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_osmline", 1, paramValues, paramLengths, paramFormats, 1);
 167
 168                 }
 169                 else
 170                 {
 171                     iResult = PQsendQueryPrepared(conn, "index_sector_places", 2, paramValues, paramLengths, paramFormats, 1);
 172                 }
 173             }
 174             if (!iResult)
 175             {
 176                 fprintf(stderr, "index_sector_places: SELECT failed: %s", PQerrorMessage(conn));
 177                 PQclear(resPlaces);
 178                 exit(EXIT_FAILURE);
 179             }
 180         }
 181         if (iSector > 0)
 182         {
 183             count = 0;
 184             rankPerSecond = 0;
 185             tuples = PQntuples(resPlaces);
 186
 187             if (tuples > 0)
 188             {
 189                 // Spawn threads
 190                 for (i = 0; i < num_threads; i++)
 191                 {
 192                     thread_data[i].res = resPlaces;
 193                     thread_data[i].tuples = tuples;
 194                     thread_data[i].count = &count;
 195                     thread_data[i].count_mutex = &count_mutex;
 196                     thread_data[i].writer = writer;
 197                     thread_data[i].writer_mutex = &writer_mutex;
 198                     if (interpolation)
 199                     {
 200                         thread_data[i].table = 0;  // use interpolations table
 201                     }
 202                     else
 203                     {
 204                         thread_data[i].table = 1;  // use placex table
 205                     }
 206                     pthread_create(&thread_data[i].thread, NULL, &nominatim_indexThread, (void *)&thread_data[i]);
 207                 }
 208
 209                 // Monitor threads to give user feedback
 210                 sleepcount = 0;
 211                 while (count < tuples)
 212                 {
 213                     usleep(1000);
 214
 215                     // Aim for one update per second
 216                     if (sleepcount++ > 500)
 217                     {
 218                         rankPerSecond = ((float)rankCountTuples + (float)count) / MAX(difftime(time(0), rankStartTime),1);
 219                         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);
 220                         sleepcount = 0;
 221                     }
 222                 }
 223
 224                 // Wait for everything to finish
 225                 for (i = 0; i < num_threads; i++)
 226                 {
 227                     pthread_join(thread_data[i].thread, NULL);
 228                 }
 229
 230                 rankCountTuples += tuples;
 231             }
 232
 233             // Finished sector
 234             rankPerSecond = (float)rankCountTuples / MAX(difftime(time(0), rankStartTime),1);
 235             fprintf(stderr, "  Done %i in %i @ %f per second - ETA (seconds): %f\n", rankCountTuples, (int)(difftime(time(0), rankStartTime)), rankPerSecond, ((float)(rankTotalTuples - rankCountTuples))/rankPerSecond);
 236
 237             PQclear(resPlaces);
 238         }
 239         if (rankTotalTuples-rankCountTuples < num_threads*20 && iSector < PQntuples(resSectors))
 240         {
 241             iSector = PQntuples(resSectors) - 1;
 242         }
 243     }
 244     // Finished rank
 245     fprintf(stderr, "\r  Done %i in %i @ %f per second - FINISHED\n\n", rankCountTuples, (int)(difftime(time(0), rankStartTime)), rankPerSecond);
 246
 247     PQclear(resSectors);
 248 }
 249
 250 void nominatim_index(int rank_min, int rank_max, int num_threads, const char *conninfo, const char *structuredoutputfile)
 251 {
 252     struct index_thread_data * thread_data;
 253
 254     PGconn *conn;
 255     PGresult * res;
 256
 257     int rank;
 258
 259     int i;
 260
 261     xmlTextWriterPtr writer;
 262     pthread_mutex_t writer_mutex = PTHREAD_MUTEX_INITIALIZER;
 263
 264     Oid pg_prepare_params[2];
 265
 266     conn = PQconnectdb(conninfo);
 267     if (PQstatus(conn) != CONNECTION_OK)
 268     {
 269         fprintf(stderr, "Connection to database failed: %s\n", PQerrorMessage(conn));
 270         exit(EXIT_FAILURE);
 271     }
 272
 273     pg_prepare_params[0] = PG_OID_INT4;
 274     res = PQprepare(conn, "index_sectors",
 275                     "select geometry_sector,count(*) from placex where rank_search = $1 and indexed_status > 0 group by geometry_sector order by geometry_sector",
 276                     1, pg_prepare_params);
 277     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 278     {
 279         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 280         exit(EXIT_FAILURE);
 281     }
 282     PQclear(res);
 283
 284     res = PQprepare(conn, "index_sectors_osmline",
