]> git.openstreetmap.org Git - nominatim.git/blobdiff - lib-php/SearchDescription.php
Revert "ICU: additional ranking by matching of normalised term"
[nominatim.git] / lib-php / SearchDescription.php
index dd20550214325b952452ec98de3bf4a96351071e..ee8bbc0c9be81304c9531760aac3751366a6a689 100644 (file)
@@ -67,35 +67,6 @@ class SearchDescription
         return $this->iSearchRank;
     }
 
-    /**
-     * 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;
-        $this->sClass = $sClass;
-        $this->sType = $sType;
-    }
-
-    /**
-     * Check if any operator is set.
-     *
-     * @return bool True, if this is a special search operation.
-     */
-    public function hasOperator()
-    {
-        return $this->iOperator != Operator::NONE;
-    }
-
     /**
      * Extract key/value pairs from a query.
      *
@@ -148,251 +119,234 @@ class SearchDescription
 
     /////////// Search building functions
 
+    /**
+     * Create a copy of this search description adding to search rank.
+     *
+     * @param integer $iTermCost  Cost to add to the current search rank.
+     *
+     * @return object Cloned search description.
+     */
+    public function clone($iTermCost)
+    {
+        $oSearch = clone $this;
+        $oSearch->iSearchRank += $iTermCost;
+
+        return $oSearch;
+    }
 
     /**
-     * Derive new searches by adding a full term to the existing search.
+     * Check if the search currently includes a name.
      *
-     * @param object $oSearchTerm  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   $bLastToken   True if the token is at the end of the query.
+     * @param bool bIncludeNonNames  If true stop-word tokens are taken into
+     *                               account, too.
      *
-     * @return SearchDescription[] List of derived search descriptions.
+     * @return bool True, if search has a name.
      */
-    public function extendWithFullTerm($oSearchTerm, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken)
+    public function hasName($bIncludeNonNames = false)
     {
-        $aNewSearches = array();
+        return !empty($this->aName)
+               || (!empty($this->aNameNonSearch) && $bIncludeNonNames);
+    }
 
-        if (($sPhraseType == '' || $sPhraseType == 'country')
-            && is_a($oSearchTerm, '\Nominatim\Token\Country')
-        ) {
-            if (!$this->sCountryCode) {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank++;
-                $oSearch->sCountryCode = $oSearchTerm->sCountryCode;
-                // Country is almost always at the end of the string
-                // - increase score for finding it anywhere else (optimisation)
-                if (!$bLastToken) {
-                    $oSearch->iSearchRank += 5;
-                    $oSearch->iNamePhrase = -1;
-                }
-                $aNewSearches[] = $oSearch;
-            }
-        } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode')
-                  && is_a($oSearchTerm, '\Nominatim\Token\Postcode')
-        ) {
-            if (!$this->sPostcode) {
-                // If we have structured search or this is the first term,
-                // make the postcode the primary search element.
-                if ($this->iOperator == Operator::NONE && $bFirstToken) {
-                    $oSearch = clone $this;
-                    $oSearch->iSearchRank++;
-                    $oSearch->iOperator = Operator::POSTCODE;
-                    $oSearch->aAddress = array_merge($this->aAddress, $this->aName);
-                    $oSearch->aName =
-                        array($oSearchTerm->iId => $oSearchTerm->sPostcode);
-                    $aNewSearches[] = $oSearch;
-                }
+    /**
+     * Check if the search currently includes an address term.
+     *
+     * @return bool True, if any address term is included, including stop-word
+     *              terms.
+     */
+    public function hasAddress()
+    {
+        return !empty($this->aAddress) || !empty($this->aAddressNonSearch);
+    }
 
