err.no Git - mapper/blob - src/osm-db-import.c

   1 /*
   2  * This file is part of mapper
   3  *
   4  * Copyright (C) 2007 Kaj-Michael Lang
   5  *
   6  * This program is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License along
  17  * with this program; if not, write to the Free Software Foundation, Inc.,
  18  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  19  */
  20
  21 /*
  22  * Routines to read OSM planet XML file and store it in a sqlite3 database.
  23  * Reads in all nodes (if used, skips nodes outside bounding box)
  24  * Special POI nodes are stored in POI table.
  25  * Place POI nodes are stored in place table.
  26  *
  27  * Ways are read in and their data (name, type, etc) are stored
  28  * in way, way_name and way_ref tables.
  29  *
  30  * Nodes used by they ways are stored in way_n2n table.
  31  *
  32  */
  33
  34 #include <stdio.h>
  35 #include <unistd.h>
  36 #include <string.h>
  37 #include <strings.h>
  38 #include <sys/types.h>
  39 #include <sys/stat.h>
  40 #include <fcntl.h>
  41 #include <math.h>
  42 #include <glib.h>
  43 #include <glib/gstdio.h>
  44 #include <sqlite3.h>
  45 #include <expat.h>
  46 #include <bzlib.h>
  47
  48 #include "osm.h"
  49 #include "latlon.h"
  50 #include "db.h"
  51 #include "osm-db-import.h"
  52
  53 #if 0
  54 #define VERBOSE
  55 #endif
  56 /* #define VERBOSE_KEYS */
  57
  58
  59 /* Use g_convert to transliterate names.. my iconv seems to be fucked so this didn't work... */
  60 /* #define TRANSLIT_NAMES */
  61
  62 #define FILE_BUFFER (128*1024)
  63
  64 static guint node_cnt=0;                /* Nodes */
  65 static guint node_skip_cnt=0;   /* Skipped nodes */
  66 static guint noded_cnt=0;               /* Nodes with (usable) data */
  67 static guint way_cnt=0;                 /* Ways */
  68 static guint way_names=0;               /* Ways with name */
  69 static guint way_refs=0;                /* Ways with ref or int_ref */
  70
  71 static guint dbnode_cnt=0;
  72 static guint dbnoded_cnt=0;
  73 static guint dbway_cnt=0;
  74
  75 /* For threaded importing */
  76 static GThread* import_thread=NULL;
  77 static GSourceFunc osm_import_progress_cb=NULL;
  78 static osm_import_data_req osm_import_req;
  79 static guint import_sid=0;
  80
  81 static gboolean is_update=FALSE;
  82 static XML_Parser xp;
  83
  84
  85 /* XML tag IDs */
  86 typedef enum {
  87         START,
  88         IN_OSM_TAG,
  89         IN_NODE_TAG,
  90         IN_WNODE_TAG,
  91         IN_WAY_TAG,
  92         IN_KEY_TAG,
  93         IN_BOUND_TAG,
  94         IN_RELATION_TAG,
  95         IN_MEMBER_TAG,
  96         END,
  97         ERROR
  98 } tag_state_t;
  99
 100 /* Parent tag type */
 101 typedef enum {
 102         IS_NONE,
 103         IS_NODE,
 104         IS_WAY,
 105         IS_RELATION
 106 } tag_parent_t;
 107
 108 /* Node types table */
 109 /* XXX: Add support for parent category */
 110 struct _nodeinfo {
 111         gchar *k, *v;
 112         node_type_t type;
 113 } nodeinfo[] = {
 114         { "amenity", "fuel",            NODE_AMENITY_FUEL },
 115         { "amenity", "parking",         NODE_AMENITY_PARKING },
 116
 117         { "amenity", "pub",                     NODE_AMENITY_PUB },
 118         { "amenity", "nightclub",       NODE_AMENITY_NIGHTCLUB },
 119         { "amenity", "biergarten",      NODE_AMENITY_PUB },
 120         { "amenity", "cafe",            NODE_AMENITY_CAFE },
 121         { "amenity", "fast_food",       NODE_AMENITY_FOOD },
 122         { "amenity", "restaurant",      NODE_AMENITY_RESTAURANT },
 123
 124         { "amenity", "telephone",       NODE_AMENITY_TELEPHONE },
 125         { "amenity", "toilets",         NODE_AMENITY_WC },
 126
 127         { "amenity", "hospital",        NODE_AMENITY_HOSPITAL },
 128         { "amenity", "doctors",         NODE_AMENITY_HOSPITAL },
 129         { "amenity", "pharmacy",        NODE_AMENITY_PHARMACY },
 130
 131         { "amenity", "post_office",     NODE_AMENITY_POST },
 132         { "amenity", "post_box",        NODE_AMENITY_POST_BOX },
 133
 134         { "amenity", "cinema",          NODE_AMENITY_CINEMA },
 135         { "amenity", "theatre",         NODE_AMENITY_THEATRE },
 136
 137         { "amenity", "atm",                     NODE_AMENITY_ATM },
 138         { "amenity", "bank",            NODE_AMENITY_BANK },
 139
 140         { "amenity", "police",          NODE_AMENITY_POLICE },
 141         { "amenity", "speed_trap",      NODE_AMENITY_SPEEDCAM },
 142         { "amenity", "speed_camera",    NODE_AMENITY_SPEEDCAM },
 143         { "amenity", "speed camera",    NODE_AMENITY_SPEEDCAM },
 144
 145         { "amenity", "place_of_worship",NODE_AMENITY_POW },
 146
 147         { "amenity", "school",          NODE_AMENITY_SCHOOL },
 148         { "amenity", "college",         NODE_AMENITY_COLLEGE },
 149         { "amenity", "university",      NODE_AMENITY_COLLEGE },
 150
 151         { "amenity", "library", NODE_AMENITY_LIBRARY },
 152         { "amenity", "townhall",        NODE_AMENITY_TOWNHALL },
 153
 154         { "amenity", "supermarket",     NODE_AMENITY_SHOP },
 155         { "amenity", "shopping_centre", NODE_AMENITY_SHOP },
 156         { "amenity", "shop",            NODE_AMENITY_SHOP },
 157         { "amenity", "shops",           NODE_AMENITY_SHOP },
 158         { "amenity", "shopping",        NODE_AMENITY_SHOP },
 159         { "amenity", "shopping_mall",NODE_AMENITY_SHOP },
 160         { "amenity", "cycle_shop",      NODE_AMENITY_SHOP },
 161         { "amenity", "bike_shop",       NODE_AMENITY_SHOP },
 162         { "amenity", "coffee_shop",     NODE_AMENITY_SHOP },
 163         { "amenity", "indoor_shopping_centre",  NODE_AMENITY_SHOP },
 164         { "amenity", "farm_shop",       NODE_AMENITY_SHOP },
 165         { "amenity", "tea_shop",        NODE_AMENITY_SHOP },
 166
 167         /* Shops */
 168         { "shop",        "supermarket", NODE_AMENITY_SHOP },
 169         { "shop",        "bakery",              NODE_AMENITY_SHOP },
 170         { "shop",        "alcohol",             NODE_AMENITY_SHOP },
 171         { "shop",        "butcher",             NODE_AMENITY_SHOP },
 172         { "shop",        "flowers",             NODE_AMENITY_SHOP },
 173         { "shop",        "clothing",    NODE_AMENITY_SHOP },
 174         { "shop",        "souvenir",    NODE_AMENITY_SHOP },
 175         { "shop",        "bicycles",    NODE_AMENITY_SHOP },
 176         { "shop",        "grocers",             NODE_AMENITY_SHOP },
 177         { "shop",        "newsagents",  NODE_AMENITY_SHOP },
 178         { "shop",        "convenience", NODE_AMENITY_SHOP },
 179         { "shop",        "bakers",              NODE_AMENITY_SHOP },
 180         { "shop",        "garden_centre",NODE_AMENITY_SHOP },
 181         { "shop",        "photography", NODE_AMENITY_SHOP },
 182         { "shop",        "general_store",NODE_AMENITY_SHOP },
 183         { "shop",        "food",                NODE_AMENITY_SHOP },
 184         { "shop",        "drinks",              NODE_AMENITY_SHOP },
 185         { "shop",        "sex",                 NODE_AMENITY_SHOP_ADULT },
