nominatim/index.c

   1 /*
   2  * triggers indexing (reparenting etc.) through setting resetting indexed_status: update placex/osmline set indexed_status = 0 where indexed_status > 0
   3  * triggers placex_update and osmline_update
   4 */
   5
   6 #include <stdio.h>
   7 #include <unistd.h>
   8 #include <stdlib.h>
   9 #include <string.h>
  10 #include <assert.h>
  11 #include <pthread.h>
  12 #include <time.h>
  13 #include <stdint.h>
  14
  15 #include <libpq-fe.h>
  16
  17 #include "nominatim.h"
  18 #include "index.h"
  19 #include "export.h"
  20 #include "postgresql.h"
  21
  22 extern int verbose;
  23
  24 void run_indexing(int rank, int interpolation, PGconn *conn, int num_threads,
  25 struct index_thread_data * thread_data, const char *structuredoutputfile)
  26 {
  27     int tuples, count, sleepcount;
  28     pthread_mutex_t count_mutex = PTHREAD_MUTEX_INITIALIZER;
  29
  30     time_t rankStartTime;
  31     int rankTotalTuples;
  32     int rankCountTuples;
  33     float rankPerSecond;
  34
  35     PGresult * resSectors;
  36     PGresult * resPlaces;
  37     PGresult * resNULL;
  38
  39     int i;
  40     int iSector;
  41     int iResult;
  42
  43     const char *paramValues[2];
  44     int         paramLengths[2];
  45     int         paramFormats[2];
  46     uint32_t    paramRank;
  47     uint32_t    paramSector;
  48     uint32_t    sector;
  49
  50     xmlTextWriterPtr writer;
  51     pthread_mutex_t writer_mutex = PTHREAD_MUTEX_INITIALIZER;
  52
  53     // Create the output file
  54     writer = NULL;
  55     if (structuredoutputfile)
  56     {
  57         writer = nominatim_exportXMLStart(structuredoutputfile);
  58     }
  59
  60     if (interpolation)
  61     {
  62         fprintf(stderr, "Starting interpolation lines (location_property_osmline)\n");
  63     }
  64     else
  65     {
  66         fprintf(stderr, "Starting rank %d\n", rank);
  67     }
  68
  69     rankCountTuples = 0;
  70     rankPerSecond = 0;
  71
  72     paramRank = PGint32(rank);
  73     paramValues[0] = (char *)&paramRank;
  74     paramLengths[0] = sizeof(paramRank);
  75     paramFormats[0] = 1;
  76
  77     if (interpolation)
  78     {
  79         resSectors = PQexecPrepared(conn, "index_sectors_osmline", 0, NULL, 0, NULL, 1);
  80     }
  81     else
  82     {
  83         resSectors = PQexecPrepared(conn, "index_sectors", 1, paramValues, paramLengths, paramFormats, 1);
  84     }
  85     if (PQresultStatus(resSectors) != PGRES_TUPLES_OK)
  86     {
  87         fprintf(stderr, "index_sectors: SELECT failed: %s", PQerrorMessage(conn));
  88         PQclear(resSectors);
  89         exit(EXIT_FAILURE);
  90     }
  91     if (PQftype(resSectors, 0) != PG_OID_INT4)
  92     {
  93         fprintf(stderr, "Sector value has unexpected type\n");
  94         PQclear(resSectors);
  95         exit(EXIT_FAILURE);
  96     }
  97     if (PQftype(resSectors, 1) != PG_OID_INT8)
  98     {
  99         fprintf(stderr, "Sector value has unexpected type\n");
 100         PQclear(resSectors);
 101         exit(EXIT_FAILURE);
 102     }
 103
 104     rankTotalTuples = 0;
 105     for (iSector = 0; iSector < PQntuples(resSectors); iSector++)
 106     {
 107         rankTotalTuples += PGint64(*((uint64_t *)PQgetvalue(resSectors, iSector, 1)));
 108     }
 109
 110     rankStartTime = time(0);
 111     for (iSector = 0; iSector <= PQntuples(resSectors); iSector++)
 112     {
 113         if (iSector > 0)
 114         {
 115             resPlaces = PQgetResult(conn);
 116             if (PQresultStatus(resPlaces) != PGRES_TUPLES_OK)
 117             {
 118                 fprintf(stderr, "index_sector_places: SELECT failed: %s", PQerrorMessage(conn));
 119                 PQclear(resPlaces);
 120                 exit(EXIT_FAILURE);
 121             }
 122             if (PQftype(resPlaces, 0) != PG_OID_INT8)
 123             {
 124                 fprintf(stderr, "Place_id value has unexpected type\n");
 125                 PQclear(resPlaces);
 126                 exit(EXIT_FAILURE);
 127             }
