foreach ($aPhrases as $iPhrase => $aPhrase) {
$aNewPhraseSearches = array();
- if ($bStructuredPhrases) $sPhraseType = $aPhraseTypes[$iPhrase];
- else $sPhraseType = '';
+ if ($bStructuredPhrases) {
+ $sPhraseType = $aPhraseTypes[$iPhrase];
+ } else {
+ $sPhraseType = '';
+ }
foreach ($aPhrase['wordsets'] as $iWordSet => $aWordset) {
// Too many permutations - too expensive
//echo "<br><b>$sToken</b>";
$aNewWordsetSearches = array();
- foreach ($aWordsetSearches as $aCurrentSearch) {
+ foreach ($aWordsetSearches as $oCurrentSearch) {
//echo "<i>";
- //var_dump($aCurrentSearch);
+ //var_dump($oCurrentSearch);
//echo "</i>";
// If the token is valid
if (isset($aValidTokens[' '.$sToken])) {
- // TODO variable should go into aCurrentSearch
- $bHavePostcode = false;
+ // Recheck if the original word shows up in the query.
+ $bWordInQuery = false;
+ if (isset($aSearchTerm['word']) && $aSearchTerm['word']) {
+ $bWordInQuery = $this->normTerm($aSearchTerm['word'])) !== false;
+ }
foreach ($aValidTokens[' '.$sToken] as $aSearchTerm) {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank']++;
- if (($sPhraseType == '' || $sPhraseType == 'country') && !empty($aSearchTerm['country_code']) && $aSearchTerm['country_code'] != '0') {
- if ($aSearch['sCountryCode'] === false) {
- $aSearch['sCountryCode'] = strtolower($aSearchTerm['country_code']);
- // Country is almost always at the end of the string - increase score for finding it anywhere else (optimisation)
- if (($iToken+1 != sizeof($aWordset) || $iPhrase+1 != sizeof($aPhrases))) {
- $aSearch['iSearchRank'] += 5;
- }
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- // If it is at the beginning, we can be almost sure that this is the wrong order
- // Increase score for all searches.
- if ($iToken == 0 && $iPhrase == 0) {
- $iGlobalRank++;
- }
- }
- } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode') {
- // We need to try the case where the postal code is the primary element (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode) so try both
- if ($aSearch['sPostcode'] === '' &&
- isset($aSearchTerm['word']) && $aSearchTerm['word'] && strpos($sNormQuery, $this->normTerm($aSearchTerm['word'])) !== false) {
- // If we have structured search or this is the first term,
- // make the postcode the primary search element.
- if (!$bHavePostcode && $aSearch['sOperator'] === '' && ($sPhraseType == 'postalcode' || ($iToken == 0 && $iPhrase == 0))) {
- $aNewSearch = $aSearch;
- $aNewSearch['sOperator'] = 'postcode';
- $aNewSearch['aAddress'] = array_merge($aNewSearch['aAddress'], $aNewSearch['aName']);
- $aNewSearch['aName'] = array($aSearchTerm['word_id'] => $aSearchTerm['word']);
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aNewSearch;
- $bHavePostcode = true;
- }
-
- // If we have a structured search or this is not the first term,
- // add the postcode as an addendum.
- if ($aSearch['sOperator'] !== 'postcode' && ($sPhraseType == 'postalcode' || sizeof($aSearch['aName']))) {
- $aSearch['sPostcode'] = $aSearchTerm['word'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- } elseif (($sPhraseType == '' || $sPhraseType == 'street') && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house') {
- if ($aSearch['sHouseNumber'] === '' && $aSearch['sOperator'] !== 'postcode') {
- $aSearch['sHouseNumber'] = $sToken;
- // sanity check: if the housenumber is not mainly made
- // up of numbers, add a penalty
- if (preg_match_all("/[^0-9]/", $sToken, $aMatches) > 2) $aSearch['iSearchRank']++;
- // also must not appear in the middle of the address
- if ($aSearch['aAddress'] || $aSearch['aAddressNonSearch']) $aSearch['iSearchRank'] += 1;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- /*
- // Fall back to not searching for this item (better than nothing)
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- */
- }
- } elseif ($sPhraseType == '' && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null) {
- // require a normalized exact match of the term
- // if we have the normalizer version of the query
- // available
- if ($aSearch['sOperator'] === ''
- && ($sNormQuery === null || !($aSearchTerm['word'] && strpos($sNormQuery, $aSearchTerm['word']) === false))) {
- $aSearch['sClass'] = $aSearchTerm['class'];
- $aSearch['sType'] = $aSearchTerm['type'];
- if ($aSearchTerm['operator'] == '') {
- $aSearch['sOperator'] = sizeof($aSearch['aName']) ? 'name' : 'near';
- $aSearch['iSearchRank'] += 2;
- } else {
- $aSearch['sOperator'] = 'near'; // near = in for the moment
- }
-
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- } elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
- if (sizeof($aSearch['aName'])) {
- if ((!$bStructuredPhrases || $iPhrase > 0) && $sPhraseType != 'country' && (!isset($aValidTokens[$sToken]) || strpos($sToken, ' ') !== false)) {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- } else {
- $aCurrentSearch['aFullNameAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1000; // skip;
- }
- } else {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- //$aSearch['iNamePhrase'] = $iPhrase;
+ $aNewSearches = $oCurrentSearch->extendWithFullTerm(
+ $aSearchTerm,
+ $bWordInQuery,
+ isset($aValidTokens[$sToken])
+ && strpos($sToken, ' ') === false,
+ $sPhraseType,
+ $iToken == 0 && $iPhrase == 0,
+ $iPhrase == 0,
+ $iToken + 1 == sizeof($aWordset)
+ && $iPhrase + 1 == sizeof($aPhrases),
+ $iGlobalRank
+ );
+
+ foreach ($aNewSearches as $oSearch) {
+ if ($oSearch->getRank() < $this->iMaxRank) {
+ $aNewWordsetSearches[] = $oSearch;
}
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
}
}
}
// Look for partial matches.
