Merge remote-tracking branch 'origin/master' into osmline

Conflicts:
	lib/Geocode.php
	lib/PlaceLookup.php
	sql/functions.sql
	sql/tables.sql
	utils/setup.php
This commit is contained in:
Markus Gail
2016-04-25 11:01:04 +02:00
26 changed files with 391 additions and 10198 deletions

0
NEWS
View File

File diff suppressed because one or more lines are too long

File diff suppressed because one or more lines are too long

View File

@@ -228,11 +228,9 @@
function loadParamArray($aParams)
{
if (isset($aParams['addressdetails'])) $this->bIncludeAddressDetails = (bool)$aParams['addressdetails'];
if ((float) CONST_Postgresql_Version > 9.2)
{
if (isset($aParams['extratags'])) $this->bIncludeExtraTags = (bool)$aParams['extratags'];
if (isset($aParams['namedetails'])) $this->bIncludeNameDetails = (bool)$aParams['namedetails'];
}
if (isset($aParams['extratags'])) $this->bIncludeExtraTags = (bool)$aParams['extratags'];
if (isset($aParams['namedetails'])) $this->bIncludeNameDetails = (bool)$aParams['namedetails'];
if (isset($aParams['bounded'])) $this->bBoundedSearch = (bool)$aParams['bounded'];
if (isset($aParams['dedupe'])) $this->bDeDupe = (bool)$aParams['dedupe'];
@@ -433,18 +431,41 @@
if (30 >= $this->iMinAddressRank && 30 <= $this->iMaxAddressRank)
{
//query also location_property_osmline and location_property_tiger and location_property_aux
//Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
//only Tiger housenumbers need to be interpolated, because they are saved as lines with start- and endnumber, the common osm housenumbers are usually saved as points
$sHousenumbers = "";
$i = 0;
$length = count($aPlaceIDs);
foreach($aPlaceIDs as $placeID => $housenumber)
if (CONST_Use_US_Tiger_Data)
{
$i++;
$sHousenumbers .= "(".$placeID.", ".$housenumber.")";
if($i<$length)
$sHousenumbers .= ", ";
//query also location_property_tiger and location_property_aux
//Tiger search only if a housenumber was searched and if it was found (i.e. aPlaceIDs[placeID] = housenumber != -1) (realized through a join)
//only Tiger housenumbers need to be interpolated, because they are saved as lines with start- and endnumber, the common osm housenumbers are usually saved as points
$sHousenumbers = "";
$i = 0;
$length = count($aPlaceIDs);
foreach($aPlaceIDs as $placeID => $housenumber)
{
$i++;
$sHousenumbers .= "(".$placeID.", ".$housenumber.")";
if($i<$length)
$sHousenumbers .= ", ";
}
$sSQL .= "union ";
$sSQL .= "select 'T' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 30 as rank_search, 30 as rank_address, min(place_id) as place_id, min(parent_place_id) as parent_place_id, 'us' as country_code";
$sSQL .= ", get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) as langaddress ";
$sSQL .= ", null as placename";
$sSQL .= ", null as ref";
if ($this->bIncludeExtraTags) $sSQL .= ", null as extra";
if ($this->bIncludeNameDetails) $sSQL .= ", null as names";
$sSQL .= ", avg(st_x(centroid)) as lon, avg(st_y(centroid)) as lat,";
$sSQL .= $sImportanceSQL."-1.15 as importance ";
$sSQL .= ", (select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(blub.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";
$sSQL .= ", null as extra_place ";
$sSQL .= " from (select place_id";
//interpolate the Tiger housenumbers here
$sSQL .= ", ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) as centroid, parent_place_id, housenumber_for_place ";
$sSQL .= "from (location_property_tiger ";
$sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id, housenumber_for_place) using(place_id)) ";
$sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here
$sSQL .= " group by place_id, housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
if (!$this->bDeDupe) $sSQL .= ", place_id ";
}
// osmline, osm_type is 'I' for Interpolation Line
$sSQL .= " union ";
@@ -469,46 +490,25 @@
$sSQL .= " group by place_id, housenumber_for_place, calculated_country_code "; //is this group by really needed?, place_id + housenumber (in combination) are unique
if (!$this->bDeDupe) $sSQL .= ", place_id ";
// tiger
$sSQL .= " union ";
$sSQL .= "select 'T' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 30 as rank_search, 30 as rank_address, min(place_id) as place_id, min(parent_place_id) as parent_place_id, 'us' as country_code";
$sSQL .= ", get_address_by_language(place_id, housenumber_for_place, $sLanguagePrefArraySQL) as langaddress ";
$sSQL .= ", null as placename";
$sSQL .= ", null as ref";
if ($this->bIncludeExtraTags) $sSQL .= ", null as extra";
if ($this->bIncludeNameDetails) $sSQL .= ", null as names";
$sSQL .= ", avg(st_x(centroid)) as lon, avg(st_y(centroid)) as lat,";
$sSQL .= $sImportanceSQL."-1.15 as importance ";
$sSQL .= ", (select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(blub.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance ";
$sSQL .= ", null as extra_place ";
$sSQL .= " from (select place_id";
//interpolate the Tiger housenumbers here
$sSQL .= ", CASE WHEN startnumber != endnumber THEN ST_LineInterpolatePoint(linegeo, (housenumber_for_place-startnumber::float)/(endnumber-startnumber)::float) ";
$sSQL .= " ELSE ST_LineInterpolatePoint(linegeo, 0.5) END as centroid";
$sSQL .= ", parent_place_id, housenumber_for_place ";
$sSQL .= " from (location_property_tiger ";
$sSQL .= " join (values ".$sHousenumbers.") as housenumbers(place_id, housenumber_for_place) using(place_id)) ";
$sSQL .= " where housenumber_for_place>=0 and 30 between $this->iMinAddressRank and $this->iMaxAddressRank) as blub"; //postgres wants an alias here
$sSQL .= " group by place_id, housenumber_for_place"; //is this group by really needed?, place_id + housenumber (in combination) are unique
if (!$this->bDeDupe) $sSQL .= ", place_id ";
// aux
$sSQL .= " union ";
$sSQL .= "select 'L' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 0 as rank_search, 0 as rank_address, min(place_id) as place_id, min(parent_place_id) as parent_place_id, 'us' as country_code, ";
$sSQL .= "get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) as langaddress, ";
$sSQL .= "null as placename, ";
$sSQL .= "null as ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "null as extra, ";
if ($this->bIncludeNameDetails) $sSQL .= "null as names, ";
$sSQL .= "avg(ST_X(centroid)) as lon, avg(ST_Y(centroid)) as lat, ";
$sSQL .= $sImportanceSQL."-1.10 as importance, ";
$sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, ";
$sSQL .= "null as extra_place ";
$sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
$sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
$sSQL .= "group by place_id";
if (!$this->bDeDupe) $sSQL .= ", place_id";
$sSQL .= ", get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
if (CONST_Use_Aux_Location_data)
{
$sSQL .= " union ";
$sSQL .= "select 'L' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, 0 as rank_search, 0 as rank_address, min(place_id) as place_id, min(parent_place_id) as parent_place_id, 'us' as country_code, ";
$sSQL .= "get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) as langaddress, ";
$sSQL .= "null as placename, ";
$sSQL .= "null as ref, ";
if ($this->bIncludeExtraTags) $sSQL .= "null as extra, ";
if ($this->bIncludeNameDetails) $sSQL .= "null as names, ";
$sSQL .= "avg(ST_X(centroid)) as lon, avg(ST_Y(centroid)) as lat, ";
$sSQL .= $sImportanceSQL."-1.10 as importance, ";
$sSQL .= "(select max(p.importance*(p.rank_address+2)) from place_addressline s, placex p where s.place_id = min(location_property_aux.parent_place_id) and p.place_id = s.address_place_id and s.isaddress and p.importance is not null) as addressimportance, ";
$sSQL .= "null as extra_place ";
$sSQL .= "from location_property_aux where place_id in ($sPlaceIDs) ";
$sSQL .= "and 30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
$sSQL .= "group by place_id";
if (!$this->bDeDupe) $sSQL .= ", place_id";
$sSQL .= ", get_address_by_language(place_id, -1, $sLanguagePrefArraySQL) ";
}
}
$sSQL .= " order by importance desc";
@@ -1487,7 +1487,7 @@
}
// If nothing found try the aux fallback table
if (!sizeof($aPlaceIDs))
if (CONST_Use_Aux_Location_data && !sizeof($aPlaceIDs))
{
$sSQL = "select place_id from location_property_aux where parent_place_id in (".$sPlaceIDs.") and housenumber = '".pg_escape_string($aSearch['sHouseNumber'])."'";
if (sizeof($this->aExcludePlaceIDs))
@@ -1498,8 +1498,10 @@
if (CONST_Debug) var_dump($sSQL);
$aPlaceIDs = $this->oDB->getCol($sSQL);
}
//if nothing found search in Tiger data for this housenumber(location_property_tiger)
if (!sizeof($aPlaceIDs))
//if nothing was found in placex or location_property_aux, then search in Tiger data for this housenumber(location_property_tiger)
$searchedHousenumber = intval($aSearch['sHouseNumber']);
if (CONST_Use_US_Tiger_Data && !sizeof($aPlaceIDs))
{
//new query for lines, not housenumbers anymore
if($searchedHousenumber%2 == 0){
@@ -1672,11 +1674,13 @@
$sSQL .= "and (placex.rank_address between $this->iMinAddressRank and $this->iMaxAddressRank ";
if (14 >= $this->iMinAddressRank && 14 <= $this->iMaxAddressRank) $sSQL .= " OR (extratags->'place') = 'city'";
if ($this->aAddressRankList) $sSQL .= " OR placex.rank_address in (".join(',',$this->aAddressRankList).")";
$sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";
$sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',',$this->aAddressRankList).")";
$sSQL .= ") UNION select place_id from location_property_osmline where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";
$sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank)";
if (CONST_Use_US_Tiger_Data)
{
$sSQL .= ") UNION select place_id from location_property_tiger where place_id in (".join(',',array_keys($aResultPlaceIDs)).") ";
$sSQL .= "and (30 between $this->iMinAddressRank and $this->iMaxAddressRank ";
if ($this->aAddressRankList) $sSQL .= " OR 30 in (".join(',',$this->aAddressRankList).")";
}
$sSQL .= ")";
if (CONST_Debug) var_dump($sSQL);
$aFilteredPlaceIDs = $this->oDB->getCol($sSQL);
$tempIDs = array();

