switch postcode tokens to new word table layout

[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 09cfe70fbf661a3e0440531310a45fb1fabbfab7..2461a1fd828698bb6786e16778b62c1b8b6077f2 100644 (file)
--- a/lib-php/tokenizer/legacy_icu_tokenizer.php
+++ b/lib-php/tokenizer/legacy_icu_tokenizer.php
@@ -47,9 +47,7 @@ class Tokenizer
 
     private function makeStandardWord($sTerm)
     {
 
     private function makeStandardWord($sTerm)
     {

-        $sNorm = ' '.$this->oTransliterator->transliterate($sTerm).' ';
-
-        return trim(str_replace(CONST_Abbreviations[0], CONST_Abbreviations[1], $sNorm));
+        return trim($this->oTransliterator->transliterate(' '.$sTerm.' '));

     }
 
 
     }
 
 


@@ -90,6 +88,7 @@ class Tokenizer
         foreach ($aPhrases as $iPhrase => $oPhrase) {
             $sNormQuery .= ','.$this->normalizeString($oPhrase->getPhrase());
             $sPhrase = $this->makeStandardWord($oPhrase->getPhrase());
         foreach ($aPhrases as $iPhrase => $oPhrase) {
             $sNormQuery .= ','.$this->normalizeString($oPhrase->getPhrase());
             $sPhrase = $this->makeStandardWord($oPhrase->getPhrase());

+            Debug::printVar('Phrase', $sPhrase);

             if (strlen($sPhrase) > 0) {
                 $aWords = explode(' ', $sPhrase);
                 Tokenizer::addTokens($aTokens, $aWords);
             if (strlen($sPhrase) > 0) {
                 $aWords = explode(' ', $sPhrase);
                 Tokenizer::addTokens($aTokens, $aWords);


@@ -121,14 +120,14 @@ class Tokenizer
 
             // Try more interpretations for Tokens that could not be matched.
             foreach ($aTokens as $sToken) {
 
             // Try more interpretations for Tokens that could not be matched.
             foreach ($aTokens as $sToken) {

-                if ($sToken[0] == ' ' && !$oValidTokens->contains($sToken)) {
-                    if (preg_match('/^ ([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {
+                if ($sToken[0] != ' ' && !$oValidTokens->contains($sToken)) {
+                    if (preg_match('/^([0-9]{5}) [0-9]{4}$/', $sToken, $aData)) {

                         // US ZIP+4 codes - merge in the 5-digit ZIP code
                         $oValidTokens->addToken(
                             $sToken,
                             new Token\Postcode(null, $aData[1], 'us')
                         );
                         // US ZIP+4 codes - merge in the 5-digit ZIP code
                         $oValidTokens->addToken(
                             $sToken,
                             new Token\Postcode(null, $aData[1], 'us')
                         );

-                    } elseif (preg_match('/^ [0-9]+$/', $sToken)) {
+                    } elseif (preg_match('/^[0-9]+$/', $sToken)) {

                         // Unknown single word token with a number.
                         // Assume it is a house number.
                         $oValidTokens->addToken(
                         // Unknown single word token with a number.
                         // Assume it is a house number.
                         $oValidTokens->addToken(


@@ -147,8 +146,8 @@ class Tokenizer
     private function addTokensFromDB(&$oValidTokens, $aTokens, $sNormQuery)
     {
         // Check which tokens we have, get the ID numbers
     private function addTokensFromDB(&$oValidTokens, $aTokens, $sNormQuery)
     {
         // Check which tokens we have, get the ID numbers

-        $sSQL = 'SELECT word_id, word_token, word, class, type, country_code,';
-        $sSQL .= ' operator, coalesce(search_name_count, 0) as count';
+        $sSQL = 'SELECT word_id, word_token, type';
+        $sSQL .= "      info->>'cc' as country, info->>'postcode' as postcode";

         $sSQL .= ' FROM word WHERE word_token in (';
         $sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
 
         $sSQL .= ' FROM word WHERE word_token in (';
         $sSQL .= join(',', $this->oDB->getDBQuotedList($aTokens)).')';
 


@@ -157,10 +156,35 @@ class Tokenizer
         $aDBWords = $this->oDB->getAll($sSQL, null, 'Could not get word tokens.');
 
         foreach ($aDBWords as $aWord) {
         $aDBWords = $this->oDB->getAll($sSQL, null, 'Could not get word tokens.');
 
         foreach ($aDBWords as $aWord) {

-            $oToken = null;

             $iId = (int) $aWord['word_id'];
 
             $iId = (int) $aWord['word_id'];
 

-            if ($aWord['class']) {
+            switch ($aWord['type']) {
+                'C':  // country name tokens
+                    if ($aWord['country'] === null
+                        || ($this->aCountryRestriction
+                            && !in_array($aWord['country'], $this->aCountryRestriction))
+                    ) {
+                        continue;
+                    }
+                    $oToken = new Token\Country($iId, $aWord['country'])
+                    break;
+                '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']
+                    ) {
+                       continue;
+                    }
+                    $oToken = new Token\Postcode($iId, $aWord['postcode'], null);
+                default:
+                    continue;
+            }
+/*          if ($aWord['class']) {

                 // Special terms need to appear in their normalized form.
                 // (postcodes are not normalized in the word table)
                 $sNormWord = $this->normalizeString($aWord['word']);
                 // Special terms need to appear in their normalized form.
                 // (postcodes are not normalized in the word table)
                 $sNormWord = $this->normalizeString($aWord['word']);


@@ -196,18 +220,21 @@ class Tokenizer
                 ) {
                     $oToken = new Token\Country($iId, $aWord['country_code']);
                 }
                 ) {
                     $oToken = new Token\Country($iId, $aWord['country_code']);
                 }

+            } elseif ($aWord['word_token'][0] == ' ') {
+                 $oToken = new Token\Word(
+                     $iId,
+                     (int) $aWord['count'],
+                     substr_count($aWord['word_token'], ' ')
+                 );

             } else {
             } else {

-                $oToken = new Token\Word(
+                $oToken = new Token\Partial(

                     $iId,
                     $iId,

-                    $aWord['word_token'][0] != ' ',
-                    (int) $aWord['count'],
-                    substr_count($aWord['word_token'], ' ')
+                    $aWord['word_token'],
+                    (int) $aWord['count']

                 );
                 );

-            }
+            }*/

 
 

-            if ($oToken) {
-                $oValidTokens->addToken($aWord['word_token'], $oToken);
-            }
+            $oValidTokens->addToken($aWord['word_token'], $oToken);

         }
     }
 
         }
     }
 


@@ -225,12 +252,10 @@ class Tokenizer
 
         for ($i = 0; $i < $iNumWords; $i++) {
             $sPhrase = $aWords[$i];
 
         for ($i = 0; $i < $iNumWords; $i++) {
             $sPhrase = $aWords[$i];

-            $aTokens[' '.$sPhrase] = ' '.$sPhrase;

             $aTokens[$sPhrase] = $sPhrase;
 
             for ($j = $i + 1; $j < $iNumWords; $j++) {
                 $sPhrase .= ' '.$aWords[$j];
             $aTokens[$sPhrase] = $sPhrase;
 
             for ($j = $i + 1; $j < $iNumWords; $j++) {
                 $sPhrase .= ' '.$aWords[$j];

-                $aTokens[' '.$sPhrase] = ' '.$sPhrase;

                 $aTokens[$sPhrase] = $sPhrase;
             }
         }
                 $aTokens[$sPhrase] = $sPhrase;
             }
         }