move word recheck into token collection

[nominatim.git] / lib / SearchDescription.php

diff --git a/lib/SearchDescription.php b/lib/SearchDescription.php
index 39f7ffca90910c9b2b416d952156982ef90fd9ec..1325a46335396401e975e72232440b3cf41d7be9 100644 (file)
--- a/lib/SearchDescription.php
+++ b/lib/SearchDescription.php
@@ -43,27 +43,55 @@ class SearchDescription
     private $iNamePhrase = -1;
 
 
     private $iNamePhrase = -1;
 
 

+    /**
+     * Create an empty search description.
+     *
+     * @param object $oContext Global context to use. Will be inherited by
+     *                         all derived search objects.
+     */

     public function __construct($oContext)
     {
         $this->oContext = $oContext;
     }
 
     public function __construct($oContext)
     {
         $this->oContext = $oContext;
     }
 

+    /**
+     * Get current search rank.
+     *
+     * The higher the search rank the lower the likelyhood that the
+     * search is a correct interpretation of the search query.
+     *
+     * @return integer Search rank.
+     */

     public function getRank()
     {
         return $this->iSearchRank;
     }
 
     public function getRank()
     {
         return $this->iSearchRank;
     }
 

+    /**
+     * Increase the search rank.
+     *
+     * @param integer $iAddRank Number of ranks to increase.
+     *
+     * @return void
+     */

     public function addToRank($iAddRank)
     {
         $this->iSearchRank += $iAddRank;
         return $this->iSearchRank;
     }
 
     public function addToRank($iAddRank)
     {
         $this->iSearchRank += $iAddRank;
         return $this->iSearchRank;
     }
 

-    public function getPostCode()
-    {
-        return $this->sPostcode;
-    }
-
+    /**
+     * Make this search a POI search.
+     *
+     * In a POI search, objects are not (only) searched by their name
+     * but also by the primary OSM key/value pair (class and type in Nominatim).
+     *
+     * @param integer $iOperator Type of POI search
+     * @param string  $sClass    Class (or OSM tag key) of POI.
+     * @param string  $sType     Type (or OSM tag value) of POI.
+     *
+     * @return void
+     */

     public function setPoiSearch($iOperator, $sClass, $sType)
     {
         $this->iOperator = $iOperator;
     public function setPoiSearch($iOperator, $sClass, $sType)
     {
         $this->iOperator = $iOperator;


@@ -71,22 +99,11 @@ class SearchDescription
         $this->sType = $sType;
     }
 
         $this->sType = $sType;
     }
 

-    public function isNamedSearch()
-    {
-        return sizeof($this->aName) > 0 || sizeof($this->aAddress) > 0;
-    }
-
-    public function isCountrySearch()
-    {
-        return $this->sCountryCode && sizeof($this->aName) == 0
-               && !$this->iOperator && !$this->oContext->hasNearPoint();
-    }
-
-    public function isPoiSearch()
-    {
-        return (bool) $this->sClass;
-    }
-
+    /**
+     * Check if this might be a full address search.
+     *
+     * @return bool True if the search contains name, address and housenumber.
+     */

     public function looksLikeFullAddress()
     {
         return sizeof($this->aName)
     public function looksLikeFullAddress()
     {
         return sizeof($this->aName)


@@ -94,38 +111,27 @@ class SearchDescription
                && preg_match('/[0-9]+/', $this->sHouseNumber);
     }
 
                && preg_match('/[0-9]+/', $this->sHouseNumber);
     }
 

-    public function isOperator($iType)
-    {
-        return $this->iOperator == $iType;
-    }
-
-    public function hasHouseNumber()
-    {
-        return (bool) $this->sHouseNumber;
-    }
-
-    private function poiTable()
-    {
-        return 'place_classtype_'.$this->sClass.'_'.$this->sType;
-    }
-
-    public function countryCodeSQL($sVar)
-    {
-        if ($this->sCountryCode) {
-            return $sVar.' = \''.$this->sCountryCode."'";
-        }
-        if ($this->oContext->sqlCountryList) {
-            return $sVar.' in '.$this->oContext->sqlCountryList;
-        }
-
-        return '';
-    }
-
+    /**
+     * Check if any operator is set.
+     *
+     * @return bool True, if this is a special search operation.
+     */

     public function hasOperator()
     {
         return $this->iOperator != Operator::NONE;
     }
 
     public function hasOperator()
     {
         return $this->iOperator != Operator::NONE;
     }
 