View File

@@ -42,18 +42,12 @@
function setIncludeExtraTags($bExtraTags = false)
{
if ((float) CONST_Postgresql_Version > 9.2)
{
$this->bExtraTags = $bExtraTags;
}
$this->bExtraTags = $bExtraTags;
}
function setIncludeNameDetails($bNameDetails = false)
{
if ((float) CONST_Postgresql_Version > 9.2)
{
$this->bNameDetails = $bNameDetails;
}
$this->bNameDetails = $bNameDetails;
}
@@ -128,7 +122,7 @@
$sLanguagePrefArraySQL = "ARRAY[".join(',',array_map("getDBQuoted", $this->aLangPrefOrder))."]";
if ($this->sType == 'tiger')
if (CONST_Use_US_Tiger_Data && $this->sType == 'tiger')
{
$sSQL = "select place_id,partition, 'T' as osm_type, place_id as osm_id, 'place' as class, 'house' as type, null as admin_level, housenumber, null as street, null as isin, postcode,";
$sSQL .= " 'us' as country_code, parent_place_id, null as linked_place_id, 30 as rank_address, 30 as rank_search,";
@@ -193,7 +187,7 @@
if ($this->bAddressDetails)
{
if ($this->sType == 'tiger' || $this->sType == 'interpolation') // to get addressdetails for interpolation lines and tiger data, the housenumber is needed
if(CONST_Use_US_Tiger_Data && $this->sType == 'tiger') // to get addressdetails for tiger data, the housenumber is needed
$aAddress = $this->getAddressNames($aPlace['housenumber']);
else
$aAddress = $this->getAddressNames();

View File

@@ -259,7 +259,7 @@
}
// Only street found? If it's in the US we can check TIGER data for nearest housenumber
if ($bIsInUnitedStates && $iMaxRank_orig >= 28 && $iPlaceID && ($aPlace['rank_search'] == 26 || $aPlace['rank_search'] == 27 ))
if (CONST_Use_US_Tiger_Data && $bIsInUnitedStates && $iMaxRank_orig >= 28 && $iPlaceID && ($aPlace['rank_search'] == 26 || $aPlace['rank_search'] == 27 ))
{
$fSearchDiam = 0.001;
$sSQL = 'SELECT place_id,parent_place_id,30 as rank_search, ST_line_locate_point(linegeo,'.$sPointSQL.') as fraction';

View File

@@ -24,3 +24,16 @@
return "'".pg_escape_string($s)."'";
}
function getPostgresVersion(&$oDB)
{
$sVersionString = $oDB->getOne('select version()');
preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches);
return (float) ($aMatches[1].'.'.$aMatches[2]);
}
function getPostgisVersion(&$oDB)
{
$sVersionString = $oDB->getOne('select postgis_full_version()');
preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
return (float) ($aMatches[1].'.'.$aMatches[2]);
}

View File

@@ -12,12 +12,6 @@
}
if ($_SERVER['REQUEST_METHOD'] == 'OPTIONS') exit;
if (CONST_ClosedForIndexing && strpos(CONST_ClosedForIndexingExceptionIPs, ','.$_SERVER["REMOTE_ADDR"].',') === false)
{
echo "Closed for re-indexing...";
exit;
}
$aBucketKeys = array();
if (isset($_SERVER["HTTP_REFERER"])) $aBucketKeys[] = str_replace('www.','',strtolower(parse_url($_SERVER["HTTP_REFERER"], PHP_URL_HOST)));

View File

@@ -55,17 +55,6 @@
}
function getBlockingProcesses()
{
$sStats = file_get_contents('/proc/stat');
if (preg_match('/procs_blocked ([0-9]+)/i', $sStats, $aMatches))
{
return (int)$aMatches[1];
}
return 0;
}
function getLoadAverage()
{
$sLoadAverage = file_get_contents('/proc/loadavg');
@@ -110,16 +99,7 @@
{
if ($a['importance'] != $b['importance'])
return ($a['importance'] > $b['importance']?-1:1);
/*
if ($a['aPointPolygon']['numfeatures'] != $b['aPointPolygon']['numfeatures'])
return ($a['aPointPolygon']['numfeatures'] > $b['aPointPolygon']['numfeatures']?-1:1);
if ($a['aPointPolygon']['area'] != $b['aPointPolygon']['area'])
return ($a['aPointPolygon']['area'] > $b['aPointPolygon']['area']?-1:1);
// if ($a['levenshtein'] != $b['levenshtein'])
// return ($a['levenshtein'] < $b['levenshtein']?-1:1);
if ($a['rank_search'] != $b['rank_search'])
return ($a['rank_search'] < $b['rank_search']?-1:1);
*/
return ($a['foundorder'] < $b['foundorder']?-1:1);
}
@@ -229,7 +209,6 @@
{
$aTokens[' '.$sWord] = ' '.$sWord;
$aTokens[$sWord] = $sWord;
//if (!strpos($sWord,' ')) $aTokens[$sWord] = $sWord;
}
}
return $aTokens;
@@ -240,34 +219,6 @@
GB Postcode functions
*/
function gbPostcodeAlphaDifference($s1, $s2)
{
$aValues = array(
'A'=>0,
'B'=>1,
'D'=>2,
'E'=>3,
'F'=>4,
'G'=>5,
'H'=>6,
'J'=>7,
'L'=>8,
'N'=>9,
'O'=>10,
'P'=>11,
'Q'=>12,
'R'=>13,
'S'=>14,
'T'=>15,
'U'=>16,
'W'=>17,
'X'=>18,
'Y'=>19,
'Z'=>20);
return abs(($aValues[$s1[0]]*21+$aValues[$s1[1]]) - ($aValues[$s2[0]]*21+$aValues[$s2[1]]));
}
function gbPostcodeCalculate($sPostcode, $sPostcodeSector, $sPostcodeEnd, &$oDB)
{
// Try an exact match on the gb_postcode table
@@ -294,66 +245,6 @@
}
function usPostcodeCalculate($sPostcode, &$oDB)
{
$iZipcode = (int)$sPostcode;
// Try an exact match on the us_zippostcode table
$sSQL = 'select zipcode, ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from us_zipcode where zipcode = '.$iZipcode;
$aNearPostcodes = $oDB->getAll($sSQL);
if (PEAR::IsError($aNearPostcodes))
{
var_dump($sSQL, $aNearPostcodes);
exit;
}
if (!sizeof($aNearPostcodes))
{
$sSQL = 'select zipcode,ST_X(ST_Centroid(geometry)) as lon,ST_Y(ST_Centroid(geometry)) as lat from us_zipcode where zipcode between '.($iZipcode-100).' and '.($iZipcode+100).' order by abs(zipcode - '.$iZipcode.') asc limit 5';
$aNearPostcodes = $oDB->getAll($sSQL);
if (PEAR::IsError($aNearPostcodes))
{
var_dump($sSQL, $aNearPostcodes);
exit;
}
}
if (!sizeof($aNearPostcodes))
{
return false;
}
$fTotalLat = 0;
$fTotalLon = 0;
$fTotalFac = 0;
foreach($aNearPostcodes as $aPostcode)
{
$iDiff = abs($aPostcode['zipcode'] - $iZipcode) + 1;
if ($iDiff == 0)
$fFac = 1;
else
$fFac = 1/($iDiff*$iDiff);
$fTotalFac += $fFac;
$fTotalLat += $aPostcode['lat'] * $fFac;
$fTotalLon += $aPostcode['lon'] * $fFac;
}
if ($fTotalFac)
{
$fLat = $fTotalLat / $fTotalFac;
$fLon = $fTotalLon / $fTotalFac;
return array(array('lat' => $fLat, 'lon' => $fLon, 'radius' => 0.2));
}
return false;
/*
$fTotalFac is a surprisingly good indicator of accuracy
$iZoom = 18 + round(log($fTotalFac,32));
$iZoom = max(13,min(18,$iZoom));
*/
}
function getClassTypes()
{
return array(
@@ -701,7 +592,6 @@
function javascript_renderData($xVal, $iOptions = 0)
{
header("Access-Control-Allow-Origin: *");
if (defined('PHP_VERSION_ID') && PHP_VERSION_ID > 50400)
$iOptions |= JSON_UNESCAPED_UNICODE;
$jsonout = json_encode($xVal, $iOptions);

