lib-php/setup/SetupClass.php

   1 <?php
   2
   3 namespace Nominatim\Setup;
   4
   5 require_once(CONST_LibDir.'/Shell.php');
   6
   7 class SetupFunctions
   8 {
   9     protected $iInstances;
  10     protected $aDSNInfo;
  11     protected $bQuiet;
  12     protected $bVerbose;
  13     protected $sIgnoreErrors;
  14     protected $bEnableDiffUpdates;
  15     protected $bEnableDebugStatements;
  16     protected $bDrop;
  17     protected $oDB = null;
  18     protected $oNominatimCmd;
  19
  20     public function __construct(array $aCMDResult)
  21     {
  22         // by default, use all but one processor, but never more than 15.
  23         $this->iInstances = isset($aCMDResult['threads'])
  24             ? $aCMDResult['threads']
  25             : (min(16, getProcessorCount()) - 1);
  26
  27         if ($this->iInstances < 1) {
  28             $this->iInstances = 1;
  29             warn('resetting threads to '.$this->iInstances);
  30         }
  31
  32         // parse database string
  33         $this->aDSNInfo = \Nominatim\DB::parseDSN(getSetting('DATABASE_DSN'));
  34         if (!isset($this->aDSNInfo['port'])) {
  35             $this->aDSNInfo['port'] = 5432;
  36         }
  37
  38         // setting member variables based on command line options stored in $aCMDResult
  39         $this->bQuiet = isset($aCMDResult['quiet']) && $aCMDResult['quiet'];
  40         $this->bVerbose = $aCMDResult['verbose'];
  41
  42         //setting default values which are not set by the update.php array
  43         if (isset($aCMDResult['ignore-errors'])) {
  44             $this->sIgnoreErrors = $aCMDResult['ignore-errors'];
  45         } else {
  46             $this->sIgnoreErrors = false;
  47         }
  48         if (isset($aCMDResult['enable-debug-statements'])) {
  49             $this->bEnableDebugStatements = $aCMDResult['enable-debug-statements'];
  50         } else {
  51             $this->bEnableDebugStatements = false;
  52         }
  53         if (isset($aCMDResult['enable-diff-updates'])) {
  54             $this->bEnableDiffUpdates = $aCMDResult['enable-diff-updates'];
  55         } else {
  56             $this->bEnableDiffUpdates = false;
  57         }
  58
  59         $this->bDrop = isset($aCMDResult['drop']) && $aCMDResult['drop'];
  60
  61         $this->oNominatimCmd = new \Nominatim\Shell(getSetting('NOMINATIM_TOOL'));
  62         if ($this->bQuiet) {
  63             $this->oNominatimCmd->addParams('--quiet');
  64         }
  65         if ($this->bVerbose) {
  66             $this->oNominatimCmd->addParams('--verbose');
  67         }
  68     }
  69
  70     public function createFunctions()
  71     {
  72         info('Create Functions');
  73
  74         // Try accessing the C module, so we know early if something is wrong
  75         $this->checkModulePresence(); // raises exception on failure
  76
  77         $this->createSqlFunctions();
  78     }
  79
  80     public function importTigerData($sTigerPath)
  81     {
  82         info('Import Tiger data');
  83
  84         $aFilenames = glob($sTigerPath.'/*.sql');
  85         info('Found '.count($aFilenames).' SQL files in path '.$sTigerPath);
  86         if (empty($aFilenames)) {
  87             warn('Tiger data import selected but no files found in path '.$sTigerPath);
  88             return;
  89         }
  90         $sTemplate = file_get_contents(CONST_SqlDir.'/tiger_import_start.sql');
  91         $sTemplate = $this->replaceSqlPatterns($sTemplate);
  92
  93         $this->pgsqlRunScript($sTemplate, false);
  94
  95         $aDBInstances = array();
  96         for ($i = 0; $i < $this->iInstances; $i++) {
  97             // https://secure.php.net/manual/en/function.pg-connect.php
  98             $DSN = getSetting('DATABASE_DSN');
  99             $DSN = preg_replace('/^pgsql:/', '', $DSN);
 100             $DSN = preg_replace('/;/', ' ', $DSN);
 101             $aDBInstances[$i] = pg_connect($DSN, PGSQL_CONNECT_FORCE_NEW | PGSQL_CONNECT_ASYNC);
 102             pg_ping($aDBInstances[$i]);
 103         }
 104
 105         foreach ($aFilenames as $sFile) {
 106             echo $sFile.': ';
 107             $hFile = fopen($sFile, 'r');
 108             $sSQL = fgets($hFile, 100000);
 109             $iLines = 0;
 110             while (true) {
 111                 for ($i = 0; $i < $this->iInstances; $i++) {
 112                     if (!pg_connection_busy($aDBInstances[$i])) {
 113                         while (pg_get_result($aDBInstances[$i]));
