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             } elseif (! $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'] && $visibility === 'private') {
 212                     continue;
 213                 } elseif ($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      * generate a regex fragment out of a tag
 254      * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
 255      * @return string generated regex fragment
 256      */
 257     private static function tag2regex($tag)
 258     {
 259         $len = strlen($tag);
 260         if(!$len || $tag === "-" || $tag === "*"){
 261             // nothing to search, return empty regex
 262             return '';
 263         }
 264         if($tag[0] === "-") {
 265             // query is negated
 266             $i = 1; // use offset to start after '-' character
 267             $regex = '(?!'; // create negative lookahead
 268         } else {
 269             $i = 0; // start at first character
 270             $regex = '(?='; // use positive lookahead
 271         }
 272         $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning
 273         // iterate over string, separating it into placeholder and content
 274         for(; $i < $len; $i++){
 275             if($tag[$i] === '*'){
 276                 // placeholder found
 277                 $regex .= '[^ ]*?';
 278             } else {
 279                 // regular characters
 280                 $offset = strpos($tag, '*', $i);
 281                 if($offset === false){
 282                     // no placeholder found, set offset to end of string
 283                     $offset = $len;
 284                 }
 285                 // subtract one, as we want to get before the placeholder or end of string
 286                 $offset -= 1;
 287                 // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
 288                 $regex .= preg_quote(substr($tag, $i, $offset - $i + 1), '/');
 289                 // move $i on
 290                 $i = $offset;
 291             }
 292         }
 293         $regex .= '(?:$| ))'; // after the tag may only be a space or the end
 294         return $regex;
 295     }
 296
 297     /**
 298      * Returns the list of links associated with a given list of tags
 299      *
 300      * You can specify one or more tags, separated by space or a comma, e.g.
 301      *  print_r($mydb->filterTags('linux programming'));
 302      *
 303      * @param string $tags          list of tags separated by commas or blank spaces.
 304      * @param bool   $casesensitive ignore case if false.
 305      * @param string $visibility    Optional: return only all/private/public links.
 306      *
 307      * @return array filtered links.
 308      */
 309     public function filterTags($tags, $casesensitive = false, $visibility = 'all')
 310     {
 311         // get single tags (we may get passed an array, even though the docs say different)
 312         $inputTags = $tags;
 313         if(!is_array($tags)) {
 314             // we got an input string, split tags
 315             $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY);
 316         }
 317
 318         if(!count($inputTags)){
 319             // no input tags
 320             return $this->noFilter($visibility);
 321         }
 322
 323         // build regex from all tags
 324         $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/';
 325         if(!$casesensitive) {
 326             // make regex case insensitive
 327             $re .= 'i';
 328         }
 329
 330         // create resulting array
 331         $filtered = array();
 332
 333         // iterate over each link
 334         foreach ($this->links as $key => $link) {
 335             // check level of visibility
 336             // ignore non private links when 'privateonly' is on.
 337             if ($visibility !== 'all') {
 338                 if (! $link['private'] && $visibility === 'private') {
 339                     continue;
 340                 } elseif ($link['private'] && $visibility === 'public') {
 341                     continue;
 342                 }
 343             }
 344             $search = $link['tags']; // build search string, start with tags of current link
 345             if(strlen(trim($link['description'])) && strpos($link['description'], '#') !== false){
 346                 // description given and at least one possible tag found
 347                 $descTags = array();
 348                 // find all tags in the form of #tag in the description
 349                 preg_match_all(
 350                     '/(?<![' . self::$HASHTAG_CHARS . '])#([' . self::$HASHTAG_CHARS . ']+?)\b/sm',
 351                     $link['description'],
 352                     $descTags
 353                 );
 354                 if(count($descTags[1])){
 355                     // there were some tags in the description, add them to the search string
 356                     $search .= ' ' . implode(' ', $descTags[1]);
 357                 }
 358             };
 359             // match regular expression with search string
 360             if(!preg_match($re, $search)){
 361                 // this entry does _not_ match our regex
 362                 continue;
 363             }
 364             $filtered[$key] = $link;
 365         }
 366         return $filtered;
 367     }
 368
 369     /**
 370      * Return only links without any tag.
 371      *
 372      * @param string $visibility return only all/private/public links.
 373      *
 374      * @return array filtered links.
 375      */
 376     public function filterUntagged($visibility)
 377     {
 378         $filtered = [];
 379         foreach ($this->links as $key => $link) {
 380             if ($visibility !== 'all') {
 381                 if (! $link['private'] && $visibility === 'private') {
 382                     continue;
 383                 } elseif ($link['private'] && $visibility === 'public') {
 384                     continue;
 385                 }
 386             }
 387
 388             if (empty(trim($link['tags']))) {
 389                 $filtered[$key] = $link;
 390             }
 391         }
 392
 393         return $filtered;
 394     }
 395
 396     /**
 397      * Returns the list of articles for a given day, chronologically sorted
 398      *
 399      * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
 400      *  print_r($mydb->filterDay('20120125'));
 401      *
 402      * @param string $day day to filter.
 403      *
 404      * @return array all link matching given day.
 405      *
 406      * @throws Exception if date format is invalid.
 407      */
 408     public function filterDay($day)
 409     {
 410         if (! checkDateFormat('Ymd', $day)) {
 411             throw new Exception('Invalid date format');
 412         }
 413
 414         $filtered = array();
 415         foreach ($this->links as $key => $l) {
 416             if ($l['created']->format('Ymd') == $day) {
 417                 $filtered[$key] = $l;
 418             }
 419         }
 420
 421         // sort by date ASC
 422         return array_reverse($filtered, true);
 423     }
 424
 425     /**
 426      * Convert a list of tags (str) to an array. Also
 427      * - handle case sensitivity.
 428      * - accepts spaces commas as separator.
 429      *
 430      * @param string $tags          string containing a list of tags.
 431      * @param bool   $casesensitive will convert everything to lowercase if false.
 432      *
 433      * @return array filtered tags string.
 434      */
 435     public static function tagsStrToArray($tags, $casesensitive)
 436     {
 437         // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
 438         $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
 439         $tagsOut = str_replace(',', ' ', $tagsOut);
 440
 441         return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
 442     }
 443 }
 444
 445 class LinkNotFoundException extends Exception
 446 {
 447     /**
 448      * LinkNotFoundException constructor.
 449      */
 450     public function __construct()
 451     {
 452         $this->message =  t('The link you are trying to reach does not exist or has been deleted.');
 453     }
 454 }