 186         { "shop",        "pharmacy",    NODE_AMENITY_PHARMACY },
 187
 188         /* Sport */
 189         { "sport"  , "swimming",        NODE_SPORT_SWIMMING },
 190         { "sport"  , "golf",            NODE_SPORT_GOLF },
 191         { "sport"  , "tennis",          NODE_SPORT_TENNIS },
 192         { "sport"  , "football",        NODE_SPORT_FOOTBALL },
 193         { "sport"  , "soccer",          NODE_SPORT_SOCCER },
 194         { "sport"  , "baskteball",      NODE_SPORT_BASKETBALL },
 195         { "sport"  , "rugby",           NODE_SPORT_RUGBY },
 196         { "sport"  , "skating",         NODE_SPORT_SKATING },
 197         { "sport"  , "hockey",          NODE_SPORT_HOCKEY },
 198         { "sport"  , "skateboard",      NODE_SPORT_SKATEBOARD },
 199         { "sport"  , "bowling",         NODE_SPORT_BOWLING },
 200         { "sport"  , "10pin",           NODE_SPORT_BOWLING },
 201         { "sport"  , "motor",           NODE_SPORT_MOTOR },
 202         { "sport"  , "shooting_range",NODE_SPORT_SHOOTING },
 203         { "sport"  , "paintball",       NODE_SPORT_PAINTBALL },
 204         { "sport"  , "horse_racing",NODE_SPORT_HORSES },
 205         { "sport"  , "horse",           NODE_SPORT_HORSES },
 206         { "sport"  , "horses",          NODE_SPORT_HORSES },
 207         { "sport"  , "dog_racing",      NODE_SPORT_DOG },
 208         { "sport"  , "pelota",          NODE_SPORT_PELOTA },
 209         { "sport"  , "racquet",         NODE_SPORT_RACQUET },
 210         { "sport"  , "equestrian",      NODE_SPORT_HORSES },
 211         { "sport"  , "baseball",        NODE_SPORT_BASEBALL },
 212         { "sport"  , "cricket",         NODE_SPORT_CRICKET },
 213         { "sport"  , "croquet",         NODE_SPORT_CROQUET },
 214         { "sport"  , "cycling",         NODE_SPORT_CYCLING },
 215         { "sport"  , "bowls",           NODE_SPORT_BOWLS },
 216         { "sport"  , "athletics",       NODE_SPORT_ATHLETICS },
 217         { "sport"  , "gymnastics",      NODE_SPORT_GYMNASTICS },
 218         { "sport"  , "multi",           NODE_SPORT_OTHER },
 219         { "leisure", "sport_centre",NODE_SPORT_CENTER },
 220
 221         /* Tourism */
 222         { "tourism", "information",     NODE_TOURISM_INFO },
 223         { "tourism", "camp_site",       NODE_TOURISM_CAMP_SITE },
 224         { "tourism", "caravan_site",NODE_TOURISM_CARAVAN_SITE },
 225         { "tourism", "picnic_site",     NODE_TOURISM_PICNIC_SITE },
 226         { "tourism", "theme_park",      NODE_TOURISM_THEME_PARK },
 227         { "tourism", "hotel",           NODE_TOURISM_HOTEL },
 228         { "tourism", "motel",           NODE_TOURISM_MOTEL },
 229         { "tourism", "hostel",          NODE_TOURISM_HOSTEL },
 230         { "tourism", "attraction",      NODE_TOURISM_ATTRACTION },
 231         { "tourism", "zoo",                     NODE_TOURISM_ATTRACTION },
 232
 233         { "historic", "ruins",          NODE_TOURISM_ATTRACTION },
 234         { "historic", "monument",       NODE_TOURISM_ATTRACTION },
 235         { "historic", "memorial",       NODE_TOURISM_ATTRACTION },
 236         { "historic", "museum",         NODE_HISTORIC_MUSEUM },
 237         { "historic", "castle",         NODE_HISTORIC_CASTLE },
 238
 239         { "railway", "station",         NODE_RAILWAY_STATION },
 240         { "railway", "halt",            NODE_RAILWAY_HALT },
 241
 242         { "aeroway", "terminal",        NODE_AIRPORT_TERMINAL },
 243
 244         /* Places */
 245         { "place", "city",                      NODE_PLACE_CITY },
 246         { "place", "town",                      NODE_PLACE_TOWN },
 247         { "place", "village",           NODE_PLACE_VILLAGE },
 248         { "place", "hamlet",            NODE_PLACE_HAMLET },
 249         { "place", "locality",          NODE_PLACE_LOCALITY },
 250         { "place", "suburb",            NODE_PLACE_SUBURB },
 251         { "place", "island",            NODE_PLACE_ISLAND },
 252
 253         { "highway", "traffic_signals", NODE_TRAFFIC_SIGNALS },
 254         { "highway", "motorway_junction",       NODE_JUNCTION },
 255         { "highway", "services",        NODE_AMENITY_PARKING },
 256         { "highway", "toll_booth",      NODE_TOLLBOOTH },
 257         { "highway", "gate",            NODE_GATE },
 258
 259         { NULL, NULL, NODE_PLAIN }
 260 };
 261
 262 /* Array to get id number and defaults for ways of different types */
 263 struct _wayinfo {
 264         gchar *k, *v;
 265         guint defspeed;
 266         way_type_t type;
 267         gboolean oneway, link, area, car, foot;
 268 } wayinfo[] = {
 269         { "highway", "motorway",120,WAY_MOTORWAY,               TRUE, FALSE, FALSE, TRUE, FALSE },
 270         { "highway", "motorway_link",120,WAY_MOTORWAY,  TRUE, TRUE, FALSE, TRUE, FALSE },
 271         { "highway", "trunk",100,WAY_TRUNK,                             FALSE, FALSE, FALSE, TRUE, FALSE },
 272         { "highway", "trunk_link",100,WAY_TRUNK,                FALSE, TRUE, FALSE, TRUE, FALSE },
 273         { "highway", "primary",80,WAY_PRIMARY,                  FALSE, FALSE, FALSE, TRUE, TRUE },
 274         { "highway", "primary_link",60,WAY_PRIMARY,             FALSE, TRUE, FALSE, TRUE, TRUE },
 275         { "highway", "secondary",80,WAY_SECONDARY,              FALSE, FALSE, FALSE, TRUE, TRUE },
 276         { "highway", "secondary_link",60,WAY_SECONDARY, FALSE, TRUE, FALSE, TRUE, TRUE },
 277         { "highway", "tertiary",60,WAY_TERTIARY,                FALSE, FALSE, FALSE, TRUE, TRUE },
 278         { "highway", "unclassified",50,WAY_UNCLASSIFIED,        FALSE, FALSE, FALSE, TRUE, TRUE },
 279         { "highway", "byway",40,WAY_UNCLASSIFIED,       FALSE, FALSE, FALSE, TRUE, TRUE },
 280         { "highway", "residential",40,WAY_RESIDENTIAL,  FALSE, FALSE, FALSE, TRUE, TRUE },
 281         { "highway", "service",20,WAY_SERVICE,                  FALSE, FALSE, FALSE, TRUE, TRUE },
 282         { "highway", "track",20,WAY_TRACK,                              FALSE, FALSE, FALSE, TRUE, TRUE },
 283         { "highway", "unsurfaced",60,WAY_TRACK,                 FALSE, FALSE, FALSE, TRUE, TRUE },
 284         { "highway", "minor",60,WAY_TRACK,                      FALSE, FALSE, FALSE, TRUE, TRUE },
 285         { "highway", "pedestrian",20,WAY_FOOTWAY,               FALSE, FALSE, FALSE, FALSE, TRUE },
 286         { "highway", "footway",1,WAY_FOOTWAY,                   FALSE, FALSE, FALSE, FALSE, TRUE },
 287         { "highway", "steps",0,WAY_FOOTWAY,                             FALSE, FALSE, FALSE, FALSE, TRUE},
 288         { "highway", "bridleway",10,WAY_FOOTWAY,                FALSE, FALSE, FALSE, FALSE, TRUE },
 289         { "highway", "cycleway",10,WAY_CYCLEWAY,                FALSE, FALSE, FALSE, FALSE, TRUE },
 290         { "railway", "rail",0,WAY_RAIL,                                 FALSE, FALSE, FALSE, FALSE, FALSE },
 291         { "aeroway", "runway",0,WAY_RUNWAY,                             FALSE, FALSE, FALSE, FALSE, FALSE },
 292         { "aeroway", "taxiway",0,WAY_TAXIWAY,                   FALSE, FALSE, FALSE, FALSE, FALSE },
