add more tests for database import

[nominatim.git] / lib-php / SearchDescription.php

diff --git a/lib-php/SearchDescription.php b/lib-php/SearchDescription.php
index 4d944bfb1ee498835c2d3178b2ea04d2439ab75d..be5623afcd3170ec58662b7e7c9ea6f8258adb83 100644 (file)
--- a/lib-php/SearchDescription.php
+++ b/lib-php/SearchDescription.php
@@ -19,6 +19,8 @@ class SearchDescription
     private $aName = array();
     /// True if the name is rare enough to force index use on name.
     private $bRareName = false;
     private $aName = array();
     /// True if the name is rare enough to force index use on name.
     private $bRareName = false;

+    /// True if the name requires to be accompanied by address terms.
+    private $bNameNeedsAddress = false;

     /// List of word ids making up the address of the object.
     private $aAddress = array();
     /// List of word ids that appear in the name but should be ignored.
     /// List of word ids making up the address of the object.
     private $aAddress = array();
     /// List of word ids that appear in the name but should be ignored.


@@ -113,6 +115,9 @@ class SearchDescription
                 return false;
             }
         }
                 return false;
             }
         }

+        if ($this->bNameNeedsAddress && empty($this->aAddress)) {
+            return false;
+        }

 
         return true;
     }
 
         return true;
     }


@@ -231,6 +236,7 @@ class SearchDescription
     {
         $this->aName[$iId] = $iId;
         $this->bRareName = $bRareName;
     {
         $this->aName[$iId] = $iId;
         $this->bRareName = $bRareName;

+        $this->bNameNeedsAddress = false;

     }
 
     /**
     }
 
     /**


@@ -240,11 +246,19 @@ class SearchDescription
      * @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 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 bool bNeedsAddress     True if the term is too unspecific to be used
+     *                               in a stand-alone search without an address
+     *                               to narrow down the search.

      * @param integer iPhraseNumber  Index of phrase, where the partial term
      *                               appears.
      */
      * @param integer iPhraseNumber  Index of phrase, where the partial term
      *                               appears.
      */

-    public function addPartialNameToken($iId, $bSearchable, $iPhraseNumber)
+    public function addPartialNameToken($iId, $bSearchable, $bNeedsAddress, $iPhraseNumber)

     {
     {

+        if (empty($this->aName)) {
+            $this->bNameNeedsAddress = $bNeedsAddress;
+        } else {
+            $this->bNameNeedsAddress &= $bNeedsAddress;
+        }

         if ($bSearchable) {
             $this->aName[$iId] = $iId;
         } else {
         if ($bSearchable) {
             $this->aName[$iId] = $iId;
         } else {


@@ -310,6 +324,7 @@ class SearchDescription
     {
         $this->aAddress = array_merge($this->aAddress, $this->aName);
         $this->bRareName = false;
     {
         $this->aAddress = array_merge($this->aAddress, $this->aName);
         $this->bRareName = false;

+        $this->bNameNeedsAddress = true;

         $this->aName = array($iId => $iId);
         $this->iNamePhrase = -1;
     }
         $this->aName = array($iId => $iId);
         $this->iNamePhrase = -1;
     }


@@ -566,32 +581,40 @@ class SearchDescription
 
         // Sort by existence of the requested house number but only if not
         // too many results are expected for the street, i.e. if the result
 
         // Sort by existence of the requested house number but only if not
         // too many results are expected for the street, i.e. if the result

-        // will be narrowed down by an address. Remeber that with ordering
+        // will be narrowed down by an address. Remember that with ordering

         // every single result has to be checked.
         if ($this->sHouseNumber && ($this->bRareName || !empty($this->aAddress) || $this->sPostcode)) {
         // every single result has to be checked.
         if ($this->sHouseNumber && ($this->bRareName || !empty($this->aAddress) || $this->sPostcode)) {