 285                     "select geometry_sector,count(*) from location_property_osmline where indexed_status > 0 group by geometry_sector order by geometry_sector",
 286                     0, NULL);
 287     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 288     {
 289         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 290         exit(EXIT_FAILURE);
 291     }
 292     PQclear(res);
 293
 294     pg_prepare_params[0] = PG_OID_INT4;
 295     res = PQprepare(conn, "index_nosectors",
 296                     "select 0::integer,count(*) from placex where rank_search = $1 and indexed_status > 0",
 297                     1, pg_prepare_params);
 298     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 299     {
 300         fprintf(stderr, "Failed preparing index_sectors: %s\n", PQerrorMessage(conn));
 301         exit(EXIT_FAILURE);
 302     }
 303     PQclear(res);
 304
 305     pg_prepare_params[0] = PG_OID_INT4;
 306     pg_prepare_params[1] = PG_OID_INT4;
 307     res = PQprepare(conn, "index_sector_places",
 308                     "select place_id from placex where rank_search = $1 and geometry_sector = $2 and indexed_status > 0",
 309                     2, pg_prepare_params);
 310     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 311     {
 312         fprintf(stderr, "Failed preparing index_sector_places: %s\n", PQerrorMessage(conn));
 313         exit(EXIT_FAILURE);
 314     }
 315     PQclear(res);
 316
 317     pg_prepare_params[0] = PG_OID_INT4;
 318     res = PQprepare(conn, "index_nosector_places",
 319                     "select place_id from placex where rank_search = $1 and indexed_status > 0 order by geometry_sector",
 320                     1, pg_prepare_params);
 321     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 322     {
 323         fprintf(stderr, "Failed preparing index_nosector_places: %s\n", PQerrorMessage(conn));
 324         exit(EXIT_FAILURE);
 325     }
 326     PQclear(res);
 327
 328     pg_prepare_params[0] = PG_OID_INT4;
 329     res = PQprepare(conn, "index_sector_places_osmline",
 330                     "select place_id from location_property_osmline where geometry_sector = $1 and indexed_status > 0",
 331                     1, pg_prepare_params);
 332     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 333     {
 334         fprintf(stderr, "Failed preparing index_sector_places: %s\n", PQerrorMessage(conn));
 335         exit(EXIT_FAILURE);
 336     }
 337     PQclear(res);
 338
 339     res = PQprepare(conn, "index_nosector_places_osmline",
 340                     "select place_id from location_property_osmline where indexed_status > 0 order by geometry_sector",
 341                     0, NULL);
 342     if (PQresultStatus(res) != PGRES_COMMAND_OK)
 343     {
 344         fprintf(stderr, "Failed preparing index_nosector_places: %s\n", PQerrorMessage(conn));
 345         exit(EXIT_FAILURE);
 346     }
 347     PQclear(res);
 348
 349     // Build the data for each thread
 350     thread_data = (struct index_thread_data *)malloc(sizeof(struct index_thread_data)*num_threads);
 351     for (i = 0; i < num_threads; i++)
 352     {
 353         thread_data[i].conn = PQconnectdb(conninfo);
 354         if (PQstatus(thread_data[i].conn) != CONNECTION_OK)
 355         {
 356             fprintf(stderr, "Connection to database failed: %s\n", PQerrorMessage(thread_data[i].conn));
 357             exit(EXIT_FAILURE);
 358         }
 359
 360         pg_prepare_params[0] = PG_OID_INT8;
 361         res = PQprepare(thread_data[i].conn, "index_placex",
 362                         "update placex set indexed_status = 0 where place_id = $1",
 363                         1, pg_prepare_params);
 364         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 365         {
 366             fprintf(stderr, "Failed preparing index_placex: %s\n", PQerrorMessage(conn));
 367             exit(EXIT_FAILURE);
 368         }
 369         PQclear(res);
 370
 371         pg_prepare_params[0] = PG_OID_INT8;
 372         res = PQprepare(thread_data[i].conn, "index_osmline",
 373                         "update location_property_osmline set indexed_status = 0 where place_id = $1",
 374                         1, pg_prepare_params);
 375         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 376         {
 377             fprintf(stderr, "Failed preparing index_osmline: %s\n", PQerrorMessage(conn));
 378             exit(EXIT_FAILURE);
 379         }
 380         PQclear(res);
 381
 382         /*res = PQexec(thread_data[i].conn, "set enable_seqscan = false");
 383         if (PQresultStatus(res) != PGRES_COMMAND_OK)
 384         {
 385             fprintf(stderr, "Failed disabling sequential scan: %s\n", PQerrorMessage(conn));
 386             exit(EXIT_FAILURE);
 387         }
 388         PQclear(res);*/
