utils/setup.php

   1 #!/usr/bin/php -Cq
   2 <?php
   3
   4         require_once(dirname(dirname(__FILE__)).'/settings/settings.php');
   5         require_once(CONST_BasePath.'/lib/init-cmd.php');
   6         ini_set('memory_limit', '800M');
   7
   8         $aCMDOptions = array(
   9                 "Create and setup nominatim search system",
  10                 array('help', 'h', 0, 1, 0, 0, false, 'Show Help'),
  11                 array('quiet', 'q', 0, 1, 0, 0, 'bool', 'Quiet output'),
  12                 array('verbose', 'v', 0, 1, 0, 0, 'bool', 'Verbose output'),
  13
  14                 array('osm-file', '', 0, 1, 1, 1, 'realpath', 'File to import'),
  15                 array('threads', '', 0, 1, 1, 1, 'int', 'Number of threads (where possible)'),
  16
  17                 array('all', '', 0, 1, 0, 0, 'bool', 'Do the complete process'),
  18
  19                 array('create-db', '', 0, 1, 0, 0, 'bool', 'Create nominatim db'),
  20                 array('setup-db', '', 0, 1, 0, 0, 'bool', 'Build a blank nominatim db'),
  21                 array('import-data', '', 0, 1, 0, 0, 'bool', 'Import a osm file'),
  22                 array('osm2pgsql-cache', '', 0, 1, 1, 1, 'int', 'Cache size used by osm2pgsql'),
  23                 array('create-functions', '', 0, 1, 0, 0, 'bool', 'Create functions'),
  24                 array('enable-diff-updates', '', 0, 1, 0, 0, 'bool', 'Turn on the code required to make diff updates work'),
  25                 array('enable-debug-statements', '', 0, 1, 0, 0, 'bool', 'Include debug warning statements in pgsql commands'),
  26                 array('ignore-errors', '', 0, 1, 0, 0, 'bool', 'Continue import even when errors in SQL are present (EXPERT)'),
  27                 array('create-tables', '', 0, 1, 0, 0, 'bool', 'Create main tables'),
  28                 array('create-partition-tables', '', 0, 1, 0, 0, 'bool', 'Create required partition tables'),
  29                 array('create-partition-functions', '', 0, 1, 0, 0, 'bool', 'Create required partition triggers'),
  30                 array('no-partitions', '', 0, 1, 0, 0, 'bool', "Do not partition search indices (speeds up import of single country extracts)"),
  31                 array('import-wikipedia-articles', '', 0, 1, 0, 0, 'bool', 'Import wikipedia article dump'),
  32                 array('load-data', '', 0, 1, 0, 0, 'bool', 'Copy data to live tables from import table'),
  33                 array('disable-token-precalc', '', 0, 1, 0, 0, 'bool', 'Disable name precalculation (EXPERT)'),
  34                 array('import-tiger-data', '', 0, 1, 0, 0, 'bool', 'Import tiger data (not included in \'all\')'),
  35                 array('calculate-postcodes', '', 0, 1, 0, 0, 'bool', 'Calculate postcode centroids'),
  36                 array('osmosis-init', '', 0, 1, 0, 0, 'bool', 'Generate default osmosis configuration'),
  37                 array('index', '', 0, 1, 0, 0, 'bool', 'Index the data'),
  38                 array('index-noanalyse', '', 0, 1, 0, 0, 'bool', 'Do not perform analyse operations during index (EXPERT)'),
  39                 array('create-search-indices', '', 0, 1, 0, 0, 'bool', 'Create additional indices required for search and update'),
  40                 array('create-website', '', 0, 1, 1, 1, 'realpath', 'Create symlinks to setup web directory'),
  41                 array('drop', '', 0, 1, 0, 0, 'bool', 'Drop tables needed for updates, making the database readonly (EXPERIMENTAL)'),
  42         );
  43         getCmdOpt($_SERVER['argv'], $aCMDOptions, $aCMDResult, true, true);
  44
  45         $bDidSomething = false;
  46
  47         // Check if osm-file is set and points to a valid file if --all or --import-data is given
  48         if ($aCMDResult['import-data'] || $aCMDResult['all'])
  49         {
  50                 if (!isset($aCMDResult['osm-file']))
  51                 {
  52                         fail('missing --osm-file for data import');
  53                 }
  54
  55                 if (!file_exists($aCMDResult['osm-file']))
  56                 {
  57                         fail('the path supplied to --osm-file does not exist');
  58                 }
  59
  60                 if (!is_readable($aCMDResult['osm-file']))
  61                 {
  62                         fail('osm-file "'.$aCMDResult['osm-file'].'" not readable');
  63                 }
  64         }
  65
  66
  67         // This is a pretty hard core default - the number of processors in the box - 1
  68         $iInstances = isset($aCMDResult['threads'])?$aCMDResult['threads']:(getProcessorCount()-1);
  69         if ($iInstances < 1)
  70         {
  71                 $iInstances = 1;
  72                 echo "WARNING: resetting threads to $iInstances\n";
  73         }
  74         if ($iInstances > getProcessorCount())
  75         {
  76                 $iInstances = getProcessorCount();
  77                 echo "WARNING: resetting threads to $iInstances\n";
  78         }
  79
  80         // Assume we can steal all the cache memory in the box (unless told otherwise)
  81         if (isset($aCMDResult['osm2pgsql-cache']))
  82         {
  83                 $iCacheMemory = $aCMDResult['osm2pgsql-cache'];
  84         }
  85         else
  86         {
  87                 $iCacheMemory = getCacheMemoryMB();
  88         }
  89
  90         $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
  91         if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
  92
  93         if ($aCMDResult['create-db'] || $aCMDResult['all'])
  94         {
  95                 echo "Create DB\n";
  96                 $bDidSomething = true;
  97                 $oDB = DB::connect(CONST_Database_DSN, false);
  98                 if (!PEAR::isError($oDB))
  99                 {
 100                         fail('database already exists ('.CONST_Database_DSN.')');
 101                 }
 102                 passthruCheckReturn('createdb -E UTF-8 -p '.$aDSNInfo['port'].' '.$aDSNInfo['database']);
 103         }
 104
 105         if ($aCMDResult['setup-db'] || $aCMDResult['all'])
 106         {
 107                 echo "Setup DB\n";
 108                 $bDidSomething = true;
 109                 // TODO: path detection, detection memory, etc.
