X-Git-Url: https://git.openstreetmap.org/nominatim.git/blobdiff_plain/30511fd3ab9d731b0afec202386fc2425a0b1b9f..77abe882ab160669d5206f94426353883750857f:/lib/SearchDescription.php diff --git a/lib/SearchDescription.php b/lib/SearchDescription.php index 7073186b..143d2864 100644 --- a/lib/SearchDescription.php +++ b/lib/SearchDescription.php @@ -43,27 +43,55 @@ class SearchDescription 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; } + /** + * 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; } + /** + * 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 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; @@ -71,22 +99,11 @@ class SearchDescription $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) @@ -94,38 +111,27 @@ class SearchDescription && 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, $sCountryList) - { - if ($this->sCountryCode) { - return $sVar.' = \''.$this->sCountryCode."'"; - } - if ($sCountryList) { - return $sVar.' in ('.$sCountryList.')'; - } - - 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; } + /** + * Extract key/value pairs from a query. + * + * Key/value pairs are recognised if they are of the form [=]. + * 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 [=]. @@ -146,6 +152,13 @@ class SearchDescription return $sQuery; } + /** + * Check if the combination of parameters is sensible. + * + * @param string[] $aCountryCodes List of country codes. + * + * @return bool True, if the search looks valid. + */ public function isValidSearch(&$aCountryCodes) { if (!sizeof($this->aName)) { @@ -166,6 +179,25 @@ class SearchDescription /////////// Search building functions + /** + * Derive new searches by adding a full term to the existing search. + * + * @param mixed[] $aSearchTerm Description of the token. + * @param bool $bWordInQuery True, if the normalised version of the word + * is contained in the query. + * @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, $bWordInQuery, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank) { $aNewSearches = array(); @@ -295,7 +327,18 @@ class SearchDescription 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']) && $aSearchTerm['word_id'])) { @@ -309,7 +352,7 @@ class SearchDescription && sizeof($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] = $iWordID; @@ -352,7 +395,7 @@ class SearchDescription if (preg_match('#^[0-9]+$#', $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] = $iWordID; @@ -367,13 +410,114 @@ class SearchDescription /////////// Query functions - public function queryCountry(&$oDB, $sViewboxSQL) + /** + * 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, $iLimit); + } + } + + if (CONST_Debug) { + echo "
Place IDs: "; + var_Dump($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 = $aFilteredPlaceIDs; + if (CONST_Debug) { + echo "
Place IDs after postcode filtering: "; + var_Dump($aPlaceIDs); + } + } + } + + return array('IDs' => $aPlaceIDs, 'houseNumber' => $iHousenumber); + } + + + private function queryCountry(&$oDB) { $sSQL = 'SELECT place_id FROM placex '; $sSQL .= "WHERE country_code='".$this->sCountryCode."'"; $sSQL .= ' AND rank_search = 4'; - if ($sViewboxSQL) { - $sSQL .= " AND ST_Intersects($sViewboxSQL, geometry)"; + if ($this->oContext->bViewboxBounded) { + $sSQL .= ' AND ST_Intersects('.$this->oContext->sqlViewboxSmall.', geometry)'; } $sSQL .= " ORDER BY st_area(geometry) DESC LIMIT 1"; @@ -382,7 +526,7 @@ class SearchDescription return chksql($oDB->getCol($sSQL)); } - public function queryNearbyPoi(&$oDB, $sCountryList, $sViewboxSQL, $sViewboxCentreSQL, $sExcludeSQL, $iLimit) + private function queryNearbyPoi(&$oDB, $iLimit) { if (!