 389
 390         nominatim_exportCreatePreparedQueries(thread_data[i].conn);
 391     }
 392
 393
 394     fprintf(stderr, "Starting indexing rank (%i to %i) using %i threads\n", rank_min, rank_max, num_threads);
 395
 396     for (rank = rank_min; rank <= rank_max; rank++)
 397     {
 398         // OSMLINE: do reindexing (=> reparenting) for interpolation lines at rank 30, but before all other objects of rank 30
 399         // reason: houses (rank 30) depend on the updated interpolation line, when reparenting (see placex_update in functions.sql)
 400         if (rank == 30)
 401         {
 402             run_indexing(rank, 1, conn, num_threads, thread_data, structuredoutputfile);
 403         }
 404         run_indexing(rank, 0, conn, num_threads, thread_data, structuredoutputfile);
 405     }
 406         // Close all connections
 407         for (i = 0; i < num_threads; i++)
 408         {
 409                 PQfinish(thread_data[i].conn);
 410         }
 411         PQfinish(conn);
 412 }
 413
 414 void *nominatim_indexThread(void * thread_data_in)
 415 {
 416     struct index_thread_data * thread_data = (struct index_thread_data * )thread_data_in;
 417     struct export_data  querySet;
 418
 419     PGresult   *res;
 420
 421     const char  *paramValues[1];
 422     int         paramLengths[1];
 423     int         paramFormats[1];
 424     uint64_t    paramPlaceID;
 425     uint64_t    place_id;
 426     time_t      updateStartTime;
 427     uint        table;
 428
 429     table = (uint)(thread_data->table);
 430
 431     while (1)
 432     {
 433         pthread_mutex_lock( thread_data->count_mutex );
 434         if (*(thread_data->count) >= thread_data->tuples)
 435         {
 436             pthread_mutex_unlock( thread_data->count_mutex );
 437             break;
 438         }
 439
 440         place_id = PGint64(*((uint64_t *)PQgetvalue(thread_data->res, *thread_data->count, 0)));
 441         (*thread_data->count)++;
 442
 443         pthread_mutex_unlock( thread_data->count_mutex );
 444
 445         if (verbose) fprintf(stderr, "  Processing place_id %ld\n", place_id);
 446
 447         updateStartTime = time(0);
 448         int done = 0;
 449
 450         if (thread_data->writer)
 451         {
 452              nominatim_exportPlaceQueries(place_id, thread_data->conn, &querySet);
 453         }
 454
 455         while(!done)
 456         {
 457             paramPlaceID = PGint64(place_id);
 458             paramValues[0] = (char *)&paramPlaceID;
 459             paramLengths[0] = sizeof(paramPlaceID);
 460             paramFormats[0] = 1;
 461             if (table == 1) // table=1 for placex
 462             {
 463                 res = PQexecPrepared(thread_data->conn, "index_placex", 1, paramValues, paramLengths, paramFormats, 1);
 464             }
 465             else // table=0 for osmline
 466             {
 467                 res = PQexecPrepared(thread_data->conn, "index_osmline", 1, paramValues, paramLengths, paramFormats, 1);
 468             }
 469             if (PQresultStatus(res) == PGRES_COMMAND_OK)
 470                 done = 1;
 471             else
 472             {
 473                 if (!strncmp(PQerrorMessage(thread_data->conn), "ERROR:  deadlock detected", 25))
 474                 {
 475                     if (table == 1)
 476                     {
 477                         fprintf(stderr, "index_placex: UPDATE failed - deadlock, retrying (%ld)\n", place_id);
 478                     }
 479                     else
 480                     {
 481                         fprintf(stderr, "index_osmline: UPDATE failed - deadlock, retrying (%ld)\n", place_id);
 482                     }
 483                     PQclear(res);
 484                     sleep(rand() % 10);
 485                 }
 486                 else
 487                 {
 488                     if (table == 1)
 489                     {
 490                         fprintf(stderr, "index_placex: UPDATE failed: %s", PQerrorMessage(thread_data->conn));
 491                     }
 492                     else
 493                     {
 494                         fprintf(stderr, "index_osmline: UPDATE failed: %s", PQerrorMessage(thread_data->conn));
 495                     }
 496                     PQclear(res);
 497                     exit(EXIT_FAILURE);
 498                 }
 499             }
 500         }
 501         PQclear(res);
 502         if (difftime(time(0), updateStartTime) > 1) fprintf(stderr, "  Slow place_id %ld\n", place_id);
 503
 504         if (thread_data->writer)
 505         {
 506             nominatim_exportPlace(place_id, thread_data->conn, thread_data->writer, thread_data->writer_mutex, &querySet);
 507             nominatim_exportFreeQueries(&querySet);
 508         }
 509     }
 510
 511     return NULL;
 512 }