 110
 111                 $oDB =& getDB();
 112
 113                 $fPostgresVersion = getPostgresVersion($oDB);
 114                 echo 'Postgres version found: '.$fPostgresVersion."\n";
 115
 116                 if ($fPostgresVersion < 9.1)
 117                 {
 118                         fail("Minimum supported version of Postgresql is 9.1.");
 119                 }
 120
 121                 pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS hstore');
 122                 pgsqlRunScript('CREATE EXTENSION IF NOT EXISTS postgis');
 123
 124                 // For extratags and namedetails the hstore_to_json converter is
 125                 // needed which is only available from Postgresql 9.3+. For older
 126                 // versions add a dummy function that returns nothing.
 127                 $iNumFunc = $oDB->getOne("select count(*) from pg_proc where proname = 'hstore_to_json'");
 128                 if (PEAR::isError($iNumFunc))
 129                 {
 130                         fail("Cannot query stored procedures.", $iNumFunc);
 131                 }
 132                 if ($iNumFunc == 0)
 133                 {
 134                         pgsqlRunScript("create function hstore_to_json(dummy hstore) returns text AS 'select null::text' language sql immutable");
 135                         echo "WARNING: Postgresql is too old. extratags and namedetails API not available.";
 136                 }
 137
 138                 $fPostgisVersion = getPostgisVersion($oDB);
 139                 echo 'Postgis version found: '.$fPostgisVersion."\n";
 140
 141                 if ($fPostgisVersion < 2.1)
 142                 {
 143                         // Function was renamed in 2.1 and throws an annoying deprecation warning
 144                         pgsqlRunScript('ALTER FUNCTION st_line_interpolate_point(geometry, double precision) RENAME TO ST_LineInterpolatePoint');
 145                 }
 146
 147                 pgsqlRunScriptFile(CONST_BasePath.'/data/country_name.sql');
 148                 pgsqlRunScriptFile(CONST_BasePath.'/data/country_naturalearthdata.sql');
 149                 pgsqlRunScriptFile(CONST_BasePath.'/data/country_osm_grid.sql');
 150                 pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_table.sql');
 151                 if (file_exists(CONST_BasePath.'/data/gb_postcode_data.sql.gz'))
 152                 {
 153                         pgsqlRunScriptFile(CONST_BasePath.'/data/gb_postcode_data.sql.gz');
 154                 }
 155                 else
 156                 {
 157                         echo "WARNING: external UK postcode table not found.\n";
 158                 }
 159                 if (CONST_Use_Extra_US_Postcodes)
 160                 {
 161                         pgsqlRunScriptFile(CONST_BasePath.'/data/us_postcode.sql');
 162                 }
 163
 164                 if ($aCMDResult['no-partitions'])
 165                 {
 166                         pgsqlRunScript('update country_name set partition = 0');
 167                 }
 168
 169                 // the following will be needed by create_functions later but
 170                 // is only defined in the subsequently called create_tables.
 171                 // Create dummies here that will be overwritten by the proper
 172                 // versions in create-tables.
 173                 pgsqlRunScript('CREATE TABLE place_boundingbox ()');
 174                 pgsqlRunScript('create type wikipedia_article_match as ()');
 175         }
 176
 177         if ($aCMDResult['import-data'] || $aCMDResult['all'])
 178         {
 179                 echo "Import\n";
 180                 $bDidSomething = true;
 181
 182                 $osm2pgsql = CONST_Osm2pgsql_Binary;
 183                 if (!file_exists($osm2pgsql))
 184                 {
 185                         echo "Please download and build osm2pgsql.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
 186                         fail("osm2pgsql not found in '$osm2pgsql'");
 187                 }
 188
 189                 if (!is_null(CONST_Osm2pgsql_Flatnode_File))
 190                 {
 191                         $osm2pgsql .= ' --flat-nodes '.CONST_Osm2pgsql_Flatnode_File;
 192                 }
 193                 if (CONST_Tablespace_Osm2pgsql_Data)
 194                         $osm2pgsql .= ' --tablespace-slim-data '.CONST_Tablespace_Osm2pgsql_Data;
 195                 if (CONST_Tablespace_Osm2pgsql_Index)
 196                         $osm2pgsql .= ' --tablespace-slim-index '.CONST_Tablespace_Osm2pgsql_Index;
 197                 if (CONST_Tablespace_Place_Data)
 198                         $osm2pgsql .= ' --tablespace-main-data '.CONST_Tablespace_Place_Data;
 199                 if (CONST_Tablespace_Place_Index)
 200                         $osm2pgsql .= ' --tablespace-main-index '.CONST_Tablespace_Place_Index;
 201                 $osm2pgsql .= ' -lsc -O gazetteer --hstore --number-processes 1';
 202                 $osm2pgsql .= ' -C '.$iCacheMemory;
 203                 $osm2pgsql .= ' -P '.$aDSNInfo['port'];
 204                 $osm2pgsql .= ' -d '.$aDSNInfo['database'].' '.$aCMDResult['osm-file'];
 205                 passthruCheckReturn($osm2pgsql);
 206
 207                 $oDB =& getDB();
 208                 $x = $oDB->getRow('select * from place limit 1');
 209                 if (PEAR::isError($x)) {
 210                         fail($x->getMessage());
 211                 }
 212                 if (!$x) fail('No Data');
 213         }
 214
 215         if ($aCMDResult['create-functions'] || $aCMDResult['all'])
 216         {
 217                 echo "Functions\n";
 218                 $bDidSomething = true;
 219                 if (!file_exists(CONST_InstallPath.'/module/nominatim.so')) fail("nominatim module not built");
 220                 create_sql_functions($aCMDResult);
 221         }
 222
 223         if ($aCMDResult['create-tables'] || $aCMDResult['all'])
 224         {
 225                 $bDidSomething = true;
 226
 227                 echo "Tables\n";
 228                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/tables.sql');
 229                 $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
 230                 $sTemplate = replace_tablespace('{ts:address-data}',
 231                                                 CONST_Tablespace_Address_Data, $sTemplate);
 232                 $sTemplate = replace_tablespace('{ts:address-index}',
 233                                                 CONST_Tablespace_Address_Index, $sTemplate);