-                // If we have a structured search or this is not the first term,
-                // add the postcode as an addendum.
-                if ($this->iOperator != Operator::POSTCODE
-                    && ($sPhraseType == 'postalcode' || !empty($this->aName))
-                ) {
-                    $oSearch = clone $this;
-                    $oSearch->iSearchRank++;
-                    $oSearch->iNamePhrase = -1;
-                    if (strlen($oSearchTerm->sPostcode) < 4) {
-                        $oSearch->iSearchRank += 4 - strlen($oSearchTerm->sPostcode);
-                    }
-                    $oSearch->sPostcode = $oSearchTerm->sPostcode;
-                    $aNewSearches[] = $oSearch;
-                }
-            }
-        } elseif (($sPhraseType == '' || $sPhraseType == 'street')
-                 && is_a($oSearchTerm, '\Nominatim\Token\HouseNumber')
-        ) {
-            if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank++;
-                $oSearch->iNamePhrase = -1;
-                $oSearch->sHouseNumber = $oSearchTerm->sToken;
-                if ($this->iOperator != Operator::NONE) {
-                    $oSearch->iSearchRank++;
-                }
-                // sanity check: if the housenumber is not mainly made
-                // up of numbers, add a penalty
-                if (preg_match('/\\d/', $oSearch->sHouseNumber) === 0
-                    || preg_match_all('/[^0-9]/', $oSearch->sHouseNumber, $aMatches) > 2) {
-                    $oSearch->iSearchRank++;
-                }
-                if (empty($oSearchTerm->iId)) {
-                    $oSearch->iSearchRank++;
-                }
-                // also must not appear in the middle of the address
-                if (!empty($this->aAddress)
-                    || (!empty($this->aAddressNonSearch))
-                    || $this->sPostcode
-                ) {
-                    $oSearch->iSearchRank++;
-                }
-                $aNewSearches[] = $oSearch;
-                // Housenumbers may appear in the name when the place has its own
-                // address terms.
-                if ($oSearchTerm->iId !== null
-                    && ($this->iNamePhrase >= 0 || empty($this->aName))
-                    && empty($this->aAddress)
-                   ) {
-                    $oSearch = clone $this;
-                    $oSearch->iSearchRank++;
-                    $oSearch->aAddress = $this->aName;
-                    $oSearch->bRareName = false;
-                    $oSearch->aName = array($oSearchTerm->iId => $oSearchTerm->iId);
-                    $aNewSearches[] = $oSearch;
-                }
-            }
-        } elseif ($sPhraseType == ''
-                  && is_a($oSearchTerm, '\Nominatim\Token\SpecialTerm')
-        ) {
-            if ($this->iOperator == Operator::NONE) {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank += 2;
-                $oSearch->iNamePhrase = -1;
-
-                $iOp = $oSearchTerm->iOperator;
-                if ($iOp == Operator::NONE) {
-                    if (!empty($this->aName) || $this->oContext->isBoundedSearch()) {
-                        $iOp = Operator::NAME;
-                    } else {
-                        $iOp = Operator::NEAR;
-                    }
-                    $oSearch->iSearchRank += 2;
-                } elseif (!$bFirstToken && !$bLastToken) {
-                    $oSearch->iSearchRank += 2;
-                }
-                if ($this->sHouseNumber) {
-                    $oSearch->iSearchRank++;
-                }
+    /**
+     * Check if a country restriction is currently included in the search.
+     *
+     * @return bool True, if a country restriction is set.
+     */
+    public function hasCountry()
+    {
+        return $this->sCountryCode !== '';
+    }
 
-                $oSearch->setPoiSearch(
-                    $iOp,
-                    $oSearchTerm->sClass,
-                    $oSearchTerm->sType
-                );
-                $aNewSearches[] = $oSearch;
-            }
-        } elseif ($sPhraseType != 'country'
-                  && is_a($oSearchTerm, '\Nominatim\Token\Word')
-        ) {
-            $iWordID = $oSearchTerm->iId;
-            // Full words can only be a name if they appear at the beginning
-            // of the phrase. In structured search the name must forcably in
-            // the first phrase. In unstructured search it may be in a later
-            // phrase when the first phrase is a house number.
-            if (!empty($this->aName) || !($bFirstPhrase || $sPhraseType == '')) {
-                if (($sPhraseType == '' || !$bFirstPhrase) && !$bHasPartial) {
-                    $oSearch = clone $this;
-                    $oSearch->iNamePhrase = -1;
-                    $oSearch->iSearchRank += 3 * $oSearchTerm->iTermCount;
-                    $oSearch->aAddress[$iWordID] = $iWordID;
-                    $aNewSearches[] = $oSearch;
-                }
-            } elseif (empty($this->aNameNonSearch)) {
-                $oSearch = clone $this;
-                $oSearch->iSearchRank++;
-                $oSearch->aName = array($iWordID => $iWordID);
-                if (CONST_Search_NameOnlySearchFrequencyThreshold) {
-                    $oSearch->bRareName =
-                        $oSearchTerm->iSearchNameCount
-                          < CONST_Search_NameOnlySearchFrequencyThreshold;
-                }
-                $aNewSearches[] = $oSearch;
-            }
-        }
+    /**
+     * Check if a postcode is currently included in the search.
+     *
+     * @return bool True, if a postcode is set.
+     */
+    public function hasPostcode()
+    {
+        return $this->sPostcode !== '';
+    }
 