View File

@@ -2,68 +2,59 @@
function logStart(&$oDB, $sType = '', $sQuery = '', $aLanguageList = array())
{
$aStartTime = explode('.',microtime(true));
$fStartTime = microtime(true);
$aStartTime = explode('.', $fStartTime);
if (!isset($aStartTime[1])) $aStartTime[1] = '0';
$sOutputFormat = '';
if (isset($_GET['format'])) $sOutputFormat = $_GET['format'];
if ($sType == 'reverse')
{
$sOutQuery = (isset($_GET['lat'])?$_GET['lat']:'').'/';
if (isset($_GET['lon'])) $sOutQuery .= $_GET['lon'];
if (isset($_GET['zoom'])) $sOutQuery .= '/'.$_GET['zoom'];
}
else
$sOutQuery = $sQuery;
$hLog = array(
date('Y-m-d H:i:s',$aStartTime[0]).'.'.$aStartTime[1],
$_SERVER["REMOTE_ADDR"],
$_SERVER['QUERY_STRING'],
$sQuery
$sOutQuery,
$sType,
$fStartTime
);
if (CONST_Log_DB)
{
// Log
if ($sType == 'search')
{
$oDB->query('insert into query_log values ('.getDBQuoted($hLog[0]).','.getDBQuoted($hLog[3]).','.getDBQuoted($hLog[1]).')');
}
$sSQL = 'insert into new_query_log (type,starttime,query,ipaddress,useragent,language,format)';
if (isset($_GET['email']))
$sUserAgent = $_GET['email'];
elseif (isset($_SERVER['HTTP_REFERER']))
$sUserAgent = $_SERVER['HTTP_REFERER'];
elseif (isset($_SERVER['HTTP_USER_AGENT']))
$sUserAgent = $_SERVER['HTTP_USER_AGENT'];
else
$sUserAgent = '';
$sSQL = 'insert into new_query_log (type,starttime,query,ipaddress,useragent,language,format,searchterm)';
$sSQL .= ' values ('.getDBQuoted($sType).','.getDBQuoted($hLog[0]).','.getDBQuoted($hLog[2]);
$sSQL .= ','.getDBQuoted($hLog[1]).','.getDBQuoted($_SERVER['HTTP_USER_AGENT']).','.getDBQuoted(join(',',$aLanguageList)).','.getDBQuoted($sOutputFormat).')';
$sSQL .= ','.getDBQuoted($hLog[1]).','.getDBQuoted($sUserAgent).','.getDBQuoted(join(',',$aLanguageList)).','.getDBQuoted($sOutputFormat).','.getDBQuoted($hLog[3]).')';
$oDB->query($sSQL);
}
if (CONST_Log_File && CONST_Log_File_ReverseLog != '')
{
if ($sType == 'reverse')
{
$aStartTime = explode('.',$hLog[0]);
file_put_contents(CONST_Log_File_ReverseLog,
$aStartTime[0].','.$aStartTime[1].','.
php_uname('n').','.
'"'.addslashes(isset($_SERVER['HTTP_REFERER'])?$_SERVER['HTTP_REFERER']:'').'",'.
'"'.addslashes($hLog[1]).'",'.
$_GET['lat'].','.
$_GET['lon'].','.
$_GET['zoom'].','.
'"'.addslashes($_SERVER['HTTP_USER_AGENT']).'",'.
'"'.addslashes($sOutputFormat).'"'."\n",
FILE_APPEND);
}
}
return $hLog;
}
function logEnd(&$oDB, $hLog, $iNumResults)
{
$aEndTime = explode('.',microtime(true));
if (!$aEndTime[1]) $aEndTime[1] = '0';
$sEndTime = date('Y-m-d H:i:s',$aEndTime[0]).'.'.$aEndTime[1];
$fEndTime = microtime(true);
if (CONST_Log_DB)
{
$sSQL = 'update query_log set endtime = '.getDBQuoted($sEndTime).', results = '.$iNumResults;
$sSQL .= ' where starttime = '.getDBQuoted($hLog[0]);
$sSQL .= ' and ipaddress = '.getDBQuoted($hLog[1]);
$sSQL .= ' and query = '.getDBQuoted($hLog[3]);
$oDB->query($sSQL);
$aEndTime = explode('.', $fEndTime);
if (!$aEndTime[1]) $aEndTime[1] = '0';
$sEndTime = date('Y-m-d H:i:s',$aEndTime[0]).'.'.$aEndTime[1];
$sSQL = 'update new_query_log set endtime = '.getDBQuoted($sEndTime).', results = '.$iNumResults;
$sSQL .= ' where starttime = '.getDBQuoted($hLog[0]);
@@ -72,19 +63,12 @@
$oDB->query($sSQL);
}
if (CONST_Log_File && CONST_Log_File_SearchLog != '')
if (CONST_Log_File)
{
$aStartTime = explode('.',$hLog[0]);
file_put_contents(CONST_Log_File_SearchLog,
$aStartTime[0].','.$aStartTime[1].','.
php_uname('n').','.
'"'.addslashes(isset($_SERVER['HTTP_REFERER'])?$_SERVER['HTTP_REFERER']:'').'",'.
'"'.addslashes($hLog[1]).'",'.
'"'.addslashes($hLog[3]).'",'.
'"'.addslashes($_SERVER['HTTP_USER_AGENT']).'",'.
'"'.addslashes((isset($_GET['format']))?$_GET['format']:'').'",'.
$iNumResults."\n",
FILE_APPEND);
$aOutdata = sprintf("[%s] %.4f %d %s \"%s\"\n",
$hLog[0], $fEndTime-$hLog[5], $iNumResults,
$hLog[4], $hLog[2]);
file_put_contents(CONST_Log_File, $aOutdata, FILE_APPEND | LOCK_EX);
}
}

View File

@@ -1,165 +0,0 @@
amenity airport
amenity arts_centre
amenity atm
amenity auditorium
amenity bank
amenity bar
amenity bench
amenity bicycle_parking
amenity bicycle_rental
amenity brothel
amenity bureau_de_change
amenity bus_station
amenity cafe
amenity car_rental
amenity car_wash
amenity casino
amenity cinema
amenity clinic
amenity club
amenity college
amenity community_centre
amenity courthouse
amenity crematorium
amenity dentist
amenity doctors
amenity dormitory
amenity drinking_water
amenity driving_school
amenity embassy
amenity emergency_phone
amenity fast_food
amenity ferry_terminal
amenity fire_hydrant
amenity fire_station
amenity fountain
amenity fuel
amenity grave_yard
amenity hall
amenity health_centre
amenity hospital
amenity hotel
amenity hunting_stand
amenity ice_cream
amenity kindergarten
amenity library
amenity market
amenity marketplace
amenity nightclub
amenity nursery
amenity nursing_home
amenity office
amenity park
amenity parking
amenity pharmacy
amenity place_of_worship
amenity police
amenity post_box
amenity post_office
amenity preschool
amenity prison
amenity pub
amenity public_building
amenity public_market
amenity reception_area
amenity restaurant
amenity retirement_home
amenity sauna
amenity school
amenity shelter
amenity shop
amenity shopping
amenity social_club
amenity studio
amenity supermarket
amenity taxi
amenity telephone
amenity theatre
amenity toilets
amenity townhall
amenity university
amenity veterinary
amenity waste_basket
amenity wifi
amenity youth_centre
boundary administrative
building apartments
building block
building bunker
building chapel
building church
building commercial
building dormitory
building entrance
building faculty
building farm
building flats
building garage
building hospital
building hotel
building house
building industrial
building office
building public
building residential
building retail
building school
building shop
building stadium
building store
building terrace
building tower
building train_station
building university
highway bridleway
highway bus_stop
highway construction
highway cycleway
highway distance_marker
highway emergency_access_point
highway footway
highway gate
highway motorway_junction
highway path
highway pedestrian
highway platform
highway primary
highway primary_link
highway raceway
highway road
highway secondary
highway secondary_link
highway services
highway steps
highway tertiary
highway track
highway trail
highway trunk
highway trunk_link
highway unsurfaced
historic archaeological_site
historic battlefield
historic building
historic castle
historic church
historic house
historic icon
historic manor
historic memorial
historic mine
historic monument
historic museum
historic ruins
historic tower
historic wayside_cross
historic wayside_shrine
historic wreck
landuse cemetery
landuse commercial
landuse construction
landuse farm
landuse farmland
landuse farmyard
landuse forest
landuse grass
landuse industrial

