Merge remote-tracking branch 'upstream/master'

[nominatim.git] / lib / setup / SetupClass.php

diff --git a/lib/setup/SetupClass.php b/lib/setup/SetupClass.php
index b6705968752abdecd81a06b8682072b1b44d7dea..2fdb3926d59ecc23d2f26ecc75e94776f3b81ca8 100755 (executable)
--- a/lib/setup/SetupClass.php
+++ b/lib/setup/SetupClass.php
@@ -80,13 +80,15 @@ class SetupFunctions
             fail('database already exists ('.CONST_Database_DSN.')');
         }
 
             fail('database already exists ('.CONST_Database_DSN.')');
         }
 

-        $sCreateDBCmd = 'createdb -E UTF-8 -p '.$this->aDSNInfo['port'].' '.$this->aDSNInfo['database'];
+        $sCreateDBCmd = 'createdb -E UTF-8'
+            .' -p '.escapeshellarg($this->aDSNInfo['port'])
+            .' '.escapeshellarg($this->aDSNInfo['database']);

         if (isset($this->aDSNInfo['username'])) {
         if (isset($this->aDSNInfo['username'])) {

-            $sCreateDBCmd .= ' -U '.$this->aDSNInfo['username'];
+            $sCreateDBCmd .= ' -U '.escapeshellarg($this->aDSNInfo['username']);

         }
 
         if (isset($this->aDSNInfo['hostspec'])) {
         }
 
         if (isset($this->aDSNInfo['hostspec'])) {

-            $sCreateDBCmd .= ' -h '.$this->aDSNInfo['hostspec'];
+            $sCreateDBCmd .= ' -h '.escapeshellarg($this->aDSNInfo['hostspec']);

         }
 
         $result = $this->runWithPgEnv($sCreateDBCmd);
         }
 
         $result = $this->runWithPgEnv($sCreateDBCmd);


@@ -137,7 +139,7 @@ class SetupFunctions
             exit(1);
         }
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
             exit(1);
         }
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');

-        $this->pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql.gz');
+        $this->pgsqlRunScriptFile(CONST_ExtraDataPath.'/country_osm_grid.sql.gz');

         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode_table.sql');
 
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
         $this->pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode_table.sql');
 


@@ -158,13 +160,6 @@ class SetupFunctions
         if ($this->bNoPartitions) {
             $this->pgsqlRunScript('update country_name set partition = 0');
         }
         if ($this->bNoPartitions) {
             $this->pgsqlRunScript('update country_name set partition = 0');
         }

-
-        // the following will be needed by createFunctions later but
-        // is only defined in the subsequently called createTables
-        // Create dummies here that will be overwritten by the proper
-        // versions in create-tables.
-        $this->pgsqlRunScript('CREATE TABLE IF NOT EXISTS place_boundingbox ()');
-        $this->pgsqlRunScript('CREATE TYPE wikipedia_article_match AS ()', false);

     }
 
     public function importData($sOSMFile)
     }
 
     public function importData($sOSMFile)


@@ -178,30 +173,30 @@ class SetupFunctions
             fail("osm2pgsql not found in '$osm2pgsql'");
         }
 
             fail("osm2pgsql not found in '$osm2pgsql'");
         }
 

-        $osm2pgsql .= ' -S '.CONST_Import_Style;
+        $osm2pgsql .= ' -S '.escapeshellarg(CONST_Import_Style);

 
         if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
 
         if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {

-            $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
+            $osm2pgsql .= ' --flat-nodes '.escapeshellarg(CONST_Osm2pgsql_Flatnode_File);

         }
 
         if (CONST_Tablespace_Osm2pgsql_Data)
         }
 
         if (CONST_Tablespace_Osm2pgsql_Data)

-            $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data;
+            $osm2pgsql .= ' --tablespace-slim-data '.escapeshellarg(CONST_Tablespace_Osm2pgsql_Data);

         if (CONST_Tablespace_Osm2pgsql_Index)
         if (CONST_Tablespace_Osm2pgsql_Index)

-            $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index;
+            $osm2pgsql .= ' --tablespace-slim-index '.escapeshellarg(CONST_Tablespace_Osm2pgsql_Index);

         if (CONST_Tablespace_Place_Data)
         if (CONST_Tablespace_Place_Data)

-            $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data;
+            $osm2pgsql .= ' --tablespace-main-data '.escapeshellarg(CONST_Tablespace_Place_Data);

         if (CONST_Tablespace_Place_Index)
         if (CONST_Tablespace_Place_Index)

-            $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index;
+            $osm2pgsql .= ' --tablespace-main-index '.escapeshellarg(CONST_Tablespace_Place_Index);

         $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1';
         $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1';

