]> err.no Git - mapper/blob - src/osm-db-import.c
Include settings.h so the banner macro hildon version works.
[mapper] / 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;
65 static guint node_skip_cnt=0;
66 static guint noded_cnt=0;
67 static guint way_cnt=0;
68 static guint way_names=0;
69 static guint way_refs=0;
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 void 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 void
394 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
438 /********************************************************************/
439
440 static void
441 print_way(way *w)
442 {
443 #ifdef VERBOSE
444 g_assert(w);
445 g_printf("Way #%d(N:%d T:%d S:%d IS: %d/%d): %s [%s:%s:%s]\n", 
446                 w->id,  
447                 g_slist_length(w->nodes), 
448                 w->type,
449                 w->data ? w->data->speed : 0,
450                 w->data ? w->data->isin_c : -1,
451                 w->data ? w->data->isin_p : -1,
452                 w->data ? w->data->name ? w->data->name : "" : "", 
453                 w->flags & W_ONEWAY ? "-" : "=", 
454                 w->flags & W_ROUNDABOUT ? "O" : "-", 
455                 w->flags & W_LINK ? "|" : " ");
456 #endif
457 }
458
459 static void
460 print_node(node *n)
461 {
462 #ifdef VERBOSE
463 g_assert(n);
464 g_printf("Node #%d: T:%d IS: %d/%d [%s]\n",
465         n->id,
466         n->type,
467         n->data ? n->data->isin_c : -1,
468         n->data ? n->data->isin_p : -1,
469         n->data ? n->data->name : "");
470 #endif
471 }
472
473 /********************************************************************/
474
475 static gboolean
476 db_insert_node(node *n)
477 {
478 gint32 lat, lon;
479
480 g_assert(n);
481
482 lat=lat2mp_int(n->lat);
483 lon=lon2mp_int(n->lon);
484
485 sqlite3_bind_int(sql.insert_node, 1, n->id);
486
487 /* Projected and integerized lat/lot */
488 sqlite3_bind_int(sql.insert_node, 2, lat);
489 sqlite3_bind_int(sql.insert_node, 3, lon);
490 /* Original */
491 sqlite3_bind_double(sql.insert_node, 4, n->lat);
492 sqlite3_bind_double(sql.insert_node, 5, n->lon);
493 sqlite3_bind_int(sql.insert_node, 6, n->type);
494
495 db_exec(db, sql.insert_node);
496
497 return TRUE;
498 }
499
500 static gboolean
501 db_insert_place(node *n)
502 {
503 g_assert(n);
504 if (!n->data)
505         return FALSE;
506 if (!n->data->name)
507         return FALSE;
508 sqlite3_bind_int(sql.insert_place, 1, n->id);
509 sqlite3_bind_int(sql.insert_place, 2, n->type);
510 sqlite3_bind_text(sql.insert_place, 3, n->data->name, -1, SQLITE_TRANSIENT);
511 sqlite3_bind_int(sql.insert_place, 4, n->data->isin_p);
512 sqlite3_bind_int(sql.insert_place, 5, n->data->isin_c);
513
514 return db_exec(db,sql.insert_place);
515 }
516
517 static gboolean
518 db_insert_poi(node *n)
519 {
520 g_assert(n);
521 sqlite3_bind_int(sql.insert_poi, 1, n->id);
522 sqlite3_bind_double(sql.insert_poi, 2, n->lat);
523 sqlite3_bind_double(sql.insert_poi, 3, n->lon);
524 if (n->data->name)
525         sqlite3_bind_text(sql.insert_poi, 4, n->data->name, -1, SQLITE_TRANSIENT);
526 else
527         sqlite3_bind_text(sql.insert_poi, 4, "", -1, SQLITE_TRANSIENT);
528 sqlite3_bind_int(sql.insert_poi, 5, n->type);
529 sqlite3_bind_int(sql.insert_poi, 6, n->type/100);
530 sqlite3_bind_int(sql.insert_poi, 7, n->data->isin_c);
531 sqlite3_bind_int(sql.insert_poi, 8, n->data->isin_p);
532
533 if (n->data->desc)
534         sqlite3_bind_text(sql.insert_poi, 9, n->data->desc, -1, SQLITE_TRANSIENT);
535 if (n->data->url)
536         sqlite3_bind_text(sql.insert_poi, 10, n->data->url, -1, SQLITE_TRANSIENT);
537 if (n->data->postal_code)
538         sqlite3_bind_text(sql.insert_poi, 11, n->data->postal_code, -1, SQLITE_TRANSIENT);
539
540 return db_exec(db,sql.insert_poi);