 293         { "natural", "water",0,WAY_WATER,                               FALSE, FALSE, TRUE, FALSE, FALSE },
 294         { "waterway", "river",0,WAY_WATER,                              FALSE, FALSE, FALSE, FALSE, FALSE },
 295         { "waterway", "canal",0,WAY_WATER,                              FALSE, FALSE, FALSE, FALSE, FALSE },
 296         { "waterway", "stream",0,WAY_WATER,                             FALSE, FALSE, FALSE, FALSE, FALSE },
 297         { "building", "*",0,WAY_UNWAYED,                                FALSE, FALSE, TRUE, FALSE, FALSE },
 298         { NULL, NULL, 0, WAY_UNWAYED, FALSE, FALSE, FALSE, FALSE, FALSE }
 299 };
 300
 301 static sqlite3 *db;
 302 tag_parent_t tag_parent=IS_NONE;
 303
 304 static GHashTable *osm_nodes;
 305 static GHashTable *osm_node_tags;
 306 static GHashTable *osm_way_tags;
 307 static GSList *osm_ways;
 308 static GSList *osm_poi;
 309
 310 static GHashTable *osm_place_country;
 311 static GHashTable *osm_place_region;
 312 static GHashTable *osm_place_city;
 313 static GHashTable *osm_place_suburb;
 314 static GHashTable *osm_place_village;
 315 static GHashTable *osm_node_isin;
 316 static GHashTable *osm_way_isin;
 317
 318 static node *cnode=NULL;
 319 static way *cway=NULL;
 320
 321 struct sql_stmt {
 322         sqlite3_stmt *insert_poi;
 323         sqlite3_stmt *delete_osm_poi;
 324
 325         sqlite3_stmt *insert_node;
 326         sqlite3_stmt *delete_nodes;
 327         sqlite3_stmt *select_node;
 328         sqlite3_stmt *update_node;
 329
 330         sqlite3_stmt *insert_way_data;
 331         sqlite3_stmt *insert_way_ref;
 332         sqlite3_stmt *insert_way_pc;
 333         sqlite3_stmt *insert_way_name;
 334         sqlite3_stmt *insert_way_names_nls;
 335         sqlite3_stmt *insert_way_n2n;
 336         sqlite3_stmt *delete_way;
 337         sqlite3_stmt *delete_way_n2n;
 338         sqlite3_stmt *delete_way_name;
 339         sqlite3_stmt *delete_way_names_nls;
 340         sqlite3_stmt *delete_way_ref;
 341         sqlite3_stmt *delete_way_pc;
 342
 343         sqlite3_stmt *insert_place;
 344         sqlite3_stmt *delete_place;
 345 };
 346 static struct sql_stmt sql;
 347
 348
 349 static struct map_bbox bbox;
 350 static gboolean use_bbox;
 351
 352 static void osm_free_way_data(way *w);
 353 static void print_way(way *w);
 354
 355 static gboolean osm_db_prepare(sqlite3 *db);
 356 static gboolean db_insert_node(node *n);
 357 static guint32 osm_find_way_place(way *w, node_type_t nt);
 358
 359 /****************************************************/
 360 /* Functions */
 361 /****************************************************/
 362
 363 static void
 364 db_finalize(void)
 365 {
 366 sqlite3_finalize(sql.insert_poi);
 367 sqlite3_finalize(sql.delete_osm_poi);
 368
 369 sqlite3_finalize(sql.insert_node);
 370 sqlite3_finalize(sql.select_node);
 371 sqlite3_finalize(sql.delete_nodes);
 372 sqlite3_finalize(sql.update_node);
 373
 374 sqlite3_finalize(sql.insert_place);
 375 sqlite3_finalize(sql.delete_place);
 376
 377 sqlite3_finalize(sql.delete_way);
 378 sqlite3_finalize(sql.insert_way_data);
 379
 380 sqlite3_finalize(sql.delete_way_name);
 381 sqlite3_finalize(sql.insert_way_name);
 382
 383 sqlite3_finalize(sql.delete_way_n2n);
 384 sqlite3_finalize(sql.insert_way_n2n);
 385
 386 sqlite3_finalize(sql.delete_way_pc);
 387 sqlite3_finalize(sql.insert_way_pc);
 388
 389 sqlite3_finalize(sql.delete_way_names_nls);
 390 sqlite3_finalize(sql.insert_way_names_nls);
 391 }
 392
 393 static gboolean
 394 osm_db_prepare(sqlite3 *db)
 395 {
 396 /* Way nodes */
 397 sqlite3_prepare_v2(db, "insert or replace into nodes (nid,ilat,ilon,rlat,rlon,l,f) values (?,?,?,?,?,0,?)", -1, &sql.insert_node, NULL);
 398 sqlite3_prepare_v2(db, "select ilat,ilon,l from nodes where nid=?", -1, &sql.select_node, NULL);
 399 sqlite3_prepare_v2(db, "delete from nodes", -1, &sql.delete_nodes, NULL);
 400 sqlite3_prepare_v2(db, "update nodes set l=l+1 where nid=?", -1, &sql.update_node, NULL);
 401
 402 /* Places */
 403 sqlite3_prepare_v2(db, "insert or replace into places (nid,type,name,isin_c,isin_p) values (?, ?, ?, ?, ?)", -1, &sql.insert_place, NULL);
 404 sqlite3_prepare_v2(db, "delete from places", -1, &sql.delete_place, NULL);
 405
 406 /* POI nodes */
 407 if (sqlite3_prepare_v2(db, "insert or replace into poi (osm_id, lat, lon, label, cat_id, public, source, priority, isin_c, isin_p, desc, url, postal_code) "
 408                                            " values (?, ?, ?, ?, ?, 1, 1, ?, ?, ?, ?, ?, ?)", -1, &sql.insert_poi, NULL)!=SQLITE_OK)
 409         g_printf("SQL: %s\n", sqlite3_errmsg(db));
 410
 411 sqlite3_prepare_v2(db, "delete from poi where osm_id>0 and source=1", -1, &sql.delete_osm_poi, NULL);
 412
 413 /* Ways */
 414 sqlite3_prepare_v2(db, "insert or replace into way (wid,nodes,type,flags,speed,isin_c,isin_p,lat,lon) values (?, ?, ?, ?, ?, ?, ?, ?, ?)", -1, &sql.insert_way_data, NULL);
 415 sqlite3_prepare_v2(db, "delete from way", -1, &sql.delete_way, NULL);
 416
 417 /* Way nodes */
 418 sqlite3_prepare_v2(db, "insert into way_n2n (wid,f,t) values (?,?,?)", -1, &sql.insert_way_n2n, NULL);
 419 sqlite3_prepare_v2(db, "delete from way_n2n where wid=?", -1, &sql.delete_way_n2n, NULL);
 420
 421 /* Way names */
 422 sqlite3_prepare_v2(db, "insert or replace into way_names (wid,name,norm) values (?, ?, ?)",  -1, &sql.insert_way_name, NULL);
 423 sqlite3_prepare_v2(db, "delete from way_names", -1, &sql.delete_way_name, NULL);
 424
 425 /* Way postal codes */
 426 sqlite3_prepare_v2(db, "insert or replace into way_pc (wid,pc) values (?, ?)",  -1, &sql.insert_way_pc, NULL);
 427 sqlite3_prepare_v2(db, "delete from way_pc", -1, &sql.delete_way_pc, NULL);
 428
 429 /* Other language names for ways */
 430 sqlite3_prepare_v2(db, "insert into way_names_nls (wid,lang,name, norm) values (?, ?, ?, ?)",  -1, &sql.insert_way_names_nls, NULL);
 431 sqlite3_prepare_v2(db, "delete from way_names_nls where wid=?", -1, &sql.delete_way_names_nls, NULL);
 432
 433 /* Way ref and int_ref */
 434 sqlite3_prepare_v2(db, "insert or replace into way_ref (rid,ref,int_ref) values (?, ?, ?)", -1, &sql.insert_way_ref, NULL);
 435 sqlite3_prepare_v2(db, "delete from way_ref", -1, &sql.delete_way_ref, NULL);
 436
 437 return TRUE;
 438 }
 439
 440 /********************************************************************/
 441
 442 static void
 443 print_way(way *w)
 444 {
 445 #ifdef VERBOSE
 446 g_assert(w);
 447 g_printf("Way #%d(N:%d T:%d S:%d IS: %d/%d): %s [%s:%s:%s]\n",
 448                 w->id,
 449                 g_slist_length(w->nodes),
 450                 w->type,
 451                 w->data ? w->data->speed : 0,
 452                 w->data ? w->data->isin_c : -1,
 453                 w->data ? w->data->isin_p : -1,
 454                 w->data ? w->data->name ? w->data->name : "" : "",
 455                 w->flags & W_ONEWAY ? "-" : "=",
 456                 w->flags & W_ROUNDABOUT ? "O" : "-",
 457                 w->flags & W_LINK ? "|" : " ");
 458 #endif
 459 }
 460
 461 static void
 462 print_node(node *n)
 463 {
 464 #ifdef VERBOSE
 465 g_assert(n);
 466 g_printf("Node #%d: T:%d IS: %d/%d [%s]\n",
