFoundException();
 134         }
 135
 136         return $filtered;
 137     }
 138
 139     /**
 140      * Returns the list of links corresponding to a full-text search
 141      *
 142      * Searches:
 143      *  - in the URLs, title and description;
 144      *  - are case-insensitive;
 145      *  - terms surrounded by quotes " are exact terms search.
 146      *  - terms starting with a dash - are excluded (except exact terms).
 147      *
 148      * Example:
 149      *    print_r($mydb->filterFulltext('hollandais'));
 150      *
 151      * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
 152      *  - allows to perform searches on Unicode text
 153      *  - see https://github.com/shaarli/Shaarli/issues/75 for examples
 154      *
 155      * @param string $searchterms search query.
 156      * @param bool   $privateonly return only private links if true.
 157      *
 158      * @return array search results.
 159      */
 160     private function filterFulltext($searchterms, $privateonly = false)
 161     {
 162         if (empty($searchterms)) {
 163             return $this->links;
 164         }
 165
 166         $filtered = array();
 167         $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8');
 168         $exactRegex = '/"([^"]+)"/';
 169         // Retrieve exact search terms.
 170         preg_match_all($exactRegex, $search, $exactSearch);
 171         $exactSearch = array_values(array_filter($exactSearch[1]));
 172
 173         // Remove exact search terms to get AND terms search.
 174         $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search)));
 175         $explodedSearchAnd = array_values(array_filter($explodedSearchAnd));
 176
 177         // Filter excluding terms and update andSearch.
 178         $excludeSearch = array();
 179         $andSearch = array();
 180         foreach ($explodedSearchAnd as $needle) {
 181             if ($needle[0] == '-' && strlen($needle) > 1) {
 182                 $excludeSearch[] = substr($needle, 1);
 183             } else {
 184                 $andSearch[] = $needle;
 185             }
 186         }
 187
 188         $keys = array('title', 'description', 'url', 'tags');
 189
 190         // Iterate over every stored link.
 191         foreach ($this->links as $link) {
 192
 193             // ignore non private links when 'privatonly' is on.
 194             if (! $link['private'] && $privateonly === true) {
 195                 continue;
 196             }
 197
 198             // Concatenate link fields to search across fields.
 199             // Adds a '\' separator for exact search terms.
 200             $content = '';
 201             foreach ($keys as $key) {
 202                 $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\';
 203             }
 204
 205             // Be optimistic
 206             $found = true;
 207
 208             // First, we look for exact term search
 209             for ($i = 0; $i < count($exactSearch) && $found; $i++) {
 210                 $found = strpos($content, $exactSearch[$i]) !== false;
 211             }
 212
 213             // Iterate over keywords, if keyword is not found,
 214             // no need to check for the others. We want all or nothing.
 215             for ($i = 0; $i < count($andSearch) && $found; $i++) {
 216                 $found = strpos($content, $andSearch[$i]) !== false;
 217             }
 218
 219             // Exclude terms.
 220             for ($i = 0; $i < count($excludeSearch) && $found; $i++) {
 221                 $found = strpos($content, $excludeSearch[$i]) === false;
 222             }
 223
 224             if ($found) {
 225                 $filtered[$link['linkdate']] = $link;
 226             }
 227         }
 228
 229         krsort($filtered);
 230         return $filtered;
 231     }
 232
 233     /**
 234      * Returns the list of links associated with a given list of tags
 235      *
 236      * You can specify one or more tags, separated by space or a comma, e.g.
 237      *  print_r($mydb->filterTags('linux programming'));
 238      *
 239      * @param string $tags          list of tags separated by commas or blank spaces.
 240      * @param bool   $casesensitive ignore case if false.
 241      * @param bool   $privateonly   returns private links only.
 242      *
 243      * @return array filtered links.
 244      */
 245     public function filterTags($tags, $casesensitive = false, $privateonly = false)
 246     {
 247         // Implode if array for clean up.
 248         $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
 249         if (empty($tags)) {
 250             return $this->links;
 251         }
 252
 253         $searchtags = self::tagsStrToArray($tags, $casesensitive);
 254         $filtered = array();
 255         if (empty($searchtags)) {
 256             return $filtered;
 257         }
 258
 259         foreach ($this->links as $link) {
 260             // ignore non private links when 'privatonly' is on.
 261             if (! $link['private'] && $privateonly === true) {
 262                 continue;
 263             }
 264
 265             $linktags = self::tagsStrToArray($link['tags'], $casesensitive);
 266
 267             $found = true;
 268             for ($i = 0 ; $i < count($searchtags) && $found; $i++) {
 269                 // Exclusive search, quit if tag found.
 270                 // Or, tag not found in the link, quit.
 271                 if (($searchtags[$i][0] == '-'
 272                         && $this->searchTagAndHashTag(substr($searchtags[$i], 1), $linktags, $link['description']))
 273                     || ($searchtags[$i][0] != '-')
 274                         && ! $this->searchTagAndHashTag($searchtags[$i], $linktags, $link['description'])
 275                 ) {
 276                     $found = false;
 277                 }
 278             }
 279
 280             if ($found) {
 281                 $filtered[$link['linkdate']] = $link;
 282             }
 283         }
 284         krsort($filtered);
 285         return $filtered;
 286     }
 287
 288     /**
 289      * Returns the list of articles for a given day, chronologically sorted
 290      *
 291      * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
 292      *  print_r($mydb->filterDay('20120125'));
 293      *
 294      * @param string $day day to filter.
 295      *
 296      * @return array all link matching given day.
 297      *
 298      * @throws Exception if date format is invalid.
 299      */
 300     public function filterDay($day)
 301     {
 302         if (! checkDateFormat('Ymd', $day)) {
 303             throw new Exception('Invalid date format');
 304         }
 305
 306         $filtered = array();
 307         foreach ($this->links as $l) {
 308             if (startsWith($l['linkdate'], $day)) {
 309                 $filtered[$l['linkdate']] = $l;
 310             }
 311         }
 312         ksort($filtered);
 313         return $filtered;
 314     }
 315
 316     /**
 317      * Check if a tag is found in the taglist, or as an hashtag in the link description.
 318      *
 319      * @param string $tag         Tag to search.
 320      * @param array  $taglist     List of tags for the current link.
 321      * @param string $description Link description.
 322      *
 323      * @return bool True if found, false otherwise.
 324      */
 325     protected function searchTagAndHashTag($tag, $taglist, $description)
 326     {
 327         if (in_array($tag, $taglist)) {
 328             return true;
 329         }
 330
 331         if (preg_match('/(^| )#'. $tag .'([^'. self::$HASHTAG_CHARS .']|$)/mui', $description) > 0) {
 332             return true;
 333         }
 334
 335         return false;
 336     }
 337
 338     /**
 339      * Convert a list of tags (str) to an array. Also
 340      * - handle case sensitivity.
 341      * - accepts spaces commas as separator.
 342      *
 343      * @param string $tags          string containing a list of tags.
 344      * @param bool   $casesensitive will convert everything to lowercase if false.
 345      *
 346      * @return array filtered tags string.
 347     */
 348     public static function tagsStrToArray($tags, $casesensitive)
 349     {
 350         // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
 351         $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
 352         $tagsOut = str_replace(',', ' ', $tagsOut);
 353
 354         return array_values(array_filter(explode(' ', trim($tagsOut)), 'strlen'));
 355     }
 356 }
 357
 358 class LinkNotFoundException extends Exception
 359 {
 360     protected $message = 'The link you are trying to reach does not exist or has been deleted.';
 361 }