Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / Geocode.php

diff --git a/lib/Geocode.php b/lib/Geocode.php
index 8276fe98d6c86d4d4d6d9ad6fafd74757bf949ce..e9b304d251abbe4c6c7cf6e09c5c130297a27953 100644 (file)
--- a/lib/Geocode.php
+++ b/lib/Geocode.php
@@ -225,14 +225,14 @@ class Geocode
         $aViewbox = $oParams->getStringList('viewboxlbrt');
         if ($aViewbox) {
             if (count($aViewbox) != 4) {
         $aViewbox = $oParams->getStringList('viewboxlbrt');
         if ($aViewbox) {
             if (count($aViewbox) != 4) {

-                userError("Bad parmater 'viewboxlbrt'. Expected 4 coordinates.");
+                userError("Bad parameter 'viewboxlbrt'. Expected 4 coordinates.");

             }
             $this->setViewbox($aViewbox);
         } else {
             $aViewbox = $oParams->getStringList('viewbox');
             if ($aViewbox) {
                 if (count($aViewbox) != 4) {
             }
             $this->setViewbox($aViewbox);
         } else {
             $aViewbox = $oParams->getStringList('viewbox');
             if ($aViewbox) {
                 if (count($aViewbox) != 4) {

-                    userError("Bad parmater 'viewbox'. Expected 4 coordinates.");
+                    userError("Bad parameter 'viewbox'. Expected 4 coordinates.");

                 }
                 $this->setViewBox($aViewbox);
             } else {
                 }
                 $this->setViewBox($aViewbox);
             } else {


@@ -303,7 +303,7 @@ class Geocode
         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 
         $this->loadStructuredAddressElement($sPostalCode, 'postalcode', 5, 11, array(5, 11));
         $this->loadStructuredAddressElement($sCountry, 'country', 4, 4, false);
 

-        if (sizeof($this->aStructuredQuery) > 0) {
+        if (!empty($this->aStructuredQuery)) {

             $this->sQuery = join(', ', $this->aStructuredQuery);
             if ($this->iMaxAddressRank < 30) {
                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';
             $this->sQuery = join(', ', $this->aStructuredQuery);
             if ($this->iMaxAddressRank < 30) {
                 $this->sAllowedTypesSQLList = '(\'place\',\'boundary\')';


@@ -317,7 +317,7 @@ class Geocode
 
         $aParams = $this->aStructuredQuery;
 
 
         $aParams = $this->aStructuredQuery;
 

-        if (sizeof($aParams) == 1) return false;
+        if (count($aParams) == 1) return false;

 
         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 
 
         $aOrderToFallback = array('postalcode', 'street', 'city', 'county', 'state');
 


@@ -375,8 +375,8 @@ class Geocode
                                     $sPhraseType,
                                     $iToken == 0 && $iPhrase == 0,
                                     $iPhrase == 0,
                                     $sPhraseType,
                                     $iToken == 0 && $iPhrase == 0,
                                     $iPhrase == 0,

-                                    $iToken + 1 == sizeof($aWordset)
-                                      && $iPhrase + 1 == sizeof($aPhrases)
+                                    $iToken + 1 == count($aWordset)
+                                      && $iPhrase + 1 == count($aPhrases)

                                 );
 
                                 foreach ($aNewSearches as $oSearch) {
                                 );
 
                                 foreach ($aNewSearches as $oSearch) {


@@ -411,7 +411,7 @@ class Geocode
                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
                 }
                     usort($aNewWordsetSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
                     $aWordsetSearches = array_slice($aNewWordsetSearches, 0, 50);
                 }

-                //var_Dump('<hr>',sizeof($aWordsetSearches)); exit;
+                //var_Dump('<hr>',count($aWordsetSearches)); exit;

 
                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));
 
                 $aNewPhraseSearches = array_merge($aNewPhraseSearches, $aNewWordsetSearches);
                 usort($aNewPhraseSearches, array('Nominatim\SearchDescription', 'bySearchRank'));


@@ -442,12 +442,10 @@ class Geocode
             $iSearchCount = 0;
             $aSearches = array();
             foreach ($aGroupedSearches as $iScore => $aNewSearches) {
             $iSearchCount = 0;
             $aSearches = array();
             foreach ($aGroupedSearches as $iScore => $aNewSearches) {

-                $iSearchCount += sizeof($aNewSearches);
+                $iSearchCount += count($aNewSearches);

                 $aSearches = array_merge($aSearches, $aNewSearches);
                 if ($iSearchCount > 50) break;
             }
                 $aSearches = array_merge($aSearches, $aNewSearches);
                 if ($iSearchCount > 50) break;
             }

-
-            //if (CONST_Debug) _debugDumpGroupedSearches($aGroupedSearches, $aValidTokens);

         }
 
         // Revisit searches, drop bad searches and give penalty to unlikely combinations.
         }
 
         // Revisit searches, drop bad searches and give penalty to unlikely combinations.


