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 => $link) {
 208             // ignore non private links when 'privatonly' is on.
 209             if ($visibility !== 'all') {
 210                 if (! $link['private'] && $visibility === 'private') {
 211                     continue;
 212                 } elseif ($link['private'] && $visibility === 'public') {
 213                     continue;
 214                 }
 215             }
 216
 217             // Concatenate link fields to search across fields.
 218             // Adds a '\' separator for exact search terms.
 219             $content = '';
 220             foreach ($keys as $key) {
 221                 $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\';
 222             }
 223
 224             // Be optimistic
 225             $found = true;
 226
 227             // First, we look for exact term search
 228             for ($i = 0; $i < count($exactSearch) && $found; $i++) {
 229                 $found = strpos($content, $exactSearch[$i]) !== false;
 230             }
 231
 232             // Iterate over keywords, if keyword is not found,
 233             // no need to check for the others. We want all or nothing.
 234             for ($i = 0; $i < count($andSearch) && $found; $i++) {
 235                 $found = strpos($content, $andSearch[$i]) !== false;
 236             }
 237
 238             // Exclude terms.
 239             for ($i = 0; $i < count($excludeSearch) && $found; $i++) {
 240                 $found = strpos($content, $excludeSearch[$i]) === false;
 241             }
 242
 243             if ($found) {
 244                 $filtered[$id] = $link;
 245             }
 246         }
 247
 248         return $filtered;
 249     }
 250
 251     /**
 252      * generate a regex fragment out of a tag
 253      * @param string $tag to to generate regexs from. may start with '-' to negate, contain '*' as wildcard
 254      * @return string generated regex fragment
 255      */
 256     private static function tag2regex($tag)
 257     {
 258         $len = strlen($tag);
 259         if (!$len || $tag === "-" || $tag === "*") {
 260             // nothing to search, return empty regex
 261             return '';
 262         }
 263         if ($tag[0] === "-") {
 264             // query is negated
 265             $i = 1; // use offset to start after '-' character
 266             $regex = '(?!'; // create negative lookahead
 267         } else {
 268             $i = 0; // start at first character
 269             $regex = '(?='; // use positive lookahead
 270         }
 271         $regex .= '.*(?:^| )'; // before tag may only be a space or the beginning
 272         // iterate over string, separating it into placeholder and content
 273         for (; $i < $len; $i++) {
 274             if ($tag[$i] === '*') {
 275                 // placeholder found
 276                 $regex .= '[^ ]*?';
 277             } else {
 278                 // regular characters
 279                 $offset = strpos($tag, '*', $i);
 280                 if ($offset === false) {
 281                     // no placeholder found, set offset to end of string
 282                     $offset = $len;
 283                 }
 284                 // subtract one, as we want to get before the placeholder or end of string
 285                 $offset -= 1;
 286                 // we got a tag name that we want to search for. escape any regex characters to prevent conflicts.
 287                 $regex .= preg_quote(substr($tag, $i, $offset - $i + 1), '/');
 288                 // move $i on
 289                 $i = $offset;
 290             }
 291         }
 292         $regex .= '(?:$| ))'; // after the tag may only be a space or the end
 293         return $regex;
 294     }
 295
 296     /**
 297      * Returns the list of links associated with a given list of tags
 298      *
 299      * You can specify one or more tags, separated by space or a comma, e.g.
 300      *  print_r($mydb->filterTags('linux programming'));
 301      *
 302      * @param string $tags          list of tags separated by commas or blank spaces.
 303      * @param bool   $casesensitive ignore case if false.
 304      * @param string $visibility    Optional: return only all/private/public links.
 305      *
 306      * @return array filtered links.
 307      */
 308     public function filterTags($tags, $casesensitive = false, $visibility = 'all')
 309     {
 310         // get single tags (we may get passed an array, even though the docs say different)
 311         $inputTags = $tags;
 312         if (!is_array($tags)) {
 313             // we got an input string, split tags
 314             $inputTags = preg_split('/(?:\s+)|,/', $inputTags, -1, PREG_SPLIT_NO_EMPTY);
 315         }
 316
 317         if (!count($inputTags)) {
 318             // no input tags
 319             return $this->noFilter($visibility);
 320         }
 321
 322         // build regex from all tags
 323         $re = '/^' . implode(array_map("self::tag2regex", $inputTags)) . '.*$/';
 324         if (!$casesensitive) {
 325             // make regex case insensitive
 326             $re .= 'i';
 327         }
 328
 329         // create resulting array
 330         $filtered = array();
 331
 332         // iterate over each link
 333         foreach ($this->links as $key => $link) {
 334             // check level of visibility
 335             // ignore non private links when 'privateonly' is on.