View File

@@ -11,6 +11,18 @@
@define('CONST_Database_Web_User', 'www-data');
@define('CONST_Max_Word_Frequency', '50000');
@define('CONST_Limit_Reindexing', true);
// Set to false to avoid importing extra postcodes for the US.
@define('CONST_Use_Extra_US_Postcodes', true);
// Set to true after importing Tiger house number data for the US.
// Note: The tables must already exist or queries will throw errors.
// After changing this setting run ./utils/setup --create-functions
// again.
@define('CONST_Use_US_Tiger_Data', false);
// Set to true after importing other external house number data.
// Note: the aux tables must already exist or queries will throw errors.
// After changing this setting run ./utils/setup --create-functions
// again.
@define('CONST_Use_Aux_Location_data', false);
// Proxy settings
@define('CONST_HTTP_Proxy', false);
@@ -19,13 +31,7 @@
@define('CONST_HTTP_Proxy_Login', '');
@define('CONST_HTTP_Proxy_Password', '');
// Software versions
@define('CONST_Postgresql_Version', '9.3'); // values: 9.0, ... , 9.4
@define('CONST_Postgis_Version', '2.1'); // values: 1.5, 2.0, 2.1
// Paths
@define('CONST_Path_Postgresql_Contrib', '/usr/share/postgresql/'.CONST_Postgresql_Version.'/contrib');
@define('CONST_Path_Postgresql_Postgis', CONST_Path_Postgresql_Contrib.'/postgis-'.CONST_Postgis_Version);
@define('CONST_Osm2pgsql_Binary', CONST_InstallPath.'/osm2pgsql/osm2pgsql');
@define('CONST_Osmosis_Binary', '/usr/bin/osmosis');
@define('CONST_Tiger_Data_Path', CONST_BasePath.'/data/tiger');
@@ -81,23 +87,22 @@
// Website settings
@define('CONST_NoAccessControl', true);
@define('CONST_ClosedForIndexing', false);
@define('CONST_ClosedForIndexingExceptionIPs', '');
@define('CONST_BlockedIPs', '');
@define('CONST_BulkUserIPs', '');
@define('CONST_BlockMessage', ''); // additional info to show for blocked IPs
@define('CONST_Website_BaseURL', 'http://'.php_uname('n').'/');
@define('CONST_Tile_Default', 'Mapnik');
// Language to assume when none is supplied with the query.
// When set to false, the local language (i.e. the name tag without suffix)
// will be used.
@define('CONST_Default_Language', false);
// Appearance of the map in the debug interface.
@define('CONST_Default_Lat', 20.0);
@define('CONST_Default_Lon', 0.0);
@define('CONST_Default_Zoom', 2);
@define('CONST_Map_Tile_URL', 'http://{s}.tile.osm.org/{z}/{x}/{y}.png');
@define('CONST_Map_Tile_Attribution', ''); // Set if tile source isn't osm.org
@define('CONST_Search_AreaPolygons_Enabled', true);
@define('CONST_Search_AreaPolygons', true);
@define('CONST_Search_BatchMode', false);
@@ -108,16 +113,19 @@
// When set to false only selected languages alloow reverse search.
@define('CONST_Search_ReversePlanForAll', true);
// Maximum number of OSM ids that may be queried at once
// for the places endpoint.
@define('CONST_Places_Max_ID_count', 50);
// Set to zero to disable polygon output
// Number of different geometry formats that may be queried in parallel.
// Set to zero to disable polygon output.
@define('CONST_PolygonOutput_MaximumTypes', 1);
// Log settings
@define('CONST_Log_DB', true);
// Set to true to log into new_query_log table.
// You should set up a cron job that regularly clears out this table.
@define('CONST_Log_DB', false);
// Set to a file name to enable logging to a file.
@define('CONST_Log_File', false);
@define('CONST_Log_File_Format', 'TODO'); // Currently hard coded
@define('CONST_Log_File_SearchLog', '');
@define('CONST_Log_File_ReverseLog', '');

6
sql/aux_tables.sql Normal file
View File

@@ -0,0 +1,6 @@
CREATE TABLE location_property_aux () INHERITS (location_property);
CREATE INDEX idx_location_property_aux_place_id ON location_property_aux USING BTREE (place_id);
CREATE INDEX idx_location_property_aux_parent_place_id ON location_property_aux USING BTREE (parent_place_id);
CREATE INDEX idx_location_property_aux_housenumber_parent_place_id ON location_property_aux USING BTREE (parent_place_id, housenumber);
GRANT SELECT ON location_property_aux TO "{www-user}";

View File

@@ -1,3 +1,15 @@
-- Splits the line at the given point and returns the two parts
-- in a multilinestring.
CREATE OR REPLACE FUNCTION split_line_on_node(line GEOMETRY, point GEOMETRY)
RETURNS GEOMETRY
AS $$
BEGIN
RETURN ST_Split(ST_Snap(line, point, 0.0005), point);
END;
$$
LANGUAGE plpgsql;
CREATE OR REPLACE FUNCTION geometry_sector(partition INTEGER, place geometry) RETURNS INTEGER
AS $$
DECLARE
@@ -990,7 +1002,7 @@ BEGIN
--DEBUG: RAISE WARNING 'placex_insert:END: % % % %',NEW.osm_type,NEW.osm_id,NEW.class,NEW.type;
RETURN NEW; -- @DIFFUPDATES@ The following is not needed until doing diff updates, and slows the main index process down
RETURN NEW; -- %DIFFUPDATES% The following is not needed until doing diff updates, and slows the main index process down
IF NEW.rank_address > 0 THEN
IF (ST_GeometryType(NEW.geometry) in ('ST_Polygon','ST_MultiPolygon') AND ST_IsValid(NEW.geometry)) THEN
@@ -1606,6 +1618,7 @@ BEGIN
NEW.parent_place_id = 0;
parent_place_id_rank = 0;
-- convert isin to array of tokenids
isin_tokens := '{}'::int[];
IF NEW.isin IS NOT NULL THEN
@@ -1645,6 +1658,7 @@ BEGIN
END IF;
END IF;
-- %NOTIGERDATA% IF 0 THEN
-- for the USA we have an additional address table. Merge in zip codes from there too
IF NEW.rank_search = 26 AND NEW.calculated_country_code = 'us' THEN
FOR location IN SELECT distinct postcode from location_property_tiger where parent_place_id = NEW.place_id LOOP
@@ -1657,6 +1671,7 @@ BEGIN
nameaddress_vector := array_merge(nameaddress_vector, ARRAY[address_street_word_id]);
END LOOP;
END IF;
-- %NOTIGERDATA% END IF;
-- RAISE WARNING 'ISIN: %', isin_tokens;
@@ -2315,7 +2330,6 @@ DECLARE
countryname HSTORE;
hadcountry BOOLEAN;
BEGIN
-- first query osmline (interpolation lines)
select parent_place_id, calculated_country_code, 30, postcode, null, 'place', 'house' from location_property_osmline
WHERE place_id = in_place_id AND in_housenumber>=startnumber AND in_housenumber <= endnumber
@@ -2323,8 +2337,9 @@ BEGIN
IF for_place_id IS NOT NULL THEN
searchhousenumber = in_housenumber::text;
END IF;
--then query tiger data
-- %NOTIGERDATA% IF 0 THEN
IF for_place_id IS NULL THEN
select parent_place_id,'us', 30, postcode, null, 'place', 'house' from location_property_tiger
WHERE place_id = in_place_id AND in_housenumber>=startnumber AND in_housenumber <= endnumber
@@ -2333,13 +2348,16 @@ BEGIN
searchhousenumber = in_housenumber::text;
END IF;
END IF;
-- %NOTIGERDATA% END IF;
-- %NOAUXDATA% IF 0 THEN
IF for_place_id IS NULL THEN
select parent_place_id,'us', housenumber, 30, postcode, null, 'place', 'house' from location_property_aux
WHERE place_id = in_place_id
INTO for_place_id,searchcountrycode, searchhousenumber, searchrankaddress, searchpostcode, searchhousename, searchclass, searchtype;
END IF;
-- %NOAUXDATA% END IF;
IF for_place_id IS NULL THEN
select parent_place_id, calculated_country_code, housenumber, rank_search, postcode, name, class, type from placex
WHERE place_id = in_place_id and rank_address = 30

View File

@@ -19,10 +19,6 @@ CREATE INDEX idx_location_area_country_place_id ON location_area_country USING B
CREATE INDEX idx_search_name_country_centroid ON search_name_country USING GIST (centroid) {ts:address-index};
-- start
CREATE INDEX idx_location_property_-partition-_centroid ON location_property_-partition- USING GIST (centroid) {ts:address-index};
-- end
DROP INDEX IF EXISTS place_id_idx;
CREATE UNIQUE INDEX idx_place_osm_unique on place using btree(osm_id,osm_type,class,type) {ts:address-index};

View File

@@ -51,11 +51,6 @@ CREATE INDEX idx_search_name_-partition-_place_id ON search_name_-partition- USI
CREATE INDEX idx_search_name_-partition-_centroid ON search_name_-partition- USING GIST (centroid) {ts:address-index};
CREATE INDEX idx_search_name_-partition-_name_vector ON search_name_-partition- USING GIN (name_vector) WITH (fastupdate = off) {ts:address-index};
CREATE TABLE location_property_-partition- () INHERITS (location_property) {ts:aux-data};
CREATE INDEX idx_location_property_-partition-_place_id ON location_property_-partition- USING BTREE (place_id) {ts:aux-index};
CREATE INDEX idx_location_property_-partition-_parent_place_id ON location_property_-partition- USING BTREE (parent_place_id) {ts:aux-index};
CREATE INDEX idx_location_property_-partition-_housenumber_parent_place_id ON location_property_-partition- USING BTREE (parent_place_id, housenumber) {ts:aux-index};
CREATE TABLE location_road_-partition- (
partition integer,
place_id BIGINT,

View File

@@ -1,14 +0,0 @@
-- Splits the line at the given point and returns the two parts
-- in a multilinestring.
CREATE OR REPLACE FUNCTION split_line_on_node(line GEOMETRY, point GEOMETRY)
RETURNS GEOMETRY
AS $$
DECLARE
frac FLOAT;
BEGIN
frac := ST_Line_Locate_Point(line, point);
RETURN ST_Collect(ST_Line_Substring(line, 0, frac),
ST_Line_Substring(line, frac, 1));
END
$$
LANGUAGE plpgsql;

View File

@@ -1,10 +0,0 @@
-- Splits the line at the given point and returns the two parts
-- in a multilinestring.
CREATE OR REPLACE FUNCTION split_line_on_node(line GEOMETRY, point GEOMETRY)
RETURNS GEOMETRY
AS $$
BEGIN
RETURN ST_Split(ST_Snap(line, point, 0.0005), point);
END;
$$
LANGUAGE plpgsql;