@@ -479,9 +477,9 @@ class Geocode
             osm_id: id of corresponding OSM object
             class: general object class (corresponds to tag key of primary OSM tag)
             type: subclass of object (corresponds to tag value of primary OSM tag)
             osm_id: id of corresponding OSM object
             class: general object class (corresponds to tag key of primary OSM tag)
             type: subclass of object (corresponds to tag value of primary OSM tag)

-            admin_level: see http://wiki.openstreetmap.org/wiki/Admin_level
+            admin_level: see https://wiki.openstreetmap.org/wiki/Admin_level

             rank_search: rank in search hierarchy
             rank_search: rank in search hierarchy

-                        (see also http://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)
+                        (see also https://wiki.openstreetmap.org/wiki/Nominatim/Development_overview#Country_to_street_level)

             rank_address: rank in address hierarchy (determines orer in address)
             place_id: internal key (may differ between different instances)
             country_code: ISO country code
             rank_address: rank in address hierarchy (determines orer in address)
             place_id: internal key (may differ between different instances)
             country_code: ISO country code


@@ -502,8 +500,11 @@ class Geocode
 
     public function lookup()
     {
 
     public function lookup()
     {

+        Debug::newFunction('Geocode::lookup');

         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 
         if (!$this->sQuery && !$this->aStructuredQuery) return array();
 

+        Debug::printDebugArray('Geocode', $this);
+

         $oCtx = new SearchContext();
 
         if ($this->aRoutePoints) {
         $oCtx = new SearchContext();
 
         if ($this->aRoutePoints) {


@@ -523,7 +524,11 @@ class Geocode
             $oCtx->setCountryList($this->aCountryCodes);
         }
 
             $oCtx->setCountryList($this->aCountryCodes);
         }
 

+        Debug::newSection('Query Preprocessing');
+

         $sNormQuery = $this->normTerm($this->sQuery);
         $sNormQuery = $this->normTerm($this->sQuery);

+        Debug::printVar('Normalized query', $sNormQuery);
+

         $sLanguagePrefArraySQL = getArraySQL(
             array_map('getDBQuoted', $this->aLangPrefOrder)
         );
         $sLanguagePrefArraySQL = getArraySQL(
             array_map('getDBQuoted', $this->aLangPrefOrder)
         );


@@ -535,9 +540,9 @@ class Geocode
 
         // 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'])) {

-            $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/', '\1illinois\2', $sQuery);
-            $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/', '\1alabama\2', $sQuery);
-            $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/', '\1louisiana\2', $sQuery);
+            $sQuery = preg_replace('/(^|,)\s*il\s*(,|$)/i', '\1illinois\2', $sQuery);
+            $sQuery = preg_replace('/(^|,)\s*al\s*(,|$)/i', '\1alabama\2', $sQuery);
+            $sQuery = preg_replace('/(^|,)\s*la\s*(,|$)/i', '\1louisiana\2', $sQuery);

         }
 
         // Do we have anything that looks like a lat/lon pair?
         }
 
         // Do we have anything that looks like a lat/lon pair?


@@ -560,6 +565,10 @@ class Geocode
                     $aSpecialTermsRaw,
                     PREG_SET_ORDER
                 );
                     $aSpecialTermsRaw,
                     PREG_SET_ORDER
                 );

+                if (!empty($aSpecialTermsRaw)) {
+                    Debug::printVar('Special terms', $aSpecialTermsRaw);
+                }
+

                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
                     if (!$sSpecialTerm) {
                 foreach ($aSpecialTermsRaw as $aSpecialTerm) {
                     $sQuery = str_replace($aSpecialTerm[0], ' ', $sQuery);
                     if (!$sSpecialTerm) {


@@ -582,7 +591,8 @@ class Geocode
                 $sSQL = 'SELECT class, type FROM word ';
                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
                 $sSQL .= '   AND class is not null AND class not in (\'place\')';
                 $sSQL = 'SELECT class, type FROM word ';
                 $sSQL .= '   WHERE word_token in (\' '.$sToken.'\')';
                 $sSQL .= '   AND class is not null AND class not in (\'place\')';