541 }
542
543 /**
544  * Update node usage count
545  */
546 static gboolean
547 db_update_node_links(node *n)
548 {
549 g_assert(n);
550 sqlite3_bind_int(sql.update_node, 1, n->id);
551
552 return db_exec(db,sql.update_node);
553 }
554
555 /**
556  * Insert way,node1,node2 triplet
557  */
558 static gboolean
559 db_insert_way_n2n(way *w, node *nf, node *nt)
560 {
561 if (!w) {
562         g_printf("NULL WAY\n");
563         return FALSE;
564 }
565
566 if (!nf) {
567         g_printf("NULL NODE 1\n");
568         return FALSE;
569 }
570
571 if (!nt) {
572         g_printf("NULL NODE 2\n");
573         return FALSE;
574 }
575
576 sqlite3_bind_int(sql.insert_way_n2n, 1, w->id);
577 sqlite3_bind_int(sql.insert_way_n2n, 2, nf->id);
578 sqlite3_bind_int(sql.insert_way_n2n, 3, nt->id);
579
580 #ifdef VERBOSE_N2N
581 g_printf("%d [%d - %d]\n", w->id, nf->id, nt->id);
582 #endif
583
584 db_exec(db,sql.insert_way_n2n);
585 db_update_node_links(nf);
586 db_update_node_links(nt);
587 return TRUE;
588 }
589
590 /**
591  * Insert way ref and int_ref
592  */
593 static gboolean 
594 db_insert_way_ref(way *w)
595 {
596 if (!w->data)
597         return TRUE;
598
599 if (!w->data->ref && !w->data->int_ref)
600         return TRUE;
601
602 way_refs++;
603
604 sqlite3_bind_int(sql.insert_way_ref, 1, w->id);
605 if (w->data->ref)
606         sqlite3_bind_text(sql.insert_way_ref, 2, w->data->ref, -1, SQLITE_TRANSIENT);
607 if (w->data->int_ref)
608         sqlite3_bind_text(sql.insert_way_ref, 3, w->data->int_ref, -1, SQLITE_TRANSIENT);
609
610 return db_exec(db,sql.insert_way_ref);
611 }
612
613 /**
614  * Insert way name
615  */
616 static gboolean
617 db_insert_way_name(way *w)
618 {
619 gchar *norm;
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 norm=g_convert(w->data->name, -1, "ASCII//TRANSLIT//IGNORE", "utf8", NULL, NULL, NULL);
633 if (norm && strcmp(w->data->name, norm)!=0) {
634         sqlite3_bind_text(sql.insert_way_name, 3, norm, -1, SQLITE_TRANSIENT);
635 }
636 if (norm)
637         g_free(norm);
638 #endif
639
640 return db_exec(db,sql.insert_way_name);
641 }
642
643 static gboolean
644 db_delete_way_names_nls(way *w)
645 {
646 sqlite3_bind_int(sql.delete_way_names_nls, 1, w->id);
647 return db_exec(db,sql.delete_way_names_nls);
648 }
649
650 static gboolean 
651 db_insert_way_pc(way *w)
652 {
653 if (!w->data)
654         return TRUE;
655 if (!w->data->postal_code)
656         return TRUE;
657
658 sqlite3_bind_int(sql.insert_way_pc, 1, w->id);
659 sqlite3_bind_text(sql.insert_way_pc, 2, w->data->postal_code, -1, SQLITE_TRANSIENT);
660
661 return db_exec(db,sql.insert_way_pc);
662 }
663
664 static gboolean
665 db_delete_way_pc(way *w)
666 {
667 sqlite3_bind_int(sql.delete_way_pc, 1, w->id);
668 return db_exec(db,sql.delete_way_pc);
669 }
670
671 static void
672 db_insert_way_names_nls_cb(gpointer key, gpointer value, gpointer user_data)
673 {
674 gchar *norm;
675
676 way *w=(way *)user_data;
677
678 sqlite3_bind_int(sql.insert_way_names_nls, 1, w->id);
679 sqlite3_bind_text(sql.insert_way_names_nls, 2, (gchar *)key, -1, SQLITE_TRANSIENT);
680 sqlite3_bind_text(sql.insert_way_names_nls, 3, (gchar *)value, -1, SQLITE_TRANSIENT);
681 #ifdef TRANSLIT_NAMES
682 norm=g_convert((gchar *value), -1, "ASCII//TRANSLIT//IGNORE", "utf8", NULL, NULL, NULL);
683 if (norm && strcmp((gchar *)value, norm)!=0) {
684         sqlite3_bind_text(sql.insert_way_names_nls, 4, norm, -1, SQLITE_TRANSIENT);
685 }
686 if (norm)
687         g_free(norm);
688 #endif
689 db_exec(db,sql.insert_way_names_nls);
690 }
691
692 static void
693 db_insert_way_names_nls(way *w)
694 {
695 if (!w->data)
696         return;
697 if (!w->data->names)
698         return;
699
700 g_hash_table_foreach(w->data->names, db_insert_way_names_nls_cb, w);
701 }