View File

@@ -1,155 +0,0 @@
drop table import_npi_log;
CREATE TABLE import_npi_log (
npiid integer,
batchend timestamp,
batchsize integer,
starttime timestamp,
endtime timestamp,
event text
);
drop table IF EXISTS word;
CREATE TABLE word (
word_id INTEGER,
word_token text,
word_trigram text,
word text,
class text,
type text,
country_code varchar(2),
search_name_count INTEGER,
operator TEXT
);
SELECT AddGeometryColumn('word', 'location', 4326, 'GEOMETRY', 2);
CREATE INDEX idx_word_word_id on word USING BTREE (word_id);
CREATE INDEX idx_word_word_token on word USING BTREE (word_token);
GRANT SELECT ON word TO "www-data" ;
DROP SEQUENCE seq_word;
CREATE SEQUENCE seq_word start 1;
drop table IF EXISTS location_property CASCADE;
CREATE TABLE location_property (
place_id BIGINT,
partition integer,
parent_place_id BIGINT,
housenumber TEXT,
postcode TEXT
);
SELECT AddGeometryColumn('location_property', 'centroid', 4326, 'POINT', 2);
CREATE TABLE location_property_aux () INHERITS (location_property);
CREATE INDEX idx_location_property_aux_place_id ON location_property_aux USING BTREE (place_id);
CREATE INDEX idx_location_property_aux_parent_place_id ON location_property_aux USING BTREE (parent_place_id);
CREATE INDEX idx_location_property_aux_housenumber_parent_place_id ON location_property_aux USING BTREE (parent_place_id, housenumber);
CREATE TABLE location_property_tiger () INHERITS (location_property);
CREATE INDEX idx_location_property_tiger_place_id ON location_property_tiger USING BTREE (place_id);
CREATE INDEX idx_location_property_tiger_parent_place_id ON location_property_tiger USING BTREE (parent_place_id);
CREATE INDEX idx_location_property_tiger_housenumber_parent_place_id ON location_property_tiger USING BTREE (parent_place_id, housenumber);
drop table IF EXISTS search_name_blank CASCADE;
CREATE TABLE search_name_blank (
place_id BIGINT,
search_rank integer,
address_rank integer,
importance FLOAT,
country_code varchar(2),
name_vector integer[],
nameaddress_vector integer[]
);
SELECT AddGeometryColumn('search_name_blank', 'centroid', 4326, 'GEOMETRY', 2);
drop table IF EXISTS search_name;
CREATE TABLE search_name () INHERITS (search_name_blank);
CREATE INDEX search_name_name_vector_idx ON search_name USING GIN (name_vector gin__int_ops) WITH (fastupdate = off);
CREATE INDEX searchnameplacesearch_search_nameaddress_vector_idx ON search_name USING GIN (nameaddress_vector gin__int_ops) WITH (fastupdate = off);
CREATE INDEX idx_search_name_centroid ON search_name USING GIST (centroid);
CREATE INDEX idx_search_name_place_id ON search_name USING BTREE (place_id);
drop table IF EXISTS place_addressline;
CREATE TABLE place_addressline (
place_id BIGINT,
address_place_id BIGINT,
fromarea boolean,
isaddress boolean,
distance float,
cached_rank_address integer
);
CREATE INDEX idx_place_addressline_place_id on place_addressline USING BTREE (place_id);
CREATE INDEX idx_place_addressline_address_place_id on place_addressline USING BTREE (address_place_id);
drop table IF EXISTS place_boundingbox CASCADE;
CREATE TABLE place_boundingbox (
place_id BIGINT,
minlat float,
maxlat float,
minlon float,
maxlon float,
numfeatures integer,
area float
);
CREATE INDEX idx_place_boundingbox_place_id on place_boundingbox USING BTREE (place_id);
SELECT AddGeometryColumn('place_boundingbox', 'outline', 4326, 'GEOMETRY', 2);
CREATE INDEX idx_place_boundingbox_outline ON place_boundingbox USING GIST (outline);
GRANT SELECT on place_boundingbox to "www-data" ;
GRANT INSERT on place_boundingbox to "www-data" ;
drop table country;
CREATE TABLE country (
country_code varchar(2),
country_name hstore,
country_default_language_code varchar(2)
);
SELECT AddGeometryColumn('country', 'geometry', 4326, 'POLYGON', 2);
insert into country select iso3166::varchar(2), 'name:en'->cntry_name, null,
ST_Transform(geometryn(the_geom, generate_series(1, numgeometries(the_geom))), 4326) from worldboundaries;
CREATE INDEX idx_country_country_code ON country USING BTREE (country_code);
CREATE INDEX idx_country_geometry ON country USING GIST (geometry);
drop table placex;
CREATE TABLE placex (
place_id BIGINT NOT NULL,
partition integer,
osm_type char(1),
osm_id INTEGER,
class TEXT NOT NULL,
type TEXT NOT NULL,
name HSTORE,
admin_level INTEGER,
housenumber TEXT,
street TEXT,
isin TEXT,
postcode TEXT,
country_code varchar(2),
extratags HSTORE,
parent_place_id BIGINT,
linked_place_id BIGINT,
rank_address INTEGER,
rank_search INTEGER,
importance FLOAT,
indexed_status INTEGER,
indexed_date TIMESTAMP,
geometry_sector INTEGER
);
SELECT AddGeometryColumn('placex', 'geometry', 4326, 'GEOMETRY', 2);
CREATE UNIQUE INDEX idx_place_id ON placex USING BTREE (place_id);
CREATE INDEX idx_placex_osmid ON placex USING BTREE (osm_type, osm_id);
CREATE INDEX idx_placex_rank_search ON placex USING BTREE (rank_search);
CREATE INDEX idx_placex_rank_address ON placex USING BTREE (rank_address);
CREATE INDEX idx_placex_geometry ON placex USING GIST (geometry);
CREATE INDEX idx_placex_parent_place_id ON placex USING BTREE (parent_place_id) where parent_place_id IS NOT NULL;
DROP SEQUENCE seq_place;
CREATE SEQUENCE seq_place start 1;
GRANT SELECT on placex to "www-data" ;
GRANT UPDATE ON placex to "www-data" ;
GRANT SELECT ON search_name to "www-data" ;
GRANT DELETE on search_name to "www-data" ;
GRANT INSERT on search_name to "www-data" ;
GRANT SELECT on place_addressline to "www-data" ;
GRANT INSERT ON place_addressline to "www-data" ;
GRANT DELETE on place_addressline to "www-data" ;
GRANT SELECT ON seq_word to "www-data" ;
GRANT UPDATE ON seq_word to "www-data" ;
GRANT INSERT ON word to "www-data" ;
GRANT SELECT on country to "www-data" ;

View File

@@ -13,29 +13,6 @@ CREATE TABLE import_osmosis_log (
event text
);
drop table if exists import_npi_log;
CREATE TABLE import_npi_log (
npiid integer,
batchend timestamp,
batchsize integer,
starttime timestamp,
endtime timestamp,
event text
);
--drop table IF EXISTS query_log;
CREATE TABLE query_log (
starttime timestamp,
query text,
ipaddress text,
endtime timestamp,
results integer
);
CREATE INDEX idx_query_log ON query_log USING BTREE (starttime);
GRANT SELECT ON query_log TO "{www-user}" ;
GRANT INSERT ON query_log TO "{www-user}" ;
GRANT UPDATE ON query_log TO "{www-user}" ;
CREATE TABLE new_query_log (
type text,
starttime timestamp,
@@ -43,6 +20,7 @@ CREATE TABLE new_query_log (
useragent text,
language text,
query text,
searchterm text,
endtime timestamp,
results integer,
format text,
@@ -56,9 +34,6 @@ GRANT SELECT ON new_query_log TO "{www-user}" ;
GRANT SELECT ON TABLE country_name TO "{www-user}";
GRANT SELECT ON TABLE gb_postcode TO "{www-user}";
create view vw_search_query_log as SELECT substr(query, 1, 50) AS query, starttime, endtime - starttime AS duration, substr(useragent, 1, 20) as
useragent, language, results, ipaddress FROM new_query_log WHERE type = 'search' ORDER BY starttime DESC;
drop table IF EXISTS word;
CREATE TABLE word (
word_id INTEGER,
@@ -129,7 +104,6 @@ CREATE UNIQUE INDEX idx_osmline_place_id ON location_property_osmline (place_id)
CREATE INDEX idx_osmline_parent_place_id ON location_property_osmline (parent_place_id) {ts:search-index};
GRANT SELECT ON location_property_osmline TO "{www-user}";
drop table IF EXISTS search_name;
CREATE TABLE search_name (
place_id BIGINT,

View File

@@ -1,9 +0,0 @@
\a
\t
\o /tmp/bigintupdate.sql
select 'alter table "'||relname||'" alter '||attname||' type bigint;' from pg_attribute join pg_class on
(attrelid = oid) where attname like '%place_id%' and attnum > 0 and relkind = 'r'::"char" and atttypid = 23
and not relname::text ~ '^.*_[0-9]+$' order by 'alter table "'||relname||'" alter '||attname||' type
bigint;';
\o
\i /tmp/bigintupdate.sql

View File

@@ -175,8 +175,10 @@ def db_template_setup():
conn = psycopg2.connect(database=world.config.template_db)
psycopg2.extras.register_hstore(conn, globally=False, unicode=True)
cur = conn.cursor()
for table in ('gb_postcode', 'us_postcode', 'us_state', 'us_statecounty'):
cur.execute('TRUNCATE TABLE %s' % (table,))
for table in ('gb_postcode', 'us_postcode'):
cur.execute("select * from pg_tables where tablename = '%s'" % (table, ))
if cur.rowcount > 0:
cur.execute('TRUNCATE TABLE %s' % (table,))
conn.commit()
conn.close()
# execute osm2pgsql on an empty file to get the right tables

