src/applications/search/ferret/PhabricatorFerretEngine.php

   1 <?php
   2
   3 abstract class PhabricatorFerretEngine extends Phobject {
   4
   5   private $fieldMap = array();
   6   private $ferretFunctions;
   7   private $templateObject;
   8
   9   abstract public function getApplicationName();
  10   abstract public function getScopeName();
  11   abstract public function newSearchEngine();
  12
  13   public function getDefaultFunctionKey() {
  14     return 'all';
  15   }
  16
  17   public function getObjectTypeRelevance() {
  18     return 1000;
  19   }
  20
  21   final public function getFunctionForName($raw_name) {
  22     if (isset($this->fieldMap[$raw_name])) {
  23       return $this->fieldMap[$raw_name];
  24     }
  25
  26     $normalized_name =
  27       FerretSearchFunction::getNormalizedFunctionName($raw_name);
  28
  29     if ($this->ferretFunctions === null) {
  30       $functions = FerretSearchFunction::newFerretSearchFunctions();
  31       $this->ferretFunctions = $functions;
  32     }
  33
  34     if (!isset($this->ferretFunctions[$normalized_name])) {
  35       throw new PhutilSearchQueryCompilerSyntaxException(
  36         pht(
  37           'Unknown search function "%s". Supported functions are: %s. '.
  38           '(To search for a term containing a colon, surround the term '.
  39           'in double quotes.)',
  40           $raw_name,
  41           implode(', ', array_keys($this->ferretFunctions))));
  42     }
  43
  44     $function = $this->ferretFunctions[$normalized_name];
  45     $this->fieldMap[$raw_name] = $function;
  46
  47     return $this->fieldMap[$raw_name];
  48   }
  49
  50   public function newStemmer() {
  51     return new PhutilSearchStemmer();
  52   }
  53
  54   public function newTermsCorpus($raw_corpus) {
  55     $term_corpus = strtr(
  56       $raw_corpus,
  57       array(
  58         '!' => ' ',
  59         '"' => ' ',
  60         '#' => ' ',
  61         '$' => ' ',
  62         '%' => ' ',
  63         '&' => ' ',
  64         '(' => ' ',
  65         ')' => ' ',
  66         '*' => ' ',
  67         '+' => ' ',
  68         ',' => ' ',
  69         '-' => ' ',
  70         '/' => ' ',
  71         ':' => ' ',
  72         ';' => ' ',
  73         '<' => ' ',
  74         '=' => ' ',
  75         '>' => ' ',
  76         '?' => ' ',
  77         '@' => ' ',
  78         '[' => ' ',
  79         '\\' => ' ',
  80         ']' => ' ',
  81         '^' => ' ',
  82         '`' => ' ',
  83         '{' => ' ',
  84         '|' => ' ',
  85         '}' => ' ',
  86         '~' => ' ',
  87         '.' => ' ',
  88         '_' => ' ',
  89         "\n" => ' ',
  90         "\r" => ' ',
  91         "\t" => ' ',
  92       ));
  93
  94     // NOTE: Single quotes divide terms only if they're at a word boundary.
  95     // In contractions, like "whom'st've", the entire word is a single term.
  96     $term_corpus = preg_replace('/(^| )[\']+/', ' ', $term_corpus);
  97     $term_corpus = preg_replace('/[\']+( |$)/', ' ', $term_corpus);
  98
  99     $term_corpus = preg_replace('/\s+/u', ' ', $term_corpus);
 100     $term_corpus = trim($term_corpus, ' ');
 101
 102     if (strlen($term_corpus)) {
 103       $term_corpus = ' '.$term_corpus.' ';
 104     }
 105
 106     return $term_corpus;
 107   }
 108
 109 /* -(  Schema  )------------------------------------------------------------- */
 110
 111   public function getDocumentTableName() {
 112     $application = $this->getApplicationName();
 113     $scope = $this->getScopeName();
 114
 115     return "{$application}_{$scope}_fdocument";
 116   }
 117
 118   public function getDocumentSchemaColumns() {
 119     return array(
 120       'id' => 'auto',
 121       'objectPHID' => 'phid',
 122       'isClosed' => 'bool',
 123       'authorPHID' => 'phid?',
 124       'ownerPHID' => 'phid?',
 125       'epochCreated' => 'epoch',
 126       'epochModified' => 'epoch',
 127     );
 128   }
 129
 130   public function getDocumentSchemaKeys() {
 131     return array(
 132       'PRIMARY' => array(
 133         'columns' => array('id'),
 134         'unique' => true,
 135       ),
 136       'key_object' => array(
 137         'columns' => array('objectPHID'),
 138         'unique' => true,
 139       ),
 140       'key_author' => array(
 141         'columns' => array('authorPHID'),
 142       ),
 143       'key_owner' => array(
 144         'columns' => array('ownerPHID'),
 145       ),
 146       'key_created' => array(
 147         'columns' => array('epochCreated'),
 148       ),
 149       'key_modified' => array(
 150         'columns' => array('epochModified'),
 151       ),
 152     );
 153   }
 154
 155   public function getFieldTableName() {
 156     $application = $this->getApplicationName();
 157     $scope = $this->getScopeName();
 158
 159     return "{$application}_{$scope}_ffield";
 160   }
 161
 162   public function getFieldSchemaColumns() {
 163     return array(
 164       'id' => 'auto',
 165       'documentID' => 'uint32',
 166       'fieldKey' => 'text4',
 167       'rawCorpus' => 'sort',
 168       'termCorpus' => 'sort',
 169       'normalCorpus' => 'sort',
 170     );
 171   }
 172
 173   public function getFieldSchemaKeys() {
 174     return array(
 175       'PRIMARY' => array(
 176         'columns' => array('id'),
 177         'unique' => true,
 178       ),
 179       'key_documentfield' => array(
 180         'columns' => array('documentID', 'fieldKey'),
 181         'unique' => true,
 182       ),
 183     );
 184   }
 185
 186   public function getNgramsTableName() {
 187     $application = $this->getApplicationName();
 188     $scope = $this->getScopeName();
 189
 190     return "{$application}_{$scope}_fngrams";
 191   }
 192
 193   public function getNgramsSchemaColumns() {
 194     return array(
 195       'id' => 'auto',
 196       'documentID' => 'uint32',
 197       'ngram' => 'char3',
 198     );
 199   }
 200
 201   public function getNgramsSchemaKeys() {
 202     return array(
 203       'PRIMARY' => array(
 204         'columns' => array('id'),
 205         'unique' => true,
 206       ),
 207       'key_ngram' => array(
 208         'columns' => array('ngram', 'documentID'),
 209       ),
 210       'key_object' => array(
 211         'columns' => array('documentID'),
 212       ),
 213     );
 214   }
 215
 216   public function getCommonNgramsTableName() {
 217     $application = $this->getApplicationName();
 218     $scope = $this->getScopeName();
 219
 220     return "{$application}_{$scope}_fngrams_common";
 221   }
 222
 223   public function getCommonNgramsSchemaColumns() {
 224     return array(
 225       'id' => 'auto',
 226       'ngram' => 'char3',
 227       'needsCollection' => 'bool',
 228     );
 229   }
 230
 231   public function getCommonNgramsSchemaKeys() {
 232     return array(
 233       'PRIMARY' => array(
 234         'columns' => array('id'),
 235         'unique' => true,
 236       ),
 237       'key_ngram' => array(
 238         'columns' => array('ngram'),
 239         'unique' => true,
 240       ),
 241       'key_collect' => array(
 242         'columns' => array('needsCollection'),
 243       ),
 244     );
 245   }
 246
 247 }