-        return $aNewSearches;
+    /**
+     * Check if a house number is set for the search.
+     *
+     * @return bool True, if a house number is set.
+     */
+    public function hasHousenumber()
+    {
+        return $this->sHouseNumber !== '';
     }
 
     /**
-     * Derive new searches by adding a partial term to the existing search.
+     * Check if a special type of place is requested.
      *
-     * @param string  $sToken             Term for the token.
-     * @param object  $oSearchTerm        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[] $aFullTokens        List of full term tokens with the
-     *                                    same name.
+     * param integer iOperator  When set, check for the particular
+     *                          operator used for the special type.
      *
-     * @return SearchDescription[] List of derived search descriptions.
+     * @return bool True, if speial type is requested or, if requested,
+     *              a special type with the given operator.
      */
-    public function extendWithPartialTerm($sToken, $oSearchTerm, $bStructuredPhrases, $iPhrase, $aFullTokens)
+    public function hasOperator($iOperator = null)
     {
-        // Only allow name terms.
-        if (!(is_a($oSearchTerm, '\Nominatim\Token\Word'))) {
-            return array();
+        return $iOperator === null ? $this->iOperator != Operator::NONE : $this->iOperator == $iOperator;
+    }
+
+    /**
+     * Add the given token to the list of terms to search for in the address.
+     *
+     * @param integer iID       ID of term to add.
+     * @param bool bSearchable  Term should be used to search for result
+     *                          (i.e. term is not a stop word).
+     */
+    public function addAddressToken($iId, $bSearchable = true)
+    {
+        if ($bSearchable) {
+            $this->aAddress[$iId] = $iId;
+        } else {
+            $this->aAddressNonSearch[$iId] = $iId;
         }
+    }
 
-        $aNewSearches = array();
-        $iWordID = $oSearchTerm->iId;
+    /**
+     * Add the given full-word token to the list of terms to search for in the
+     * name.
+     *
+     * @param interger iId    ID of term to add.
+     * @param bool bRareName  True if the term is infrequent enough to not
+     *                        require other constraints for efficient search.
+     */
+    public function addNameToken($iId, $bRareName)
+    {
+        $this->aName[$iId] = $iId;
+        $this->bRareName = $bRareName;
+    }
 
-        if ((!$bStructuredPhrases || $iPhrase > 0)
-            && (!empty($this->aName))
-        ) {
-            $oSearch = clone $this;
-            $oSearch->iSearchRank++;
-            if (preg_match('#^[0-9 ]+$#', $sToken)) {
-                $oSearch->iSearchRank++;
-            }
-            if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
-                $oSearch->aAddress[$iWordID] = $iWordID;
-            } else {
-                $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
-                if (!empty($aFullTokens)) {
-                    $oSearch->iSearchRank++;
-                }
-            }
-            $aNewSearches[] = $oSearch;
+    /**
+     * Add the given partial token to the list of terms to search for in
+     * the name.
+     *
+     * @param integer iID            ID of term to add.
+     * @param bool bSearchable       Term should be used to search for result
+     *                               (i.e. term is not a stop word).
+     * @param integer iPhraseNumber  Index of phrase, where the partial term
+     *                               appears.
+     */
+    public function addPartialNameToken($iId, $bSearchable, $iPhraseNumber)
+    {
+        if ($bSearchable) {
+            $this->aName[$iId] = $iId;
+        } else {
+            $this->aNameNonSearch[$iId] = $iId;
         }
+        $this->iNamePhrase = $iPhraseNumber;
+    }
 