View File

@@ -24,7 +24,6 @@
array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
array('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'),
array('create-minimal-tables', '', 0, 1, 0, 0, 'bool', 'Create minimal main tables'),
array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
@@ -37,7 +36,6 @@
array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
array('index-output', '', 0, 1, 1, 1, 'string', 'File to dump index information to'),
array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'),
@@ -92,8 +90,6 @@
$aDSNInfo = DB::parseDSN(CONST_Database_DSN);
if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
$fPostgisVersion = (float) CONST_Postgis_Version;
if ($aCMDResult['create-db'] || $aCMDResult['all'])
{
echo "Create DB\n";
@@ -114,40 +110,39 @@
$oDB =& getDB();
$sVersionString = $oDB->getOne('select version()');
preg_match('#PostgreSQL ([0-9]+)[.]([0-9]+)[^0-9]#', $sVersionString, $aMatches);
if (CONST_Postgresql_Version != $aMatches[1].'.'.$aMatches[2])
$fPostgresVersion = getPostgresVersion($oDB);
echo 'Postgres version found: '.$fPostgresVersion."\n";
if ($fPostgresVersion < 9.1)
{
echo "ERROR: PostgreSQL version is not correct. Expected ".CONST_Postgresql_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
exit;
fail("Minimum supported version of Postgresql is 9.1.");
}
passthru('createlang plpgsql -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
$pgver = (float) CONST_Postgresql_Version;
if ($pgver < 9.1) {
pgsqlRunScriptFile(CONST_Path_Postgresql_Contrib.'/hstore.sql');
pgsqlRunScriptFile(CONST_BasePath.'/sql/hstore_compatability_9_0.sql');
} else {
pgsqlRunScript('CREATE EXTENSION hstore');
pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore');
pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis');
// For extratags and namedetails the hstore_to_json converter is
// needed which is only available from Postgresql 9.3+. For older
// versions add a dummy function that returns nothing.
$iNumFunc = $oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'");
if (PEAR::isError($iNumFunc))
{
fail("Cannot query stored procedures.", $iNumFunc);
}
if ($iNumFunc == 0)
{
pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable");
echo "WARNING: Postgresql is too old. extratags and namedetails API not available.";
}
if ($fPostgisVersion < 2.0) {
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/postgis.sql');
pgsqlRunScriptFile(CONST_Path_Postgresql_Postgis.'/spatial_ref_sys.sql');
} else {
pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis');
}
if ($fPostgisVersion < 2.1) {
$fPostgisVersion = getPostgisVersion($oDB);
echo 'Postgis version found: '.$fPostgisVersion."\n";
if ($fPostgisVersion < 2.1)
{
// Function was renamed in 2.1 and throws an annoying deprecation warning
pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
}
$sVersionString = $oDB->getOne('select postgis_full_version()');
preg_match('#POSTGIS="([0-9]+)[.]([0-9]+)[.]([0-9]+)( r([0-9]+))?"#', $sVersionString, $aMatches);
if (CONST_Postgis_Version != $aMatches[1].'.'.$aMatches[2])
{
echo "ERROR: PostGIS version is not correct. Expected ".CONST_Postgis_Version." found ".$aMatches[1].'.'.$aMatches[2]."\n";
exit;
}
pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
@@ -161,9 +156,10 @@
{
echo "WARNING: external UK postcode table not found.\n";
}
pgsqlRunScriptFile(CONST_BasePath.'/data/us_statecounty.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_state.sql');
pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
if (CONST_Use_Extra_US_Postcodes)
{
pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
}
if ($aCMDResult['no-partitions'])
{
@@ -221,51 +217,7 @@
echo "Functions\n";
$bDidSomething = true;
if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
$sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/module', $sTemplate);
if ($aCMDResult['enable-diff-updates']) $sTemplate = str_replace('RETURN NEW; -- @DIFFUPDATES@', '--', $sTemplate);
if ($aCMDResult['enable-debug-statements']) $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
if (CONST_Limit_Reindexing) $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
pgsqlRunScript($sTemplate);
if ($fPostgisVersion < 2.0) {
echo "Helper functions for postgis < 2.0\n";
$sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_15_aux.sql');
} else {
echo "Helper functions for postgis >= 2.0\n";
$sTemplate = file_get_contents(CONST_BasePath.'/sql/postgis_20_aux.sql');
}
pgsqlRunScript($sTemplate);
}
if ($aCMDResult['create-minimal-tables'])
{
echo "Minimal Tables\n";
$bDidSomething = true;
pgsqlRunScriptFile(CONST_BasePath.'/sql/tables-minimal.sql');
$sScript = '';
// Backstop the import process - easliest possible import id
$sScript .= "insert into import_npi_log values (18022);\n";
$hFile = @fopen(CONST_BasePath.'/settings/partitionedtags.def', "r");
if (!$hFile) fail('unable to open list of partitions: '.CONST_BasePath.'/settings/partitionedtags.def');
while (($sLine = fgets($hFile, 4096)) !== false && $sLine && substr($sLine,0,1) !='#')
{
list($sClass, $sType) = explode(' ', trim($sLine));
$sScript .= "create table place_classtype_".$sClass."_".$sType." as ";
$sScript .= "select place_id as place_id,geometry as centroid from placex limit 0;\n";
$sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_centroid ";
$sScript .= "ON place_classtype_".$sClass."_".$sType." USING GIST (centroid);\n";
$sScript .= "CREATE INDEX idx_place_classtype_".$sClass."_".$sType."_place_id ";
$sScript .= "ON place_classtype_".$sClass."_".$sType." USING btree(place_id);\n";
}
fclose($hFile);
pgsqlRunScript($sScript);
create_sql_functions($aCMDResult);
}
if ($aCMDResult['create-tables'] || $aCMDResult['all'])
@@ -291,10 +243,7 @@
// re-run the functions
echo "Functions\n";
$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
$sTemplate = str_replace('{modulepath}',
CONST_InstallPath.'/module', $sTemplate);
pgsqlRunScript($sTemplate);
create_sql_functions($aCMDResult);
}
if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
@@ -597,10 +546,14 @@
$sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
$sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
$sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
$sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
if (CONST_Use_Extra_US_Postcodes)
{
$sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
$sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
$sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
}
}
if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables
@@ -701,7 +654,6 @@
{
$bDidSomething = true;
$sOutputFile = '';
if (isset($aCMDResult['index-output'])) $sOutputFile = ' -F '.$aCMDResult['index-output'];
$sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
passthruCheckReturn($sBaseCmd.' -R 4');
if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
@@ -714,14 +666,6 @@
{
echo "Search indices\n";
$bDidSomething = true;
$oDB =& getDB();
$sSQL = 'select distinct partition from country_name';
$aPartitions = $oDB->getCol($sSQL);
if (PEAR::isError($aPartitions))
{
fail($aPartitions->getMessage());
}
if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
$sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
$sTemplate = replace_tablespace('{ts:address-index}',
@@ -730,16 +674,6 @@
CONST_Tablespace_Search_Index, $sTemplate);
$sTemplate = replace_tablespace('{ts:aux-index}',
CONST_Tablespace_Aux_Index, $sTemplate);
preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
foreach($aMatches as $aMatch)
{
$sResult = '';
foreach($aPartitions as $sPartitionName)
{
$sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
}
$sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
}
pgsqlRunScript($sTemplate);
}
@@ -1011,3 +945,31 @@
return $sSql;
}
function create_sql_functions($aCMDResult)
{
$sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
$sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/module', $sTemplate);
if ($aCMDResult['enable-diff-updates'])
{
$sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate);
}
if ($aCMDResult['enable-debug-statements'])
{
$sTemplate = str_replace('--DEBUG:', '', $sTemplate);
}
if (CONST_Limit_Reindexing)
{
$sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
}
if (!CONST_Use_US_Tiger_Data)
{
$sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate);
}
if (!CONST_Use_Aux_Location_data)
{
$sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
}
pgsqlRunScript($sTemplate);
}

