safety-screen/Nominatim
1
0
Fork 0
forked from hans/Nominatim
Code Pull Requests Activity

fix indentation and misc errors according to PSR2 coding style guide

Browse Source
This commit is contained in:
Marc Tobias Metten
2016-09-14 03:16:46 +02:00
parent 74c06ef877
commit 4a321487f3
21 changed files with 406 additions and 374 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
utils/blocks.php
View File

@@ -35,9 +35,9 @@ if ($aResult['list']) {
printf(
" %-40s | %12s | %7s | %13s | %31s | %8s\n",
$sKey,
$aDetails['totalBlocks'],
$aDetails['totalBlocks'],
(int)$aDetails['currentBucketSize'],
$aDetails['currentlyBlocked']?'Y':'N',
$aDetails['currentlyBlocked']?'Y':'N',
date("r", $aDetails['lastBlockTimestamp']),
$aDetails['isSleeping']?'Y':'N'
);

370
utils/importWikipedia.php
View File

@@ -90,12 +90,14 @@ EOD;
$oDB->query($sSQL);
}
function degreesAndMinutesToDecimal($iDegrees, $iMinutes = 0, $fSeconds = 0, $sNSEW = 'N')
{
$sNSEW = strtoupper($sNSEW);
return ($sNSEW == 'S' || $sNSEW == 'W'?-1:1) * ((float)$iDegrees + (float)$iMinutes/60 + (float)$fSeconds/3600);
}
function _parseWikipediaContent($sPageText)
{
$sPageText = str_replace("\n", ' ', $sPageText);
@@ -113,83 +115,83 @@ function _parseWikipediaContent($sPageText)
$aState = array('body');
foreach ($aPageText as $i => $sPart) {
switch ($sPart) {
case '{{':
array_unshift($aTemplateStack, array('', array()));
array_unshift($aState, 'template');
break;
case '}}':
if ($aState[0] == 'template' || $aState[0] == 'templateparam') {
$aTemplate = array_shift($aTemplateStack);
array_shift($aState);
case '{{':
array_unshift($aTemplateStack, array('', array()));
array_unshift($aState, 'template');
break;
case '}}':
if ($aState[0] == 'template' || $aState[0] == 'templateparam') {
$aTemplate = array_shift($aTemplateStack);
array_shift($aState);
$aTemplates[] = $aTemplate;
}
break;
case '[[':
$sLinkPage = '';
$sLinkSyn = '';
array_unshift($aState, 'link');
break;
case ']]':
if ($aState[0] == 'link' || $aState[0] == 'linksynonim') {
if (!$sLinkSyn) $sLinkSyn = $sLinkPage;
if (substr($sLinkPage, 0, 6) == 'Image:') $sLinkSyn = substr($sLinkPage, 6);
$aLinks[] = array($sLinkPage, $sLinkSyn);
array_shift($aState);
switch ($aState[0]) {
case 'template':
$aTemplateStack[0][0] .= trim($sPart);
break;
case 'templateparam':
$aTemplateStack[0][1][0] .= $sLinkSyn;
break;
case 'link':
$sLinkPage .= trim($sPart);
break;
case 'linksynonim':
$sLinkSyn .= $sPart;
break;
case 'body':
$sPageBody .= $sLinkSyn;
break;
default:
var_dump($aState, $sPageName, $aTemplateStack, $sPart, $aPageText);
fail('unknown state');
$aTemplates[] = $aTemplate;
}
}
break;
case '|':
if ($aState[0] == 'template' || $aState[0] == 'templateparam') {
// Create a new template paramater
$aState[0] = 'templateparam';
array_unshift($aTemplateStack[0][1], '');
}
if ($aState[0] == 'link') $aState[0] = 'linksynonim';
break;
default:
switch ($aState[0]) {
case 'template':
$aTemplateStack[0][0] .= trim($sPart);
break;
case 'templateparam':
$aTemplateStack[0][1][0] .= $sPart;
case '[[':
$sLinkPage = '';
$sLinkSyn = '';
array_unshift($aState, 'link');
break;
case 'link':
$sLinkPage .= trim($sPart);
case ']]':
if ($aState[0] == 'link' || $aState[0] == 'linksynonim') {
if (!$sLinkSyn) $sLinkSyn = $sLinkPage;