 114                         $sSQL = fgets($hFile, 100000);
 115                         if (!$sSQL) break 2;
 116                         if (!pg_send_query($aDBInstances[$i], $sSQL)) fail(pg_last_error($aDBInstances[$i]));
 117                         $iLines++;
 118                         if ($iLines == 1000) {
 119                             echo '.';
 120                             $iLines = 0;
 121                         }
 122                     }
 123                 }
 124                 usleep(10);
 125             }
 126             fclose($hFile);
 127
 128             $bAnyBusy = true;
 129             while ($bAnyBusy) {
 130                 $bAnyBusy = false;
 131                 for ($i = 0; $i < $this->iInstances; $i++) {
 132                     if (pg_connection_busy($aDBInstances[$i])) $bAnyBusy = true;
 133                 }
 134                 usleep(10);
 135             }
 136             echo "\n";
 137         }
 138
 139         for ($i = 0; $i < $this->iInstances; $i++) {
 140             pg_close($aDBInstances[$i]);
 141         }
 142
 143         info('Creating indexes on Tiger data');
 144         $sTemplate = file_get_contents(CONST_SqlDir.'/tiger_import_finish.sql');
 145         $sTemplate = $this->replaceSqlPatterns($sTemplate);
 146
 147         $this->pgsqlRunScript($sTemplate, false);
 148     }
 149
 150     public function calculatePostcodes($bCMDResultAll)
 151     {
 152         info('Calculate Postcodes');
 153         $this->pgsqlRunScriptFile(CONST_SqlDir.'/postcode_tables.sql');
 154
 155         $sPostcodeFilename = CONST_InstallDir.'/gb_postcode_data.sql.gz';
 156         if (file_exists($sPostcodeFilename)) {
 157             $this->pgsqlRunScriptFile($sPostcodeFilename);
 158         } else {
 159             warn('optional external GB postcode table file ('.$sPostcodeFilename.') not found. Skipping.');
 160         }
 161
 162         $sPostcodeFilename = CONST_InstallDir.'/us_postcode_data.sql.gz';
 163         if (file_exists($sPostcodeFilename)) {
 164             $this->pgsqlRunScriptFile($sPostcodeFilename);
 165         } else {
 166             warn('optional external US postcode table file ('.$sPostcodeFilename.') not found. Skipping.');
 167         }
 168
 169
 170         $this->db()->exec('TRUNCATE location_postcode');
 171
 172         $sSQL  = 'INSERT INTO location_postcode';
 173         $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) ';
 174         $sSQL .= "SELECT nextval('seq_place'), 1, country_code,";
 175         $sSQL .= "       upper(trim (both ' ' from address->'postcode')) as pc,";
 176         $sSQL .= '       ST_Centroid(ST_Collect(ST_Centroid(geometry)))';
 177         $sSQL .= '  FROM placex';
 178         $sSQL .= " WHERE address ? 'postcode' AND address->'postcode' NOT SIMILAR TO '%(,|;)%'";
 179         $sSQL .= '       AND geometry IS NOT null';
 180         $sSQL .= ' GROUP BY country_code, pc';
 181         $this->db()->exec($sSQL);
 182
 183         // only add postcodes that are not yet available in OSM
 184         $sSQL  = 'INSERT INTO location_postcode';
 185         $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) ';
 186         $sSQL .= "SELECT nextval('seq_place'), 1, 'us', postcode,";
 187         $sSQL .= '       ST_SetSRID(ST_Point(x,y),4326)';
 188         $sSQL .= '  FROM us_postcode WHERE postcode NOT IN';
 189         $sSQL .= '        (SELECT postcode FROM location_postcode';
 190         $sSQL .= "          WHERE country_code = 'us')";
 191         $this->db()->exec($sSQL);
 192
 193         // add missing postcodes for GB (if available)
 194         $sSQL  = 'INSERT INTO location_postcode';
 195         $sSQL .= ' (place_id, indexed_status, country_code, postcode, geometry) ';
 196         $sSQL .= "SELECT nextval('seq_place'), 1, 'gb', postcode, geometry";
 197         $sSQL .= '  FROM gb_postcode WHERE postcode NOT IN';
 198         $sSQL .= '           (SELECT postcode FROM location_postcode';
 199         $sSQL .= "             WHERE country_code = 'gb')";
 200         $this->db()->exec($sSQL);
 201
 202         if (!$bCMDResultAll) {
 203             $sSQL = "DELETE FROM word WHERE class='place' and type='postcode'";
 204             $sSQL .= 'and word NOT IN (SELECT postcode FROM location_postcode)';
 205             $this->db()->exec($sSQL);
 206         }
 207
 208         $sSQL = 'SELECT count(getorcreate_postcode_id(v)) FROM ';
 209         $sSQL .= '(SELECT distinct(postcode) as v FROM location_postcode) p';