$this->sClass) { return array(); @@ -393,22 +537,21 @@ class SearchDescription $sSQL = 'SELECT count(*) FROM pg_tables WHERE tablename = \''.$sPoiTable."'"; if (chksql($oDB->getOne($sSQL))) { $sSQL = 'SELECT place_id FROM '.$sPoiTable.' ct'; - if ($sCountryList) { + if ($this->oContext->sqlCountryList) { $sSQL .= ' JOIN placex USING (place_id)'; } if ($this->oContext->hasNearPoint()) { $sSQL .= ' WHERE '.$this->oContext->withinSQL('ct.centroid'); - } else { - $sSQL .= " WHERE ST_Contains($sViewboxSQL, ct.centroid)"; - } - if ($sCountryList) { - $sSQL .= " AND country_code in ($sCountryList)"; + } elseif ($this->oContext->bViewboxBounded) { + $sSQL .= ' WHERE ST_Contains('.$this->oContext->sqlViewboxSmall.', ct.centroid)'; } - if ($sExcludeSQL) { - $sSQL .= ' AND place_id not in ('.$sExcludeSQL.')'; + if ($this->oContext->sqlCountryList) { + $sSQL .= ' AND country_code in '.$this->oContext->sqlCountryList; } - if ($sViewboxCentreSQL) { - $sSQL .= " ORDER BY ST_Distance($sViewboxCentreSQL, ct.centroid) ASC"; + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); + if ($this->oContext->sqlViewboxCentre) { + $sSQL .= ' ORDER BY ST_Distance('; + $sSQL .= $this->oContext->sqlViewboxCentre.', ct.centroid) ASC'; } elseif ($this->oContext->hasNearPoint()) { $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('ct.centroid').' ASC'; } @@ -422,8 +565,8 @@ class SearchDescription $sSQL .= 'class=\''.$this->sClass."' and type='".$this->sType."'"; $sSQL .= ' AND '.$this->oContext->withinSQL('geometry'); $sSQL .= ' AND linked_place_id is null'; - if ($sCountryList) { - $sSQL .= " AND country_code in ($sCountryList)"; + if ($this->oContext->sqlCountryList) { + $sSQL .= ' AND country_code in '.$this->oContext->sqlCountryList; } $sSQL .= ' ORDER BY '.$this->oContext->distanceSQL('centroid')." ASC"; $sSQL .= " LIMIT $iLimit"; @@ -434,7 +577,7 @@ class SearchDescription return array(); } - public function queryPostcode(&$oDB, $sCountryList, $iLimit) + private function queryPostcode(&$oDB, $iLimit) { $sSQL = 'SELECT p.place_id FROM location_postcode p '; @@ -448,10 +591,8 @@ class SearchDescription } $sSQL .= "p.postcode = '".reset($this->aName)."'"; - $sCountryTerm = $this->countryCodeSQL('p.country_code', $sCountryList); - if ($sCountryTerm) { - $sSQL .= ' AND '.$sCountryTerm; - } + $sSQL .= $this->countryCodeSQL(' AND p.country_code'); + $sSQL .= $this->oContext->excludeSQL(' AND p.place_id'); $sSQL .= " LIMIT $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -459,7 +600,7 @@ class SearchDescription return chksql($oDB->getCol($sSQL)); } - public function queryNamedPlace(&$oDB, $aWordFrequencyScores, $sCountryList, $iMinAddressRank, $iMaxAddressRank, $sExcludeSQL, $sViewboxSmall, $sViewboxLarge, $iLimit) + private function queryNamedPlace(&$oDB, $aWordFrequencyScores, $iMinAddressRank, $iMaxAddressRank, $iLimit) { $aTerms = array(); $aOrder = array(); @@ -506,7 +647,7 @@ class SearchDescription } } - $sCountryTerm = $this->countryCodeSQL('country_code', $sCountryList); + $sCountryTerm = $this->countryCodeSQL('country_code'); if ($sCountryTerm) { $aTerms[] = $sCountryTerm; } @@ -533,12 +674,13 @@ class SearchDescription } } + $sExcludeSQL = $this->oContext->excludeSQL('place_id'); if ($sExcludeSQL) { - $aTerms[] = 'place_id not in ('.$sExcludeSQL.')'; + $aTerms[] = $sExcludeSQL; } - if ($sViewboxSmall) { - $aTerms[] = 'centroid && '.$sViewboxSmall; + if ($this->oContext->bViewboxBounded) { + $aTerms[] = 'centroid && '.