 234                 $sTemplate = replace_tablespace('{ts:search-data}',
 235                                                 CONST_Tablespace_Search_Data, $sTemplate);
 236                 $sTemplate = replace_tablespace('{ts:search-index}',
 237                                                 CONST_Tablespace_Search_Index, $sTemplate);
 238                 $sTemplate = replace_tablespace('{ts:aux-data}',
 239                                                 CONST_Tablespace_Aux_Data, $sTemplate);
 240                 $sTemplate = replace_tablespace('{ts:aux-index}',
 241                                                 CONST_Tablespace_Aux_Index, $sTemplate);
 242                 pgsqlRunScript($sTemplate, false);
 243
 244                 // re-run the functions
 245                 echo "Functions\n";
 246                 create_sql_functions($aCMDResult);
 247         }
 248
 249         if ($aCMDResult['create-partition-tables'] || $aCMDResult['all'])
 250         {
 251                 echo "Partition Tables\n";
 252                 $bDidSomething = true;
 253                 $oDB =& getDB();
 254                 $sSQL = 'select distinct partition from country_name';
 255                 $aPartitions = $oDB->getCol($sSQL);
 256                 if (PEAR::isError($aPartitions))
 257                 {
 258                         fail($aPartitions->getMessage());
 259                 }
 260                 if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 261
 262                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-tables.src.sql');
 263                 $sTemplate = replace_tablespace('{ts:address-data}',
 264                                                 CONST_Tablespace_Address_Data, $sTemplate);
 265                 $sTemplate = replace_tablespace('{ts:address-index}',
 266                                                 CONST_Tablespace_Address_Index, $sTemplate);
 267                 $sTemplate = replace_tablespace('{ts:search-data}',
 268                                                 CONST_Tablespace_Search_Data, $sTemplate);
 269                 $sTemplate = replace_tablespace('{ts:search-index}',
 270                                                 CONST_Tablespace_Search_Index, $sTemplate);
 271                 $sTemplate = replace_tablespace('{ts:aux-data}',
 272                                                 CONST_Tablespace_Aux_Data, $sTemplate);
 273                 $sTemplate = replace_tablespace('{ts:aux-index}',
 274                                                 CONST_Tablespace_Aux_Index, $sTemplate);
 275                 preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
 276                 foreach($aMatches as $aMatch)
 277                 {
 278                         $sResult = '';
 279                         foreach($aPartitions as $sPartitionName)
 280                         {
 281                                 $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
 282                         }
 283                         $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
 284                 }
 285
 286                 pgsqlRunScript($sTemplate);
 287         }
 288
 289
 290         if ($aCMDResult['create-partition-functions'] || $aCMDResult['all'])
 291         {
 292                 echo "Partition Functions\n";
 293                 $bDidSomething = true;
 294                 $oDB =& getDB();
 295                 $sSQL = 'select distinct partition from country_name';
 296                 $aPartitions = $oDB->getCol($sSQL);
 297                 if (PEAR::isError($aPartitions))
 298                 {
 299                         fail($aPartitions->getMessage());
 300                 }
 301                 if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 302
 303                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/partition-functions.src.sql');
 304                 preg_match_all('#^-- start(.*?)^-- end#ms', $sTemplate, $aMatches, PREG_SET_ORDER);
 305                 foreach($aMatches as $aMatch)
 306                 {
 307                         $sResult = '';
 308                         foreach($aPartitions as $sPartitionName)
 309                         {
 310                                 $sResult .= str_replace('-partition-', $sPartitionName, $aMatch[1]);
 311                         }
 312                         $sTemplate = str_replace($aMatch[0], $sResult, $sTemplate);
 313                 }
 314
 315                 pgsqlRunScript($sTemplate);
 316         }
 317
 318         if ($aCMDResult['import-wikipedia-articles'] || $aCMDResult['all'])
 319         {
 320                 $bDidSomething = true;
 321                 $sWikiArticlesFile = CONST_BasePath.'/data/wikipedia_article.sql.bin';
 322                 $sWikiRedirectsFile = CONST_BasePath.'/data/wikipedia_redirect.sql.bin';
 323                 if (file_exists($sWikiArticlesFile))
 324                 {
 325                         echo "Importing wikipedia articles...";
 326                         pgsqlRunDropAndRestore($sWikiArticlesFile);
 327                         echo "...done\n";
 328                 }
 329                 else
 330                 {
 331                         echo "WARNING: wikipedia article dump file not found - places will have default importance\n";
 332                 }
 333                 if (file_exists($sWikiRedirectsFile))
 334                 {
 335                         echo "Importing wikipedia redirects...";
 336                         pgsqlRunDropAndRestore($sWikiRedirectsFile);
 337                         echo "...done\n";
 338                 }
 339                 else
 340                 {
 341                         echo "WARNING: wikipedia redirect dump file not found - some place importance values may be missing\n";
 342                 }
 343         }
 344
 345
 346         if ($aCMDResult['load-data'] || $aCMDResult['all'])
 347         {
 348                 echo "Drop old Data\n";
 349                 $bDidSomething = true;
 350
 351                 $oDB =& getDB();
 352                 if (!pg_query($oDB->connection, 'TRUNCATE word')) fail(pg_last_error($oDB->connection));
 353                 echo '.';
 354                 if (!pg_query($oDB->connection, 'TRUNCATE placex')) fail(pg_last_error($oDB->connection));
 355                 echo '.';
 356                 if (!pg_query($oDB->connection, 'TRUNCATE location_property_osmline')) fail(pg_last_error($oDB->connection));
 357                 echo '.';
 358                 if (!pg_query($oDB->connection, 'TRUNCATE place_addressline')) fail(pg_last_error($oDB->connection));