+    /**
+     * Extract key/value pairs from a query.
+     *
+     * Key/value pairs are recognised if they are of the form [<key>=<value>].
+     * If multiple terms of this kind are found then all terms are removed
+     * but only the first is used for search.
+     *
+     * @param string $sQuery Original query string.
+     *
+     * @return string The query string with the special search patterns removed.
+     */

     public function extractKeyValuePairs($sQuery)
     {
         // Search for terms of kind [<key>=<value>].
     public function extractKeyValuePairs($sQuery)
     {
         // Search for terms of kind [<key>=<value>].


@@ -146,18 +152,20 @@ class SearchDescription
         return $sQuery;
     }
 
         return $sQuery;
     }
 

-    public function isValidSearch(&$aCountryCodes)
+    /**
+     * Check if the combination of parameters is sensible.
+     *
+     * @return bool True, if the search looks valid.
+     */
+    public function isValidSearch()

     {
         if (!sizeof($this->aName)) {
             if ($this->sHouseNumber) {
                 return false;
             }
     {
         if (!sizeof($this->aName)) {
             if ($this->sHouseNumber) {
                 return false;
             }

-        }
-        if ($aCountryCodes
-            && $this->sCountryCode
-            && !in_array($this->sCountryCode, $aCountryCodes)
-        ) {
-            return false;
+            if (!$this->sClass && !$this->sCountryCode) {
+                return false;
+            }

         }
 
         return true;
         }
 
         return true;


@@ -166,7 +174,24 @@ class SearchDescription
     /////////// Search building functions
 
 
     /////////// Search building functions
 
 

-    public function extendWithFullTerm($aSearchTerm, $bWordInQuery, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank)
+    /**
+     * Derive new searches by adding a full term to the existing search.
+     *
+     * @param mixed[] $aSearchTerm  Description of the token.
+     * @param bool    $bHasPartial  True if there are also tokens of partial terms
+     *                              with the same name.
+     * @param string  $sPhraseType  Type of phrase the token is contained in.
+     * @param bool    $bFirstToken  True if the token is at the beginning of the
+     *                              query.
+     * @param bool    $bFirstPhrase True if the token is in the first phrase of
+     *                              the query.
+     * @param bool    $bLastToken   True if the token is at the end of the query.
+     * @param integer $iGlobalRank  Changable ranking of all searches in the
+     *                              batch.
+     *
+     * @return SearchDescription[] List of derived search descriptions.
+     */
+    public function extendWithFullTerm($aSearchTerm, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank)

     {
         $aNewSearches = array();
 
     {
         $aNewSearches = array();
 


@@ -197,7 +222,8 @@ class SearchDescription
             // We need to try the case where the postal code is the primary element
             // (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode)
             // so try both.
             // We need to try the case where the postal code is the primary element
             // (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode)
             // so try both.

-            if (!$this->sPostcode && $bWordInQuery
+            if (!$this->sPostcode
+                && $aSearchTerm['word']

                 && pg_escape_string($aSearchTerm['word']) == $aSearchTerm['word']
             ) {
                 // If we have structured search or this is the first term,
                 && pg_escape_string($aSearchTerm['word']) == $aSearchTerm['word']
             ) {
                 // If we have structured search or this is the first term,


@@ -246,16 +272,8 @@ class SearchDescription
                 }
                 $aNewSearches[] = $oSearch;
             }
                 }
                 $aNewSearches[] = $oSearch;
             }

