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