 359                 echo '.';
 360                 if (!pg_query($oDB->connection, 'TRUNCATE place_boundingbox')) fail(pg_last_error($oDB->connection));
 361                 echo '.';
 362                 if (!pg_query($oDB->connection, 'TRUNCATE location_area')) fail(pg_last_error($oDB->connection));
 363                 echo '.';
 364                 if (!pg_query($oDB->connection, 'TRUNCATE search_name')) fail(pg_last_error($oDB->connection));
 365                 echo '.';
 366                 if (!pg_query($oDB->connection, 'TRUNCATE search_name_blank')) fail(pg_last_error($oDB->connection));
 367                 echo '.';
 368                 if (!pg_query($oDB->connection, 'DROP SEQUENCE seq_place')) fail(pg_last_error($oDB->connection));
 369                 echo '.';
 370                 if (!pg_query($oDB->connection, 'CREATE SEQUENCE seq_place start 100000')) fail(pg_last_error($oDB->connection));
 371                 echo '.';
 372
 373                 $sSQL = 'select distinct partition from country_name';
 374                 $aPartitions = $oDB->getCol($sSQL);
 375                 if (PEAR::isError($aPartitions))
 376                 {
 377                         fail($aPartitions->getMessage());
 378                 }
 379                 if (!$aCMDResult['no-partitions']) $aPartitions[] = 0;
 380                 foreach($aPartitions as $sPartition)
 381                 {
 382                         if (!pg_query($oDB->connection, 'TRUNCATE location_road_'.$sPartition)) fail(pg_last_error($oDB->connection));
 383                         echo '.';
 384                 }
 385
 386                 // used by getorcreate_word_id to ignore frequent partial words
 387                 if (!pg_query($oDB->connection, 'CREATE OR REPLACE FUNCTION get_maxwordfreq() RETURNS integer AS $$ SELECT '.CONST_Max_Word_Frequency.' as maxwordfreq; $$ LANGUAGE SQL IMMUTABLE')) fail(pg_last_error($oDB->connection));
 388                 echo ".\n";
 389
 390                 // pre-create the word list
 391                 if (!$aCMDResult['disable-token-precalc'])
 392                 {
 393                         echo "Loading word list\n";
 394                         pgsqlRunScriptFile(CONST_BasePath.'/data/words.sql');
 395                 }
 396
 397                 echo "Load Data\n";
 398                 $aDBInstances = array();
 399                 $iLoadThreads = max(1, $iInstances - 1);
 400                 for($i = 0; $i < $iLoadThreads; $i++)
 401                 {
 402                         $aDBInstances[$i] =& getDB(true);
 403                         $sSQL = 'insert into placex (osm_type, osm_id, class, type, name, admin_level, ';
 404                         $sSQL .= 'housenumber, street, addr_place, isin, postcode, country_code, extratags, ';
 405                         $sSQL .= 'geometry) select * from place where osm_id % '.$iLoadThreads.' = '.$i;
 406                         $sSQL .= " and not (class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString')";
 407                         if ($aCMDResult['verbose']) echo "$sSQL\n";
 408                         if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 409                 }
 410                 // last thread for interpolation lines
 411                 $aDBInstances[$iLoadThreads] =& getDB(true);
 412                 $sSQL = 'select insert_osmline (osm_id, housenumber, street, addr_place, postcode, country_code, ';
 413                 $sSQL .= 'geometry) from place where ';
 414                 $sSQL .= "class='place' and type='houses' and osm_type='W' and ST_GeometryType(geometry) = 'ST_LineString'";
 415                 if ($aCMDResult['verbose']) echo "$sSQL\n";
 416                 if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 417
 418                 $bAnyBusy = true;
 419                 while($bAnyBusy)
 420                 {
 421                         $bAnyBusy = false;
 422                         for($i = 0; $i <= $iLoadThreads; $i++)
 423                         {
 424                                 if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
 425                         }
 426                         sleep(1);
 427                         echo '.';
 428                 }
 429                 echo "\n";
 430                 echo "Reanalysing database...\n";
 431                 pgsqlRunScript('ANALYSE');
 432         }
 433
 434         if ($aCMDResult['import-tiger-data'])
 435         {
 436                 $bDidSomething = true;
 437
 438                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_start.sql');
 439                 $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
 440                 $sTemplate = replace_tablespace('{ts:aux-data}',
 441                                                 CONST_Tablespace_Aux_Data, $sTemplate);
 442                 $sTemplate = replace_tablespace('{ts:aux-index}',
 443                                                 CONST_Tablespace_Aux_Index, $sTemplate);
 444                 pgsqlRunScript($sTemplate, false);
 445
 446                 $aDBInstances = array();
 447                 for($i = 0; $i < $iInstances; $i++)
 448                 {
 449                         $aDBInstances[$i] =& getDB(true);
 450                 }
 451
 452                 foreach(glob(CONST_Tiger_Data_Path.'/*.sql') as $sFile)
 453                 {
 454                         echo $sFile.': ';
 455                         $hFile = fopen($sFile, "r");
 456                         $sSQL = fgets($hFile, 100000);
 457                         $iLines = 0;
 458
 459                         while(true)
 460                         {
 461                                 for($i = 0; $i < $iInstances; $i++)
 462                                 {
 463                                         if (!pg_connection_busy($aDBInstances[$i]->connection))
 464                                         {
 465                                                 while(pg_get_result($aDBInstances[$i]->connection));
 466                                                 $sSQL = fgets($hFile, 100000);
 467                                                 if (!$sSQL) break 2;
 468                                                 if (!pg_send_query($aDBInstances[$i]->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 469                                                 $iLines++;
 470                                                 if ($iLines == 1000)
 471                                                 {
 472                                                         echo ".";