// Note that there is no point in adding country terms here
- // because country are omitted in the address.
+ // because country is omitted in the address.
if (isset($aValidTokens[$sToken]) && $sPhraseType != 'country') {
// Allow searching for a word - but at extra cost
foreach ($aValidTokens[$sToken] as $aSearchTerm) {
- if (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
- if ((!$bStructuredPhrases || $iPhrase > 0) && sizeof($aCurrentSearch['aName']) && strpos($sToken, ' ') === false) {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) {
- $aSearch['aAddress'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- } elseif (isset($aValidTokens[' '.$sToken])) { // revert to the token version?
- $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- $aSearch['iSearchRank'] += 1;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- foreach ($aValidTokens[' '.$sToken] as $aSearchTermToken) {
- if (empty($aSearchTermToken['country_code'])
- && empty($aSearchTermToken['lat'])
- && empty($aSearchTermToken['class'])
- ) {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- $aSearch['aAddress'][$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
- } else {
- $aSearch['aAddressNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
- }
- }
-
- if ((!$aCurrentSearch['sPostcode'] && !$aCurrentSearch['aAddress'] && !$aCurrentSearch['aAddressNonSearch'])
- && (!sizeof($aCurrentSearch['aName']) || $aCurrentSearch['iNamePhrase'] == $iPhrase)) {
- $aSearch = $aCurrentSearch;
- $aSearch['iSearchRank'] += 1;
- if (!sizeof($aCurrentSearch['aName'])) $aSearch['iSearchRank'] += 1;
- if (preg_match('#^[0-9]+$#', $sToken)) $aSearch['iSearchRank'] += 2;
- if ($aWordFrequencyScores[$aSearchTerm['word_id']] < CONST_Max_Word_Frequency) {
- $aSearch['aName'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- } else {
- $aSearch['aNameNonSearch'][$aSearchTerm['word_id']] = $aSearchTerm['word_id'];
- }
- $aSearch['iNamePhrase'] = $iPhrase;
- if ($aSearch['iSearchRank'] < $this->iMaxRank) $aNewWordsetSearches[] = $aSearch;
+ $aNewSearches = $oCurrentSearch->extendWithPartialTerm(
+ $aSearchTerm,
+ $bStructuredPhrases,
+ $iPhrase,
+ $aWordFrequencyScores,
+ isset($aValidTokens[' '.$sToken]) ? $aValidTokens[' '.$sToken] : array()
+ );
+
+ foreach ($aNewSearches as $oSearch) {
+ if ($oSearch->getRank() < $this->iMaxRank) {
+ $aNewWordsetSearches[] = $oSearch;
}
}
+
}
- } else {
- // Allow skipping a word - but at EXTREAM cost
- //$aSearch = $aCurrentSearch;
- //$aSearch['iSearchRank']+=100;
- //$aNewWordsetSearches[] = $aSearch;
}
}
// Sort and cut
// Re-group the searches by their score, junk anything over 20 as just not worth trying
$aGroupedSearches = array();
foreach ($aNewPhraseSearches as $aSearch) {
- if ($aSearch['iSearchRank'] < $this->iMaxRank) {
- if (!isset($aGroupedSearches[$aSearch['iSearchRank']])) $aGroupedSearches[$aSearch['iSearchRank']] = array();
- $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+ $iRank = $aSearch->getRank();
+ if ($iRank < $this->iMaxRank) {
+ if (!isset($aGroupedSearches[$iRank])) {
+ $aGroupedSearches[$iRank] = array();
+ }
+ $aGroupedSearches[$iRank][] = $aSearch;
}
}
ksort($aGroupedSearches);
// Revisit searches, drop bad searches and give penalty to unlikely combinations.