-        $osm2pgsql .= ' -C '.$this->iCacheMemory;
-        $osm2pgsql .= ' -P '.$this->aDSNInfo['port'];
+        $osm2pgsql .= ' -C '.escapeshellarg($this->iCacheMemory);
+        $osm2pgsql .= ' -P '.escapeshellarg($this->aDSNInfo['port']);

         if (isset($this->aDSNInfo['username'])) {
         if (isset($this->aDSNInfo['username'])) {

-            $osm2pgsql .= ' -U '.$this->aDSNInfo['username'];
+            $osm2pgsql .= ' -U '.escapeshellarg($this->aDSNInfo['username']);

         }
         if (isset($this->aDSNInfo['hostspec'])) {
         }
         if (isset($this->aDSNInfo['hostspec'])) {

-            $osm2pgsql .= ' -H '.$this->aDSNInfo['hostspec'];
+            $osm2pgsql .= ' -H '.escapeshellarg($this->aDSNInfo['hostspec']);

         }
         }

-        $osm2pgsql .= ' -d '.$this->aDSNInfo['database'].' '.$sOSMFile;
+        $osm2pgsql .= ' -d '.escapeshellarg($this->aDSNInfo['database']).' '.escapeshellarg($sOSMFile);

 
         $this->runWithPgEnv($osm2pgsql);
 
 
         $this->runWithPgEnv($osm2pgsql);
 


@@ -321,19 +316,14 @@ class SetupFunctions
 
     public function importWikipediaArticles()
     {
 
     public function importWikipediaArticles()
     {

-        $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikipedia_article.sql.bin';
-        $sWikiRedirectsFile = CONST_Wikipedia_Data_Path.'/wikipedia_redirect.sql.bin';
+        $sWikiArticlesFile = CONST_Wikipedia_Data_Path.'/wikimedia-importance.sql.gz';

         if (file_exists($sWikiArticlesFile)) {
         if (file_exists($sWikiArticlesFile)) {

-            info('Importing wikipedia articles');
-            $this->pgsqlRunDropAndRestore($sWikiArticlesFile);
-        } else {
-            warn('wikipedia article dump file not found - places will have default importance');
-        }
-        if (file_exists($sWikiRedirectsFile)) {
-            info('Importing wikipedia redirects');
-            $this->pgsqlRunDropAndRestore($sWikiRedirectsFile);
+            info('Importing wikipedia articles and redirects');
+            $this->pgExec('DROP TABLE IF EXISTS wikipedia_article');
+            $this->pgExec('DROP TABLE IF EXISTS wikipedia_redirect');
+            $this->pgsqlRunScriptFile($sWikiArticlesFile);

         } else {
         } else {

-            warn('wikipedia redirect dump file not found - some place importance values may be missing');
+            warn('wikipedia importance dump file not found - places will have default importance');

         }
     }
 
         }
     }
 


@@ -349,8 +339,6 @@ class SetupFunctions
         echo '.';
         $this->pgExec('TRUNCATE place_addressline');
         echo '.';
         echo '.';
         $this->pgExec('TRUNCATE place_addressline');
         echo '.';

-        $this->pgExec('TRUNCATE place_boundingbox');
-        echo '.';

         $this->pgExec('TRUNCATE location_area');
         echo '.';
         if (!$this->dbReverseOnly()) {
         $this->pgExec('TRUNCATE location_area');
         echo '.';
         if (!$this->dbReverseOnly()) {


@@ -468,6 +456,10 @@ class SetupFunctions
     {
         info('Import Tiger data');
 
     {
         info('Import Tiger data');
 

+        $aFilenames = glob(CONST_Tiger_Data_Path.'/*.sql');
+        info('Found '.count($aFilenames).' SQL files in path '.CONST_Tiger_Data_Path);
+        if (empty($aFilenames)) return;
+

         $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
         $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
         $sTemplate = $this->replaceTablespace(
         $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
         $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
         $sTemplate = $this->replaceTablespace(


@@ -492,7 +484,7 @@ class SetupFunctions
             pg_ping($aDBInstances[$i]);
         }
 
             pg_ping($aDBInstances[$i]);
         }
 

-        foreach (glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile) {
+        foreach ($aFilenames as $sFile) {

             echo $sFile.': ';
             $hFile = fopen($sFile, 'r');
             $sSQL = fgets($hFile, 100000);
             echo $sFile.': ';
             $hFile = fopen($sFile, 'r');
             $sSQL = fgets($hFile, 100000);


@@ -595,13 +587,15 @@ class SetupFunctions
     public function index($bIndexNoanalyse)
     {
         $sOutputFile = '';
     public function index($bIndexNoanalyse)
     {
         $sOutputFile = '';

-        $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$this->aDSNInfo['database'].' -P '
-            .$this->aDSNInfo['port'].' -t '.$this->iInstances.$sOutputFile;
+        $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i'
+            .' -d '.escapeshellarg($this->aDSNInfo['database'])
+            .' -P '.escapeshellarg($this->aDSNInfo['port'])
+            .' -t '.escapeshellarg($this->iInstances.$sOutputFile);

         if (isset($this->aDSNInfo['hostspec'])) {
         if (isset($this->aDSNInfo['hostspec'])) {