 473                                                         $iLines = 0;
 474                                                 }
 475                                         }
 476                                 }
 477                                 usleep(10);
 478                         }
 479
 480                         fclose($hFile);
 481
 482                         $bAnyBusy = true;
 483                         while($bAnyBusy)
 484                         {
 485                                 $bAnyBusy = false;
 486                                 for($i = 0; $i < $iInstances; $i++)
 487                                 {
 488                                         if (pg_connection_busy($aDBInstances[$i]->connection)) $bAnyBusy = true;
 489                                 }
 490                                 usleep(10);
 491                         }
 492                         echo "\n";
 493                 }
 494
 495                 echo "Creating indexes\n";
 496                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/tiger_import_finish.sql');
 497                 $sTemplate = str_replace('{www-user}', CONST_Database_Web_User, $sTemplate);
 498                 $sTemplate = replace_tablespace('{ts:aux-data}',
 499                                                 CONST_Tablespace_Aux_Data, $sTemplate);
 500                 $sTemplate = replace_tablespace('{ts:aux-index}',
 501                                                 CONST_Tablespace_Aux_Index, $sTemplate);
 502                 pgsqlRunScript($sTemplate, false);
 503         }
 504
 505         if ($aCMDResult['calculate-postcodes'] || $aCMDResult['all'])
 506         {
 507                 $bDidSomething = true;
 508                 $oDB =& getDB();
 509                 if (!pg_query($oDB->connection, 'DELETE from placex where osm_type=\'P\'')) fail(pg_last_error($oDB->connection));
 510                 $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
 511                 $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,calculated_country_code,";
 512                 $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from (select calculated_country_code,postcode,";
 513                 $sSQL .= "avg(st_x(st_centroid(geometry))) as x,avg(st_y(st_centroid(geometry))) as y ";
 514                 $sSQL .= "from placex where postcode is not null group by calculated_country_code,postcode) as x";
 515                 if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 516
 517                 if (CONST_Use_Extra_US_Postcodes)
 518                 {
 519                         $sSQL = "insert into placex (osm_type,osm_id,class,type,postcode,calculated_country_code,geometry) ";
 520                         $sSQL .= "select 'P',nextval('seq_postcodes'),'place','postcode',postcode,'us',";
 521                         $sSQL .= "ST_SetSRID(ST_Point(x,y),4326) as geometry from us_postcode";
 522                         if (!pg_query($oDB->connection, $sSQL)) fail(pg_last_error($oDB->connection));
 523                 }
 524         }
 525
 526         if ($aCMDResult['osmosis-init'] || ($aCMDResult['all'] && !$aCMDResult['drop'])) // no use doing osmosis-init when dropping update tables
 527         {
 528                 $bDidSomething = true;
 529                 $oDB =& getDB();
 530
 531                 if (!file_exists(CONST_Osmosis_Binary))
 532                 {
 533                         echo "Please download osmosis.\nIf it is already installed, check the path in your local settings (settings/local.php) file.\n";
 534                         if (!$aCMDResult['all'])
 535                         {
 536                                 fail("osmosis not found in '".CONST_Osmosis_Binary."'");
 537                         }
 538                 }
 539                 else
 540                 {
 541                         if (file_exists(CONST_InstallPath.'/settings/configuration.txt'))
 542                         {
 543                                 echo "settings/configuration.txt already exists\n";
 544                         }
 545                         else
 546                         {
 547                                 passthru(CONST_Osmosis_Binary.' --read-replication-interval-init '.CONST_InstallPath.'/settings');
 548                                 // update osmosis configuration.txt with our settings
 549                                 passthru("sed -i 's!baseUrl=.*!baseUrl=".CONST_Replication_Url."!' ".CONST_InstallPath.'/settings/configuration.txt');
 550                                 passthru("sed -i 's:maxInterval = .*:maxInterval = ".CONST_Replication_MaxInterval.":' ".CONST_InstallPath.'/settings/configuration.txt');
 551                         }
 552
 553                         // Find the last node in the DB
 554                         $iLastOSMID = $oDB->getOne("select max(osm_id) from place where osm_type = 'N'");
 555
 556                         // Lookup the timestamp that node was created (less 3 hours for margin for changsets to be closed)
 557                         $sLastNodeURL = 'http://www.openstreetmap.org/api/0.6/node/'.$iLastOSMID."/1";
 558                         $sLastNodeXML = file_get_contents($sLastNodeURL);
 559                         preg_match('#timestamp="(([0-9]{4})-([0-9]{2})-([0-9]{2})T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)"#', $sLastNodeXML, $aLastNodeDate);
 560                         $iLastNodeTimestamp = strtotime($aLastNodeDate[1]) - (3*60*60);
 561
 562                         // Search for the correct state file - uses file timestamps so need to sort by date descending
 563                         $sRepURL = CONST_Replication_Url."/";
 564                         $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
 565                         // download.geofabrik.de:    <a href="000/">000/</a></td><td align="right">26-Feb-2013 11:53  </td>
 566                         // planet.openstreetmap.org: <a href="273/">273/</a>                    2013-03-11 07:41    -
 567                         preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
 568                         if ($aRepMatches)
 569                         {
 570                                 $aPrevRepMatch = false;
 571                                 foreach($aRepMatches as $aRepMatch)
 572                                 {
 573                                         if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
 574                                         $aPrevRepMatch = $aRepMatch;
 575                                 }
 576                                 if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