-        } elseif ($sPhraseType == ''
-                  && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null
-        ) {
-            // require a normalized exact match of the term
-            // if we have the normalizer version of the query
-            // available
-            if ($this->iOperator == Operator::NONE
-                && (isset($aSearchTerm['word']) && $aSearchTerm['word'])
-                && $bWordInQuery
-            ) {
+        } elseif ($sPhraseType == '' && $aSearchTerm['class']) {
+            if ($this->iOperator == Operator::NONE) {

                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
 
                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
 


@@ -270,7 +288,10 @@ class SearchDescription
                 $oSearch->setPoiSearch($iOp, $aSearchTerm['class'], $aSearchTerm['type']);
                 $aNewSearches[] = $oSearch;
             }
                 $oSearch->setPoiSearch($iOp, $aSearchTerm['class'], $aSearchTerm['type']);
                 $aNewSearches[] = $oSearch;
             }

-        } elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
+        } elseif (isset($aSearchTerm['word_id'])
+                  && $aSearchTerm['word_id']
+                  && $sPhraseType != 'country'
+        ) {

             $iWordID = $aSearchTerm['word_id'];
             if (sizeof($this->aName)) {
                 if (($sPhraseType == '' || !$bFirstPhrase)
             $iWordID = $aSearchTerm['word_id'];
             if (sizeof($this->aName)) {
                 if (($sPhraseType == '' || !$bFirstPhrase)


@@ -295,7 +316,18 @@ class SearchDescription
         return $aNewSearches;
     }
 
         return $aNewSearches;
     }
 

-    public function extendWithPartialTerm($aSearchTerm, $bStructuredPhrases, $iPhrase, &$aWordFrequencyScores, $aFullTokens)
+    /**
+     * Derive new searches by adding a partial term to the existing search.
+     *
+     * @param mixed[] $aSearchTerm          Description of the token.
+     * @param bool    $bStructuredPhrases   True if the search is structured.
+     * @param integer $iPhrase              Number of the phrase the token is in.
+     * @param array[] $aFullTokens          List of full term tokens with the
+     *                                      same name.
+     *
+     * @return SearchDescription[] List of derived search descriptions.
+     */
+    public function extendWithPartialTerm($aSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)

     {
         // Only allow name terms.
         if (!(isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])) {
     {
         // Only allow name terms.
         if (!(isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])) {


@@ -309,7 +341,7 @@ class SearchDescription
             && sizeof($this->aName)
             && strpos($aSearchTerm['word_token'], ' ') === false
         ) {
             && sizeof($this->aName)
             && strpos($aSearchTerm['word_token'], ' ') === false
         ) {

-            if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
+            if ($aSearchTerm['search_name_count'] + 1 < CONST_Max_Word_Frequency) {

                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
                 $oSearch->aAddress[$iWordID] = $iWordID;
                 $oSearch = clone $this;
                 $oSearch->iSearchRank++;
                 $oSearch->aAddress[$iWordID] = $iWordID;


@@ -352,7 +384,7 @@ class SearchDescription
             if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
                 $oSearch->iSearchRank += 2;
             }
             if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
                 $oSearch->iSearchRank += 2;
             }

-            if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
+            if ($aSearchTerm['search_name_count'] + 1 < CONST_Max_Word_Frequency) {

                 $oSearch->aName[$iWordID] = $iWordID;
             } else {
                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;
                 $oSearch->aName[$iWordID] = $iWordID;
             } else {
                 $oSearch->aNameNonSearch[$iWordID] = $iWordID;


@@ -367,7 +399,108 @@ class SearchDescription
     /////////// Query functions
 
 
     /////////// Query functions
 
 