-        if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
-            && ((empty($this->aName) && empty($this->aNameNonSearch)) || $this->iNamePhrase == $iPhrase)
-            && strpos($sToken, ' ') === false
-        ) {
-            $oSearch = clone $this;
-            $oSearch->iSearchRank++;
-            if (empty($this->aName) && empty($this->aNameNonSearch)) {
-                $oSearch->iSearchRank++;
-            }
-            if (preg_match('#^[0-9 ]+$#', $sToken)) {
-                $oSearch->iSearchRank++;
-            }
-            if ($oSearchTerm->iSearchNameCount < CONST_Max_Word_Frequency) {
-                if (empty($this->aName)
-                    && CONST_Search_NameOnlySearchFrequencyThreshold
-                ) {
-                    $oSearch->bRareName =
-                        $oSearchTerm->iSearchNameCount
-                          < CONST_Search_NameOnlySearchFrequencyThreshold;
-                } else {
-                    $oSearch->bRareName = false;
-                }
-                $oSearch->aName[$iWordID] = $iWordID;
-            } else {
-                if (!empty($aFullTokens)) {
-                    $oSearch->iSearchRank++;
-                }
-                $oSearch->aNameNonSearch[$iWordID] = $iWordID;
-            }
-            $oSearch->iNamePhrase = $iPhrase;
-            $aNewSearches[] = $oSearch;
-        }
+    /**
+     * Set country restriction for the search.
+     *
+     * @param string sCountryCode  Country code of country to restrict search to.
+     */
+    public function setCountry($sCountryCode)
+    {
+        $this->sCountryCode = $sCountryCode;
+        $this->iNamePhrase = -1;
+    }
+
+    /**
+     * Set postcode search constraint.
+     *
+     * @param string sPostcode  Postcode the result should have.
+     */
+    public function setPostcode($sPostcode)
+    {
+        $this->sPostcode = $sPostcode;
+        $this->iNamePhrase = -1;
+    }
+
+    /**
+     * Make this search a search for a postcode object.
+     *
+     * @param integer iId       Token Id for the postcode.
+     * @param string sPostcode  Postcode to look for.
+     */
+    public function setPostcodeAsName($iId, $sPostcode)
+    {
+        $this->iOperator = Operator::POSTCODE;
+        $this->aAddress = array_merge($this->aAddress, $this->aName);
+        $this->aName = array($iId => $sPostcode);
+        $this->bRareName = true;
+        $this->iNamePhrase = -1;
+    }
+
+    /**
+     * Set house number search cnstraint.
+     *
+     * @param string sNumber  House number the result should have.
+     */
+    public function setHousenumber($sNumber)
+    {
+        $this->sHouseNumber = $sNumber;
+        $this->iNamePhrase = -1;
+    }
+
+    /**
+     * Make this search a search for a house number.
+     *
+     * @param integer iId  Token Id for the house number.
+     */
+    public function setHousenumberAsName($iId)
+    {
+        $this->aAddress = array_merge($this->aAddress, $this->aName);
+        $this->bRareName = false;
+        $this->aName = array($iId => $iId);
+        $this->iNamePhrase = -1;
+    }
+
+    /**
+     * 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;
+        $this->sClass = $sClass;
+        $this->sType = $sType;
+        $this->iNamePhrase = -1;
+    }
 
-        return $aNewSearches;
+    public function getNamePhrase()
+    {
+        return $this->iNamePhrase;
+    }
+
+    /**
+     * Get the global search context.
+     *
+     * @return object  Objects of global search constraints.
+     */
+    public function getContext()
+    {
+        return $this->oContext;
     }
 
     /////////// Query functions
@@ -413,7 +367,6 @@ class SearchDescription
     public function query(&$oDB, $iMinRank, $iMaxRank, $iLimit)
     {
         $aResults = array();
-        $iHousenumber = -1;
 
         if ($this->sCountryCode
             && empty($this->aName)
@@ -446,23 +399,24 @@ class SearchDescription
 
             // Now search for housenumber, if housenumber provided. Can be zero.
             if (($this->sHouseNumber || $this->sHouseNumber === '0') && !empty($aResults)) {
+                $aHnResults = $this->queryHouseNumber($oDB, $aResults);
+
                 // Downgrade the rank of the street results, they are missing
-                // the housenumber.
+                // the housenumber. Also drop POI places (rank 30) here, they
+                // cannot be a parent place and therefore must not be shown
+                // as a result for a search with a missing housenumber.
                 foreach ($aResults as $oRes) {
-                    if ($oRes->iAddressRank >= 26) {
-                        $oRes->iResultRank++;
-                    } else {
-                        $oRes->iResultRank += 2;
+                    if ($oRes->iAddressRank < 28) {
+                        if ($oRes->iAddressRank >= 26) {
+                            $oRes->iResultRank++;
+                        } else {
+                            $oRes->iResultRank += 2;
+                        }
+                        $aHnResults[$oRes->iId] = $oRes;
                     }
                 }
 
-                $aHnResults = $this->queryHouseNumber($oDB, $aResults);
-
-                if (!empty($aHnResults)) {
-                    foreach ($aHnResults as $oRes) {
-                        $aResults[$oRes->iId] = $oRes;
-                    }
-                }
+                $aResults = $aHnResults;
             }
 
