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 string $visibility    Optional: return only all/private/public links
  55      * @param string $untaggedonly  Optional: return only untagged links. Applies only if $type includes FILTER_TAG
  56      *
  57      * @return array filtered link list.
  58      */
  59     public function filter($type, $request, $casesensitive = false, $visibility = 'all', $untaggedonly = false)
  60     {
  61         if (! in_array($visibility, ['all', 'public', 'private'])) {
  62             $visibility = 'all';
  63         }
  64
  65         switch($type) {
  66             case self::$FILTER_HASH:
  67                 return $this->filterSmallHash($request);
  68             case self::$FILTER_TAG | self::$FILTER_TEXT: // == "vuotext"
  69                 $noRequest = empty($request) || (empty($request[0]) && empty($request[1]));
  70                 if ($noRequest) {
  71                     if ($untaggedonly) {
  72                         return $this->filterUntagged($visibility);
  73                     }
  74                     return $this->noFilter($visibility);
  75                 }
  76                 if ($untaggedonly) {
  77                     $filtered = $this->filterUntagged($visibility);
  78                 } else {
  79                     $filtered = $this->links;
  80                 }
  81                 if (!empty($request[0])) {
  82                     $filtered = (new LinkFilter($filtered))->filterTags($request[0], $casesensitive, $visibility);
  83                 }
  84                 if (!empty($request[1])) {
  85                     $filtered = (new LinkFilter($filtered))->filterFulltext($request[1], $visibility);
  86                 }
  87                 return $filtered;
  88             case self::$FILTER_TEXT:
  89                 return $this->filterFulltext($request, $visibility);
  90             case self::$FILTER_TAG:
  91                 if ($untaggedonly) {
  92                     return $this->filterUntagged($visibility);
  93                 } else {
  94                     return $this->filterTags($request, $casesensitive, $visibility);
  95                 }
  96             case self::$FILTER_DAY:
  97                 return $this->filterDay($request);
  98             default:
  99                 return $this->noFilter($visibility);
 100         }
 101     }
 102
 103     /**
 104      * Unknown filter, but handle private only.
 105      *
 106      * @param string $visibility Optional: return only all/private/public links
 107      *
 108      * @return array filtered links.
 109      */
 110     private function noFilter($visibility = 'all')
 111     {
 112         if ($visibility === 'all') {
 113             return $this->links;
 114         }
 115
 116         $out = array();
 117         foreach ($this->links as $key => $value) {
 118             if ($value['private'] && $visibility === 'private') {
 119                 $out[$key] = $value;
 120             } else if (! $value['private'] && $visibility === 'public') {
 121                 $out[$key] = $value;
 122             }
 123         }
 124
 125         return $out;
 126     }
 127
 128     /**
 129      * Returns the shaare corresponding to a smallHash.
 130      *
 131      * @param string $smallHash permalink hash.
 132      *
 133      * @return array $filtered array containing permalink data.
 134      *
 135      * @throws LinkNotFoundException if the smallhash doesn't match any link.
 136      */
 137     private function filterSmallHash($smallHash)
 138     {
 139         $filtered = array();
 140         foreach ($this->links as $key => $l) {
 141             if ($smallHash == $l['shorturl']) {
 142                 // Yes, this is ugly and slow
 143                 $filtered[$key] = $l;
 144                 return $filtered;
 145             }
 146         }
 147
 148         if (empty($filtered)) {
 149             throw new LinkNotFoundException();
 150         }
 151
 152         return $filtered;
 153     }
 154
 155     /**
 156      * Returns the list of links corresponding to a full-text search
 157      *
 158      * Searches:
 159      *  - in the URLs, title and description;
 160      *  - are case-insensitive;
 161      *  - terms surrounded by quotes " are exact terms search.
 162      *  - terms starting with a dash - are excluded (except exact terms).
 163      *
 164      * Example:
 165      *    print_r($mydb->filterFulltext('hollandais'));
 166      *
 167      * mb_convert_case($val, MB_CASE_LOWER, 'UTF-8')
 168      *  - allows to perform searches on Unicode text
 169      *  - see https://github.com/shaarli/Shaarli/issues/75 for examples
 170      *
 171      * @param string $searchterms search query.