-    public function queryCountry(&$oDB)
+    /**
+     * Query database for places that match this search.
+     *
+     * @param object  $oDB                  Database connection to use.
+     * @param mixed[] $aWordFrequencyScores Number of times tokens appears
+     *                                      overall in a planet database.
+     * @param mixed[] $aExactMatchCache     Saves number of exact matches.
+     * @param integer $iMinRank             Minimum address rank to restrict
+     *                                      search to.
+     * @param integer $iMaxRank             Maximum address rank to restrict
+     *                                      search to.
+     * @param integer $iLimit               Maximum number of results.
+     *
+     * @return mixed[] An array with two fields: IDs contains the list of
+     *                 matching place IDs and houseNumber the houseNumber
+     *                 if appicable or -1 if not.
+     */
+    public function query(&$oDB, &$aWordFrequencyScores, &$aExactMatchCache, $iMinRank, $iMaxRank, $iLimit)
+    {
+        $aPlaceIDs = array();
+        $iHousenumber = -1;
+
+        if ($this->sCountryCode
+            && !sizeof($this->aName)
+            && !$this->iOperator
+            && !$this->sClass
+            && !$this->oContext->hasNearPoint()
+        ) {
+            // Just looking for a country - look it up
+            if (4 >= $iMinRank && 4 <= $iMaxRank) {
+                $aPlaceIDs = $this->queryCountry($oDB);
+            }
+        } elseif (!sizeof($this->aName) && !sizeof($this->aAddress)) {
+            // Neither name nor address? Then we must be
+            // looking for a POI in a geographic area.
+            if ($this->oContext->isBoundedSearch()) {
+                $aPlaceIDs = $this->queryNearbyPoi($oDB, $iLimit);
+            }
+        } elseif ($this->iOperator == Operator::POSTCODE) {
+            // looking for postcode
+            $aPlaceIDs = $this->queryPostcode($oDB, $iLimit);
+        } else {
+            // Ordinary search:
+            // First search for places according to name and address.
+            $aNamedPlaceIDs = $this->queryNamedPlace(
+                $oDB,
+                $aWordFrequencyScores,
+                $iMinRank,
+                $iMaxRank,
+                $iLimit
+            );
+
+            if (sizeof($aNamedPlaceIDs)) {
+                foreach ($aNamedPlaceIDs as $aRow) {
+                    $aPlaceIDs[] = $aRow['place_id'];
+                    $aExactMatchCache[$aRow['place_id']] = $aRow['exactmatch'];
+                }
+            }
+
+            //now search for housenumber, if housenumber provided
+            if ($this->sHouseNumber && sizeof($aPlaceIDs)) {
+                $aResult = $this->queryHouseNumber($oDB, $aPlaceIDs, $iLimit);
+
+                if (sizeof($aResult)) {
+                    $iHousenumber = $aResult['iHouseNumber'];
+                    $aPlaceIDs = $aResult['aPlaceIDs'];
+                } elseif (!$this->looksLikeFullAddress()) {
+                    $aPlaceIDs = array();
+                }
+            }
+
+            // finally get POIs if requested
+            if ($this->sClass && sizeof($aPlaceIDs)) {
+                $aPlaceIDs = $this->queryPoiByOperator($oDB, $aPlaceIDs, $iLimit);
+            }
+        }
+
+        if (CONST_Debug) {
+            echo "<br><b>Place IDs:</b> ";
+            var_Dump($aPlaceIDs);
+        }
+
+        if (sizeof($aPlaceIDs) && $this->sPostcode) {
+            $sSQL = 'SELECT place_id FROM placex';
+            $sSQL .= ' WHERE place_id in ('.join(',', $aPlaceIDs).')';
+            $sSQL .= " AND postcode = '".$this->sPostcode."'";
+            if (CONST_Debug) var_dump($sSQL);
+            $aFilteredPlaceIDs = chksql($oDB->getCol($sSQL));
+            if ($aFilteredPlaceIDs) {
+                $aPlaceIDs = $aFilteredPlaceIDs;
+                if (CONST_Debug) {
+                    echo "<br><b>Place IDs after postcode filtering:</b> ";
+                    var_Dump($aPlaceIDs);
+                }
+            }
+        }
+
+        return array('IDs' => $aPlaceIDs, 'houseNumber' => $iHousenumber);
+    }
+
+
+    private function queryCountry(&$oDB)