 210         $this->db()->exec($sSQL);
 211     }
 212
 213     public function createSearchIndices()
 214     {
 215         info('Create Search indices');
 216
 217         $sSQL = 'SELECT relname FROM pg_class, pg_index ';
 218         $sSQL .= 'WHERE pg_index.indisvalid = false AND pg_index.indexrelid = pg_class.oid';
 219         $aInvalidIndices = $this->db()->getCol($sSQL);
 220
 221         foreach ($aInvalidIndices as $sIndexName) {
 222             info("Cleaning up invalid index $sIndexName");
 223             $this->db()->exec("DROP INDEX $sIndexName;");
 224         }
 225
 226         $sTemplate = file_get_contents(CONST_SqlDir.'/indices.src.sql');
 227         if (!$this->bDrop) {
 228             $sTemplate .= file_get_contents(CONST_SqlDir.'/indices_updates.src.sql');
 229         }
 230         if (!$this->dbReverseOnly()) {
 231             $sTemplate .= file_get_contents(CONST_SqlDir.'/indices_search.src.sql');
 232         }
 233         $sTemplate = $this->replaceSqlPatterns($sTemplate);
 234
 235         $this->pgsqlRunScript($sTemplate);
 236     }
 237
 238     public function createCountryNames()
 239     {
 240         info('Create search index for default country names');
 241
 242         $this->pgsqlRunScript("select getorcreate_country(make_standard_name('uk'), 'gb')");
 243         $this->pgsqlRunScript("select getorcreate_country(make_standard_name('united states'), 'us')");
 244         $this->pgsqlRunScript('select count(*) from (select getorcreate_country(make_standard_name(country_code), country_code) from country_name where country_code is not null) as x');
 245         $this->pgsqlRunScript("select count(*) from (select getorcreate_country(make_standard_name(name->'name'), country_code) from country_name where name ? 'name') as x");
 246         $sSQL = 'select count(*) from (select getorcreate_country(make_standard_name(v),'
 247             .'country_code) from (select country_code, skeys(name) as k, svals(name) as v from country_name) x where k ';
 248         $sLanguages = getSetting('LANGUAGES');
 249         if ($sLanguages) {
 250             $sSQL .= 'in ';
 251             $sDelim = '(';
 252             foreach (explode(',', $sLanguages) as $sLang) {
 253                 $sSQL .= $sDelim."'name:$sLang'";
 254                 $sDelim = ',';
 255             }
 256             $sSQL .= ')';
 257         } else {
 258             // all include all simple name tags
 259             $sSQL .= "like 'name:%'";
 260         }
 261         $sSQL .= ') v';
 262         $this->pgsqlRunScript($sSQL);
 263     }
 264
 265     /**
 266      * Return the connection to the database.
 267      *
 268      * @return Database object.
 269      *
 270      * Creates a new connection if none exists yet. Otherwise reuses the
 271      * already established connection.
 272      */
 273     private function db()
 274     {
 275         if (is_null($this->oDB)) {
 276             $this->oDB = new \Nominatim\DB();
 277             $this->oDB->connect();
 278         }
 279
 280         return $this->oDB;
 281     }
 282
 283     private function pgsqlRunScript($sScript, $bfatal = true)
 284     {
 285         runSQLScript(
 286             $sScript,
 287             $bfatal,
 288             $this->bVerbose,
 289             $this->sIgnoreErrors
 290         );
 291     }
 292
 293     private function createSqlFunctions()
 294     {
 295         $oCmd = (clone($this->oNominatimCmd))
 296                 ->addParams('refresh', '--functions');
 297
 298         if (!$this->bEnableDiffUpdates) {
 299             $oCmd->addParams('--no-diff-updates');
 300         }
 301
 302         if ($this->bEnableDebugStatements) {
 303             $oCmd->addParams('--enable-debug-statements');
 304         }
 305
 306         $oCmd->run(!$this->sIgnoreErrors);
 307     }
 308
 309     private function pgsqlRunScriptFile($sFilename)
 310     {
 311         if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 312
 313         $oCmd = (new \Nominatim\Shell('psql'))
 314                 ->addParams('--port', $this->aDSNInfo['port'])
 315                 ->addParams('--dbname', $this->aDSNInfo['database']);
 316
 317         if (!$this->bVerbose) {
 318             $oCmd->addParams('--quiet');
 319         }
 320         if (isset($this->aDSNInfo['hostspec'])) {
 321             $oCmd->addParams('--host', $this->aDSNInfo['hostspec']);
 322         }
 323         if (isset($this->aDSNInfo['username'])) {
 324             $oCmd->addParams('--username', $this->aDSNInfo['username']);