-                if (CONST_Debug) var_Dump($sSQL);
+
+                Debug::printSQL($sSQL);

                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
                 $aNewSearches = array();
                 foreach ($aSearches as $oSearch) {
                 $aSearchWords = chksql($this->oDB->getAll($sSQL));
                 $aNewSearches = array();
                 foreach ($aSearches as $oSearch) {


@@ -609,10 +619,15 @@ class Geocode
                 $bStructuredPhrases = false;
             }
 
                 $bStructuredPhrases = false;
             }
 

+            Debug::printDebugArray('Search context', $oCtx);
+            Debug::printDebugArray('Base search', empty($aSearches) ? null : $aSearches[0]);
+            Debug::printVar('Final query phrases', $aInPhrases);
+

             // Convert each phrase to standard form
             // Create a list of standard words
             // Get all 'sets' of words
             // Generate a complete list of all
             // Convert each phrase to standard form
             // Create a list of standard words
             // Get all 'sets' of words
             // Generate a complete list of all

+            Debug::newSection('Tokenization');

             $aTokens = array();
             $aPhrases = array();
             foreach ($aInPhrases as $iPhrase => $sPhrase) {
             $aTokens = array();
             $aPhrases = array();
             foreach ($aInPhrases as $iPhrase => $sPhrase) {


@@ -627,20 +642,22 @@ class Geocode
                 }
             }
 
                 }
             }
 

-            if (sizeof($aTokens)) {
+            Debug::printDebugTable('Phrases', $aPhrases);
+            Debug::printVar('Tokens', $aTokens);
+
+            if (!empty($aTokens)) {

                 // Check which tokens we have, get the ID numbers
                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
                 $sSQL .= ' FROM word ';
                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 
                 // Check which tokens we have, get the ID numbers
                 $sSQL = 'SELECT word_id, word_token, word, class, type, country_code, operator, search_name_count';
                 $sSQL .= ' FROM word ';
                 $sSQL .= ' WHERE word_token in ('.join(',', array_map('getDBQuoted', $aTokens)).')';
 

-                if (CONST_Debug) var_Dump($sSQL);
+                Debug::printSQL($sSQL);

 
                 $aValidTokens = array();
                 $aDatabaseWords = chksql(
                     $this->oDB->getAll($sSQL),
                     'Could not get word tokens.'
                 );
 
                 $aValidTokens = array();
                 $aDatabaseWords = chksql(
                     $this->oDB->getAll($sSQL),
                     'Could not get word tokens.'
                 );

-                $aWordFrequencyScores = array();

                 foreach ($aDatabaseWords as $aToken) {
                     // Filter country tokens that do not match restricted countries.
                     if ($this->aCountryCodes
                 foreach ($aDatabaseWords as $aToken) {
                     // Filter country tokens that do not match restricted countries.
                     if ($this->aCountryCodes


@@ -663,9 +680,7 @@ class Geocode
                     } else {
                         $aValidTokens[$aToken['word_token']] = array($aToken);
                     }
                     } else {
                         $aValidTokens[$aToken['word_token']] = array($aToken);
                     }

-                    $aWordFrequencyScores[$aToken['word_id']] = $aToken['search_name_count'] + 1;

                 }
                 }

-                if (CONST_Debug) var_Dump($aPhrases, $aValidTokens);

 
                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
                 foreach ($aTokens as $sToken) {
 
                 // US ZIP+4 codes - if there is no token, merge in the 5-digit ZIP code
                 foreach ($aTokens as $sToken) {


@@ -690,9 +705,11 @@ class Geocode
                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
                     }
                 }
                         $aValidTokens[' '.$sToken] = array(array('class' => 'place', 'type' => 'house', 'word_token' => ' '.$sToken));
                     }
                 }

+                Debug::printGroupTable('Valid Tokens', $aValidTokens);

 
                 // Any words that have failed completely?
                 // TODO: suggestions
 
                 // Any words that have failed completely?
                 // TODO: suggestions

+                Debug::newSection('Search candidates');

 
                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 
 
                 $aGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, $bStructuredPhrases);
 


@@ -702,8 +719,8 @@ class Geocode
                     // because order in the address doesn't matter.
                     $aPhrases = array_reverse($aPhrases);
                     $aPhrases[0]->invertWordSets();
                     // because order in the address doesn't matter.
                     $aPhrases = array_reverse($aPhrases);
                     $aPhrases[0]->invertWordSets();