702
703 /**
704  * Insert all data for the given way
705  * - name
706  * - ref
707  * - nodes
708  * 
709  */
710 static gboolean
711 db_insert_way(way *w)
712 {
713 GSList *iter;
714 guint ncnt;
715 node *wmn;
716
717 if (!w)
718         return FALSE;
719
720 /* Skip things we don't use (yet) */
721 if (w->type==WAY_UNWAYED || w->type>WAY_ROAD_END)
722         return TRUE;
723
724 /* Insert nodes */
725 for (iter=w->nodes; iter!=NULL; iter=iter->next) {
726         if (!iter->next)
727                 break;
728         db_insert_way_n2n(w, iter->data, iter->next->data);
729 }
730
731 if (w->id==0)
732         return FALSE;
733
734 if (w->data) {
735         w->data->isin_p=osm_find_way_place(w, NODE_PLACE_CITY);
736         w->data->isin_c=osm_find_way_place(w, NODE_PLACE_COUNTRY);
737 }
738
739 print_way(w);
740
741 sqlite3_bind_int(sql.insert_way_data, 1, w->id);
742 sqlite3_bind_int(sql.insert_way_data, 2, w->ncnt);
743 sqlite3_bind_int(sql.insert_way_data, 3, w->type);
744 sqlite3_bind_int(sql.insert_way_data, 4, w->flags);
745 if (w->data) {
746         sqlite3_bind_int(sql.insert_way_data, 5, w->data->speed);
747         sqlite3_bind_int(sql.insert_way_data, 6, w->data->isin_c);
748         sqlite3_bind_int(sql.insert_way_data, 7, w->data->isin_p);
749 }
750
751 /* Get middle node, use it as a rough way location */
752 ncnt=g_slist_length(w->nodes);
753 if (ncnt>1) {
754         wmn=g_slist_nth_data(w->nodes, ncnt/2);
755         if (wmn) {
756                 sqlite3_bind_double(sql.insert_way_data, 8, wmn->lat);
757                 sqlite3_bind_double(sql.insert_way_data, 9, wmn->lon);
758         } else {
759                 g_printerr("Failed to get way middlepoint node for location information!\n");
760         }
761 }
762
763 db_exec(db,sql.insert_way_data);
764
765 db_insert_way_ref(w);
766 db_insert_way_name(w);
767 db_insert_way_names_nls(w);
768 db_insert_way_pc(w);
769
770 osm_free_way_data(w);
771 return TRUE;
772 }
773
774 /********************************************************************/
775
776 static gchar *
777 get_attr_key_value(const gchar **p, gchar *key)
778 {
779 gchar **d;
780
781 d=p;
782 while (*d!=NULL) {
783         if (strncmp(*d, key, strlen(key))==0) {
784                 d++;
785                 return *d;
786         }
787         d++;
788         d++;
789 }
790 return NULL;
791 }
792
793 static tag_state_t 
794 check_tag(const gchar *tag)
795 {
796 if (strcmp(tag,"node")==0) return IN_NODE_TAG;
797 else if (strcmp(tag,"nd")==0) return IN_WNODE_TAG;
798 else if (strcmp(tag,"way")==0) return IN_WAY_TAG;
799 else if (strcmp(tag,"tag")==0) return IN_KEY_TAG;
800 else if (strcmp(tag,"osm")==0) return IN_OSM_TAG;
801 else if (strcmp(tag,"bound")==0) return IN_BOUND_TAG;
802 else if (strcmp(tag,"relation")==0) return IN_RELATION_TAG;
803 else if (strcmp(tag,"member")==0) return IN_MEMBER_TAG;
804 else return ERROR;
805 }
806
807 static void
808 find_nls_names(gpointer key, gpointer value, gpointer user_data)
809 {
810 gchar *k, *v;
811 gchar *tmp;
812 GHashTable *nls;
813
814 k=(gchar *)key;
815 v=(gchar *)value;
816 nls=(GHashTable *)user_data;
817
818 /* Check if it is a name key, return if not. */
819 if (g_str_has_prefix(k, "name:")==FALSE)
820         return;
821
822 tmp=g_strrstr(k, ":");
823 if (!tmp)
824         return;
825 tmp++; /* skip : */
826 if (*tmp==0)
827         return;
828 g_hash_table_insert(nls, g_strdup(tmp), g_strdup(v));
829 #ifdef VERBOSE
830 g_printf("NLS(%s): [%s]\n", tmp, v);
831 #endif
832 }
833
834 /********************************************************************/
835
836 static void
837 node_print (node *n)
838 {
839 g_assert(n);
840 if (n->data) {
841         g_printf("N: %d [%f:%f][%s](%d)\n", 
842                 n->id, n->lat, n->lon, 