 172      * @param string $visibility Optional: return only all/private/public links.
 173      *
 174      * @return array search results.
 175      */
 176     private function filterFulltext($searchterms, $visibility = 'all')
 177     {
 178         if (empty($searchterms)) {
 179             return $this->noFilter($visibility);
 180         }
 181
 182         $filtered = array();
 183         $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8');
 184         $exactRegex = '/"([^"]+)"/';
 185         // Retrieve exact search terms.
 186         preg_match_all($exactRegex, $search, $exactSearch);
 187         $exactSearch = array_values(array_filter($exactSearch[1]));
 188
 189         // Remove exact search terms to get AND terms search.
 190         $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search)));
 191         $explodedSearchAnd = array_values(array_filter($explodedSearchAnd));
 192
 193         // Filter excluding terms and update andSearch.
 194         $excludeSearch = array();
 195         $andSearch = array();
 196         foreach ($explodedSearchAnd as $needle) {
 197             if ($needle[0] == '-' && strlen($needle) > 1) {
 198                 $excludeSearch[] = substr($needle, 1);
 199             } else {
 200                 $andSearch[] = $needle;
 201             }
 202         }
 203
 204         $keys = array('title', 'description', 'url', 'tags');
 205
 206         // Iterate over every stored link.
 207         foreach ($this->links as $id => $link) {
 208
 209             // ignore non private links when 'privatonly' is on.
 210             if ($visibility !== 'all') {
 211                 if (! $link['private'] && $visibility === 'private') {
 212                     continue;
 213                 } else if ($link['private'] && $visibility === 'public') {
 214                     continue;
 215                 }
 216             }
 217
 218             // Concatenate link fields to search across fields.
 219             // Adds a '\' separator for exact search terms.
 220             $content = '';
 221             foreach ($keys as $key) {
 222                 $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\';
 223             }
 224
 225             // Be optimistic
 226             $found = true;
 227
 228             // First, we look for exact term search
 229             for ($i = 0; $i < count($exactSearch) && $found; $i++) {
 230                 $found = strpos($content, $exactSearch[$i]) !== false;
 231             }
 232
 233             // Iterate over keywords, if keyword is not found,
 234             // no need to check for the others. We want all or nothing.
 235             for ($i = 0; $i < count($andSearch) && $found; $i++) {
 236                 $found = strpos($content, $andSearch[$i]) !== false;
 237             }
 238
 239             // Exclude terms.
 240             for ($i = 0; $i < count($excludeSearch) && $found; $i++) {
 241                 $found = strpos($content, $excludeSearch[$i]) === false;
 242             }
 243
 244             if ($found) {
 245                 $filtered[$id] = $link;
 246             }
 247         }
 248
 249         return $filtered;
 250     }
 251
 252     /**
 253      * Returns the list of links associated with a given list of tags
 254      *
 255      * You can specify one or more tags, separated by space or a comma, e.g.
 256      *  print_r($mydb->filterTags('linux programming'));
 257      *
 258      * @param string $tags          list of tags separated by commas or blank spaces.
 259      * @param bool   $casesensitive ignore case if false.
 260      * @param string $visibility    Optional: return only all/private/public links.
 261      *
 262      * @return array filtered links.
 263      */
 264     public function filterTags($tags, $casesensitive = false, $visibility = 'all')
 265     {
 266         // Implode if array for clean up.
 267         $tags = is_array($tags) ? trim(implode(' ', $tags)) : $tags;
 268         if (empty($tags)) {
 269             return $this->noFilter($visibility);
 270         }
 271
 272         $searchtags = self::tagsStrToArray($tags, $casesensitive);
 273         $filtered = array();
 274         if (empty($searchtags)) {
 275             return $filtered;
 276         }
 277
 278         foreach ($this->links as $key => $link) {
 279             // ignore non private links when 'privatonly' is on.
