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