View File

@@ -11,18 +11,11 @@
array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
array('max-load', '', 0, 1, 1, 1, 'float', 'Maximum load average - indexing is paused if this is exceeded'),
array('max-blocking', '', 0, 1, 1, 1, 'int', 'Maximum blocking processes - indexing is aborted / paused if this is exceeded'),
array('import-osmosis', '', 0, 1, 0, 0, 'bool', 'Import using osmosis'),
array('import-osmosis-all', '', 0, 1, 0, 0, 'bool', 'Import using osmosis forever'),
array('no-npi', '', 0, 1, 0, 0, 'bool', 'Do not write npi index files'),
array('no-npi', '', 0, 1, 0, 0, 'bool', '(obsolate)'),
array('no-index', '', 0, 1, 0, 0, 'bool', 'Do not index the new data'),
array('import-npi-all', '', 0, 1, 0, 0, 'bool', 'Import npi pre-indexed files'),
array('import-hourly', '', 0, 1, 0, 0, 'bool', 'Import hourly diffs'),
array('import-daily', '', 0, 1, 0, 0, 'bool', 'Import daily diffs'),
array('import-all', '', 0, 1, 0, 0, 'bool', 'Import all available files'),
array('import-file', '', 0, 1, 1, 1, 'realpath', 'Re-import data from an OSM file'),
@@ -37,34 +30,14 @@
array('index', '', 0, 1, 0, 0, 'bool', 'Index'),
array('index-rank', '', 0, 1, 1, 1, 'int', 'Rank to start indexing from'),
array('index-instances', '', 0, 1, 1, 1, 'int', 'Number of indexing instances (threads)'),
array('index-estrate', '', 0, 1, 1, 1, 'int', 'Estimated indexed items per second (def:30)'),
array('deduplicate', '', 0, 1, 0, 0, 'bool', 'Deduplicate tokens'),
);
getCmdOpt($_SERVER['argv'], $aCMDOptions, $aResult, true, true);
if ($aResult['import-hourly'] + $aResult['import-daily'] + isset($aResult['import-diff']) > 1)
{
showUsage($aCMDOptions, true, 'Select either import of hourly or daily');
}
if (!isset($aResult['index-instances'])) $aResult['index-instances'] = 1;
if (!isset($aResult['index-rank'])) $aResult['index-rank'] = 0;
/*
// Lock to prevent multiple copies running
if (exec('/bin/ps uww | grep '.basename(__FILE__).' | grep -v /dev/null | grep -v grep -c', $aOutput2, $iResult) > 1)
{
fail("Copy already running\n");
}
if (!isset($aResult['max-load'])) $aResult['max-load'] = 1.9;
if (!isset($aResult['max-blocking'])) $aResult['max-blocking'] = 3;
if (getBlockingProcesses() > $aResult['max-blocking'])
{
fail("Too many blocking processes for import\n");
}
*/
date_default_timezone_set('Etc/UTC');
$oDB =& getDB();
@@ -86,153 +59,112 @@
}
$bFirst = true;
$bContinue = $aResult['import-all'];
while ($bContinue || $bFirst)
if (isset($aResult['import-diff']))
{
$bFirst = false;
if ($aResult['import-hourly'])
// import diff directly (e.g. from osmosis --rri)
$sNextFile = $aResult['import-diff'];
if (!file_exists($sNextFile))
{
// Mirror the hourly diffs
exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/hourly');
$sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDDHH24\')||\'-\'||TO_CHAR(lastimportdate+\'1 hour\'::interval,\'YYYYMMDDHH24\')||\'.osc.gz\' from import_status');
$sNextFile = $sMirrorDir.'planet.openstreetmap.org/hourly/'.$sNextFile;
$sUpdateSQL = 'update import_status set lastimportdate = lastimportdate+\'1 hour\'::interval';
fail("Cannot open $sNextFile\n");
}
if ($aResult['import-daily'])
// Import the file
$sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
echo $sCMD."\n";
exec($sCMD, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
// Mirror the daily diffs
exec('wget --quiet --mirror -l 1 -P '.$sMirrorDir.' http://planet.openstreetmap.org/daily');
$sNextFile = $oDB->getOne('select TO_CHAR(lastimportdate,\'YYYYMMDD\')||\'-\'||TO_CHAR(lastimportdate+\'1 day\'::interval,\'YYYYMMDD\')||\'.osc.gz\' from import_status');
$sNextFile = $sMirrorDir.'planet.openstreetmap.org/daily/'.$sNextFile;
$sUpdateSQL = 'update import_status set lastimportdate = lastimportdate::date + 1';
}
if (isset($aResult['import-diff']))
{
// import diff directly (e.g. from osmosis --rri)
$sNextFile = $aResult['import-diff'];
if (!file_exists($sNextFile))
{
fail("Cannot open $sNextFile\n");
}
// Don't update the import status - we don't know what this file contains
$sUpdateSQL = 'update import_status set lastimportdate = now() where false';
fail("Error from osm2pgsql, $iErrorLevel\n");
}
// Missing file is not an error - it might not be created yet
if (($aResult['import-hourly'] || $aResult['import-daily'] || isset($aResult['import-diff'])) && file_exists($sNextFile))
{
// Import the file
$sCMD = $sOsm2pgsqlCmd.' '.$sNextFile;
echo $sCMD."\n";
exec($sCMD, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
fail("Error from osm2pgsql, $iErrorLevel\n");
}
// Move the date onwards
$oDB->query($sUpdateSQL);
}
else
{
$bContinue = false;
}
// Don't update the import status - we don't know what this file contains
}
$bModifyXML = false;
$sModifyXMLstr = '';
$bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
$sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
$bHaveDiff = false;
if (isset($aResult['import-file']) && $aResult['import-file'])
{
$bModifyXML = true;
$bHaveDiff = true;
$sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
echo $sCMD."\n";
exec($sCMD, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
fail("Error converting osm to osc, osmosis returned: $iErrorLevel\n");
}
}
$bUseOSMApi = isset($aResult['import-from-main-api']) && $aResult['import-from-main-api'];
$sContentURL = '';
if (isset($aResult['import-node']) && $aResult['import-node'])
{
$bModifyXML = true;
if ($bUseOSMApi)
{
$sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node']);
$sContentURL = 'http://www.openstreetmap.org/api/0.6/node/'.$aResult['import-node'];
}
else
{
$sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;');
$sContentURL = 'http://overpass-api.de/api/interpreter?data=node('.$aResult['import-node'].');out%20meta;';
}
}
if (isset($aResult['import-way']) && $aResult['import-way'])
{
$bModifyXML = true;
if ($bUseOSMApi)
{
$sCmd = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
$sContentURL = 'http://www.openstreetmap.org/api/0.6/way/'.$aResult['import-way'].'/full';
}
else
{
$sCmd = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
$sContentURL = 'http://overpass-api.de/api/interpreter?data=(way('.$aResult['import-way'].');node(w););out%20meta;';
}
$sModifyXMLstr = file_get_contents($sCmd);
}
if (isset($aResult['import-relation']) && $aResult['import-relation'])
{
$bModifyXML = true;
if ($bUseOSMApi)
{
$sModifyXMLstr = file_get_contents('http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full');
$sContentURLsModifyXMLstr = 'http://www.openstreetmap.org/api/0.6/relation/'.$aResult['import-relation'].'/full';
}
else
{
$sModifyXMLstr = file_get_contents('http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;');
$sContentURL = 'http://overpass-api.de/api/interpreter?data=((rel('.$aResult['import-relation'].');way(r);node(w));node(r));out%20meta;';
}
}
if ($bModifyXML)
if ($sContentURL)
{
// derive change from normal osm file with osmosis
$sTemporaryFile = CONST_BasePath.'/data/osmosischange.osc';
if (isset($aResult['import-file']) && $aResult['import-file'])
{
$sCMD = CONST_Osmosis_Binary.' --read-xml \''.$aResult['import-file'].'\' --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
echo $sCMD."\n";
exec($sCMD, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
fail("Error converting osm to osc, osmosis returned: $iErrorLevel\n");
}
}
else
{
$aSpec = array(
0 => array("pipe", "r"), // stdin
1 => array("pipe", "w"), // stdout
2 => array("pipe", "w") // stderr
);
$sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
echo $sCMD."\n";
$hProc = proc_open($sCMD, $aSpec, $aPipes);
if (!is_resource($hProc))
{
fail("Error converting osm to osc, osmosis failed\n");
}
fwrite($aPipes[0], $sModifyXMLstr);
fclose($aPipes[0]);
$sOut = stream_get_contents($aPipes[1]);
if ($aResult['verbose']) echo $sOut;
fclose($aPipes[1]);
$sErrors = stream_get_contents($aPipes[2]);
if ($aResult['verbose']) echo $sErrors;
fclose($aPipes[2]);
if ($iError = proc_close($hProc))
{
echo "Error converting osm to osc, osmosis returned: $iError\n";
echo $sOut;
echo $sErrors;
exit(-1);
}
}
$sModifyXMLstr = file_get_contents($sContentURL);
$bHaveDiff = true;
$aSpec = array(
0 => array("pipe", "r"), // stdin
1 => array("pipe", "w"), // stdout
2 => array("pipe", "w") // stderr
);
$sCMD = CONST_Osmosis_Binary.' --read-xml - --read-empty --derive-change --write-xml-change '.$sTemporaryFile;
echo $sCMD."\n";
$hProc = proc_open($sCMD, $aSpec, $aPipes);
if (!is_resource($hProc))
{
fail("Error converting osm to osc, osmosis failed\n");
}
fwrite($aPipes[0], $sModifyXMLstr);
fclose($aPipes[0]);
$sOut = stream_get_contents($aPipes[1]);
if ($aResult['verbose']) echo $sOut;
fclose($aPipes[1]);
$sErrors = stream_get_contents($aPipes[2]);
if ($aResult['verbose']) echo $sErrors;
fclose($aPipes[2]);
if ($iError = proc_close($hProc))
{
echo $sOut;
echo $sErrors;
fail("Error converting osm to osc, osmosis returned: $iError\n");
}
}
if ($bHaveDiff)
{
// import generated change file
$sCMD = $sOsm2pgsqlCmd.' '.$sTemporaryFile;
echo $sCMD."\n";
@@ -246,19 +178,19 @@
if ($aResult['deduplicate'])
{
$pgver = (float) CONST_Postgresql_Version;
if ($pgver < 9.3) {
if (getPostgresVersion() < 9.3)
{
fail("ERROR: deduplicate is only currently supported in postgresql 9.3");
}
$oDB =& getDB();
$sSQL = 'select partition from country_name order by country_code';
$aPartitions = $oDB->getCol($sSQL);
if (PEAR::isError($aPartitions))
{
fail($aPartitions->getMessage());
}
$aPartitions[] = 0;
$oDB =& getDB();
$sSQL = 'select partition from country_name order by country_code';