 128             resNULL = PQgetResult(conn);
 129             if (resNULL != NULL)
 130             {
 131                 fprintf(stderr, "Unexpected non-null response\n");
 132                 exit(EXIT_FAILURE);
 133             }
 134         }
 135
 136         if (iSector < PQntuples(resSectors))
 137         {
 138             sector = PGint32(*((uint32_t *)PQgetvalue(resSectors, iSector, 0)));
 139 //                fprintf(stderr, "\n Starting sector %d size %ld\n", sector, PGint64(*((uint64_t *)PQgetvalue(resSectors, iSector, 1))));
 140
 141             // Get all the place_id's for this sector
 142             paramRank = PGint32(rank);
 143             paramValues[0] = (char *)&paramRank;
 144             paramLengths[0] = sizeof(paramRank);
 145             paramFormats[0] = 1;
 146             paramSector = PGint32(sector);
 147             paramValues[1] = (char *)&paramSector;
 148             paramLengths[1] = sizeof(paramSector);
 149             paramFormats[1] = 1;
 150             if (rankTotalTuples-rankCountTuples < num_threads*1000)
 151             {
 152                 // no sectors
 153                 if (interpolation)
 154                 {
 155                     iResult = PQsendQueryPrepared(conn, "index_nosector_places_osmline", 0, NULL, 0, NULL, 1);
 156                 }
 157                 else
 158                 {
 159                     iResult = PQsendQueryPrepared(conn, "index_nosector_places", 1, paramValues, paramLengths, paramFormats, 1);
 160                 }
 161             }
 162             else
 163             {
 164                 if (interpolation)
 165                 {
 166                     iResult = PQsendQueryPrepared(conn, "index_sector_places", 2, paramValues, paramLengths, paramFormats, 1);
 167                 }
 168                 else
 169                 {
 170                     iResult = PQsendQueryPrepared(conn, "index_sector_places_osmline", 1, paramValues, paramLengths, paramFormats, 1);
 171                 }
 172             }
 173             if (!iResult)
 174             {
 175                 fprintf(stderr, "index_sector_places: SELECT failed: %s", PQerrorMessage(conn));
 176                 PQclear(resPlaces);
 177                 exit(EXIT_FAILURE);
 178             }
 179         }
 180         if (iSector > 0)
 181         {
 182             count = 0;
 183             rankPerSecond = 0;
 184             tuples = PQntuples(resPlaces);
 185
 186             if (tuples > 0)
 187             {
 188                 // Spawn threads
 189                 for (i = 0; i < num_threads; i++)
 190                 {
 191                     thread_data[i].res = resPlaces;
 192                     thread_data[i].tuples = tuples;
 193                     thread_data[i].count = &count;
 194                     thread_data[i].count_mutex = &count_mutex;
 195                     thread_data[i].writer = writer;
 196                     thread_data[i].writer_mutex = &writer_mutex;
 197                     if (interpolation)
 198                     {
 199                         thread_data[i].table = 0;  // use interpolations table
 200                     }
 201                     else
 202                     {
 203                         thread_data[i].table = 1;  // use placex table
 204                     }
 205                     pthread_create(&thread_data[i].thread, NULL, &nominatim_indexThread, (void *)&thread_data[i]);
 206                 }
 207
 208                 // Monitor threads to give user feedback
 209                 sleepcount = 0;
 210                 while (count < tuples)
 211                 {
 212                     usleep(1000);
 213
 214                     // Aim for one update per second
 215                     if (sleepcount++ > 500)
 216                     {
 217                         rankPerSecond = ((float)rankCountTuples + (float)count) / MAX(difftime(time(0), rankStartTime),1);
 218                         fprintf(stderr, "  Done %i in %i @ %f per second - Rank %i ETA (seconds): %f\n", (rankCountTuples + count), (int)(difftime(time(0), rankStartTime)), rankPerSecond, rank, ((float)(rankTotalTuples - (rankCountTuples + count)))/rankPerSecond);