$aGroupedSearches = array();
- foreach ($aSearches as $aSearch) {
- if (!$aSearch['aName']) {
- if ($aSearch['sHouseNumber']) {
- continue;
- }
- }
- if ($this->aCountryCodes && $aSearch['sCountryCode']
- && !in_array($aSearch['sCountryCode'], $this->aCountryCodes)) {
+ foreach ($aSearches as $oSearch) {
+ if (!$oSearch->isValidSearch()) {
continue;
}
- $aSearch['iSearchRank'] += $iGlobalRank;
- $aGroupedSearches[$aSearch['iSearchRank']][] = $aSearch;
+ $iRank = $oSearch->addToRank($iGlobalRank);
+ if (!isset($aGroupedSearches[$iRank]) {
+ $aGroupedSearches[$iRank] = array();
+ }
+ $aGroupedSearches[$iRank][] = $oSearch;
}
ksort($aGroupedSearches);
return $this->iSearchRank;
}
+ public function addToRank($iAddRank)
+ {
+ $this->iSearchRank += $iAddRank;
+ return $this->iSearchRank;
+ }
+
public function getPostCode()
{
return $this->sPostcode;
return $sQuery;
}
+ public function isValidSearch(&$aCountryCodes)
+ {
+ if (!sizeof($this->aName)) {
+ if ($this->sHouseNumber) {
+ return false;
+ }
+ }
+ if ($aCountryCodes
+ && $this->sCounrtyCode
+ && !in_array($this->sCountryCode, $aCountryCodes)
+ ) {
+ return false;
+ }
+
+ return true;
+ }
+
+ /////////// Search building functions
+
+ public function extendWithFullTerm($aSearchTerm, $bWordInQuery, $bHasPartial, $sPhraseType, $bFirstToken, $bFirstPhrase, $bLastToken, &$iGlobalRank)
+ {
+ $aNewSearches = array();
+
+ if (($sPhraseType == '' || $sPhraseType == 'country')
+ && !empty($aSearchTerm['country_code'])
+ && $aSearchTerm['country_code'] != '0'
+ ) {
+ if (!$this->sCountryCode) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->sCountryCode = $aSearchTerm['country_code'];
+ // Country is almost always at the end of the string
+ // - increase score for finding it anywhere else (optimisation)
+ if (!$bLastToken) {
+ $oSearch->iSearchRank += 5;
+ }
+ $aNewSearches[] = $oSearch;
+
+ // If it is at the beginning, we can be almost sure that
+ // the terms are in the wrong order. Increase score for all searches.
+ if ($bFirstToken) {
+ $iGlobalRank++;
+ }
+ }
+ } elseif (($sPhraseType == '' || $sPhraseType == 'postalcode')
+ && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'postcode'
+ ) {
+ // We need to try the case where the postal code is the primary element
+ // (i.e. no way to tell if it is (postalcode, city) OR (city, postalcode)
+ // so try both.
+ if (!$this->sPostcode && $bWordInQuery) {
+ // If we have structured search or this is the first term,
+ // make the postcode the primary search element.
+ if ($this->iOperator == Operator::NONE
+ && ($sPhraseType == 'postalcode' || $bFirstToken)
+ ) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->iOperator = Operator::POSTCODE;
+ $oSearch->aAddress = array_merge($this->aAddress, $this->aName);
+ $oSearch->aName =
+ array($aSearchTerm['word_id'] => $aSearchTerm['word']);
+ $aNewSearches[] = $oSearch;
+ }
+
+ // If we have a structured search or this is not the first term,
+ // add the postcode as an addendum.