if (substr($sLinkPage, 0, 6) == 'Image:') $sLinkSyn = substr($sLinkPage, 6);
$aLinks[] = array($sLinkPage, $sLinkSyn);
array_shift($aState);
switch ($aState[0]) {
case 'template':
$aTemplateStack[0][0] .= trim($sPart);
break;
case 'templateparam':
$aTemplateStack[0][1][0] .= $sLinkSyn;
break;
case 'link':
$sLinkPage .= trim($sPart);
break;
case 'linksynonim':
$sLinkSyn .= $sPart;
break;
case 'body':
$sPageBody .= $sLinkSyn;
break;
default:
var_dump($aState, $sPageName, $aTemplateStack, $sPart, $aPageText);
fail('unknown state');
}
}
break;
case 'linksynonim':
$sLinkSyn .= $sPart;
break;
case 'body':
$sPageBody .= $sPart;
case '|':
if ($aState[0] == 'template' || $aState[0] == 'templateparam') {
// Create a new template paramater
$aState[0] = 'templateparam';
array_unshift($aTemplateStack[0][1], '');
}
if ($aState[0] == 'link') $aState[0] = 'linksynonim';
break;
default:
var_dump($aState, $aPageText);
fail('unknown state');
}
break;
switch ($aState[0]) {
case 'template':
$aTemplateStack[0][0] .= trim($sPart);
break;
case 'templateparam':
$aTemplateStack[0][1][0] .= $sPart;
break;
case 'link':
$sLinkPage .= trim($sPart);
break;
case 'linksynonim':
$sLinkSyn .= $sPart;
break;
case 'body':
$sPageBody .= $sPart;
break;
default:
var_dump($aState, $aPageText);
fail('unknown state');
}
break;
}
}
return $aTemplates;
@@ -201,7 +203,7 @@ function _templatesToProperties($aTemplates)
foreach ($aTemplates as $iTemplate => $aTemplate) {
$aParams = array();
foreach (array_reverse($aTemplate[1]) as $iParam => $sParam) {
if (($iPos = strpos($sParam, '=')) === FALSE) {
if (($iPos = strpos($sParam, '=')) === false) {
$aParams[] = trim($sParam);
} else {
$aParams[trim(substr($sParam, 0, $iPos))] = trim(substr($sParam, $iPos+1));
@@ -224,7 +226,7 @@ function _templatesToProperties($aTemplates)
if (!isset($aPageProperties['sWebsite']) && isset($aParams['website']) && $aParams['website']) {
if (preg_match('#^\\[?([^ \\]]+)[^\\]]*\\]?$#', $aParams['website'], $aMatch)) {
$aPageProperties['sWebsite'] = $aMatch[1];
if (strpos($aPageProperties['sWebsite'], ':/'.'/') === FALSE) {
if (strpos($aPageProperties['sWebsite'], ':/'.'/') === false) {
$aPageProperties['sWebsite'] = 'http:/'.'/'.$aPageProperties['sWebsite'];
}
}
@@ -306,7 +308,7 @@ function _templatesToProperties($aTemplates)
if (isset($aCMDResult['parse-wikipedia'])) {
$oDB =& getDB();
$aArticleNames = $oDB->getCol('select page_title from content where page_namespace = 0 and page_id %10 = '.$aCMDResult['parse-wikipedia'].' and (page_content ilike \'%{{Coord%\' or (page_content ilike \'%lat%\' and page_content ilike \'%lon%\'))');
// $aArticleNames = $oDB->getCol($sSQL = 'select page_title from content where page_namespace = 0 and (page_content ilike \'%{{Coord%\' or (page_content ilike \'%lat%\' and page_content ilike \'%lon%\')) and page_title in (\'Virginia\')');
// $aArticleNames = $oDB->getCol($sSQL = 'select page_title from content where page_namespace = 0 and (page_content ilike \'%{{Coord%\' or (page_content ilike \'%lat%\' and page_content ilike \'%lon%\')) and page_title in (\'Virginia\')');
foreach ($aArticleNames as $sArticleName) {
$sPageText = $oDB->getOne('select page_content from content where page_namespace = 0 and page_title = \''.pg_escape_string($sArticleName).'\'');