 219                         sleepcount = 0;
 220                     }
 221                 }
 222
 223                 // Wait for everything to finish
 224                 for (i = 0; i < num_threads; i++)
 225                 {
 226                     pthread_join(thread_data[i].thread, NULL);
 227                 }
 228
 229                 rankCountTuples += tuples;
 230             }
 231
 232             // Finished sector
 233             rankPerSecond = (float)rankCountTuples / MAX(difftime(time(0), rankStartTime),1);
 234             fprintf(stderr, "  Done %i in %i @ %f per second - ETA (seconds): %f\n", rankCountTuples, (int)(difftime(time(0), rankStartTime)), rankPerSecond, ((float)(rankTotalTuples - rankCountTuples))/rankPerSecond);
 235
 236             PQclear(resPlaces);
 237         }
 238         if (rankTotalTuples-rankCountTuples < num_threads*20 && iSector < PQntuples(resSectors))
 239         {
 240             iSector = PQntuples(resSectors) - 1;
 241         }
 242     }
 243     // Finished rank
 244     fprintf(stderr, "\r  Done %i in %i @ %f per second - FINISHED\n\n", rankCountTuples, (int)(difftime(time(0), rankStartTime)), rankPerSecond);
 245
 246     PQclear(resSectors);
 247
 248
 249 }
 250
 251 void nominatim_index(int rank_min, int rank_max, int num_threads, const char *conninfo, const char *structuredoutputfile)
 252 {
 253     struct index_thread_data * thread_data;
 254
 255     PGconn *conn;
 256     PGresult * res;
 257
 258     int rank;
 259
 260     int i;
 261
 262     xmlTextWriterPtr writer;
 263     pthread_mutex_t writer_mutex = PTHREAD_MUTEX_INITIALIZER;
 264
 265     Oid pg_prepare_params[2];
 266
 267     conn = PQconnectdb(conninfo);
 268     if (PQstatus(conn) != CONNECTION_OK)
 269     {
 270         fprintf(stderr, "Connection to database failed: %s\n", PQerrorMessage(conn));
 271         exit(EXIT_FAILURE);
 272     }
 273
 274     pg_prepare_params[0] = PG_OID_INT4;
 275     res = PQprepare(conn, "index_sectors",
 276                     "select geometry_sector,count(*) from placex where rank_search = $1 and indexed_status > 0 group by geometry_sector order by geometry_sector",
 277                     1, pg_prepare_params);
 278     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 279     {
 280         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 281         exit(EXIT_FAILURE);
 282     }
 283     PQclear(res);
 284
 285     res = PQprepare(conn, "index_sectors_osmline",
 286                     "select geometry_sector,count(*) from location_property_osmline where indexed_status > 0 group by geometry_sector order by geometry_sector",
 287                     0, NULL);
 288     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 289     {
 290         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 291         exit(EXIT_FAILURE);
 292     }
 293     PQclear(res);
 294
 295     pg_prepare_params[0] = PG_OID_INT4;
 296     res = PQprepare(conn, "index_nosectors",
 297                     "select 0::integer,count(*) from placex where rank_search = $1 and indexed_status > 0",
 298                     1, pg_prepare_params);
 299     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 300     {
 301         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 302         exit(EXIT_FAILURE);
 303     }
 304     PQclear(res);
 305
 306     pg_prepare_params[0] = PG_OID_INT4;
 307     pg_prepare_params[1] = PG_OID_INT4;
 308     res = PQprepare(conn, "index_sector_places",
 309                     "select place_id from placex where rank_search = $1 and geometry_sector = $2 and indexed_status > 0",
 310                     2, pg_prepare_params);
 311     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 312     {
 313         fprintf(stderr, "Failed preparing index_sector_places: %s\n", PQerrorMessage(conn));
 314         exit(EXIT_FAILURE);
 315     }
 316     PQclear(res);
 317
 318     pg_prepare_params[0] = PG_OID_INT4;
 319     res = PQprepare(conn, "index_nosector_places",