             // finally get POIs if requested
@@ -670,7 +624,7 @@ class SearchDescription
             $aOrder[] = $this->oContext->distanceSQL('centroid');
         } elseif ($this->sPostcode) {
             if (empty($this->aAddress)) {
-                $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."' AND ST_DWithin(search_name.centroid, p.geometry, 0.1))";
+                $aTerms[] = "EXISTS(SELECT place_id FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."' AND ST_DWithin(search_name.centroid, p.geometry, 0.12))";
             } else {
                 $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."')";
             }
@@ -742,16 +696,33 @@ class SearchDescription
     private function queryHouseNumber(&$oDB, $aRoadPlaceIDs)
     {
         $aResults = array();
-        $sPlaceIDs = Result::joinIdsByTable($aRoadPlaceIDs, Result::TABLE_PLACEX);
+        $sRoadPlaceIDs = Result::joinIdsByTableMaxRank(
+            $aRoadPlaceIDs,
+            Result::TABLE_PLACEX,
+            27
+        );
+        $sPOIPlaceIDs = Result::joinIdsByTableMinRank(
+            $aRoadPlaceIDs,
+            Result::TABLE_PLACEX,
+            30
+        );
 
-        if (!$sPlaceIDs) {
+        $aIDCondition = array();
+        if ($sRoadPlaceIDs) {
+            $aIDCondition[] = 'parent_place_id in ('.$sRoadPlaceIDs.')';
+        }
+        if ($sPOIPlaceIDs) {
+            $aIDCondition[] = 'place_id in ('.$sPOIPlaceIDs.')';
+        }
+
+        if (empty($aIDCondition)) {
             return $aResults;
         }
 
         $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
-        $sSQL = 'SELECT place_id FROM placex ';
-        $sSQL .= 'WHERE parent_place_id in ('.$sPlaceIDs.')';
-        $sSQL .= "  AND housenumber ~* E'".$sHouseNumberRegex."'";
+        $sSQL = 'SELECT place_id FROM placex WHERE';
+        $sSQL .= "  housenumber ~* E'".$sHouseNumberRegex."'";
+        $sSQL .= ' AND ('.join(' OR ', $aIDCondition).')';
         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 
         Debug::printSQL($sSQL);
@@ -763,11 +734,11 @@ class SearchDescription
 
         $bIsIntHouseNumber= (bool) preg_match('/[0-9]+/', $this->sHouseNumber);
         $iHousenumber = intval($this->sHouseNumber);
-        if ($bIsIntHouseNumber && empty($aResults)) {
+        if ($bIsIntHouseNumber && $sRoadPlaceIDs && empty($aResults)) {
             // if nothing found, search in the interpolation line table
             $sSQL = 'SELECT distinct place_id FROM location_property_osmline';
             $sSQL .= ' WHERE startnumber is not NULL';
-            $sSQL .= '  AND parent_place_id in ('.$sPlaceIDs.') AND (';
+            $sSQL .= '  AND parent_place_id in ('.$sRoadPlaceIDs.') AND (';
             if ($iHousenumber % 2 == 0) {
                 // If housenumber is even, look for housenumber in streets
                 // with interpolationtype even or all.
@@ -790,24 +761,10 @@ class SearchDescription
             }
         }
 
-        // If nothing found try the aux fallback table
-        if (CONST_Use_Aux_Location_data && empty($aResults)) {
-            $sSQL = 'SELECT place_id FROM location_property_aux';
-            $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.')';
-            $sSQL .= " AND housenumber = '".$this->sHouseNumber."'";
-            $sSQL .= $this->oContext->excludeSQL(' AND place_id');
-
-            Debug::printSQL($sSQL);
-
-            foreach ($oDB->getCol($sSQL) as $iPlaceId) {
-                $aResults[$iPlaceId] = new Result($iPlaceId, Result::TABLE_AUX);
-            }
-        }
-
         // If nothing found then search in Tiger data (location_property_tiger)
-        if (CONST_Use_US_Tiger_Data && $bIsIntHouseNumber && empty($aResults)) {
+        if (CONST_Use_US_Tiger_Data && $sRoadPlaceIDs && $bIsIntHouseNumber && empty($aResults)) {
             $sSQL = 'SELECT place_id FROM location_property_tiger';
-            $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.') and (';
+            $sSQL .= ' WHERE parent_place_id in ('.$sRoadPlaceIDs.') and (';
             if ($iHousenumber % 2 == 0) {
                 $sSQL .= "interpolationtype='even'";
             } else {