$aP = _templatesToProperties(_parseWikipediaContent($sPageText));
@@ -342,16 +344,18 @@ if (isset($aCMDResult['parse-wikipedia'])) {
}
}
function nominatimXMLStart($hParser, $sName, $aAttr)
{
global $aNominatRecords;
switch ($sName) {
global $aNominatRecords;
switch ($sName) {
case 'PLACE':
$aNominatRecords[] = $aAttr;
break;
}
$aNominatRecords[] = $aAttr;
break;
}
}
function nominatimXMLEnd($hParser, $sName)
{
}
@@ -373,104 +377,104 @@ if (isset($aCMDResult['link'])) {
$fMaxDist = 0.0000001;
$bUnknown = false;
switch (strtolower($aRecord['infobox_type'])) {
case 'former country':
continue 2;
case 'sea':
$fMaxDist = 60; // effectively turn it off
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'country':
case 'island':
case 'islands':
case 'continent':
$fMaxDist = 60; // effectively turn it off
$sURL .= "&featuretype=country";
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'prefecture japan':
$aRecord['name'] = trim(str_replace(' Prefecture', ' ', $aRecord['name']));
break;
case 'state':
case '#us state':
case 'county':
case 'u.s. state':
case 'u.s. state symbols':
case 'german state':
case 'province or territory of canada':
case 'indian jurisdiction':
case 'province':
case 'french region':
case 'region of italy':
case 'kommune':
case '#australia state or territory':
case 'russian federal subject':
$fMaxDist = 4;
$sURL .= "&featuretype=state";
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'protected area':
$fMaxDist = 1;
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'settlement':
$bUnknown = true;
break;
case 'french commune':
case 'italian comune':
case 'uk place':
case 'italian comune':
case 'australian place':
case 'german place':
case '#geobox':
case 'u.s. county':
case 'municipality':
case 'city japan':
case 'russian inhabited locality':
case 'finnish municipality/land area':
case 'england county':
case 'israel municipality':
case 'russian city':
case 'city':
$fMaxDist = 0.2;
$sURL .= "&featuretype=settlement";
$sURL .= "&viewbox=".($aRecord['lon']-0.5).",".($aRecord['lat']+0.5).",".($aRecord['lon']+0.5).",".($aRecord['lat']-0.5);
break;
case 'mountain':
case 'mountain pass':
case 'river':
case 'lake':
case 'airport':
$fMaxDist = 0.2;
$sURL .= "&viewbox=".($aRecord['lon']-0.5).",".($aRecord['lat']+0.5).",".($aRecord['lon']+0.5).",".($aRecord['lat']-0.5);
break;
case 'ship begin':
$fMaxDist = 0.1;
$aTypes = array('wreck');
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
break;
case 'road':
case 'university':
case 'company':
case 'department':
$fMaxDist = 0.005;
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
$sURL .= "&bounded=1";
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
break;
default:
$bUnknown = true;
$fMaxDist = 0.005;
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
// $sURL .= "&bounded=1";
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
echo "-- Unknown: ".$aRecord['infobox_type']."\n";
break;
case 'former country':
continue 2;
case 'sea':
$fMaxDist = 60; // effectively turn it off
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'country':
case 'island':
case 'islands':
case 'continent':
$fMaxDist = 60; // effectively turn it off
$sURL .= "&featuretype=country";