 320                     "select place_id from placex where rank_search = $1 and indexed_status > 0 order by geometry_sector",
 321                     1, pg_prepare_params);
 322     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 323     {
 324         fprintf(stderr, "Failed preparing index_nosector_places: %s\n", PQerrorMessage(conn));
 325         exit(EXIT_FAILURE);
 326     }
 327     PQclear(res);
 328
 329     pg_prepare_params[0] = PG_OID_INT4;
 330     res = PQprepare(conn, "index_sector_places_osmline",
 331                     "select place_id from location_property_osmline where geometry_sector = $1 and indexed_status > 0",
 332                     1, pg_prepare_params);
 333     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 334     {
 335         fprintf(stderr, "Failed preparing index_sector_places: %s\n", PQerrorMessage(conn));
 336         exit(EXIT_FAILURE);
 337     }
 338     PQclear(res);
 339
 340     res = PQprepare(conn, "index_nosector_places_osmline",
 341                     "select place_id from location_property_osmline where indexed_status > 0 order by geometry_sector",
 342                     0, NULL);
 343     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 344     {
 345         fprintf(stderr, "Failed preparing index_nosector_places: %s\n", PQerrorMessage(conn));
 346         exit(EXIT_FAILURE);
 347     }
 348     PQclear(res);
 349
 350     // Build the data for each thread
 351     thread_data = (struct index_thread_data *)malloc(sizeof(struct index_thread_data)*num_threads);
 352     for (i = 0; i < num_threads; i++)
 353     {
 354         thread_data[i].conn = PQconnectdb(conninfo);
 355         if (PQstatus(thread_data[i].conn) != CONNECTION_OK)
 356         {
 357             fprintf(stderr, "Connection to database failed: %s\n", PQerrorMessage(thread_data[i].conn));
 358             exit(EXIT_FAILURE);
 359         }
 360
 361         pg_prepare_params[0] = PG_OID_INT8;
 362         res = PQprepare(thread_data[i].conn, "index_placex",
 363                         "update placex set indexed_status = 0 where place_id = $1",
 364                         1, pg_prepare_params);
 365         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 366         {
 367             fprintf(stderr, "Failed preparing index_placex: %s\n", PQerrorMessage(conn));
 368             exit(EXIT_FAILURE);
 369         }
 370         PQclear(res);
 371
 372         pg_prepare_params[0] = PG_OID_INT8;
 373         res = PQprepare(thread_data[i].conn, "index_osmline",
 374                         "update location_property_osmline set indexed_status = 0 where place_id = $1",
 375                         1, pg_prepare_params);
 376         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 377         {
 378             fprintf(stderr, "Failed preparing index_osmline: %s\n", PQerrorMessage(conn));
 379             exit(EXIT_FAILURE);
 380         }
 381         PQclear(res);
 382
 383         /*res = PQexec(thread_data[i].conn, "set enable_seqscan = false");
 384         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 385         {
 386             fprintf(stderr, "Failed disabling sequential scan: %s\n", PQerrorMessage(conn));
 387             exit(EXIT_FAILURE);
 388         }
 389         PQclear(res);*/
 390
 391         nominatim_exportCreatePreparedQueries(thread_data[i].conn);
 392     }
 393
 394
 395     fprintf(stderr, "Starting indexing rank (%i to %i) using %i threads\n", rank_min, rank_max, num_threads);
 396
 397     for (rank = rank_min; rank <= rank_max; rank++)
 398     {
 399         // OSMLINE: do reindexing (=> reparenting) for interpolation lines at rank 30, but before all other objects of rank 30
 400         // reason: houses (rank 30) depend on the updated interpolation line, when reparenting (see placex_update in functions.sql)
 401         if (rank == 30)
 402         {
 403             run_indexing(rank, 1, conn, num_threads, thread_data, structuredoutputfile);
 404         }
 405         run_indexing(rank, 0, conn, num_threads, thread_data, structuredoutputfile);