 577
 578                                 $sRepURL .= $aRepMatch[1];
 579                                 $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
 580                                 preg_match_all('#<a href="[0-9]{3}/">([0-9]{3}/)</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
 581                                 $aPrevRepMatch = false;
 582                                 foreach($aRepMatches as $aRepMatch)
 583                                 {
 584                                         if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
 585                                         $aPrevRepMatch = $aRepMatch;
 586                                 }
 587                                 if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
 588
 589                                 $sRepURL .= $aRepMatch[1];
 590                                 $sRep = file_get_contents($sRepURL."?C=M;O=D;F=1");
 591                                 preg_match_all('#<a href="[0-9]{3}.state.txt">([0-9]{3}).state.txt</a>\s*([-0-9a-zA-Z]+ [0-9]{2}:[0-9]{2})#', $sRep, $aRepMatches, PREG_SET_ORDER);
 592                                 $aPrevRepMatch = false;
 593                                 foreach($aRepMatches as $aRepMatch)
 594                                 {
 595                                         if (strtotime($aRepMatch[2]) < $iLastNodeTimestamp) break;
 596                                         $aPrevRepMatch = $aRepMatch;
 597                                 }
 598                                 if ($aPrevRepMatch) $aRepMatch = $aPrevRepMatch;
 599
 600                                 $sRepURL .= $aRepMatch[1].'.state.txt';
 601                                 echo "Getting state file: $sRepURL\n";
 602                                 $sStateFile = file_get_contents($sRepURL);
 603                                 if (!$sStateFile || strlen($sStateFile) > 1000) fail("unable to obtain state file");
 604                                 file_put_contents(CONST_InstallPath.'/settings/state.txt', $sStateFile);
 605                                 echo "Updating DB status\n";
 606                                 pg_query($oDB->connection, 'TRUNCATE import_status');
 607                                 $sSQL = "INSERT INTO import_status VALUES('".$aRepMatch[2]."')";
 608                                 pg_query($oDB->connection, $sSQL);
 609                         }
 610                         else
 611                         {
 612                                 if (!$aCMDResult['all'])
 613                                 {
 614                                         fail("Cannot read state file directory.");
 615                                 }
 616                         }
 617                 }
 618         }
 619
 620         if ($aCMDResult['index'] || $aCMDResult['all'])
 621         {
 622                 $bDidSomething = true;
 623                 $sOutputFile = '';
 624                 $sBaseCmd = CONST_InstallPath.'/nominatim/nominatim -i -d '.$aDSNInfo['database'].' -P '.$aDSNInfo['port'].' -t '.$iInstances.$sOutputFile;
 625                 passthruCheckReturn($sBaseCmd.' -R 4');
 626                 if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
 627                 passthruCheckReturn($sBaseCmd.' -r 5 -R 25');
 628                 if (!$aCMDResult['index-noanalyse']) pgsqlRunScript('ANALYSE');
 629                 passthruCheckReturn($sBaseCmd.' -r 26');
 630         }
 631
 632         if ($aCMDResult['create-search-indices'] || $aCMDResult['all'])
 633         {
 634                 echo "Search indices\n";
 635                 $bDidSomething = true;
 636
 637                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/indices.src.sql');
 638                 $sTemplate = replace_tablespace('{ts:address-index}',
 639                                                 CONST_Tablespace_Address_Index, $sTemplate);
 640                 $sTemplate = replace_tablespace('{ts:search-index}',
 641                                                 CONST_Tablespace_Search_Index, $sTemplate);
 642                 $sTemplate = replace_tablespace('{ts:aux-index}',
 643                                                 CONST_Tablespace_Aux_Index, $sTemplate);
 644
 645                 pgsqlRunScript($sTemplate);
 646         }
 647
 648         if (isset($aCMDResult['create-website']))
 649         {
 650                 $bDidSomething = true;
 651                 $sTargetDir = $aCMDResult['create-website'];
 652                 if (!is_dir($sTargetDir))
 653                 {
 654                         echo "You must create the website directory before calling this function.\n";
 655                         fail("Target directory does not exist.");
 656                 }
 657
 658                 @symlink(CONST_InstallPath.'/website/details.php', $sTargetDir.'/details.php');
 659                 @symlink(CONST_InstallPath.'/website/reverse.php', $sTargetDir.'/reverse.php');
 660                 @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/search.php');
 661                 @symlink(CONST_InstallPath.'/website/search.php', $sTargetDir.'/index.php');
 662                 @symlink(CONST_InstallPath.'/website/lookup.php', $sTargetDir.'/lookup.php');
 663                 @symlink(CONST_InstallPath.'/website/deletable.php', $sTargetDir.'/deletable.php');
 664                 @symlink(CONST_InstallPath.'/website/polygons.php', $sTargetDir.'/polygons.php');
 665                 @symlink(CONST_InstallPath.'/website/status.php', $sTargetDir.'/status.php');
 666                 @symlink(CONST_BasePath.'/website/images', $sTargetDir.'/images');
 667                 @symlink(CONST_BasePath.'/website/js', $sTargetDir.'/js');
 668                 @symlink(CONST_BasePath.'/website/css', $sTargetDir.'/css');
 669                 echo "Symlinks created\n";
 670
 671                 $sTestFile = @file_get_contents(CONST_Website_BaseURL.'js/nominatim-ui.js');
 672                 if (!$sTestFile)
 673                 {
 674                         echo "\nWARNING: Unable to access the website at ".CONST_Website_BaseURL."\n";
 675                         echo "You may want to update settings/local.php with @define('CONST_Website_BaseURL', 'http://[HOST]/[PATH]/');\n";
 676                 }
 677         }
 678
 679         if ($aCMDResult['drop'])
 680         {
 681                 // The implementation is potentially a bit dangerous because it uses
 682                 // a positive selection of tables to keep, and deletes everything else.