 467         n->id,
 468         n->type,
 469         n->data ? n->data->isin_c : -1,
 470         n->data ? n->data->isin_p : -1,
 471         n->data ? n->data->name : "");
 472 #endif
 473 }
 474
 475 /********************************************************************/
 476
 477 static gboolean
 478 db_insert_node(node *n)
 479 {
 480 gint32 lat, lon;
 481
 482 g_assert(n);
 483
 484 lat=lat2mp_int(n->lat);
 485 lon=lon2mp_int(n->lon);
 486
 487 sqlite3_bind_int(sql.insert_node, 1, n->id);
 488
 489 /* Projected and integerized lat/lot */
 490 sqlite3_bind_int(sql.insert_node, 2, lat);
 491 sqlite3_bind_int(sql.insert_node, 3, lon);
 492 /* Original */
 493 sqlite3_bind_double(sql.insert_node, 4, n->lat);
 494 sqlite3_bind_double(sql.insert_node, 5, n->lon);
 495 sqlite3_bind_int(sql.insert_node, 6, n->type);
 496
 497 db_exec(db, sql.insert_node);
 498
 499 return TRUE;
 500 }
 501
 502 static gboolean
 503 db_insert_place(node *n)
 504 {
 505 g_assert(n);
 506 if (!n->data)
 507         return FALSE;
 508 if (!n->data->name)
 509         return FALSE;
 510 sqlite3_bind_int(sql.insert_place, 1, n->id);
 511 sqlite3_bind_int(sql.insert_place, 2, n->type);
 512 sqlite3_bind_text(sql.insert_place, 3, n->data->name, -1, SQLITE_TRANSIENT);
 513 sqlite3_bind_int(sql.insert_place, 4, n->data->isin_p);
 514 sqlite3_bind_int(sql.insert_place, 5, n->data->isin_c);
 515
 516 return db_exec(db,sql.insert_place);
 517 }
 518
 519 static gboolean
 520 db_insert_poi(node *n)
 521 {
 522 g_assert(n);
 523 sqlite3_bind_int(sql.insert_poi, 1, n->id);
 524 sqlite3_bind_double(sql.insert_poi, 2, n->lat);
 525 sqlite3_bind_double(sql.insert_poi, 3, n->lon);
 526 if (n->data->name)
 527         sqlite3_bind_text(sql.insert_poi, 4, n->data->name, -1, SQLITE_TRANSIENT);
 528 else
 529         sqlite3_bind_text(sql.insert_poi, 4, "", -1, SQLITE_TRANSIENT);
 530 sqlite3_bind_int(sql.insert_poi, 5, n->type);
 531 sqlite3_bind_int(sql.insert_poi, 6, n->type/100);
 532 sqlite3_bind_int(sql.insert_poi, 7, n->data->isin_c);
 533 sqlite3_bind_int(sql.insert_poi, 8, n->data->isin_p);
 534
 535 if (n->data->desc)
 536         sqlite3_bind_text(sql.insert_poi, 9, n->data->desc, -1, SQLITE_TRANSIENT);
 537 if (n->data->url)
 538         sqlite3_bind_text(sql.insert_poi, 10, n->data->url, -1, SQLITE_TRANSIENT);
 539 if (n->data->postal_code)
 540         sqlite3_bind_text(sql.insert_poi, 11, n->data->postal_code, -1, SQLITE_TRANSIENT);
 541
 542 return db_exec(db,sql.insert_poi);
 543 }
 544
 545 /**
 546  * Update node usage count
 547  */
 548 static gboolean
 549 db_update_node_links(node *n)
 550 {
 551 g_assert(n);
 552 sqlite3_bind_int(sql.update_node, 1, n->id);
 553
 554 return db_exec(db,sql.update_node);
 555 }
 556
 557 /**
 558  * Insert way,node1,node2 triplet
 559  */
 560 static gboolean
 561 db_insert_way_n2n(way *w, node *nf, node *nt)
 562 {
 563 if (!w) {
 564         g_printf("NULL WAY\n");
 565         return FALSE;
 566 }
 567
 568 if (!nf) {
 569         g_printf("NULL NODE 1\n");
 570         return FALSE;
 571 }
 572
 573 if (!nt) {
 574         g_printf("NULL NODE 2\n");
 575         return FALSE;
 576 }
 577
 578 sqlite3_bind_int(sql.insert_way_n2n, 1, w->id);
 579 sqlite3_bind_int(sql.insert_way_n2n, 2, nf->id);
 580 sqlite3_bind_int(sql.insert_way_n2n, 3, nt->id);
 581
 582 #ifdef VERBOSE_N2N
 583 g_printf("%d [%d - %d]\n", w->id, nf->id, nt->id);
 584 #endif
 585
 586 db_exec(db,sql.insert_way_n2n);
 587 db_update_node_links(nf);
 588 db_update_node_links(nt);
 589 return TRUE;
 590 }
 591
 592 /**
 593  * Insert way ref and int_ref
 594  */
 595 static gboolean
 596 db_insert_way_ref(way *w)
 597 {
 598 if (!w->data)
 599         return TRUE;
 600
 601 if (!w->data->ref && !w->data->int_ref)
 602         return TRUE;
 603
 604 way_refs++;
 605
 606 sqlite3_bind_int(sql.insert_way_ref, 1, w->id);
 607 if (w->data->ref)
 608         sqlite3_bind_text(sql.insert_way_ref, 2, w->data->ref, -1, SQLITE_TRANSIENT);
 609 if (w->data->int_ref)
 610         sqlite3_bind_text(sql.insert_way_ref, 3, w->data->int_ref, -1, SQLITE_TRANSIENT);
 611
 612 return db_exec(db,sql.insert_way_ref);
 613 }
 614
 615 /**
 616  * Insert way name
 617  */
 618 static gboolean
 619 db_insert_way_name(way *w)
 620 {
 621 if (!w->data)
 622         return TRUE;
 623 if (!w->data->name)
 624         return TRUE;
 625
 626 way_names++;
 627
 628 sqlite3_bind_int(sql.insert_way_name, 1, w->id);
 629 sqlite3_bind_text(sql.insert_way_name, 2, w->data->name, -1, SQLITE_TRANSIENT);
 630
 631 #ifdef TRANSLIT_NAMES
 632 {
 633         gchar *norm;
 634         norm=g_convert(w->data->name, -1, "ASCII//TRANSLIT//IGNORE", "utf8", NULL, NULL, NULL);
 635         if (norm && strcmp(w->data->name, norm)!=0) {
 636                 sqlite3_bind_text(sql.insert_way_name, 3, norm, -1, SQLITE_TRANSIENT);
 637         }
 638         if (norm)
 639                 g_free(norm);
 640 }
 641 #endif
 642
 643 return db_exec(db,sql.insert_way_name);
 644 }
 645
 646 static gboolean
 647 db_delete_way_names_nls(way *w)
 648 {
 649 sqlite3_bind_int(sql.delete_way_names_nls, 1, w->id);
 650 return db_exec(db,sql.delete_way_names_nls);
 651 }
 652
 653 static gboolean
 654 db_insert_way_pc(way *w)
 655 {
 656 if (!w->data)
 657         return TRUE;
 658 if (!w->data->postal_code)
 659         return TRUE;
 660
 661 sqlite3_bind_int(sql.insert_way_pc, 1, w->id);
 662 sqlite3_bind_text(sql.insert_way_pc, 2, w->data->postal_code, -1, SQLITE_TRANSIENT);
 663
 664 return db_exec(db,sql.insert_way_pc);
 665 }
 666
 667 static gboolean
 668 db_delete_way_pc(way *w)
 669 {
 670 sqlite3_bind_int(sql.delete_way_pc, 1, w->id);
 671 return db_exec(db,sql.delete_way_pc);
 672 }
 673
 674 static void
 675 db_insert_way_names_nls_cb(gpointer key, gpointer value, gpointer user_data)
 676 {
 677 way *w=(way *)user_data;
 678
 679 sqlite3_bind_int(sql.insert_way_names_nls, 1, w->id);
 680 sqlite3_bind_text(sql.insert_way_names_nls, 2, (gchar *)key, -1, SQLITE_TRANSIENT);
 681 sqlite3_bind_text(sql.insert_way_names_nls, 3, (gchar *)value, -1, SQLITE_TRANSIENT);
 682 #ifdef TRANSLIT_NAMES
 683 {
 684         gchar *norm;
 685         norm=g_convert((gchar *value), -1, "ASCII//TRANSLIT//IGNORE", "utf8", NULL, NULL, NULL);
 686         if (norm && strcmp((gchar *)value, norm)!=0) {
 687                 sqlite3_bind_text(sql.insert_way_names_nls, 4, norm, -1, SQLITE_TRANSIENT);
 688         }
 689         if (norm)
 690                 g_free(norm);
 691 }
 692 #endif
 693 db_exec(db,sql.insert_way_names_nls);
 694 }
 695
 696 static void
 697 db_insert_way_names_nls(way *w)
 698 {
 699 if (!w->data)
 700         return;
 701 if (!w->data->names)
 702         return;
 703
 704 g_hash_table_foreach(w->data->names, db_insert_way_names_nls_cb, w);
 705 }
 706
 707 /**
 708  * Insert all data for the given way
 709  * - name
 710  * - ref
 711  * - nodes
 712  *
 713  */
 714 static gboolean
 715 db_insert_way(way *w)
 716 {
 717 GSList *iter;
 718 guint ncnt;
 719 node *wmn;
 720
 721 if (!w)
 722         return FALSE;
 723
 724 /* Skip things we don't use (yet) */
 725 if (w->type==WAY_UNWAYED || w->type>WAY_ROAD_END)
