Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 979f1ed86fb4a046b015a85ffad22e65d7380889..52f66e5b9a5b229d9e9af18e71817477322642b2 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -45,6 +45,11 @@
                        $this->oDB =& $oDB;
                }
 
                        $this->oDB =& $oDB;
                }
 

+               function setReverseInPlan($bReverse)
+               {
+                       $this->bReverseInPlan = $bReverse;
+               }
+

                function setLanguagePreference($aLangPref)
                {
                        $this->aLangPrefOrder = $aLangPref;
                function setLanguagePreference($aLangPref)
                {
                        $this->aLangPrefOrder = $aLangPref;


@@ -192,6 +197,20 @@
                        return $this->sQuery;
                }
 
                        return $this->sQuery;
                }
 

+               function loadStructuredAddressElement($sValue, $sKey, $iNewMinAddressRank, $iNewMaxAddressRank, $aItemListValues)
+               {
+                       $sValue = trim($sValue);
+                       if (!$sValue) return false;
+                       $this->aStructuredQuery[$sKey] = $sValue;
+                       if ($this->iMinAddressRank == 0 && $this->iMaxAddressRank == 30)
+                       {
+                               $this->iMinAddressRank = $iNewMinAddressRank;
+                               $this->iMaxAddressRank = $iNewMaxAddressRank;
+                       }
+                       if ($aItemListValues) $this->aAddressRankList = array_merge($this->aAddressRankList, $aItemListValues);
+                       return true;
+               }
+

                function setStructuredQuery($sAmentiy = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
                {
                        $this->sQuery = false;
                function setStructuredQuery($sAmentiy = false, $sStreet = false, $sCity = false, $sCounty = false, $sState = false, $sCountry = false, $sPostalCode = false)
                {
                        $this->sQuery = false;


@@ -199,13 +218,13 @@
                        $this->aStructuredQuery = array();
                        $this->sAllowedTypesSQLList = '';
 
                        $this->aStructuredQuery = array();
                        $this->sAllowedTypesSQLList = '';
 

-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sAmentiy, 'amenity', 26, 30, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sStreet, 'street', 26, 30, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCity, 'city', 14, 24, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCounty, 'county', 9, 13, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sState, 'state', 8, 8, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sCountry, 'country', 4, 4, false);
-                       loadStructuredAddressElement($this->aStructuredQuery, $this->iMinAddressRank, $this->iMaxAddressRank, $this->aAddressRankList, $sPostalCode, 'postalcode' , 5, 11, array(5, 11));
+                       $this->loadStructuredAddressElement($sAmentiy, 'amenity', 26, 30, false);
+                       $this->loadStructuredAddressElement($sStreet, 'street', 26, 30, false);
+                       $this->loadStructuredAddressElement($sCity, 'city', 14, 24, false);
+                       $this->loadStructuredAddressElement($sCounty, 'county', 9, 13, false);
+                       $this->loadStructuredAddressElement($sState, 'state', 8, 8, false);
+                       $this->loadStructuredAddressElement($sPostalCode, 'postalcode' , 5, 11, array(5, 11));
+                       $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);

 
                        if (sizeof($this->aStructuredQuery) > 0) 
                        {
 
                        if (sizeof($this->aStructuredQuery) > 0) 
                        {


@@ -242,7 +261,7 @@
                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
                        $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                        if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'";
                        $sSQL .= "from placex where place_id in ($sPlaceIDs) ";
                        $sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
                        if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'";

-                       if ($aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$aAddressRankList).")";
+                       if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$this->aAddressRankList).")";

                        $sSQL .= ") ";
                        if ($this->sAllowedTypesSQLList) $sSQL .= "and placex.class in $this->sAllowedTypesSQLList ";
                        $sSQL .= "and linked_place_id is null ";
                        $sSQL .= ") ";
                        if ($this->sAllowedTypesSQLList) $sSQL .= "and placex.class in $this->sAllowedTypesSQLList ";
                        $sSQL .= "and linked_place_id is null ";


@@ -311,7 +330,8 @@
                        }
 
                        // Hack to make it handle "new york, ny" (and variants) correctly
                        }
 