 406     }
 407
 408
 409     if (rank == 30)
 410     {
 411         // Close all connections
 412         for (i = 0; i < num_threads; i++)
 413         {
 414             PQfinish(thread_data[i].conn);
 415         }
 416         PQfinish(conn);
 417     }
 418 }
 419
 420 void *nominatim_indexThread(void * thread_data_in)
 421 {
 422     struct index_thread_data * thread_data = (struct index_thread_data * )thread_data_in;
 423     struct export_data  querySet;
 424
 425     PGresult   *res;
 426
 427     const char  *paramValues[1];
 428     int         paramLengths[1];
 429     int         paramFormats[1];
 430     uint64_t    paramPlaceID;
 431     uint64_t    place_id;
 432     time_t      updateStartTime;
 433     uint        table;
 434
 435     table = (uint)(thread_data->table);
 436
 437     while (1)
 438     {
 439         pthread_mutex_lock( thread_data->count_mutex );
 440         if (*(thread_data->count) >= thread_data->tuples)
 441         {
 442             pthread_mutex_unlock( thread_data->count_mutex );
 443             break;
 444         }
 445
 446         place_id = PGint64(*((uint64_t *)PQgetvalue(thread_data->res, *thread_data->count, 0)));
 447         (*thread_data->count)++;
 448
 449         pthread_mutex_unlock( thread_data->count_mutex );
 450
 451         if (verbose) fprintf(stderr, "  Processing place_id %ld\n", place_id);
 452
 453         updateStartTime = time(0);
 454         int done = 0;
 455
 456         if (thread_data->writer)
 457         {
 458              nominatim_exportPlaceQueries(place_id, thread_data->conn, &querySet);
 459         }
 460
 461         while(!done)
 462         {
 463             paramPlaceID = PGint64(place_id);
 464             paramValues[0] = (char *)&paramPlaceID;
 465             paramLengths[0] = sizeof(paramPlaceID);
 466             paramFormats[0] = 1;
 467             if (table == 1) // table=1 for placex
 468             {
 469                 res = PQexecPrepared(thread_data->conn, "index_placex", 1, paramValues, paramLengths, paramFormats, 1);
 470             }
 471             else // table=0 for osmline
 472             {
 473                 res = PQexecPrepared(thread_data->conn, "index_osmline", 1, paramValues, paramLengths, paramFormats, 1);
 474             }
 475             if (PQresultStatus(res) == PGRES_COMMAND_OK)
 476                 done = 1;
 477             else
 478             {
 479                 if (!strncmp(PQerrorMessage(thread_data->conn), "ERROR:  deadlock detected", 25))
 480                 {
 481                     if (table == 1)
 482                     {
 483                         fprintf(stderr, "index_placex: UPDATE failed - deadlock, retrying (%ld)\n", place_id);
 484                     }
 485                     else
 486                     {
 487                         fprintf(stderr, "index_osmline: UPDATE failed - deadlock, retrying (%ld)\n", place_id);
 488                     }
 489                     PQclear(res);
 490                     sleep(rand() % 10);
 491                 }
 492                 else
 493                 {
 494                     if (table == 1)
 495                     {
 496                         fprintf(stderr, "index_placex: UPDATE failed: %s", PQerrorMessage(thread_data->conn));
 497                     }
 498                     else
 499                     {
 500                         fprintf(stderr, "index_osmline: UPDATE failed: %s", PQerrorMessage(thread_data->conn));
 501                     }
 502                     PQclear(res);
 503                     exit(EXIT_FAILURE);
 504                 }
 505             }
 506         }
 507         PQclear(res);
 508         if (difftime(time(0), updateStartTime) > 1) fprintf(stderr, "  Slow place_id %ld\n", place_id);
 509
 510         if (thread_data->writer)
 511         {
 512             nominatim_exportPlace(place_id, thread_data->conn, thread_data->writer, thread_data->writer_mutex, &querySet);
 513             nominatim_exportFreeQueries(&querySet);
 514         }
 515     }
 516
 517     return NULL;
 518 }