-            $sBaseCmd .= ' -H '.$this->aDSNInfo['hostspec'];
+            $sBaseCmd .= ' -H '.escapeshellarg($this->aDSNInfo['hostspec']);

         }
         if (isset($this->aDSNInfo['username'])) {
         }
         if (isset($this->aDSNInfo['username'])) {

-            $sBaseCmd .= ' -U '.$this->aDSNInfo['username'];
+            $sBaseCmd .= ' -U '.escapeshellarg($this->aDSNInfo['username']);

         }
 
         info('Index ranks 0 - 4');
         }
 
         info('Index ranks 0 - 4');


@@ -725,9 +719,7 @@ class SetupFunctions
         }
         foreach ($aDropTables as $sDrop) {
             if ($this->bVerbose) echo "Dropping table $sDrop\n";
         }
         foreach ($aDropTables as $sDrop) {
             if ($this->bVerbose) echo "Dropping table $sDrop\n";

-            $this->oDB->exec("DROP TABLE $sDrop CASCADE");
-            // ignore warnings/errors as they might be caused by a table having
-            // been deleted already by CASCADE
+            $this->oDB->exec("DROP TABLE IF EXISTS $sDrop CASCADE");

         }
 
         if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {
         }
 
         if (!is_null(CONST_Osm2pgsql_Flatnode_File) && CONST_Osm2pgsql_Flatnode_File) {


@@ -738,22 +730,6 @@ class SetupFunctions
         }
     }
 
         }
     }
 

-    private function pgsqlRunDropAndRestore($sDumpFile)
-    {
-        $sCMD = 'pg_restore -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database'].' --no-owner -Fc --clean '.$sDumpFile;
-        if ($this->oDB->getPostgresVersion() >= 9.04) {
-            $sCMD .= ' --if-exists';
-        }
-        if (isset($this->aDSNInfo['hostspec'])) {
-            $sCMD .= ' -h '.$this->aDSNInfo['hostspec'];
-        }
-        if (isset($this->aDSNInfo['username'])) {
-            $sCMD .= ' -U '.$this->aDSNInfo['username'];
-        }
-
-        $this->runWithPgEnv($sCMD);
-    }
-

     private function pgsqlRunScript($sScript, $bfatal = true)
     {
         runSQLScript(
     private function pgsqlRunScript($sScript, $bfatal = true)
     {
         runSQLScript(


@@ -812,15 +788,17 @@ class SetupFunctions
     {
         if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 
     {
         if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 

-        $sCMD = 'psql -p '.$this->aDSNInfo['port'].' -d '.$this->aDSNInfo['database'];
+        $sCMD = 'psql'
+            .' -p '.escapeshellarg($this->aDSNInfo['port'])
+            .' -d '.escapeshellarg($this->aDSNInfo['database']);

         if (!$this->bVerbose) {
             $sCMD .= ' -q';
         }
         if (isset($this->aDSNInfo['hostspec'])) {
         if (!$this->bVerbose) {
             $sCMD .= ' -q';
         }
         if (isset($this->aDSNInfo['hostspec'])) {

-            $sCMD .= ' -h '.$this->aDSNInfo['hostspec'];
+            $sCMD .= ' -h '.escapeshellarg($this->aDSNInfo['hostspec']);

         }
         if (isset($this->aDSNInfo['username'])) {
         }
         if (isset($this->aDSNInfo['username'])) {

-            $sCMD .= ' -U '.$this->aDSNInfo['username'];
+            $sCMD .= ' -U '.escapeshellarg($this->aDSNInfo['username']);

         }
         $aProcEnv = null;
         if (isset($this->aDSNInfo['password'])) {
         }
         $aProcEnv = null;
         if (isset($this->aDSNInfo['password'])) {


@@ -833,12 +811,12 @@ class SetupFunctions
                              1 => array('pipe', 'w'),
                              2 => array('file', '/dev/null', 'a')
                             );
                              1 => array('pipe', 'w'),
                              2 => array('file', '/dev/null', 'a')
                             );

-            $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes);
+            $hGzipProcess = proc_open('zcat '.escapeshellarg($sFilename), $aDescriptors, $ahGzipPipes);

             if (!is_resource($hGzipProcess)) fail('unable to start zcat');
             $aReadPipe = $ahGzipPipes[1];
             fclose($ahGzipPipes[0]);
         } else {
             if (!is_resource($hGzipProcess)) fail('unable to start zcat');
             $aReadPipe = $ahGzipPipes[1];
             fclose($ahGzipPipes[0]);
         } else {

-            $sCMD .= ' -f '.$sFilename;
+            $sCMD .= ' -f '.escapeshellarg($sFilename);

             $aReadPipe = array('pipe', 'r');
         }
         $aDescriptors = array(
             $aReadPipe = array('pipe', 'r');
         }
         $aDescriptors = array(