                        // Hack to make it handle "new york, ny" (and variants) correctly

-                       $sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery);
+                       //$sQuery = str_ireplace(array('New York, ny','new york, new york', 'New York ny','new york new york'), 'new york city, ny', $this->sQuery);
+                       $sQuery = $this->sQuery;

 
                        // Conflicts between US state abreviations and various words for 'the' in different languages
                        if (isset($this->aLangPrefOrder['name:en']))
 
                        // Conflicts between US state abreviations and various words for 'the' in different languages
                        if (isset($this->aLangPrefOrder['name:en']))


@@ -625,8 +645,11 @@
                                                if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
                                                else $sPhraseType = '';
 
                                                if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
                                                else $sPhraseType = '';
 

-                                               foreach($aPhrases[$iPhrase]['wordsets'] as $aWordset)
+                                               foreach($aPhrases[$iPhrase]['wordsets'] as $iWordSet => $aWordset)

                                                {
                                                {

+                                                       // Too many permutations - too expensive
+                                                       if ($iWordSet > 120) break;
+

                                                        $aWordsetSearches = $aSearches;
 
                                                        // Add all words from this wordset
                                                        $aWordsetSearches = $aSearches;
 
                                                        // Add all words from this wordset


@@ -1061,7 +1084,7 @@
                                                        // TODO: filter out the pointless search terms (2 letter name tokens and less)
                                                        // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";
                                                        // TODO: filter out the pointless search terms (2 letter name tokens and less)
                                                        // they might be right - but they are just too darned expensive to run
                                                        if (sizeof($aSearch['aName'])) $aTerms[] = "name_vector @> ARRAY[".join($aSearch['aName'],",")."]";

-                                                       if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";
+                                                       //if (sizeof($aSearch['aNameNonSearch'])) $aTerms[] = "array_cat(name_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aNameNonSearch'],",")."]";

                                                        if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
                                                        {
                                                                // For infrequent name terms disable index usage for address
                                                        if (sizeof($aSearch['aAddress']) && $aSearch['aName'] != $aSearch['aAddress'])
                                                        {
                                                                // For infrequent name terms disable index usage for address


@@ -1069,12 +1092,13 @@
                                                                                sizeof($aSearch['aName']) == 1 &&
                                                                                $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
                                                                {
                                                                                sizeof($aSearch['aName']) == 1 &&
                                                                                $aWordFrequencyScores[$aSearch['aName'][reset($aSearch['aName'])]] < CONST_Search_NameOnlySearchFrequencyThreshold)
                                                                {

-                                                                       $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
+                                                                       //$aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join(array_merge($aSearch['aAddress'],$aSearch['aAddressNonSearch']),",")."]";
+                                                                       $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddress'],",")."]";

                                                                }
                                                                else
                                                                {
                                                                        $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";
                                                                }
                                                                else
                                                                {
                                                                        $aTerms[] = "nameaddress_vector @> ARRAY[".join($aSearch['aAddress'],",")."]";

-                                                                       if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";
+                                                                       //if (sizeof($aSearch['aAddressNonSearch'])) $aTerms[] = "array_cat(nameaddress_vector,ARRAY[]::integer[]) @> ARRAY[".join($aSearch['aAddressNonSearch'],",")."]";

                                                                }
                                                        }
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";
                                                                }
                                                        }
                                                        if ($aSearch['sCountryCode']) $aTerms[] = "country_code = '".pg_escape_string($aSearch['sCountryCode'])."'";


@@ -1413,10 +1437,12 @@
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }

+                            /*

                                                        elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }
                                                        elseif (preg_match('#MULTIPOLYGON\\(\\(\\(([- 0-9.,]+)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                preg_match_all('/(-?[0-9.]+) (-?[0-9.]+)/',$aMatch[1],$aPolyPoints,PREG_SET_ORDER);
                                                        }

+                            */

                                                        elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                $fRadius = 0.01;
                                                        elseif (preg_match('#POINT\\((-?[0-9.]+) (-?[0-9.]+)\\)#',$aPointPolygon['astext'],$aMatch))
                                                        {
                                                                $fRadius = 0.01;