$aPartitions = $oDB->getCol($sSQL);
if (PEAR::isError($aPartitions))
{
fail($aPartitions->getMessage());
}
$aPartitions[] = 0;
$sSQL = "select word_token,count(*) from word where substr(word_token, 1, 1) = ' ' and class is null and type is null and country_code is null group by word_token having count(*) > 1 order by word_token";
$aDuplicateTokens = $oDB->getAll($sSQL);
@@ -341,7 +273,6 @@
exit(1);
}
}
}
}
@@ -358,92 +289,81 @@
}
$sImportFile = CONST_BasePath.'/data/osmosischange.osc';
$sOsmosisCMD = CONST_Osmosis_Binary;
$sOsmosisConfigDirectory = CONST_InstallPath.'/settings';
$sCMDDownload = $sOsmosisCMD.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
$sCMDCheckReplicationLag = $sOsmosisCMD.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
$sCMDDownload = CONST_Osmosis_Binary.' --read-replication-interval workingDirectory='.$sOsmosisConfigDirectory.' --simplify-change --write-xml-change '.$sImportFile;
$sCMDCheckReplicationLag = CONST_Osmosis_Binary.' -q --read-replication-lag workingDirectory='.$sOsmosisConfigDirectory;
$sCMDImport = $sOsm2pgsqlCmd.' '.$sImportFile;
$sCMDIndex = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$aResult['index-instances'];
if (!$aResult['no-npi']) {
$sCMDIndex .= '-F ';
}
while(true)
{
$fStartTime = time();
$iFileSize = 1001;
// Logic behind this is that osm2pgsql locks the database quite a bit
// So it is better to import lots of small files
// But indexing works most efficiently on large amounts of data
// So do lots of small imports and a BIG index
// while($aResult['import-osmosis-all'] && $iFileSize > 1000)
// {
if (!file_exists($sImportFile))
if (!file_exists($sImportFile))
{
// First check if there are new updates published (except for minutelies - there's always new diffs to process)
if ( CONST_Replication_Update_Interval > 60 )
{
// First check if there are new updates published (except for minutelies - there's always new diffs to process)
if ( CONST_Replication_Update_Interval > 60 )
{
unset($aReplicationLag);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
{
if ($iErrorLevel)
{
echo "Error: $iErrorLevel. ";
echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
}
else
{
echo ".";
}
sleep(CONST_Replication_Recheck_Interval);
unset($aReplicationLag);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
while ($iErrorLevel > 0 || $aReplicationLag[0] < 1)
{
if ($iErrorLevel)
{
echo "Error: $iErrorLevel. ";
echo "Re-trying: ".$sCMDCheckReplicationLag." in ".CONST_Replication_Recheck_Interval." secs\n";
}
else
{
echo ".";
}
sleep(CONST_Replication_Recheck_Interval);
unset($aReplicationLag);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
}
// There are new replication files - use osmosis to download the file
echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
}
$fStartTime = time();
$fCMDStartTime = time();
echo $sCMDDownload."\n";
exec($sCMDDownload, $sJunk, $iErrorLevel);
while ($iErrorLevel > 0)
{
echo "Error: $iErrorLevel\n";
sleep(60);
echo 'Re-trying: '.$sCMDDownload."\n";
exec($sCMDDownload, $sJunk, $iErrorLevel);
}
$iFileSize = filesize($sImportFile);
$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
$sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
var_Dump($sSQL);
$oDB->query($sSQL);
echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
// There are new replication files - use osmosis to download the file
echo "\n".date('Y-m-d H:i:s')." Replication Delay is ".$aReplicationLag[0]."\n";
}
$iFileSize = filesize($sImportFile);
$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
// Import the file
$fStartTime = time();
$fCMDStartTime = time();
echo $sCMDImport."\n";
exec($sCMDImport, $sJunk, $iErrorLevel);
if ($iErrorLevel)
echo $sCMDDownload."\n";
exec($sCMDDownload, $sJunk, $iErrorLevel);
while ($iErrorLevel > 0)
{
echo "Error: $iErrorLevel\n";
exit($iErrorLevel);
sleep(60);
echo 'Re-trying: '.$sCMDDownload."\n";
exec($sCMDDownload, $sJunk, $iErrorLevel);
}
$sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
$iFileSize = filesize($sImportFile);
$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
$sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osmosis')";
var_Dump($sSQL);
$oDB->query($sSQL);
echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
echo date('Y-m-d H:i:s')." Completed osmosis step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
}
// Archive for debug?
unlink($sImportFile);
// }
$iFileSize = filesize($sImportFile);
$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
// Import the file
$fCMDStartTime = time();
echo $sCMDImport."\n";
exec($sCMDImport, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
echo "Error: $iErrorLevel\n";
exit($iErrorLevel);
}
$sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','osm2pgsql')";
var_Dump($sSQL);
$oDB->query($sSQL);
echo date('Y-m-d H:i:s')." Completed osm2pgsql step for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
// Archive for debug?
unlink($sImportFile);
$sBatchEnd = getosmosistimestamp($sOsmosisConfigDirectory);
@@ -451,31 +371,6 @@
$sThisIndexCmd = $sCMDIndex;
$fCMDStartTime = time();
if (!$aResult['no-npi'])
{
$iFileID = $oDB->getOne('select nextval(\'file\')');
if (PEAR::isError($iFileID))
{
echo $iFileID->getMessage()."\n";
exit(-1);
}
$sFileDir = CONST_BasePath.'/export/diff/';
$sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
$sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
$sThisIndexCmd .= $sFileDir;
$sThisIndexCmd .= '/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT);
$sThisIndexCmd .= ".npi.out";
preg_match('#^([0-9]{4})-([0-9]{2})-([0-9]{2})#', $sBatchEnd, $aBatchMatch);
$sFileDir = CONST_BasePath.'/export/index/';
$sFileDir .= $aBatchMatch[1].'/'.$aBatchMatch[2];
if (!is_dir($sFileDir)) mkdir($sFileDir, 0777, true);
file_put_contents($sFileDir.'/'.$aBatchMatch[3].'.idx', "$sBatchEnd\t$iFileID\n", FILE_APPEND);
}
if (!$aResult['no-index'])
{
echo "$sThisIndexCmd\n";
@@ -485,25 +380,6 @@
echo "Error: $iErrorLevel\n";
exit($iErrorLevel);
}
if (!$aResult['no-npi'])
{
$sFileDir = CONST_BasePath.'/export/diff/';
$sFileDir .= str_pad(floor($iFileID/1000000), 3, '0', STR_PAD_LEFT);
$sFileDir .= '/'.str_pad(floor($iFileID/1000) % 1000, 3, '0', STR_PAD_LEFT);
$sThisIndexCmd = 'bzip2 -z9 '.$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out";
echo "$sThisIndexCmd\n";
exec($sThisIndexCmd, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
echo "Error: $iErrorLevel\n";
exit($iErrorLevel);
}
rename($sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.out.bz2",
$sFileDir.'/'.str_pad($iFileID % 1000, 3, '0', STR_PAD_LEFT).".npi.bz2");
}
}
$sSQL = "INSERT INTO import_osmosis_log values ('$sBatchEnd',$iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','index')";
@@ -529,55 +405,6 @@
echo date('Y-m-d H:i:s')." Sleeping $iSleep seconds\n";
sleep($iSleep);
}
}
if ($aResult['import-npi-all'])
{
$iNPIID = $oDB->getOne('select max(npiid) from import_npi_log');
if (PEAR::isError($iNPIID))
{
var_dump($iNPIID);
exit(1);
}
$sConfigDirectory = CONST_InstallPath.'/settings';
$sCMDImportTemplate = CONST_InstallPath.'/nominatim/nominatim -d gazetteer -P 5433 -I -T '.CONST_BasePath.'/settings/partitionedtags.def -F ';
while(true)
{
$fStartTime = time();
$iNPIID++;
$sImportFile = CONST_BasePath.'/export/diff/';
$sImportFile .= str_pad(floor($iNPIID/1000000), 3, '0', STR_PAD_LEFT);
$sImportFile .= '/'.str_pad(floor($iNPIID/1000) % 1000, 3, '0', STR_PAD_LEFT);
$sImportFile .= '/'.str_pad($iNPIID % 1000, 3, '0', STR_PAD_LEFT);
$sImportFile .= ".npi";
while(!file_exists($sImportFile) && !file_exists($sImportFile.'.bz2'))
{
echo "sleep (waiting for $sImportFile)\n";
sleep(10);
}
if (file_exists($sImportFile.'.bz2')) $sImportFile .= '.bz2';
$iFileSize = filesize($sImportFile);
// Import the file
$fCMDStartTime = time();
$sCMDImport = $sCMDImportTemplate . $sImportFile;
echo $sCMDImport."\n";
exec($sCMDImport, $sJunk, $iErrorLevel);
if ($iErrorLevel)
{
fail("Error: $iErrorLevel\n");
}
$sBatchEnd = $iNPIID;
echo "Completed for $sBatchEnd in ".round((time()-$fCMDStartTime)/60,2)." minutes\n";
$sSQL = "INSERT INTO import_npi_log values ($iNPIID, null, $iFileSize,'".date('Y-m-d H:i:s',$fCMDStartTime)."','".date('Y-m-d H:i:s')."','import')";
var_Dump($sSQL);
$oDB->query($sSQL);
}
}
function getosmosistimestamp($sOsmosisConfigDirectory)

View File

@@ -56,10 +56,17 @@
$iPlaceID = (int)$_GET['place_id'];
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_tiger where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_aux where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;
if (CONST_Use_US_Tiger_Data)
{
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_tiger where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;
}
if (CONST_Use_Aux_Location_data)
{
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_aux where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;
}
$hLog = logStart($oDB, 'details', $_SERVER['QUERY_STRING'], $aLangPrefOrder);

View File

@@ -47,13 +47,13 @@
$iPlaceID = (int)$_GET['place_id'];
$sAuxHouseNumber = false;
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_tiger where place_id = '.$iPlaceID);
if ($iParentPlaceID)
if (CONST_Use_US_Tiger_Data)
{
$iPlaceID = $iParentPlaceID;
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_tiger where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;
}
else
if (CONST_Use_Aux_Location_data)
{
$iParentPlaceID = $oDB->getOne('select parent_place_id from location_property_aux where place_id = '.$iPlaceID);
if ($iParentPlaceID) $iPlaceID = $iParentPlaceID;