 683                 // Including any tables that the unsuspecting user might have manually
 684                 // created. USE AT YOUR OWN PERIL.
 685                 $bDidSomething = true;
 686
 687                 // tables we want to keep. everything else goes.
 688                 $aKeepTables = array(
 689                    "*columns",
 690                    "import_polygon_*",
 691                    "import_status",
 692                    "place_addressline",
 693                    "location_property*",
 694                    "placex",
 695                    "search_name",
 696                    "seq_*",
 697                    "word",
 698                    "query_log",
 699                    "new_query_log",
 700                    "gb_postcode",
 701                    "spatial_ref_sys",
 702                    "country_name",
 703                    "place_classtype_*"
 704                 );
 705
 706                 $oDB =& getDB();
 707                 $aDropTables = array();
 708                 $aHaveTables = $oDB->getCol("SELECT tablename FROM pg_tables WHERE schemaname='public'");
 709                 if (PEAR::isError($aHaveTables))
 710                 {
 711                         fail($aPartitions->getMessage());
 712                 }
 713                 foreach($aHaveTables as $sTable)
 714                 {
 715                         $bFound = false;
 716                         foreach ($aKeepTables as $sKeep)
 717                         {
 718                                 if (fnmatch($sKeep, $sTable))
 719                                 {
 720                                         $bFound = true;
 721                                         break;
 722                                 }
 723                         }
 724                         if (!$bFound) array_push($aDropTables, $sTable);
 725                 }
 726
 727                 foreach ($aDropTables as $sDrop)
 728                 {
 729                         if ($aCMDResult['verbose']) echo "dropping table $sDrop\n";
 730                         @pg_query($oDB->connection, "DROP TABLE $sDrop CASCADE");
 731                         // ignore warnings/errors as they might be caused by a table having
 732                         // been deleted already by CASCADE
 733                 }
 734
 735                 if (!is_null(CONST_Osm2pgsql_Flatnode_File))
 736                 {
 737                         if ($aCMDResult['verbose']) echo "deleting ".CONST_Osm2pgsql_Flatnode_File."\n";
 738                         unlink(CONST_Osm2pgsql_Flatnode_File);
 739                 }
 740         }
 741
 742         if (!$bDidSomething)
 743         {
 744                 showUsage($aCMDOptions, true);
 745         }
 746         else
 747         {
 748                 echo "Setup finished.\n";
 749         }
 750
 751         function pgsqlRunScriptFile($sFilename)
 752         {
 753                 if (!file_exists($sFilename)) fail('unable to find '.$sFilename);
 754
 755                 // Convert database DSN to psql parameters
 756                 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 757                 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 758                 $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
 759
 760                 $ahGzipPipes = null;
 761                 if (preg_match('/\\.gz$/', $sFilename))
 762                 {
 763                         $aDescriptors = array(
 764                                 0 => array('pipe', 'r'),
 765                                 1 => array('pipe', 'w'),
 766                                 2 => array('file', '/dev/null', 'a')
 767                         );
 768                         $hGzipProcess = proc_open('zcat '.$sFilename, $aDescriptors, $ahGzipPipes);
 769                         if (!is_resource($hGzipProcess)) fail('unable to start zcat');
 770                         $aReadPipe = $ahGzipPipes[1];
 771                         fclose($ahGzipPipes[0]);
 772                 }
 773                 else
 774                 {
 775                         $sCMD .= ' -f '.$sFilename;
 776                         $aReadPipe = array('pipe', 'r');
 777                 }
 778
 779                 $aDescriptors = array(
 780                         0 => $aReadPipe,
 781                         1 => array('pipe', 'w'),
 782                         2 => array('file', '/dev/null', 'a')
 783                 );
 784                 $ahPipes = null;
 785                 $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
 786                 if (!is_resource($hProcess)) fail('unable to start pgsql');
 787
 788
 789                 // TODO: error checking
 790                 while(!feof($ahPipes[1]))
 791                 {
 792                         echo fread($ahPipes[1], 4096);
 793                 }
 794                 fclose($ahPipes[1]);
 795
 796                 $iReturn = proc_close($hProcess);
 797                 if ($iReturn > 0)
 798                 {
 799                         fail("pgsql returned with error code ($iReturn)");
 800                 }
 801                 if ($ahGzipPipes)
 802                 {
 803                         fclose($ahGzipPipes[1]);
 804                         proc_close($hGzipProcess);
 805                 }
 806
 807         }
 808
 809         function pgsqlRunScript($sScript, $bfatal = true)
 810         {
 811                 global $aCMDResult;
 812                 // Convert database DSN to psql parameters