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'prefecture japan':
$aRecord['name'] = trim(str_replace(' Prefecture', ' ', $aRecord['name']));
break;
case 'state':
case '#us state':
case 'county':
case 'u.s. state':
case 'u.s. state symbols':
case 'german state':
case 'province or territory of canada':
case 'indian jurisdiction':
case 'province':
case 'french region':
case 'region of italy':
case 'kommune':
case '#australia state or territory':
case 'russian federal subject':
$fMaxDist = 4;
$sURL .= "&featuretype=state";
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'protected area':
$fMaxDist = 1;
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
$sURL .= "&viewbox=".($aRecord['lon']-$fMaxDist).",".($aRecord['lat']+$fMaxDist).",".($aRecord['lon']+$fMaxDist).",".($aRecord['lat']-$fMaxDist);
break;
case 'settlement':
$bUnknown = true;
break;
case 'french commune':
case 'italian comune':
case 'uk place':
case 'italian comune':
case 'australian place':
case 'german place':
case '#geobox':
case 'u.s. county':
case 'municipality':
case 'city japan':
case 'russian inhabited locality':
case 'finnish municipality/land area':
case 'england county':
case 'israel municipality':
case 'russian city':
case 'city':
$fMaxDist = 0.2;
$sURL .= "&featuretype=settlement";
$sURL .= "&viewbox=".($aRecord['lon']-0.5).",".($aRecord['lat']+0.5).",".($aRecord['lon']+0.5).",".($aRecord['lat']-0.5);
break;
case 'mountain':
case 'mountain pass':
case 'river':
case 'lake':
case 'airport':
$fMaxDist = 0.2;
$sURL .= "&viewbox=".($aRecord['lon']-0.5).",".($aRecord['lat']+0.5).",".($aRecord['lon']+0.5).",".($aRecord['lat']-0.5);
break;
case 'ship begin':
$fMaxDist = 0.1;
$aTypes = array('wreck');
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
break;
case 'road':
case 'university':
case 'company':
case 'department':
$fMaxDist = 0.005;
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
$sURL .= "&bounded=1";
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
break;
default:
$bUnknown = true;
$fMaxDist = 0.005;
$sURL .= "&viewbox=".($aRecord['lon']-0.01).",".($aRecord['lat']+0.01).",".($aRecord['lon']+0.01).",".($aRecord['lat']-0.01);
// $sURL .= "&bounded=1";
$sURL .= "&nearlat=".$aRecord['lat'];
$sURL .= "&nearlon=".$aRecord['lon'];
echo "-- Unknown: ".$aRecord['infobox_type']."\n";
break;
}
$sNameURL = $sURL.'&q='.urlencode($aRecord['name']);
@@ -494,7 +498,7 @@ if (isset($aCMDResult['link'])) {
$hXMLParser = xml_parser_create();
xml_set_element_handler($hXMLParser, 'nominatimXMLStart', 'nominatimXMLEnd');
xml_parse($hXMLParser, $sXML, true);
xml_parser_free($hXMLParser);#
xml_parser_free($hXMLParser);
}
}
@@ -522,9 +526,15 @@ if (isset($aCMDResult['link'])) {
} else {
$sSQL = "update wikipedia_article set osm_type=";
switch ($aNominatRecords[$i]['OSM_TYPE']) {
case 'relation': $sSQL .= "'R'"; break;
case 'way': $sSQL .= "'W'"; break;
case 'node': $sSQL .= "'N'"; break;
case 'relation':
$sSQL .= "'R'";
break;
case 'way':
$sSQL .= "'W'";
break;
case 'node':
$sSQL .= "'N'";
break;
}
$sSQL .= ", osm_id=".$aNominatRecords[$i]['OSM_ID']." where language = '".pg_escape_string($aRecord['language'])."' and title = '".pg_escape_string($aRecord['title'])."'";
$oDB->query($sSQL);

