adapt special terms lookup to new word table

[nominatim.git] / lib-php / tokenizer / legacy_icu_tokenizer.php

diff --git a/lib-php/tokenizer/legacy_icu_tokenizer.php b/lib-php/tokenizer/legacy_icu_tokenizer.php
index ea445f23039fc0b03a8197dd4db3edef90c1a47c..796635eeb7dbebe788e727b59994a9d990aa9dea 100644 (file)
--- a/lib-php/tokenizer/legacy_icu_tokenizer.php
+++ b/lib-php/tokenizer/legacy_icu_tokenizer.php
@@ -19,7 +19,7 @@ class Tokenizer
 
     public function checkStatus()
     {
-        $sSQL = "SELECT word_id FROM word WHERE word_token IN (' a')";
+        $sSQL = "SELECT word_id FROM word WHERE word_token == 'a'";
         $iWordID = $this->oDB->getOne($sSQL);
         if ($iWordID === false) {
             throw new Exception('Query failed', 703);
@@ -55,9 +55,8 @@ class Tokenizer
     {
         $aResults = array();
 
-        $sSQL = 'SELECT word_id, class, type FROM word ';
-        $sSQL .= '   WHERE word_token = \' \' || :term';
-        $sSQL .= '   AND class is not null AND class not in (\'place\')';
+        $sSQL = "SELECT word_id, info->>'class' as class, info->>'type' as type ";
+        $sSQL .= '   FROM word WHERE word_token = :term and type = \'S\'';
 
         Debug::printVar('Term', $sTerm);
         Debug::printSQL($sSQL);
@@ -147,7 +146,10 @@ class Tokenizer
     {
         // Check which tokens we have, get the ID numbers
         $sSQL = 'SELECT word_id, word_token, type';
-        $sSQL .= "      info->>'cc' as country";
+        $sSQL .= "      info->>'cc' as country, info->>'postcode' as postcode,";
+        $sSQL .= "      info->>'op' as operator,";
+        $sSQL .= "      info->>'class' as class, info->>'type' as type,";
+        $sSQL .= "      info->>'count' as count";
         $sSQL .= ' FROM word WHERE word_token in (';
         $sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
 
@@ -156,6 +158,8 @@ class Tokenizer
         $aDBWords = $this->oDB->getAll($sSQL, null, 'Could not get word tokens.');
 
         foreach ($aDBWords as $aWord) {
+            $iId = (int) $aWord['word_id'];
+
             switch ($aWord['type']) {
                 'C':  // country name tokens
                     if ($aWord['country'] === null
@@ -166,60 +170,53 @@ class Tokenizer
                     }
                     $oToken = new Token\Country($iId, $aWord['country'])
                     break;
-                default:
-                    continue;
-            }
-/*            $iId = (int) $aWord['word_id'];
-
-            if ($aWord['class']) {
-                // Special terms need to appear in their normalized form.
-                // (postcodes are not normalized in the word table)
-                $sNormWord = $this->normalizeString($aWord['word']);
-                if ($aWord['word'] && strpos($sNormQuery, $sNormWord) === false) {
-                    continue;
-                }
-
-                if ($aWord['class'] == 'place' && $aWord['type'] == 'house') {
-                    $oToken = new Token\HouseNumber($iId, trim($aWord['word_token']));
-                } elseif ($aWord['class'] == 'place' && $aWord['type'] == 'postcode') {
-                    if ($aWord['word']
-                        && pg_escape_string($aWord['word']) == $aWord['word']
+                'H':  // house number tokens
+                    $oToken = new Token\HouseNumber($iId, $aWord['word_token']);
+                    break;
+                'P':  // postcode tokens
+                    // Postcodes are not normalized, so they may have content
+                    // that makes SQL injection possible. Reject postcodes
+                    // that would need special escaping.
+                    if ($aWord['postcode'] === null
+                        || pg_escape_string($aWord['postcode']) == $aWord['postcode']
                     ) {
-                        $oToken = new Token\Postcode(
-                            $iId,
-                            $aWord['word'],
-                            $aWord['country_code']
-                        );
+                       continue;
+                    }
+                    $sNormPostcode = $this->normalizeString($aWord['postcode']);
+                    if (strpos($sNormQuery, $sNormPostcode) === false) {
+                        continue;
+                    }
+                    $oToken = new Token\Postcode($iId, $aWord['postcode'], null);
+                    break;
+                'S':  // tokens for classification terms (special phrases)
+                    if ($aWord['class'] === null || $aWord['type'] === null
+                    ) {
+                        continue;
                     }
-                } else {
-                    // near and in operator the same at the moment
                     $oToken = new Token\SpecialTerm(
                         $iId,
                         $aWord['class'],
                         $aWord['type'],
-                        $aWord['operator'] ? Operator::NEAR : Operator::NONE
+                        $aWord['op'] ? Operator::NEAR : Operator::NONE
                     );
-                }
-            } elseif ($aWord['country_code']) {
-                // Filter country tokens that do not match restricted countries.
-                if (!$this->aCountryRestriction
-                    || in_array($aWord['country_code'], $this->aCountryRestriction)
-                ) {
-                    $oToken = new Token\Country($iId, $aWord['country_code']);
-                }
-            } elseif ($aWord['word_token'][0] == ' ') {
-                 $oToken = new Token\Word(
-                     $iId,
-                     (int) $aWord['count'],
-                     substr_count($aWord['word_token'], ' ')
-                 );
-            } else {
-                $oToken = new Token\Partial(
-                    $iId,
-                    $aWord['word_token'],
-                    (int) $aWord['count']
-                );
-            }*/
+                    break;
+                'W': // full-word tokens
+                    $oToken = new Token\Word(
+                        $iId,
+                        (int) $aWord['count'],
+                        substr_count($aWord['word_token'], ' ')
+                    );
+                    break;
+                'w':  // partial word terms
+                    $oToken = new Token\Partial(
+                        $iId,
+                        $aWord['word_token'],
+                        (int) $aWord['count']
+                    );
+                    break;
+                default:
+                    continue;
+            }
 
             $oValidTokens->addToken($aWord['word_token'], $oToken);
         }