 813                 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 814                 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 815                 $sCMD = 'psql -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'];
 816                 if ($bfatal && !$aCMDResult['ignore-errors'])
 817                         $sCMD .= ' -v ON_ERROR_STOP=1';
 818                 $aDescriptors = array(
 819                         0 => array('pipe', 'r'),
 820                         1 => STDOUT,
 821                         2 => STDERR
 822                 );
 823                 $ahPipes = null;
 824                 $hProcess = @proc_open($sCMD, $aDescriptors, $ahPipes);
 825                 if (!is_resource($hProcess)) fail('unable to start pgsql');
 826
 827                 while(strlen($sScript))
 828                 {
 829                         $written = fwrite($ahPipes[0], $sScript);
 830                         if ($written <= 0) break;
 831                         $sScript = substr($sScript, $written);
 832                 }
 833                 fclose($ahPipes[0]);
 834                 $iReturn = proc_close($hProcess);
 835                 if ($bfatal && $iReturn > 0)
 836                 {
 837                         fail("pgsql returned with error code ($iReturn)");
 838                 }
 839         }
 840
 841         function pgsqlRunRestoreData($sDumpFile)
 842         {
 843                 // Convert database DSN to psql parameters
 844                 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 845                 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 846                 $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc -a '.$sDumpFile;
 847
 848                 $aDescriptors = array(
 849                         0 => array('pipe', 'r'),
 850                         1 => array('pipe', 'w'),
 851                         2 => array('file', '/dev/null', 'a')
 852                 );
 853                 $ahPipes = null;
 854                 $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
 855                 if (!is_resource($hProcess)) fail('unable to start pg_restore');
 856
 857                 fclose($ahPipes[0]);
 858
 859                 // TODO: error checking
 860                 while(!feof($ahPipes[1]))
 861                 {
 862                         echo fread($ahPipes[1], 4096);
 863                 }
 864                 fclose($ahPipes[1]);
 865
 866                 $iReturn = proc_close($hProcess);
 867         }
 868
 869         function pgsqlRunDropAndRestore($sDumpFile)
 870         {
 871                 // Convert database DSN to psql parameters
 872                 $aDSNInfo = DB::parseDSN(CONST_Database_DSN);
 873                 if (!isset($aDSNInfo['port']) || !$aDSNInfo['port']) $aDSNInfo['port'] = 5432;
 874                 $sCMD = 'pg_restore -p '.$aDSNInfo['port'].' -d '.$aDSNInfo['database'].' -Fc --clean '.$sDumpFile;
 875
 876                 $aDescriptors = array(
 877                         0 => array('pipe', 'r'),
 878                         1 => array('pipe', 'w'),
 879                         2 => array('file', '/dev/null', 'a')
 880                 );
 881                 $ahPipes = null;
 882                 $hProcess = proc_open($sCMD, $aDescriptors, $ahPipes);
 883                 if (!is_resource($hProcess)) fail('unable to start pg_restore');
 884
 885                 fclose($ahPipes[0]);
 886
 887                 // TODO: error checking
 888                 while(!feof($ahPipes[1]))
 889                 {
 890                         echo fread($ahPipes[1], 4096);
 891                 }
 892                 fclose($ahPipes[1]);
 893
 894                 $iReturn = proc_close($hProcess);
 895         }
 896
 897         function passthruCheckReturn($cmd)
 898         {
 899                 $result = -1;
 900                 passthru($cmd, $result);
 901                 if ($result != 0) fail('Error executing external command: '.$cmd);
 902         }
 903
 904         function replace_tablespace($sTemplate, $sTablespace, $sSql)
 905         {
 906                 if ($sTablespace)
 907                         $sSql = str_replace($sTemplate, 'TABLESPACE "'.$sTablespace.'"',
 908                                             $sSql);
 909                 else
 910                         $sSql = str_replace($sTemplate, '', $sSql);
 911
 912                 return $sSql;
 913         }
 914
 915         function create_sql_functions($aCMDResult)
 916         {
 917                 $sTemplate = file_get_contents(CONST_BasePath.'/sql/functions.sql');
 918                 $sTemplate = str_replace('{modulepath}', CONST_InstallPath.'/module', $sTemplate);
 919                 if ($aCMDResult['enable-diff-updates'])
 920                 {
 921                         $sTemplate = str_replace('RETURN NEW; -- %DIFFUPDATES%', '--', $sTemplate);
 922                 }
 923                 if ($aCMDResult['enable-debug-statements'])
 924                 {
 925                         $sTemplate = str_replace('--DEBUG:', '', $sTemplate);
 926                 }
 927                 if (CONST_Limit_Reindexing)
 928                 {
 929                         $sTemplate = str_replace('--LIMIT INDEXING:', '', $sTemplate);
 930                 }
 931                 if (!CONST_Use_US_Tiger_Data)
 932                 {
 933                         $sTemplate = str_replace('-- %NOTIGERDATA% ', '', $sTemplate);
 934                 }
 935                 if (!CONST_Use_Aux_Location_data)
 936                 {
 937                         $sTemplate = str_replace('-- %NOAUXDATA% ', '', $sTemplate);
 938                 }
 939                 pgsqlRunScript($sTemplate);
 940
 941         }
 942