$this->oContext->sqlViewboxSmall; } if ($this->oContext->hasNearPoint()) { @@ -550,12 +692,7 @@ class SearchDescription } else { $sImportanceSQL = '(CASE WHEN importance = 0 OR importance IS NULL THEN 0.75-(search_rank::float/40) ELSE importance END)'; } - if ($sViewboxSmall) { - $sImportanceSQL .= " * CASE WHEN ST_Contains($sViewboxSmall, centroid) THEN 1 ELSE 0.5 END"; - } - if ($sViewboxLarge) { - $sImportanceSQL .= " * CASE WHEN ST_Contains($sViewboxLarge, centroid) THEN 1 ELSE 0.5 END"; - } + $sImportanceSQL .= $this->oContext->viewboxImportanceSQL('centroid'); $aOrder[] = "$sImportanceSQL DESC"; if (sizeof($this->aFullNameAddress)) { @@ -593,8 +730,7 @@ class SearchDescription return array(); } - - public function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $sExcludeSQL, $iLimit) + private function queryHouseNumber(&$oDB, $aRoadPlaceIDs, $iLimit) { $sPlaceIDs = join(',', $aRoadPlaceIDs); @@ -602,9 +738,7 @@ class SearchDescription $sSQL = 'SELECT place_id FROM placex '; $sSQL .= 'WHERE parent_place_id in ('.$sPlaceIDs.')'; $sSQL .= " AND transliteration(housenumber) ~* E'".$sHouseNumberRegex."'"; - if ($sExcludeSQL) { - $sSQL .= ' AND place_id not in ('.$sExcludeSQL.')'; - } + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); $sSQL .= " LIMIT $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -633,10 +767,7 @@ class SearchDescription $sSQL .= " or interpolationtype='all') and "; $sSQL .= $iHousenumber.">=startnumber and "; $sSQL .= $iHousenumber."<=endnumber"; - - if ($sExcludeSQL) { - $sSQL .= ' AND place_id not in ('.$sExcludeSQL.')'; - } + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -653,9 +784,7 @@ class SearchDescription $sSQL = 'SELECT place_id FROM location_property_aux'; $sSQL .= ' WHERE parent_place_id in ('.$sPlaceIDs.')'; $sSQL .= " AND housenumber = '".$this->sHouseNumber."'"; - if ($sExcludeSQL) { - $sSQL .= " AND place_id not in ($sExcludeSQL)"; - } + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -679,10 +808,7 @@ class SearchDescription $sSQL .= " or interpolationtype='all') and "; $sSQL .= $iHousenumber.">=startnumber and "; $sSQL .= $iHousenumber."<=endnumber"; - - if ($sExcludeSQL) { - $sSQL .= ' AND place_id not in ('.$sExcludeSQL.')'; - } + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); $sSQL .= " limit $iLimit"; if (CONST_Debug) var_dump($sSQL); @@ -698,7 +824,7 @@ class SearchDescription } - public function queryPoiByOperator(&$oDB, $aParentIDs, $sExcludeSQL, $iLimit) + private function queryPoiByOperator(&$oDB, $aParentIDs, $iLimit) { $sPlaceIDs = join(',', $aParentIDs); $aClassPlaceIDs = array(); @@ -711,6 +837,7 @@ class SearchDescription $sSQL .= " AND class='".$this->sClass."' "; $sSQL .= " AND type='".$this->sType."'"; $sSQL .= " AND linked_place_id is null"; + $sSQL .= $this->oContext->excludeSQL(' AND place_id'); $sSQL .= " ORDER BY rank_search ASC "; $sSQL .= " LIMIT $iLimit"; @@ -787,9 +914,7 @@ class SearchDescription $sSQL .= " WHERE ST_Contains('$sPlaceGeom', l.centroid)"; } - if ($sExcludeSQL) { - $sSQL .= ' AND l.place_id not in ('.$sExcludeSQL.')'; - } + $sSQL .= $this->oContext->excludeSQL(' AND l.place_id'); $sSQL .= 'limit 300) i '; if ($sOrderBySQL) { $sSQL .= 'order by order_term asc'; @@ -820,9 +945,7 @@ class SearchDescription $sSQL .= " AND ST_DWithin(l.geometry, f.centroid, $fRange)"; $sSQL .= " AND l.class='".$this->sClass."'"; $sSQL .= " AND l.type='".$this->sType."'"; - if ($sExcludeSQL) { - $sSQL .= " AND l.place_id not in (".$sExcludeSQL.")"; - } + $sSQL .= $this->oContext->excludeSQL(' AND l.place_id'); if ($sOrderBySQL) { $sSQL .= "ORDER BY orderterm ASC"; } @@ -838,6 +961,22 @@ class SearchDescription 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