 280             if ($visibility !== 'all') {
 281                 if (! $link['private'] && $visibility === 'private') {
 282                     continue;
 283                 } else if ($link['private'] && $visibility === 'public') {
 284                     continue;
 285                 }
 286             }
 287
 288             $linktags = self::tagsStrToArray($link['tags'], $casesensitive);
 289
 290             $found = true;
 291             for ($i = 0 ; $i < count($searchtags) && $found; $i++) {
 292                 // Exclusive search, quit if tag found.
 293                 // Or, tag not found in the link, quit.
 294                 if (($searchtags[$i][0] == '-'
 295                         && $this->searchTagAndHashTag(substr($searchtags[$i], 1), $linktags, $link['description']))
 296                     || ($searchtags[$i][0] != '-')
 297                         && ! $this->searchTagAndHashTag($searchtags[$i], $linktags, $link['description'])
 298                 ) {
 299                     $found = false;
 300                 }
 301             }
 302
 303             if ($found) {
 304                 $filtered[$key] = $link;
 305             }
 306         }
 307         return $filtered;
 308     }
 309
 310     /**
 311      * Return only links without any tag.
 312      *
 313      * @param string $visibility return only all/private/public links.
 314      *
 315      * @return array filtered links.
 316      */
 317     public function filterUntagged($visibility)
 318     {
 319         $filtered = [];
 320         foreach ($this->links as $key => $link) {
 321             if ($visibility !== 'all') {
 322                 if (! $link['private'] && $visibility === 'private') {
 323                     continue;
 324                 } else if ($link['private'] && $visibility === 'public') {
 325                     continue;
 326                 }
 327             }
 328
 329             if (empty(trim($link['tags']))) {
 330                 $filtered[$key] = $link;
 331             }
 332         }
 333
 334         return $filtered;
 335     }
 336
 337     /**
 338      * Returns the list of articles for a given day, chronologically sorted
 339      *
 340      * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
 341      *  print_r($mydb->filterDay('20120125'));
 342      *
 343      * @param string $day day to filter.
 344      *
 345      * @return array all link matching given day.
 346      *
 347      * @throws Exception if date format is invalid.
 348      */
 349     public function filterDay($day)
 350     {
 351         if (! checkDateFormat('Ymd', $day)) {
 352             throw new Exception('Invalid date format');
 353         }
 354
 355         $filtered = array();
 356         foreach ($this->links as $key => $l) {
 357             if ($l['created']->format('Ymd') == $day) {
 358                 $filtered[$key] = $l;
 359             }
 360         }
 361
 362         // sort by date ASC
 363         return array_reverse($filtered, true);
 364     }
 365
 366     /**
 367      * Check if a tag is found in the taglist, or as an hashtag in the link description.
 368      *
 369      * @param string $tag         Tag to search.
 370      * @param array  $taglist     List of tags for the current link.
 371      * @param string $description Link description.
 372      *
 373      * @return bool True if found, false otherwise.
 374      */
 375     protected function searchTagAndHashTag($tag, $taglist, $description)
 376     {
 377         if (in_array($tag, $taglist)) {
 378             return true;
 379         }
 380
 381         if (preg_match('/(^| )#'. $tag .'([^'. self::$HASHTAG_CHARS .']|$)/mui', $description) > 0) {
 382             return true;
 383         }
 384
 385         return false;
 386     }
 387
 388     /**
 389      * Convert a list of tags (str) to an array. Also
 390      * - handle case sensitivity.
 391      * - accepts spaces commas as separator.
 392      *
 393      * @param string $tags          string containing a list of tags.
 394      * @param bool   $casesensitive will convert everything to lowercase if false.
 395      *
 396      * @return array filtered tags string.
 397      */
 398     public static function tagsStrToArray($tags, $casesensitive)
 399     {
 400         // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
 401         $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
 402         $tagsOut = str_replace(',', ' ', $tagsOut);
 403
 404         return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
 405     }
 406 }
 407
 408 class LinkNotFoundException extends Exception
 409 {
 410     protected $message = 'The link you are trying to reach does not exist or has been deleted.';
 411 }