 726         return TRUE;
 727
 728 /* Insert nodes */
 729 for (iter=w->nodes; iter!=NULL; iter=iter->next) {
 730         if (!iter->next)
 731                 break;
 732         db_insert_way_n2n(w, iter->data, iter->next->data);
 733 }
 734
 735 if (w->id==0)
 736         return FALSE;
 737
 738 if (w->data) {
 739         w->data->isin_p=osm_find_way_place(w, NODE_PLACE_CITY);
 740         w->data->isin_c=osm_find_way_place(w, NODE_PLACE_COUNTRY);
 741 }
 742
 743 print_way(w);
 744
 745 sqlite3_bind_int(sql.insert_way_data, 1, w->id);
 746 sqlite3_bind_int(sql.insert_way_data, 2, w->ncnt);
 747 sqlite3_bind_int(sql.insert_way_data, 3, w->type);
 748 sqlite3_bind_int(sql.insert_way_data, 4, w->flags);
 749 if (w->data) {
 750         sqlite3_bind_int(sql.insert_way_data, 5, w->data->speed);
 751         sqlite3_bind_int(sql.insert_way_data, 6, w->data->isin_c);
 752         sqlite3_bind_int(sql.insert_way_data, 7, w->data->isin_p);
 753 }
 754
 755 /* Get middle node, use it as a rough way location */
 756 ncnt=g_slist_length(w->nodes);
 757 if (ncnt>1) {
 758         wmn=g_slist_nth_data(w->nodes, ncnt/2);
 759         if (wmn) {
 760                 sqlite3_bind_double(sql.insert_way_data, 8, wmn->lat);
 761                 sqlite3_bind_double(sql.insert_way_data, 9, wmn->lon);
 762         } else {
 763                 g_printerr("Failed to get way middlepoint node for location information!\n");
 764         }
 765 }
 766
 767 db_exec(db,sql.insert_way_data);
 768
 769 db_insert_way_ref(w);
 770 db_insert_way_name(w);
 771 db_insert_way_names_nls(w);
 772 db_insert_way_pc(w);
 773
 774 osm_free_way_data(w);
 775 return TRUE;
 776 }
 777
 778 /********************************************************************/
 779
 780 static gchar *
 781 get_attr_key_value(const gchar **p, gchar *key)
 782 {
 783 gchar **d;
 784
 785 d=p;
 786 while (*d!=NULL) {
 787         if (strncmp(*d, key, strlen(key))==0) {
 788                 d++;
 789                 return *d;
 790         }
 791         d++;
 792         d++;
 793 }
 794 return NULL;
 795 }
 796
 797 static tag_state_t
 798 check_tag(const gchar *tag)
 799 {
 800 if (strcmp(tag,"node")==0) return IN_NODE_TAG;
 801 else if (strcmp(tag,"nd")==0) return IN_WNODE_TAG;
 802 else if (strcmp(tag,"way")==0) return IN_WAY_TAG;
 803 else if (strcmp(tag,"tag")==0) return IN_KEY_TAG;
 804 else if (strcmp(tag,"osm")==0) return IN_OSM_TAG;
 805 else if (strcmp(tag,"bound")==0) return IN_BOUND_TAG;
 806 else if (strcmp(tag,"relation")==0) return IN_RELATION_TAG;
 807 else if (strcmp(tag,"member")==0) return IN_MEMBER_TAG;
 808 else return ERROR;
 809 }
 810
 811 static void
 812 find_nls_names(gpointer key, gpointer value, gpointer user_data)
 813 {
 814 gchar *k, *v;
 815 gchar *tmp;
 816 GHashTable *nls;
 817
 818 k=(gchar *)key;
 819 v=(gchar *)value;
 820 nls=(GHashTable *)user_data;
 821
 822 /* Check if it is a name key, return if not. */
 823 if (g_str_has_prefix(k, "name:")==FALSE)
 824         return;
 825
 826 tmp=g_strrstr(k, ":");
 827 if (!tmp)
 828         return;
 829 tmp++; /* skip : */
 830 if (*tmp==0)
 831         return;
 832 g_hash_table_insert(nls, g_strdup(tmp), g_strdup(v));
 833 #ifdef VERBOSE
 834 g_printf("NLS(%s): [%s]\n", tmp, v);
 835 #endif
 836 }
 837
 838 /********************************************************************/
 839
 840 static void
 841 node_print (node *n)
 842 {
 843 g_assert(n);
 844 if (n->data) {
 845         g_printf("N: %d [%f:%f][%s](%d)\n",
 846                 n->id, n->lat, n->lon,
 847                 n->data->name ? n->data->name : "-",
 848                 n->type);
 849 } else {
 850         g_printf("N: %d [%f:%f]\n",
 851                 n->id, n->lat, n->lon);
 852 }
 853 }
 854
 855 #ifdef DEBUG
 856 static void
 857 dump_array(const gchar **p)
 858 {
 859 char **d;
 860
 861 d=p;
 862 while (*d!=NULL) {
 863         g_printf("[%s]", *d);
 864         d++;
 865 }
 866 g_print("\n");
 867 }
 868 #endif
 869
 870 static inline gboolean
 871 osm_node_check_box(gdouble nlat, gdouble nlon)
 872 {
 873 if (use_bbox==FALSE)
 874         return TRUE;
 875 return (nlat > bbox.lat_min && nlat < bbox.lat_max && nlon > bbox.lon_min && nlon < bbox.lon_max) ? TRUE : FALSE;
 876 }
 877
 878 static void
 879 osm_new_node_data(node *n)
 880 {
 881 if (n==NULL)
 882         return;
 883 if (n->data!=NULL)
 884         return;
 885 n->data=g_slice_new(node_data);
 886 n->data->name=NULL;
 887 n->data->url=NULL;
 888 n->data->desc=NULL;
 889 n->data->postal_code=NULL;
 890 n->type=NODE_PLAIN;
 891 noded_cnt++;
 892 }
 893
 894 static void
 895 osm_free_node_data(node *n)
 896 {
 897 g_assert(n);
 898 g_assert(n->data);
 899 if (n->data->name)
 900         g_free(n->data->name);
 901 if (n->data->url)
 902         g_free(n->data->url);
 903 if (n->data->desc)
 904         g_free(n->data->desc);
 905 if (n->data->postal_code)
 906         g_free(n->data->postal_code);
 907 g_slice_free(node_data, n->data);
 908 n->data=NULL;
 909 noded_cnt--;
 910 }
 911
 912 static node *
 913 osm_new_node(gint id, gdouble lat, gdouble lon)
 914 {
 915 node *n=NULL;
 916
 917 n=g_slice_new(node);
 918 g_assert(n);
 919 n->id=id;
 920 n->lat=lat;
 921 n->lon=lon;
 922 n->data=(node_data *)NULL;
 923 return n;
 924 }
 925
 926 static void
 927 osm_free_node(node *n)
 928 {
 929 g_assert(n);
 930 g_slice_free(node, n);
 931 }
 932
 933 static node *
 934 osm_find_node(guint32 nid)
 935 {
 936 node *n;
 937
 938 g_assert(osm_nodes);
 939 return g_hash_table_lookup(osm_nodes, GINT_TO_POINTER(nid));
 940 }
 941
 942 static void
 943 osm_new_way_data(way *w)
 944 {
 945 if (w==NULL)
 946         return;
 947 if (w->data!=NULL)
 948         return;
 949
 950 w->data=g_slice_new(way_data);
 951 w->data->name=NULL;
 952 w->data->names=NULL;
 953 w->data->ref=NULL;
 954 w->data->int_ref=NULL;
 955 w->data->postal_code=NULL;
 956 w->data->layer=0;
 957 w->data->speed=0;
 958 }
 959
 960 static void
 961 osm_free_way_data(way *w)
 962 {
 963 g_assert(w);
 964 if (!w->data)
 965         return;
 966 if (w->data->name)
 967         g_free(w->data->name);
 968 if (w->data->ref)
 969         g_free(w->data->ref);
 970 if (w->data->int_ref)
 971         g_free(w->data->int_ref);
 972 g_slice_free(way_data, w->data);
 973 w->data=NULL;
 974 }
 975
 976 static way *
 977 osm_new_way(gint id)
 978 {
 979 way *w;
 980
 981 w=g_slice_new(way);
 982 g_assert(w);
 983 w->id=id;
 984 w->nodes=NULL;
 985 w->type=WAY_UNWAYED;
 986 w->data=NULL;
 987 w->ncnt=0;
 988 w->flags=0;
 989
 990 /* Add to list of ways */
 991 return w;
 992 }
 993
 994 static void
 995 osm_free_way(way *w)
 996 {
 997 if (w->nodes)
 998         g_slist_free(w->nodes);
 999 g_slice_free(way, w);
1000 }
1001
1002 static void
1003 osm_way_add_to_list(way *w)
1004 {
1005 g_assert(w);
1006 osm_ways=g_slist_prepend(osm_ways, w);
1007 }
1008
1009 static void
1010 osm_way_new_node(way *w, gint nid)
1011 {
1012 node *n;
1013
1014 g_assert(w);
1015 n=osm_find_node(nid);
1016 w->nodes=g_slist_prepend(w->nodes, n);
1017 w->ncnt++;
1018 }
1019
1020 /**
1021  * Search the place hash table for the location of the node.