-                    if (sizeof($aPhrases) > 1) {
-                        $aPhrases[sizeof($aPhrases)-1]->invertWordSets();
+                    if (count($aPhrases) > 1) {
+                        $aPhrases[count($aPhrases)-1]->invertWordSets();

                     }
                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 
                     }
                     $aReverseGroupedSearches = $this->getGroupedSearches($aSearches, $aPhrases, $aValidTokens, false);
 


@@ -738,7 +755,7 @@ class Geocode
                     $sHash = serialize($aSearch);
                     if (isset($aSearchHash[$sHash])) {
                         unset($aGroupedSearches[$iGroup][$iSearch]);
                     $sHash = serialize($aSearch);
                     if (isset($aSearchHash[$sHash])) {
                         unset($aGroupedSearches[$iGroup][$iSearch]);

-                        if (sizeof($aGroupedSearches[$iGroup]) == 0) unset($aGroupedSearches[$iGroup]);
+                        if (empty($aGroupedSearches[$iGroup])) unset($aGroupedSearches[$iGroup]);

                     } else {
                         $aSearchHash[$sHash] = 1;
                     }
                     } else {
                         $aSearchHash[$sHash] = 1;
                     }


@@ -762,7 +779,6 @@ class Geocode
 
                     $aResults += $oSearch->query(
                         $this->oDB,
 
                     $aResults += $oSearch->query(
                         $this->oDB,

-                        $aWordFrequencyScores,

                         $this->iMinAddressRank,
                         $this->iMaxAddressRank,
                         $this->iLimit
                         $this->iMinAddressRank,
                         $this->iMaxAddressRank,
                         $this->iLimit


@@ -771,7 +787,7 @@ class Geocode
                     if ($iQueryLoop > 20) break;
                 }
 
                     if ($iQueryLoop > 20) break;
                 }
 

-                if (sizeof($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {
+                if (!empty($aResults) && ($this->iMinAddressRank != 0 || $this->iMaxAddressRank != 30)) {

                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
                     // reduces the number of place ids, like a filter
                     // rank_address is 30 for interpolated housenumbers
                     // Need to verify passes rank limits before dropping out of the loop (yuk!)
                     // reduces the number of place ids, like a filter
                     // rank_address is 30 for interpolated housenumbers


@@ -806,7 +822,7 @@ class Geocode
                     $aFilteredIDs = array();
                     if ($aFilterSql) {
                         $sSQL = join(' UNION ', $aFilterSql);
                     $aFilteredIDs = array();
                     if ($aFilterSql) {
                         $sSQL = join(' UNION ', $aFilterSql);

-                        if (CONST_Debug) var_dump($sSQL);
+                        Debug::printSQL($sSQL);

                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
                     }
 
                         $aFilteredIDs = chksql($this->oDB->getCol($sSQL));
                     }
 


@@ -824,7 +840,7 @@ class Geocode
                     $aResults = $tempIDs;
                 }
 
                     $aResults = $tempIDs;
                 }
 

-                if (sizeof($aResults)) break;
+                if (!empty($aResults)) break;

                 if ($iGroupLoop > 4) break;
                 if ($iQueryLoop > 30) break;
             }
                 if ($iGroupLoop > 4) break;
                 if ($iQueryLoop > 30) break;
             }


@@ -835,7 +851,7 @@ class Geocode
 
             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 
 
             $oLookup = $oReverse->lookupPoint($oCtx->sqlNear, false);
 

-            if (CONST_Debug) var_dump('Reverse search', $aLookup);
+            Debug::printVar('Reverse search', $oLookup);

 
             if ($oLookup) {
                 $aResults = array($oLookup->iId => $oLookup);
 
             if ($oLookup) {
                 $aResults = array($oLookup->iId => $oLookup);


@@ -843,7 +859,7 @@ class Geocode
         }
 
         // No results? Done
         }
 
         // No results? Done

-        if (!sizeof($aResults)) {
+        if (empty($aResults)) {

             if ($this->bFallback) {
                 if ($this->fallbackStructuredQuery()) {
                     return $this->lookup();
             if ($this->bFallback) {
                 if ($this->fallbackStructuredQuery()) {
                     return $this->lookup();