7
utils/server_compare.php
View File

@@ -1,8 +1,7 @@
#!/usr/bin/php -Cq
<?php
// Apache log file
$sFile = "sample.log.txt";
$sFile = "sample.log.txt"; // Apache log file
$sHost1 = 'http://mq-open-search-lm02.ihost.aol.com:8000/nominatim/v1';
$sHost2 = 'http://mq-open-search-lm03.ihost.aol.com:8000/nominatim/v1';
@@ -50,13 +49,13 @@ while (($sLine = fgets($hFile, 10000)) !== false) {
$sRes = $sURL1.":\n";
for ($j = 0; $j < strlen($sRes1); $j+=40) {
$sRes .= substr($sRes1, $j, 40)."\n";
$sRes .= substr($sRes1, $j, 40)."\n";
}
file_put_contents('log/'.$i.'.1', $sRes);
$sRes = $sURL2.":\n";
for ($j = 0; $j < strlen($sRes2); $j+=40) {
$sRes .= substr($sRes2, $j, 40)."\n";
$sRes .= substr($sRes2, $j, 40)."\n";
}
file_put_contents('log/'.$i.'.2', $sRes);
}

9
utils/setup.php
View File

@@ -94,8 +94,9 @@ if ($aCMDResult['create-db'] || $aCMDResult['all']) {
if ($aCMDResult['setup-db'] || $aCMDResult['all']) {
echo "Setup DB\n";
$bDidSomething = true;
// TODO: path detection, detection memory, etc.
// TODO: path detection, detection memory, etc.
//
$oDB =& getDB();
$fPostgresVersion = getPostgresVersion($oDB);
@@ -649,6 +650,7 @@ if (!$bDidSomething) {
echo "Setup finished.\n";
}
function pgsqlRunScriptFile($sFilename)
{
if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
@@ -698,7 +700,6 @@ function pgsqlRunScriptFile($sFilename)
fclose($ahGzipPipes[1]);
proc_close($hGzipProcess);
}
}
function pgsqlRunScript($sScript, $bfatal = true)
@@ -712,7 +713,7 @@ function pgsqlRunScript($sScript, $bfatal = true)
$sCMD .= ' -v ON_ERROR_STOP=1';
$aDescriptors = array(
0 => array('pipe', 'r'),
1 => STDOUT,
1 => STDOUT,
2 => STDERR
);
$ahPipes = null;
@@ -844,6 +845,4 @@ function create_sql_functions($aCMDResult)
$sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
}
pgsqlRunScript($sTemplate);
}

14
utils/specialphrases.php
View File

@@ -42,24 +42,24 @@ if ($aCMDResult['wiki-import']) {
$sLabel = trim($aMatch[1]);
$sClass = trim($aMatch[2]);
$sType = trim($aMatch[3]);
# hack around a bug where building=yes was imported with
# quotes into the wiki
// hack around a bug where building=yes was imported with
// quotes into the wiki
$sType = preg_replace('/&quot;/', '', $sType);
# sanity check, in case somebody added garbage in the wiki
// sanity check, in case somebody added garbage in the wiki
if (preg_match('/^\\w+$/', $sClass) < 1
|| preg_match('/^\\w+$/', $sType) < 1
) {
trigger_error("Bad class/type for language $sLanguage: $sClass=$sType");
exit;
}
# blacklisting: disallow certain class/type combinations
// blacklisting: disallow certain class/type combinations
if (isset($aTagsBlacklist[$sClass]) && in_array($sType, $aTagsBlacklist[$sClass])) {
# fwrite(STDERR, "Blacklisted: ".$sClass."/".$sType."\n");
// fwrite(STDERR, "Blacklisted: ".$sClass."/".$sType."\n");
continue;
}
# whitelisting: if class is in whitelist, allow only tags in the list
// whitelisting: if class is in whitelist, allow only tags in the list
if (isset($aTagsWhitelist[$sClass]) && !in_array($sType, $aTagsWhitelist[$sClass])) {
# fwrite(STDERR, "Non-Whitelisted: ".$sClass."/".$sType."\n");
// fwrite(STDERR, "Non-Whitelisted: ".$sClass."/".$sType."\n");
continue;
}
$aPairs[$sClass.'|'.$sType] = array($sClass, $sType);

5
utils/update.php
View File

@@ -237,7 +237,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
// First check if there are new updates published (except for minutelies - there's always new diffs to process)
if (CONST_Replication_Update_Interval > 60) {
unset($aReplicationLag);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
while ($iErrorLevel > 0 || $aReplicationLag[0] < 1) {
if ($iErrorLevel) {
echo "Error: $iErrorLevel. ";
@@ -247,7 +247,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
}
sleep(CONST_Replication_Recheck_Interval);
unset($aReplicationLag);
exec($sCMDCheckReplicationLag, $aReplicationLag, $iErrorLevel);
exec($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";
@@ -326,6 +326,7 @@ if ($aResult['import-osmosis'] || $aResult['import-osmosis-all']) {
}
}
function getosmosistimestamp($sOsmosisConfigDirectory)
{
$sStateFile = file_get_contents($sOsmosisConfigDirectory.'/state.txt');

1
utils/warm.php
View File

@@ -67,4 +67,3 @@ if (!$aResult['reverse-only']) {
else echo ".";
}
}