1022  *
1023  */
1024 static guint32
1025 osm_find_node_place(node *n)
1026 {
1027 node *t;
1028 gchar **isin;
1029 gchar **place;
1030
1031 if (!n->data)
1032         return 0;
1033
1034 isin=g_hash_table_lookup(osm_node_isin, GINT_TO_POINTER(n->id));
1035
1036 if (!isin)
1037         return 0;
1038
1039 place=isin;
1040 while (*place!=NULL) {
1041         gchar *ps;
1042         ps=g_strstrip(*place);
1043 #ifdef VERBOSE
1044         g_printf("Checking (%d) [%s] in [%s]\n",n->type, n->data->name, ps);
1045 #endif
1046         switch (n->type) {
1047         case NODE_PLACE_CITY:
1048         case NODE_PLACE_TOWN:
1049         case NODE_PLACE_VILLAGE:
1050         case NODE_PLACE_HAMLET:
1051                 t=g_hash_table_lookup(osm_place_region, ps);
1052                 if (t)
1053                         return t->id;
1054                 t=g_hash_table_lookup(osm_place_country, ps);
1055                 if (t)
1056                         return t->id;
1057         break;
1058         case NODE_PLACE_SUBURB:
1059         case NODE_PLACE_LOCALITY:
1060                 t=g_hash_table_lookup(osm_place_city, ps);
1061                 if (t)
1062                         return t->id;
1063         break;
1064         case NODE_PLACE_ISLAND:
1065                 return 0;
1066         break;
1067         default:
1068                 t=g_hash_table_lookup(osm_place_city, ps);
1069                 if (t)
1070                         return t->id;
1071         break;
1072         }
1073         place++;
1074 }
1075
1076 return 0;
1077 }
1078
1079 static guint32
1080 osm_find_way_place(way *w, node_type_t nt)
1081 {
1082 gchar **isin;
1083 gchar **place;
1084
1085 isin=g_hash_table_lookup(osm_way_isin, GINT_TO_POINTER(w->id));
1086 if (!isin)
1087         return 0;
1088
1089 place=isin;
1090 while (*place!=NULL) {
1091         node *t;
1092         gchar *ps;
1093
1094         ps=g_strstrip(*place);
1095
1096 #ifdef VERBOSE
1097         g_printf("Checking (%d) in [%s]\n",w->id, ps);
1098 #endif
1099 switch (nt) {
1100         case NODE_PLACE_CITY:
1101         case NODE_PLACE_TOWN:
1102         case NODE_PLACE_VILLAGE:
1103         case NODE_PLACE_HAMLET:
1104         case NODE_PLACE_LOCALITY:
1105                 t=g_hash_table_lookup(osm_place_city, ps);
1106                 if (t)
1107                         return t->id;
1108         break;
1109         case NODE_PLACE_COUNTRY:
1110                 t=g_hash_table_lookup(osm_place_country, ps);
1111                 if (t)
1112                         return t->id;
1113         break;
1114         default:
1115                 g_assert_not_reached();
1116         break;
1117         }
1118         place++;
1119 }
1120
1121 return 0;
1122 }
1123
1124 /***********************************************************************/
1125
1126 static void
1127 osm_node_save_node(gint key, gpointer value, gpointer user_data)
1128 {
1129 node *n=(node *)value;
1130
1131 dbnode_cnt++;
1132 db_insert_node(n);
1133 if (dbnode_cnt % 20000==0)
1134         g_printf("\rNodes: %f%% (%u/%u)\n",
1135                 ((float)dbnode_cnt/(float)(node_cnt-node_skip_cnt))*100,
1136                 dbnode_cnt, node_cnt);
1137 }
1138
1139 /**
1140  * Check node type and insert as POI or Place
1141  * Discard extra data after insert.
1142  */
1143 static gboolean
1144 osm_node_save_poi(node *n, gpointer user_data)
1145 {
1146 if (!n) {
1147         g_printerr("ERROR: null poi\n");
1148         return FALSE;
1149 }
1150
1151 if (!n->data) {
1152         g_printerr("POI node with no data ?\n");
1153         return FALSE;
1154 }
1155
1156 n->data->isin_p=osm_find_node_place(n);
1157 n->data->isin_c=0;
1158
1159 if (n->type>NODE_POI_START && n->type<NODE_POI_END) {
1160         print_node(n);
1161         db_insert_poi(n);
1162         osm_free_node_data(n);
1163 } else if (n->type>NODE_PLACE_START && n->type<NODE_PLACE_END) {
1164         print_node(n);
1165         db_insert_place(n);
1166 } else {
1167         osm_free_node_data(n);
1168         return FALSE;
1169 }
1170
1171 return TRUE;
1172 }
1173
1174 static gboolean
1175 osm_planet_poi_clear_nodes(void)
1176 {
1177 g_print("Removing old OSM POIs...\n");
1178 db_transaction_begin(db);
1179 sqlite3_step(sql.delete_osm_poi);
1180 sqlite3_step(sql.delete_place);
1181 return db_transaction_commit(db);
1182 }
1183
1184 static gboolean
1185 osm_planet_poi_save_nodes(void)
1186 {
1187 g_print("Storing new POIs...\n");
1188 db_transaction_begin(db);
1189 g_slist_foreach(osm_poi, osm_node_save_poi, NULL);
1190 g_slist_free(osm_poi);
1191 return db_transaction_commit(db);
1192 }
1193
1194 /*********************************************************************/
1195
1196 static void
1197 osm_planet_clear_nodes(void)
1198 {
1199 g_print("Clearing old nodes\n");
1200 sqlite3_step(sql.delete_nodes);
1201 }
1202
1203 static gboolean
1204 osm_planet_save_nodes(void)
1205 {
1206 g_print("Storing nodes...\n");
1207 db_transaction_begin(db);
1208 g_hash_table_foreach(osm_nodes, osm_node_save_node, NULL);
1209 return db_transaction_commit(db);
1210 }
1211
1212 /*********************************************************************/
1213
1214 static void
1215 osm_way_save(way *value, gpointer user_data)
1216 {
1217 dbway_cnt++;
1218 db_insert_way(value);
1219 if (dbway_cnt % 15000==0 && dbway_cnt>0) {
1220                 g_printf("\rWays: %f%% (%u/%u)\n",
1221                         (((float)dbway_cnt/(float)way_cnt)*100),
1222                         dbway_cnt, way_cnt);
1223                 print_way(value);
1224 }
1225 }
1226
1227 static void
1228 osm_planet_clear_ways(void)
1229 {
1230 g_print("Clearing old data\n");
1231 sqlite3_step(sql.delete_way);
1232 sqlite3_step(sql.delete_way_name);
1233 sqlite3_step(sql.delete_way_ref);
1234 sqlite3_step(sql.delete_way_n2n);
1235 }
1236
1237 static gboolean
1238 osm_planet_save_ways(void)
1239 {
1240 g_print("Inserting new ways\n");
1241 db_transaction_begin(db);
1242 g_slist_foreach(osm_ways, osm_way_save, NULL);
1243 return db_transaction_commit(db);
1244 }
1245
1246 /*********************************************************************/
1247
1248 static void
1249 osm_planet_save_all_nodes(void)
1250 {
1251 g_printf("Saving planet nodes to database:\n");
1252
1253 osm_planet_poi_clear_nodes();
1254 osm_planet_poi_save_nodes();
1255
1256 if (!is_update) {
1257         osm_planet_clear_nodes();
1258         osm_planet_clear_ways();
1259 }
1260 osm_planet_save_nodes();
1261 }
1262
1263 static void
1264 osm_planet_save_all_ways(void)
1265 {
1266 g_printf("Saving planet way to database:\n");
1267
1268 osm_planet_save_ways();
1269 }
1270
1271 /***********************************************************************/
1272
1273 static void
1274 _osm_tag_start(void *userData, const char *name, const char **atts)
1275 {
1276 tag_state_t t;
1277 gchar *k, *v;
1278 guint32 id, ndref;
1279 gdouble nlat, nlon;
1280
1281 t=check_tag(name);
1282 switch (t) {
1283         case IN_OSM_TAG:
1284                 g_printf("Starting...\n");
1285         break;
1286         case IN_NODE_TAG:
1287                 tag_parent=IS_NODE;
1288                 node_cnt++;
1289
1290                 id=atoi(get_attr_key_value(atts, "id"));
1291                 nlat=atof(get_attr_key_value(atts, "lat"));
1292                 nlon=atof(get_attr_key_value(atts, "lon"));
1293
1294                 cnode=osm_new_node(id, nlat, nlon);
1295                 osm_node_tags=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1296         break;
1297         case IN_WAY_TAG:
1298                 tag_parent=IS_WAY;
1299                 way_cnt++;
1300                 id=atoi(get_attr_key_value(atts, "id"));
1301                 cway=osm_new_way(id);
1302                 osm_way_tags=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1303         break;
1304         case IN_WNODE_TAG:
1305                 ndref=atoi(get_attr_key_value(atts, "ref"));
1306                 if (use_bbox==TRUE) {
1307                         if (osm_find_node(ndref)==NULL) {
1308                                 cway->id=0;
1309                                 return;
1310                         }
1311                 }
1312                 osm_way_new_node(cway, ndref);
1313         break;
1314         case IN_KEY_TAG:
1315                 k=get_attr_key_value(atts, "k");
1316                 if (strcmp(k,"created_by")==0)
1317                         return;
1318                 if (strcmp(k,"source")==0)
1319                         return;
1320
1321                 v=get_attr_key_value(atts, "v");
1322 #ifdef VERBOSE_KEYS
1323                 g_printf("TAG: K=[%s] V=[%s]\n", k, v);
1324 #endif
1325
1326                 switch (tag_parent) {
1327                 case IS_NONE:
1328                         g_printf("Tag key/value pair but unknown owner\n");
1329                 break;
1330                 case IS_NODE:
1331                 {
1332                         if (!osm_node_tags)
1333                                 return;
1334
1335                         /* Insert key/value pairs into hash table */
1336                         if (cnode==NULL) {
1337                                 g_printerr("In node tags but node is NULL!\n");
1338                                 return;
1339                         }
1340                         g_hash_table_insert(osm_node_tags, g_strdup(k), g_strdup(v));
1341                 }
1342                 break;
1343                 case IS_WAY:
1344                 {
1345                         if (cway==NULL) {
1346                                 g_printerr("In way tags but way is NULL!\n");
1347                                 return;
1348                         }
1349                         g_hash_table_insert(osm_way_tags, g_strdup(k), g_strdup(v));
1350                         osm_new_way_data(cway);
1351                 }
1352                 break;
1353                 case IS_RELATION:
1354
1355                 break;
1356                 }
1357         break;
1358         case IN_BOUND_TAG:
1359                 /* Ignore for now */
1360                 g_printf("Ignoring bound tag\n");
1361         break;
1362         case IN_RELATION_TAG:
1363                 tag_parent=IS_RELATION;
1364
1365         break;
1366         case IN_MEMBER_TAG:
1367
1368         break;
1369         default:
1370                 tag_parent=IS_NONE;
1371                 g_printf("Unknown tag: %s\n", name);
1372         break;
1373 }
1374 }
1375
1376 static void
1377 _osm_tag_end(void *userData, const char *name)
1378 {
1379 tag_state_t t;
1380 gchar *v;
1381 guint i;
1382 t=check_tag(name);
1383 switch (t) {
1384         case IN_NODE_TAG:
1385
1386                 if (node_cnt % 25000==0) {
1387                         g_printf("Nodes: %u of %u, POIs: %u, Outside box: %u\n",
1388                                 node_cnt-node_skip_cnt, node_cnt, noded_cnt, node_skip_cnt);
1389                 }
1390
1391                 if (!osm_node_tags)
1392                         return;
1393
1394                 osm_new_node_data(cnode);
1395
1396                 for (i=0; nodeinfo[i].k; i++) {
1397                         v=g_hash_table_lookup(osm_node_tags, nodeinfo[i].k);
1398                         if (!v)
1399                                 continue;
1400                         if (strcasecmp (v, nodeinfo[i].v)==0) {
1401                                 cnode->type=nodeinfo[i].type;
1402                                 break;
1403                         }
1404                 }
1405
1406                 /* Check if node is inside bounding box, if not skip it.