 325         }
 326         if (isset($this->aDSNInfo['password'])) {
 327             $oCmd->addEnvPair('PGPASSWORD', $this->aDSNInfo['password']);
 328         }
 329         $ahGzipPipes = null;
 330         if (preg_match('/\\.gz$/', $sFilename)) {
 331             $aDescriptors = array(
 332                              0 => array('pipe', 'r'),
 333                              1 => array('pipe', 'w'),
 334                              2 => array('file', '/dev/null', 'a')
 335                             );
 336             $oZcatCmd = new \Nominatim\Shell('zcat', $sFilename);
 337
 338             $hGzipProcess = proc_open($oZcatCmd->escapedCmd(), $aDescriptors, $ahGzipPipes);
 339             if (!is_resource($hGzipProcess)) fail('unable to start zcat');
 340             $aReadPipe = $ahGzipPipes[1];
 341             fclose($ahGzipPipes[0]);
 342         } else {
 343             $oCmd->addParams('--file', $sFilename);
 344             $aReadPipe = array('pipe', 'r');
 345         }
 346         $aDescriptors = array(
 347                          0 => $aReadPipe,
 348                          1 => array('pipe', 'w'),
 349                          2 => array('file', '/dev/null', 'a')
 350                         );
 351         $ahPipes = null;
 352
 353         $hProcess = proc_open($oCmd->escapedCmd(), $aDescriptors, $ahPipes, null, $oCmd->aEnv);
 354         if (!is_resource($hProcess)) fail('unable to start pgsql');
 355         // TODO: error checking
 356         while (!feof($ahPipes[1])) {
 357             echo fread($ahPipes[1], 4096);
 358         }
 359         fclose($ahPipes[1]);
 360         $iReturn = proc_close($hProcess);
 361         if ($iReturn > 0) {
 362             fail("pgsql returned with error code ($iReturn)");
 363         }
 364         if ($ahGzipPipes) {
 365             fclose($ahGzipPipes[1]);
 366             proc_close($hGzipProcess);
 367         }
 368     }
 369
 370     private function replaceSqlPatterns($sSql)
 371     {
 372         $sSql = str_replace('{www-user}', getSetting('DATABASE_WEBUSER'), $sSql);
 373
 374         $aPatterns = array(
 375                       '{ts:address-data}' => getSetting('TABLESPACE_ADDRESS_DATA'),
 376                       '{ts:address-index}' => getSetting('TABLESPACE_ADDRESS_INDEX'),
 377                       '{ts:search-data}' => getSetting('TABLESPACE_SEARCH_DATA'),
 378                       '{ts:search-index}' =>  getSetting('TABLESPACE_SEARCH_INDEX'),
 379                       '{ts:aux-data}' =>  getSetting('TABLESPACE_AUX_DATA'),
 380                       '{ts:aux-index}' =>  getSetting('TABLESPACE_AUX_INDEX')
 381         );
 382
 383         foreach ($aPatterns as $sPattern => $sTablespace) {
 384             if ($sTablespace) {
 385                 $sSql = str_replace($sPattern, 'TABLESPACE "'.$sTablespace.'"', $sSql);
 386             } else {
 387                 $sSql = str_replace($sPattern, '', $sSql);
 388             }
 389         }
 390
 391         return $sSql;
 392     }
 393
 394     /**
 395      * Drop table with the given name if it exists.
 396      *
 397      * @param string $sName Name of table to remove.
 398      *
 399      * @return null
 400      */
 401     private function dropTable($sName)
 402     {
 403         if ($this->bVerbose) echo "Dropping table $sName\n";
 404         $this->db()->deleteTable($sName);
 405     }
 406
 407     /**
 408      * Check if the database is in reverse-only mode.
 409      *
 410      * @return True if there is no search_name table and infrastructure.
 411      */
 412     private function dbReverseOnly()
 413     {
 414         return !($this->db()->tableExists('search_name'));
 415     }
 416
 417     /**
 418      * Try accessing the C module, so we know early if something is wrong.
 419      *
 420      * Raises Nominatim\DatabaseError on failure
 421      */
 422     private function checkModulePresence()
 423     {
 424         $sModulePath = getSetting('DATABASE_MODULE_PATH', CONST_InstallDir.'/module');
 425         $sSQL = "CREATE FUNCTION nominatim_test_import_func(text) RETURNS text AS '";
 426         $sSQL .= $sModulePath . "/nominatim.so', 'transliteration' LANGUAGE c IMMUTABLE STRICT";
 427         $sSQL .= ';DROP FUNCTION nominatim_test_import_func(text);';
 428
 429         $oDB = new \Nominatim\DB();
 430         $oDB->connect();
 431         $oDB->exec($sSQL, null, 'Database server failed to load '.$sModulePath.'/nominatim.so module');
 432     }
 433 }