     {
         $sSQL = 'SELECT place_id FROM placex ';
         $sSQL .= "WHERE country_code='".$this->sCountryCode."'";
     {
         $sSQL = 'SELECT place_id FROM placex ';
         $sSQL .= "WHERE country_code='".$this->sCountryCode."'";


@@ -382,7 +515,7 @@ class SearchDescription
         return chksql($oDB->getCol($sSQL));
     }
 
         return chksql($oDB->getCol($sSQL));
     }
 

-    public function queryNearbyPoi(&$oDB, $iLimit)
+    private function queryNearbyPoi(&$oDB, $iLimit)

     {
         if (!$this->sClass) {
             return array();
     {
         if (!$this->sClass) {
             return array();


@@ -398,7 +531,7 @@ class SearchDescription
             }
             if ($this->oContext->hasNearPoint()) {
                 $sSQL .= ' WHERE '.$this->oContext->withinSQL('ct.centroid');
             }
             if ($this->oContext->hasNearPoint()) {
                 $sSQL .= ' WHERE '.$this->oContext->withinSQL('ct.centroid');

-            } else if ($this->oContext->bViewboxBounded) {
+            } elseif ($this->oContext->bViewboxBounded) {

                 $sSQL .= ' WHERE ST_Contains('.$this->oContext->sqlViewboxSmall.', ct.centroid)';
             }
             if ($this->oContext->sqlCountryList) {
                 $sSQL .= ' WHERE ST_Contains('.$this->oContext->sqlViewboxSmall.', ct.centroid)';
             }
             if ($this->oContext->sqlCountryList) {


@@ -433,7 +566,7 @@ class SearchDescription
         return array();
     }
 
         return array();
     }
 

-    public function queryPostcode(&$oDB, $iLimit)
+    private function queryPostcode(&$oDB, $iLimit)

     {
         $sSQL = 'SELECT p.place_id FROM location_postcode p ';
 
     {
         $sSQL = 'SELECT p.place_id FROM location_postcode p ';
 


@@ -456,7 +589,7 @@ class SearchDescription
         return chksql($oDB->getCol($sSQL));
     }
 
         return chksql($oDB->getCol($sSQL));
     }
 

-    public function queryNamedPlace(&$oDB, $aWordFrequencyScores, $iMinAddressRank, $iMaxAddressRank, $iLimit)
+    private function queryNamedPlace(&$oDB, $aWordFrequencyScores, $iMinAddressRank, $iMaxAddressRank, $iLimit)

     {
         $aTerms = array();
         $aOrder = array();
     {
         $aTerms = array();
         $aOrder = array();


@@ -586,7 +719,7 @@ class SearchDescription
         return array();
     }
 
         return array();
     }
 

-    public function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $iLimit)
+    private function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $iLimit)

     {
         $sPlaceIDs = join(',', $aRoadPlaceIDs);
 
     {
         $sPlaceIDs = join(',', $aRoadPlaceIDs);
 


@@ -680,7 +813,7 @@ class SearchDescription
     }
 
 
     }
 
 

-    public function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit)
+    private function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit)

     {
         $sPlaceIDs = join(',', $aParentIDs);
         $aClassPlaceIDs = array();
     {
         $sPlaceIDs = join(',', $aParentIDs);
         $aClassPlaceIDs = array();


@@ -817,6 +950,22 @@ class SearchDescription
         return $aClassPlaceIDs;
     }
 
         return $aClassPlaceIDs;
     }
 

+    private function poiTable()
+    {
+        return 'place_classtype_'.$this->sClass.'_'.$this->sType;
+    }
+
+    private function countryCodeSQL($sVar)
+    {
+        if ($this->sCountryCode) {
+            return $sVar.' = \''.$this->sCountryCode."'";
+        }
+        if ($this->oContext->sqlCountryList) {
+            return $sVar.' in '.$this->oContext->sqlCountryList;
+        }
+
+        return '';
+    }

 
     /////////// Sort functions
 
 
     /////////// Sort functions