1407                  * But keep it if it's something we might need for other nodes:
1408                  * - Places (for is_in)
1409                  * - ...
1410                  */
1411                 if ((osm_node_check_box(cnode->lat, cnode->lon)==FALSE) &&
1412                                 (cnode->type<NODE_PLACE_START)) {
1413                         osm_free_node_data(cnode);
1414                         osm_free_node(cnode);
1415                         g_hash_table_destroy(osm_node_tags);
1416                         node_skip_cnt++;
1417                         return;
1418                 }
1419
1420                 g_hash_table_insert(osm_nodes, GINT_TO_POINTER(cnode->id), cnode);
1421
1422                 if (cnode->type!=NODE_PLAIN) {
1423                         cnode->data->name=NULL;
1424                         v=g_hash_table_lookup(osm_node_tags, "name");
1425                         if (v)
1426                                 cnode->data->name=g_strstrip(g_utf8_normalize(v, -1, G_NORMALIZE_ALL_COMPOSE));
1427                         v=g_hash_table_lookup(osm_node_tags, "note");
1428                         if (v)
1429                                 cnode->data->desc=g_strstrip(g_strdup(v));
1430                         v=g_hash_table_lookup(osm_node_tags, "postal_code");
1431                         if (v)
1432                                 cnode->data->postal_code=g_strstrip(g_strdup(v));
1433
1434                         /* Links */
1435                         v=g_hash_table_lookup(osm_node_tags, "url");
1436                         if (v) {
1437                                 cnode->data->url=g_strstrip(g_strdup(v));
1438                         } else {
1439                                 v=g_hash_table_lookup(osm_node_tags, "wikipedia");
1440                                 if (v && strncmp(v,"http:", 5)==0)
1441                                         cnode->data->url=g_strstrip(g_strdup(v));
1442                         }
1443                 }
1444
1445                 cnode->data->isin_c=0;
1446                 cnode->data->isin_p=0;
1447                 v=g_hash_table_lookup(osm_node_tags, "is_in");
1448                 if (v) {
1449                         gchar **isin;
1450                         isin=g_strsplit(v, ",", 10);
1451                         g_hash_table_insert(osm_node_isin, GINT_TO_POINTER(cnode->id), isin);
1452                 }
1453
1454                 if (cnode->type==NODE_PLAIN) {
1455                         osm_free_node_data(cnode);
1456                 } else {
1457                         osm_poi=g_slist_prepend(osm_poi, cnode);
1458                         if (cnode->data->name) {
1459                                 switch (cnode->type) {
1460                                 case NODE_PLACE_COUNTRY:
1461                                         g_hash_table_insert(osm_place_country, cnode->data->name, cnode);
1462                                 break;
1463                                 case NODE_PLACE_CITY:
1464                                 case NODE_PLACE_TOWN:
1465                                         g_hash_table_insert(osm_place_city, cnode->data->name, cnode);
1466                                 break;
1467                                 case NODE_PLACE_SUBURB:
1468                                         g_hash_table_insert(osm_place_suburb, cnode->data->name, cnode);
1469                                 break;
1470                                 case NODE_PLACE_VILLAGE:
1471                                 case NODE_PLACE_HAMLET:
1472                                 case NODE_PLACE_LOCALITY:
1473                                         g_hash_table_insert(osm_place_village, cnode->data->name, cnode);
1474                                 break;
1475                                 case NODE_PLACE_ISLAND:
1476                                         /* Ignore for now */
1477                                 break;
1478                                 default:;
1479                                 }
1480                         }
1481                 }
1482                 g_hash_table_destroy(osm_node_tags);
1483                 cnode=NULL;
1484         break;
1485         case IN_WAY_TAG:
1486                 if (way_cnt % 1024==0) {
1487                         g_printf("\rWays: %d\n", way_cnt);
1488                 }
1489
1490                 cway->nodes=g_slist_reverse(cway->nodes);
1491
1492                 for (i=0; wayinfo[i].k; i++) {
1493                         v=g_hash_table_lookup(osm_way_tags, wayinfo[i].k);
1494                         if (!v)
1495                                 continue;
1496                         if (strcasecmp (v, wayinfo[i].v)==0) {
1497                                 if (wayinfo[i].link==TRUE)
1498                                         cway->flags|=W_LINK;
1499                                 if (wayinfo[i].area==TRUE)
1500                                         cway->flags|=W_AREA;
1501                                 if (wayinfo[i].oneway==TRUE)
1502                                         cway->flags|=W_ONEWAY;
1503                                 cway->type=wayinfo[i].type;
1504                                 if (cway->data->speed==0)
1505                                         cway->data->speed=wayinfo[i].defspeed;
1506                                 break;
1507                         }
1508                 }
1509
1510                 v=g_hash_table_lookup(osm_way_tags, "name");
1511                 if (v) {
1512                         cway->data->name=g_utf8_normalize(v, -1, G_NORMALIZE_ALL_COMPOSE);
1513                         /* Try to find other language names */
1514                         cway->data->names=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1515                         g_hash_table_foreach(osm_way_tags, find_nls_names, cway->data->names);
1516                         if (g_hash_table_size(cway->data->names)==0) {
1517                                 g_hash_table_destroy(cway->data->names);
1518                                 cway->data->names=NULL;
1519                         }
1520                 }
1521
1522                 v=g_hash_table_lookup(osm_way_tags, "ref");
1523                 if (v)
1524                         cway->data->ref=g_strdup(v);
1525                 v=g_hash_table_lookup(osm_way_tags, "int_ref");
1526                 if (v)
1527                         cway->data->int_ref=g_strdup(v);
1528
1529                 v=g_hash_table_lookup(osm_way_tags, "postal_code");
1530                 if (v)
1531                         cway->data->postal_code=g_strdup(v);
1532
1533                 /* XXX: somehow handle the silly -1 'reversed' oneway */
1534                 v=g_hash_table_lookup(osm_way_tags, "oneway");
1535                 if (v)
1536                         cway->flags|=W_ONEWAY;
1537
1538                 v=g_hash_table_lookup(osm_way_tags, "noexit");
1539                 if (v)
1540                         cway->flags|=W_NOEXIT;
1541
1542                 v=g_hash_table_lookup(osm_way_tags, "speedlimit");
1543                 if (v)
1544                         cway->data->speed=atoi(v);
1545                 v=g_hash_table_lookup(osm_way_tags, "maxspeed");
1546                 if (v)
1547                         cway->data->speed=atoi(v);
1548
1549 #if 0
1550                 v=g_hash_table_lookup(osm_way_tags, "layer");
1551                 if (v)
1552                         cway->data->layer=atoi(v);
1553 #endif
1554
1555                 v=g_hash_table_lookup(osm_way_tags, "junction");
1556                 if (v && strcasecmp(v,"roundabout")==0) {
1557                         cway->flags|=W_ROUNDABOUT;
1558                         cway->flags|=W_ONEWAY;
1559                 } else if (v && strcasecmp(v,"mini_roundabout")==0) {
1560                         cway->flags|=W_ROUNDABOUT;
1561                         cway->flags|=W_ONEWAY;
1562                 }
1563
1564                 /* XXX: Should check keys */
1565                 v=g_hash_table_lookup(osm_way_tags, "access");
1566                 if (v && (strcasecmp(v, "private")==0)) {
1567                         cway->flags|=W_NOACCESS;
1568                 }
1569
1570                 print_way(cway);
1571
1572                 v=g_hash_table_lookup(osm_way_tags, "is_in");
1573                 if (v) {