 336             if ($visibility !== 'all') {
 337                 if (! $link['private'] && $visibility === 'private') {
 338                     continue;
 339                 } elseif ($link['private'] && $visibility === 'public') {
 340                     continue;
 341                 }
 342             }
 343             $search = $link['tags']; // build search string, start with tags of current link
 344             if (strlen(trim($link['description'])) && strpos($link['description'], '#') !== false) {
 345                 // description given and at least one possible tag found
 346                 $descTags = array();
 347                 // find all tags in the form of #tag in the description
 348                 preg_match_all(
 349                     '/(?<![' . self::$HASHTAG_CHARS . '])#([' . self::$HASHTAG_CHARS . ']+?)\b/sm',
 350                     $link['description'],
 351                     $descTags
 352                 );
 353                 if (count($descTags[1])) {
 354                     // there were some tags in the description, add them to the search string
 355                     $search .= ' ' . implode(' ', $descTags[1]);
 356                 }
 357             };
 358             // match regular expression with search string
 359             if (!preg_match($re, $search)) {
 360                 // this entry does _not_ match our regex
 361                 continue;
 362             }
 363             $filtered[$key] = $link;
 364         }
 365         return $filtered;
 366     }
 367
 368     /**
 369      * Return only links without any tag.
 370      *
 371      * @param string $visibility return only all/private/public links.
 372      *
 373      * @return array filtered links.
 374      */
 375     public function filterUntagged($visibility)
 376     {
 377         $filtered = [];
 378         foreach ($this->links as $key => $link) {
 379             if ($visibility !== 'all') {
 380                 if (! $link['private'] && $visibility === 'private') {
 381                     continue;
 382                 } elseif ($link['private'] && $visibility === 'public') {
 383                     continue;
 384                 }
 385             }
 386
 387             if (empty(trim($link['tags']))) {
 388                 $filtered[$key] = $link;
 389             }
 390         }
 391
 392         return $filtered;
 393     }
 394
 395     /**
 396      * Returns the list of articles for a given day, chronologically sorted
 397      *
 398      * Day must be in the form 'YYYYMMDD' (e.g. '20120125'), e.g.
 399      *  print_r($mydb->filterDay('20120125'));
 400      *
 401      * @param string $day day to filter.
 402      *
 403      * @return array all link matching given day.
 404      *
 405      * @throws Exception if date format is invalid.
 406      */
 407     public function filterDay($day)
 408     {
 409         if (! checkDateFormat('Ymd', $day)) {
 410             throw new Exception('Invalid date format');
 411         }
 412
 413         $filtered = array();
 414         foreach ($this->links as $key => $l) {
 415             if ($l['created']->format('Ymd') == $day) {
 416                 $filtered[$key] = $l;
 417             }
 418         }
 419
 420         // sort by date ASC
 421         return array_reverse($filtered, true);
 422     }
 423
 424     /**
 425      * Convert a list of tags (str) to an array. Also
 426      * - handle case sensitivity.
 427      * - accepts spaces commas as separator.
 428      *
 429      * @param string $tags          string containing a list of tags.
 430      * @param bool   $casesensitive will convert everything to lowercase if false.
 431      *
 432      * @return array filtered tags string.
 433      */
 434     public static function tagsStrToArray($tags, $casesensitive)
 435     {
 436         // We use UTF-8 conversion to handle various graphemes (i.e. cyrillic, or greek)
 437         $tagsOut = $casesensitive ? $tags : mb_convert_case($tags, MB_CASE_LOWER, 'UTF-8');
 438         $tagsOut = str_replace(',', ' ', $tagsOut);
 439
 440         return preg_split('/\s+/', $tagsOut, -1, PREG_SPLIT_NO_EMPTY);
 441     }
 442 }
 443
 444 class LinkNotFoundException extends Exception
 445 {
 446     /**
 447      * LinkNotFoundException constructor.
 448      */
 449     public function __construct()
 450     {
 451         $this->message =  t('The link you are trying to reach does not exist or has been deleted.');
 452     }
 453 }