+ if ($this->iOperator != Operator::POSTCODE
+ && ($sPhraseType == 'postalcode' || sizeof($this->aName))
+ ) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->sPostcode = $aSearchTerm['word'];
+ $aNewSearches[] = $oSearch;
+ }
+ }
+ } elseif (($sPhraseType == '' || $sPhraseType == 'street')
+ && $aSearchTerm['class'] == 'place' && $aSearchTerm['type'] == 'house'
+ ) {
+ if (!$this->sHouseNumber && $this->iOperator != Operator::POSTCODE) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->sHouseNumber = trim($aSearchTerm['word_token']);
+ // sanity check: if the housenumber is not mainly made
+ // up of numbers, add a penalty
+ if (preg_match_all("/[^0-9]/", $oSearch->sHouseNumber, $aMatches) > 2) {
+ $oSearch->iSearchRank++;
+ }
+ // also must not appear in the middle of the address
+ if (sizeof($this->aAddress) || sizeof($this->aAddressNonSearch)) {
+ $oSearch->iSearchRank++;
+ }
+ $aNewSearches[] = $oSearch;
+ }
+ } elseif ($sPhraseType == ''
+ && $aSearchTerm['class'] !== '' && $aSearchTerm['class'] !== null
+ ) {
+ // require a normalized exact match of the term
+ // if we have the normalizer version of the query
+ // available
+ if ($this->iOperator == Operator::NONE
+ && (isset($aSearchTerm['word']) && $aSearchTerm['word'])
+ && $bWordInQuery
+ ) {
+ $oSearch = clone this;
+ $oSearch->iSearchRank++;
+
+ $iOp = Operator::NEAR; // near == in for the moment
+ if ($aSearchTerm['operator'] == '') {
+ if (sizeof($this->aName)) {
+ $iOp = Operator::NAME;
+ }
+ $oSearch->iSearchRank += 2;
+ }
+
+ $oSearch->setPoiSearch($iOp, $aSearchTerm['class'], $aSearchTerm['type']);
+ $aNewWordsetSearches[] = $oSearch;
+ }
+ } elseif (isset($aSearchTerm['word_id']) && $aSearchTerm['word_id']) {
+ $iWordID = $aSearchTerm['word_id'];
+ if (sizeof($this->aName)) {
+ if (($sPhraseType == '' || !$bFirstPhrase)
+ && $sPhraseType != 'country'))
+ && !$bHasPartial
+ ) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->aAddress[$iWordID] = $iWordID;
+ );
+ $aNewSearches[] = $oSearch;
+ }
+ else {
+ $this->aFullNameAddress[$iWordID] = $iWordID;
+ }
+ } else {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->aName = array($iWordID => $iWordID);
+ $aNewSearches[] = $oSearch;
+ }
+ }
+
+ return $aNewSearches;
+ }
+
+ public function extendWithPartialTerm($aSearchTerm, $bStructuredPhrases, $iPhrase, &$aWordFrequencyScores, $aFullTokens)
+ {
+ // Only allow name terms.
+ if (!(isset($aSearchTerm['word_id']) && $aSearchTerm['word_id'])) {
+ return array();
+ }
+
+ $aNewSearches = array();
+ $iWordID = $aSearchTerm['word_id'];
+
+ if ((!$bStructuredPhrases || $iPhrase > 0)
+ && sizeof($this->aName)
+ && strpos($aSearchTerm['word_token'], ' ') === false
+ ) {
+ if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
+ $oSearch = clone this;
+ $oSearch->iSearchRank++;
+ $oSearch->aAddress[$iWordID] = $iWordID;
+ $aNewSearches[] = $oSearch;
+ } else {
+ $oSearch = clone this;
+ $oSearch->iSearchRank++;
+ $oSearch->aAddressNonSearch[$iWordID] = $iWordID;
+ if (preg_match('#^[0-9]+$#', $aSearchTerm['word_token'])) {
+ $oSearch->iSearchRank += 2;
+ }
+ if (sizeof($aFullTokens) {
+ $oSearch->iSearchRank++;
+ }
+ $aNewSearches[] = $oSearch;
+
+ // revert to the token version?
+ foreach ($aFullTokens as $aSearchTermToken) {
+ if (empty($aSearchTermToken['country_code'])
+ && empty($aSearchTermToken['lat'])
+ && empty($aSearchTermToken['class'])
+ ) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ $oSearch->aAddress[$aSearchTermToken['word_id']] = $aSearchTermToken['word_id'];
+ $aNewSearches[] = $oSearch;
+ }
+ }
+ }
+ }
+
+ if ((!$this->sPostcode && !$this->aAddress && !$this->aAddressNonSearch)
+ && (!sizeof($this->aName) || $this->iNamePhrase == $iPhrase)
+ ) {
+ $oSearch = clone $this;
+ $oSearch->iSearchRank++;
+ if (!sizeof($this->aName)) {
+ $aSearch->iSearchRank += 1;
+ }
+ if (preg_match('#^[0-9]+$#', $sSerchTerm['word_token')) {
+ $oSearch->iSearchRank += 2;
+ }
+ if ($aWordFrequencyScores[$iWordID] < CONST_Max_Word_Frequency) {
+ $oSearch->aName[$iWordID] = $iWordID;
+ } else {
+ $aSearch->aNameNonSearch[$iWordID] = $iWordID;
+ }
+ $oSearch->iNamePhrase = $iPhrase;
+ $aNewSearches[] = $aSearch;
+ }
+
+ return $aNewSearches;
+ }
+
+ /////////// Query functions
+
public function queryCountry(&$oDB, $sViewboxSQL)
{
$sSQL = 'SELECT place_id FROM placex ';
projects / nominatim.git / commitdiff