1574                         gchar **isin;
1575                         isin=g_strsplit(v, ",", 10);
1576                         g_hash_table_insert(osm_way_isin, GINT_TO_POINTER(cway->id), isin);
1577                 }
1578
1579                 if (cway->data && cway->data->name==NULL && cway->data->ref==NULL &&
1580                         cway->data->int_ref==NULL && cway->data->layer==0 && cway->data->speed==0)
1581                         osm_free_way_data(cway);
1582
1583                 if (cway->id!=0)
1584                         osm_way_add_to_list(cway);
1585                 else
1586                         osm_free_way(cway);
1587
1588                 cway=NULL;
1589                 g_hash_table_destroy(osm_way_tags);
1590         break;
1591         case IN_BOUND_TAG:
1592                 /* */
1593         break;
1594         case IN_OSM_TAG:
1595                 g_printf("\nPlanet loaded.\n");
1596         break;
1597         default:;
1598 }
1599 }
1600
1601 /************************************************************************/
1602
1603 static void
1604 storage_init(void)
1605 {
1606 osm_nodes=g_hash_table_new(g_direct_hash, g_direct_equal);
1607
1608 osm_place_country=g_hash_table_new(g_str_hash, g_str_equal);
1609 osm_place_city=g_hash_table_new(g_str_hash, g_str_equal);
1610 osm_place_suburb=g_hash_table_new(g_str_hash, g_str_equal);
1611 osm_place_village=g_hash_table_new(g_str_hash, g_str_equal);
1612 osm_place_region=g_hash_table_new(g_str_hash, g_str_equal);
1613 osm_node_isin=g_hash_table_new(g_direct_hash, g_direct_equal);
1614 osm_way_isin=g_hash_table_new(g_direct_hash, g_direct_equal);
1615 }
1616
1617 static void
1618 storage_free(void)
1619 {
1620 g_hash_table_destroy(osm_nodes);
1621
1622 g_hash_table_destroy(osm_place_country);
1623 g_hash_table_destroy(osm_place_city);
1624 g_hash_table_destroy(osm_place_suburb);
1625 g_hash_table_destroy(osm_place_village);
1626 g_hash_table_destroy(osm_place_region);
1627 g_hash_table_destroy(osm_node_isin);
1628 }
1629
1630 /************************************************************************/
1631
1632 static gint
1633 print_fail(const gchar *msg, gint ret)
1634 {
1635 g_printerr("ERROR: %s\n", msg);
1636 return ret;
1637 }
1638
1639 /************************************************************************/
1640
1641 static void
1642 print_memory_usage(void)
1643 {
1644 g_print("Memory usage per item:\n");
1645 g_printf("Node  size: %d\n", (gint)sizeof(node));
1646 g_printf("NodeD size: %d\n", (gint)sizeof(node_data));
1647 g_printf("Way   size: %d\n", (gint)sizeof(way));
1648 g_printf("WayD  size: %d\n", (gint)sizeof(way_data));
1649 }
1650
1651 /************************************************************************
1652  * Public inteface
1653  ************************************************************************/
1654
1655 void
1656 osm_planet_parser_init(void)
1657 {
1658 xp=XML_ParserCreate(NULL);
1659 XML_SetElementHandler(xp, _osm_tag_start, _osm_tag_end);
1660 storage_init();
1661 }
1662
1663 void
1664 osm_planet_parser_deinit(void)
1665 {
1666 XML_ParserFree(xp);
1667 storage_free();
1668 }
1669
1670 gboolean
1671 osm_planet_parse_buffer(gchar *buffer, size_t r)
1672 {
1673 if (XML_Parse(xp, buffer, r, r>0 ? 0:1) == XML_STATUS_ERROR) {
1674         g_printerr("Parse error at line %d:\n%s\n",
1675                 (gint)XML_GetCurrentLineNumber(xp),
1676                 XML_ErrorString(XML_GetErrorCode(xp)));
1677         return FALSE;
1678 }
1679 return TRUE;
1680 }
1681
1682 gboolean
1683 osm_planet_parse_file(gchar *pfile)
1684 {
1685 FILE *f;
1686 BZFILE *b;
1687 int bzerror;
1688 int r;
1689 gchar buffer[FILE_BUFFER];
1690 gboolean res=TRUE;
1691
1692 f=fopen(pfile, "r");
1693 if (!f) {
1694         perror("fopen failed\n");
1695         return FALSE;
1696 }
1697
1698 b=BZ2_bzReadOpen(&bzerror, f, 0, 0, NULL, 0);
1699 if (bzerror != BZ_OK) {
1700         g_printf("BZ2_bzReadOpen failed\n");
1701         BZ2_bzReadClose(&bzerror, b);
1702         return FALSE;
1703 }
1704
1705 do {
1706         r=BZ2_bzRead(&bzerror, b, buffer, FILE_BUFFER);
1707         if ((bzerror!=BZ_STREAM_END) && (bzerror!=BZ_OK)) {
1708                 res=FALSE;
1709                 break;
1710         }
1711         if (!osm_planet_parse_buffer(buffer, r)) {
1712                 res=FALSE;
1713                 break;
1714         }
1715 } while (bzerror==BZ_OK);
1716
1717 BZ2_bzReadClose(&bzerror, b);
1718 fclose(f);
1719 return res;
1720 }
1721
1722 /**
1723  * Set up bounding box for import.
1724  *
1725  */
1726 void
1727 osm_import_set_bbox(gboolean use_bb, gdouble latmin, gdouble lonmin, gdouble latmax, gdouble lonmax)
1728 {
1729 use_bbox=use_bb;
1730 bbox.lat_min=latmin;
1731 bbox.lon_min=lonmin;
1732 bbox.lat_max=latmax;
1733 bbox.lon_max=lonmax;
1734 g_printf("Skipping data outside of box: %f,%f - %f,%f\n",
1735         bbox.lat_min, bbox.lon_min,     bbox.lat_max, bbox.lon_max);
1736 }
1737
1738 static void
1739 osm_print_import_stats(void)
1740 {
1741 g_printf("Total nodes %d, POIs: %d and Ways %d.\n",     node_cnt, noded_cnt, way_cnt);
1742 g_printf("Cities/Towns: %d\n", g_hash_table_size(osm_place_city));
1743 g_printf("Villages/Hamlets: %d\n", g_hash_table_size(osm_place_village));
1744 g_printf("Suburbs: %d\n", g_hash_table_size(osm_place_suburb));
1745 g_printf("Nodes: %d\n", g_hash_table_size(osm_nodes));
1746 }
1747
1748 /**
1749  * Simple helper to do all preparations and importing from planet -> database
1750  *
1751  */
1752 gboolean
1753 osm_import(const gchar *planet, const gchar *database)
1754 {
1755 if (db_connect(&db, database)!=TRUE) {
1756         g_printerr("Database open failed: %s", database);
1757         return FALSE;
1758 }
1759
1760 if (!osm_db_create(db)) {
1761         g_printerr("Failed to create osm tables or indexes\n");
1762         return FALSE;
1763 }
1764
1765 if (!osm_db_prepare(db)) {
1766         g_printerr("Failed to prepare sql statements\n");
1767         return FALSE;
1768 }
1769
1770 osm_planet_parser_init();
1771
1772 if (osm_planet_parse_file(planet)==FALSE) {
1773         g_printerr("Failed to parse file: %s\n", planet);
1774         return FALSE;
1775 }
1776
1777 osm_print_import_stats();
1778
1779 osm_planet_save_all_nodes();
1780 osm_planet_save_all_ways();
1781 osm_planet_parser_deinit();
1782 db_finalize();
1783 db_close(&db);
1784 g_printf("All done.\n");
1785 return TRUE;
1786 }
1787
1788 static gpointer
1789 osm_import_thread(gpointer user_data)
1790 {
1791 gboolean r;
1792 osm_import_data_req *req=(osm_import_data_req *)user_data;
1793
1794 g_assert(req);
1795 g_assert(req->planet);
1796 g_assert(req->db);
1797
1798 osm_import_progress_cb=req->progress_cb!=NULL ? req->progress_cb : NULL;
1799
1800 r=osm_import(req->planet, req->db);
1801 g_debug("OSM import result: %d", r);
1802
1803 g_free(req->planet);
1804 g_free(req->db);
1805
1806 if (req->done_cb!=NULL)
1807         g_idle_add(req->done_cb, GINT_TO_POINTER(r==TRUE ? 0 : 1));
1808
1809 return r==TRUE ? 0 : 1;
1810 }
1811
1812 /**
1813  * Helper to start an import in the background using a thread.
1814  *
1815  * Two callback can be given, one for progress feedback and one when the operation is done.
1816  * Done callback must call the join function.
1817  * Only one import thread can run at a time.
1818  *
1819  */
1820 gboolean
1821 osm_import_bg(const gchar *planet, const gchar *database, GSourceFunc progress_cb, GSourceFunc done_cb)
1822 {
1823 GError *error=NULL;
1824
1825 g_return_val_if_fail(import_thread==NULL, FALSE);
1826
1827 osm_import_req.planet=g_strdup(planet);
1828 osm_import_req.db=g_strdup(database);
1829 osm_import_req.progress_cb=progress_cb;
1830 osm_import_req.done_cb=done_cb;
1831
1832 import_thread=g_thread_create(osm_import_thread, &osm_import_req, TRUE, &error);
1833 if (import_thread==NULL) {
1834         g_free(osm_import_req.planet);
1835         g_free(osm_import_req.db);
1836         g_printerr("Import thread creation failed.\n");
1837         return FALSE;
1838 }
1839 if (osm_import_progress_cb!=NULL)
1840         import_sid=g_timeout_add(1000, osm_import_progress_cb, NULL);
1841 return TRUE;
1842 }
1843
1844 gint
1845 osm_import_join_bg(void)
1846 {
1847 gint ret;
1848 g_assert(import_thread!=NULL);
1849
1850 if (import_sid!=0)
1851         g_source_remove(import_sid);
1852 ret=g_thread_join(import_thread);
1853 import_thread=NULL;
1854 return ret;
1855 }