-            $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
-            $aOrder[] = ' (';
-            $aOrder[0] .= 'EXISTS(';
-            $aOrder[0] .= '  SELECT place_id';
-            $aOrder[0] .= '  FROM placex';
-            $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
-            $aOrder[0] .= "    AND housenumber ~* E'".$sHouseNumberRegex."'";
-            $aOrder[0] .= '  LIMIT 1';
-            $aOrder[0] .= ') ';
-            // also housenumbers from interpolation lines table are needed
-            if (preg_match('/[0-9]+/', $this->sHouseNumber)) {
-                $iHouseNumber = intval($this->sHouseNumber);
-                $aOrder[0] .= 'OR EXISTS(';
-                $aOrder[0] .= '  SELECT place_id ';
-                $aOrder[0] .= '  FROM location_property_osmline ';
-                $aOrder[0] .= '  WHERE parent_place_id = search_name.place_id';
-                $aOrder[0] .= '    AND startnumber is not NULL';
-                $aOrder[0] .= '    AND '.$iHouseNumber.'>=startnumber ';
-                $aOrder[0] .= '    AND '.$iHouseNumber.'<=endnumber ';
-                $aOrder[0] .= '  LIMIT 1';
-                $aOrder[0] .= ')';
+            $sHouseNumberRegex = $oDB->getDBQuoted('\\\\m'.$this->sHouseNumber.'\\\\M');
+
+            // Housenumbers on streets and places.
+            $sChildHnr = 'SELECT * FROM placex WHERE parent_place_id = search_name.place_id';
+            $sChildHnr .= '    AND housenumber ~* E'.$sHouseNumberRegex;
+            // Interpolations on streets and places.
+            if (preg_match('/^[0-9]+$/', $this->sHouseNumber)) {
+                $sIpolHnr = 'WHERE parent_place_id = search_name.place_id ';
+                $sIpolHnr .= '  AND startnumber is not NULL';
+                $sIpolHnr .= '    AND '.$this->sHouseNumber.'>=startnumber ';
+                $sIpolHnr .= '    AND '.$this->sHouseNumber.'<=endnumber ';
+            } else {
+                $sIpolHnr = false;

             }
             }

-            $aOrder[0] .= ') DESC';
+            // Housenumbers on the object iteself for unlisted places.
+            $sSelfHnr = 'SELECT * FROM placex WHERE place_id = search_name.place_id';
+            $sSelfHnr .= '    AND housenumber ~* E'.$sHouseNumberRegex;
+
+            $sSql = '(CASE WHEN address_rank = 30 THEN EXISTS('.$sSelfHnr.') ';
+            $sSql .= ' ELSE EXISTS('.$sChildHnr.') ';
+            if ($sIpolHnr) {
+                $sSql .= 'OR EXISTS(SELECT * FROM location_property_osmline '.$sIpolHnr.') ';
+                if (CONST_Use_US_Tiger_Data) {
+                    $sSql .= "OR (country_code = 'us' AND ";
+                    $sSql .= '    EXISTS(SELECT * FROM location_property_tiger '.$sIpolHnr.')) ';
+                }
+            }
+            $sSql .= 'END) DESC';
+
+
+            $aOrder[] = $sSql;

         }
 
         if (!empty($this->aName)) {
         }
 
         if (!empty($this->aName)) {


@@ -624,7 +647,7 @@ class SearchDescription
             $aOrder[] = $this->oContext->distanceSQL('centroid');
         } elseif ($this->sPostcode) {
             if (empty($this->aAddress)) {
             $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."')";
             }
             } else {
                 $aOrder[] = "(SELECT min(ST_Distance(search_name.centroid, p.geometry)) FROM location_postcode p WHERE p.postcode = '".$this->sPostcode."')";
             }


@@ -719,9 +742,9 @@ class SearchDescription
             return $aResults;
         }
 
             return $aResults;
         }
 

-        $sHouseNumberRegex = '\\\\m'.$this->sHouseNumber.'\\\\M';
+        $sHouseNumberRegex = $oDB->getDBQuoted('\\\\m'.$this->sHouseNumber.'\\\\M');

         $sSQL = 'SELECT place_id FROM placex WHERE';
         $sSQL = 'SELECT place_id FROM placex WHERE';

-        $sSQL .= "  housenumber ~* E'".$sHouseNumberRegex."'";
+        $sSQL .= '  housenumber ~* E'.$sHouseNumberRegex;

         $sSQL .= ' AND ('.join(' OR ', $aIDCondition).')';
         $sSQL .= $this->oContext->excludeSQL(' AND place_id');
 
         $sSQL .= ' AND ('.join(' OR ', $aIDCondition).')';
         $sSQL .= $this->oContext->excludeSQL(' AND place_id');