@@ -870,10 +886,7 @@ class Geocode
             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
         }
 
             if (!preg_match('/[\pL\pN]/', $sWord)) unset($aRecheckWords[$i]);
         }
 

-        if (CONST_Debug) {
-            echo '<i>Recheck words:<\i>';
-            var_dump($aRecheckWords);
-        }
+        Debug::printVar('Recheck words', $aRecheckWords);

 
         foreach ($aSearchResults as $iIdx => $aResult) {
             // Default
 
         foreach ($aSearchResults as $iIdx => $aResult) {
             // Default


@@ -921,12 +934,12 @@ class Geocode
                 $aResult['importance'] = 0.001;
                 $aResult['foundorder'] = $aResult['addressimportance'];
             } else {
                 $aResult['importance'] = 0.001;
                 $aResult['foundorder'] = $aResult['addressimportance'];
             } else {

-                // Adjust importance for the number of exact string matches in the result
+                $aResult['importance'] = max(0.001, $aResult['importance']);

                 $aResult['importance'] *= $this->viewboxImportanceFactor(
                     $aResult['lon'],
                     $aResult['lat']
                 );
                 $aResult['importance'] *= $this->viewboxImportanceFactor(
                     $aResult['lon'],
                     $aResult['lat']
                 );

-                $aResult['importance'] = max(0.001, $aResult['importance']);
+                // Adjust importance for the number of exact string matches in the result

                 $iCountWords = 0;
                 $sAddress = $aResult['langaddress'];
                 foreach ($aRecheckWords as $i => $sWord) {
                 $iCountWords = 0;
                 $sAddress = $aResult['langaddress'];
                 foreach ($aRecheckWords as $i => $sWord) {


@@ -952,18 +965,16 @@ class Geocode
                     $aResult['foundorder'] += 0.01;
                 }
             }
                     $aResult['foundorder'] += 0.01;
                 }
             }

-            if (CONST_Debug) var_dump($aResult);

             $aSearchResults[$iIdx] = $aResult;
         }
         uasort($aSearchResults, 'byImportance');
             $aSearchResults[$iIdx] = $aResult;
         }
         uasort($aSearchResults, 'byImportance');

+        Debug::printVar('Pre-filter results', $aSearchResults);

 
         $aOSMIDDone = array();
         $aClassTypeNameDone = array();
         $aToFilter = $aSearchResults;
         $aSearchResults = array();
 
 
         $aOSMIDDone = array();
         $aClassTypeNameDone = array();
         $aToFilter = $aSearchResults;
         $aSearchResults = array();
 

-        if (CONST_Debug) var_dump($aToFilter);
-

         $bFirst = true;
         foreach ($aToFilter as $aResult) {
             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];
         $bFirst = true;
         foreach ($aToFilter as $aResult) {
             $this->aExcludePlaceIDs[$aResult['place_id']] = $aResult['place_id'];


@@ -982,10 +993,33 @@ class Geocode
             }
 
             // Absolute limit on number of results
             }
 
             // Absolute limit on number of results

-            if (sizeof($aSearchResults) >= $this->iFinalLimit) break;
+            if (count($aSearchResults) >= $this->iFinalLimit) break;

         }
 
         }
 

-        if (CONST_Debug) var_dump($aSearchResults);
+        Debug::printVar('Post-filter results', $aSearchResults);

         return $aSearchResults;
     } // end lookup()
         return $aSearchResults;
     } // end lookup()

+
+    public function debugInfo()
+    {
+        return array(
+                'Query' => $this->sQuery,
+                'Structured query' => $this->aStructuredQuery,
+                'Name keys' => Debug::fmtArrayVals($this->aLangPrefOrder),
+                'Include address' => $this->bIncludeAddressDetails,
+                'Excluded place IDs' => Debug::fmtArrayVals($this->aExcludePlaceIDs),
+                'Try reversed query'=> $this->bReverseInPlan,
+                'Limit (for searches)' => $this->iLimit,
+                'Limit (for results)'=> $this->iFinalLimit,
+                'Country codes' => Debug::fmtArrayVals($this->aCountryCodes),
+                'Bounded search' => $this->bBoundedSearch,
+                'Viewbox' => Debug::fmtArrayVals($this->aViewBox),
+                'Route points' => Debug::fmtArrayVals($this->aRoutePoints),
+                'Route width' => $this->aRouteWidth,
+                'Max rank' => $this->iMaxRank,
+                'Min address rank' => $this->iMinAddressRank,
+                'Max address rank' => $this->iMaxAddressRank,
+                'Address rank list' => Debug::fmtArrayVals($this->aAddressRankList)
+               );
+    }

 } // end class
 } // end class