843                 n->data->name ? n->data->name : "-", 
844                 n->type);
845 } else {
846         g_printf("N: %d [%f:%f]\n",
847                 n->id, n->lat, n->lon);
848 }
849 }
850
851 #ifdef DEBUG
852 static void 
853 dump_array(const gchar **p)
854 {
855 char **d;
856
857 d=p;
858 while (*d!=NULL) {
859         g_printf("[%s]", *d);
860         d++;
861 }
862 g_print("\n");
863 }
864 #endif
865
866 static inline gboolean
867 osm_node_check_box(gdouble nlat, gdouble nlon)
868 {
869 if (use_bbox==FALSE)
870         return TRUE;
871 return (nlat > bbox.lat_min && nlat < bbox.lat_max && nlon > bbox.lon_min && nlon < bbox.lon_max) ? TRUE : FALSE;
872 }
873
874 static void
875 osm_new_node_data(node *n)
876 {
877 if (n==NULL) 
878         return;
879 if (n->data!=NULL) 
880         return;
881 n->data=g_slice_new(node_data);
882 n->data->name=NULL;
883 n->data->url=NULL;
884 n->data->desc=NULL;
885 n->data->postal_code=NULL;
886 n->type=NODE_PLAIN;
887 noded_cnt++;
888 }
889
890 static void
891 osm_free_node_data(node *n)
892 {
893 g_assert(n);
894 g_assert(n->data);
895 if (n->data->name)
896         g_free(n->data->name);
897 if (n->data->url)
898         g_free(n->data->url);
899 if (n->data->desc)
900         g_free(n->data->desc);
901 if (n->data->postal_code)
902         g_free(n->data->postal_code);
903 g_slice_free(node_data, n->data);
904 n->data=NULL;
905 noded_cnt--;
906 }
907
908 static node *
909 osm_new_node(gint id, gdouble lat, gdouble lon)
910 {
911 node *n=NULL;
912
913 n=g_slice_new(node);
914 g_assert(n);
915 n->id=id;
916 n->lat=lat;
917 n->lon=lon;
918 n->data=(node_data *)NULL;
919 return n;
920 }
921
922 static void
923 osm_free_node(node *n)
924 {
925 g_assert(n);
926 g_slice_free(node, n);
927 }
928
929 static node *
930 osm_find_node(guint32 nid)
931 {
932 node *n;
933
934 g_assert(osm_nodes);
935 return g_hash_table_lookup(osm_nodes, GINT_TO_POINTER(nid));
936 }
937
938 static void
939 osm_new_way_data(way *w)
940 {
941 if (w==NULL) 
942         return;
943 if (w->data!=NULL) 
944         return;
945
946 w->data=g_slice_new(way_data);
947 w->data->name=NULL;
948 w->data->names=NULL;
949 w->data->ref=NULL;
950 w->data->int_ref=NULL;
951 w->data->postal_code=NULL;
952 w->data->layer=0;
953 w->data->speed=0;
954 }
955
956 static void
957 osm_free_way_data(way *w)
958 {
959 g_assert(w);
960 if (!w->data)
961         return;
962 if (w->data->name)
963         g_free(w->data->name);
964 if (w->data->ref)
965         g_free(w->data->ref);
966 if (w->data->int_ref)
967         g_free(w->data->int_ref);
968 g_slice_free(way_data, w->data);
969 w->data=NULL;
970 }
971
972 static way *
973 osm_new_way(gint id)
974 {
975 way *w;
976
977 w=g_slice_new(way);
978 g_assert(w);
979 w->id=id;
980 w->nodes=NULL;
981 w->type=WAY_UNWAYED;
982 w->data=NULL;
983 w->ncnt=0;
984 w->flags=0;
985
986 /* Add to list of ways */
987 return w;
988 }
989
990 static void
991 osm_free_way(way *w)
992 {
993 if (w->nodes)
994         g_slist_free(w->nodes);
995 g_slice_free(way, w);
996 }
997
998 static void
999 osm_way_add_to_list(way *w)
1000 {
1001 g_assert(w);
1002 osm_ways=g_slist_prepend(osm_ways, w);
1003 }
1004
1005 static void
1006 osm_way_new_node(way *w, gint nid)
1007 {
1008 node *n;
1009
1010 g_assert(w);
1011 n=osm_find_node(nid);
1012 w->nodes=g_slist_prepend(w->nodes, n);
1013 w->ncnt++;
1014 }
1015
1016 /**
1017  * Search the place hash table for the location of the node.
1018  *
1019  */
1020 static guint32 
1021 osm_find_node_place(node *n)
1022 {
1023 node *t;
1024 gchar **isin;
1025 gchar **place;
1026
1027 if (!n->data)
1028         return 0;
1029
1030 isin=g_hash_table_lookup(osm_node_isin, GINT_TO_POINTER(n->id));
1031
1032 if (!isin)
1033         return 0;
1034
1035 place=isin;
1036 while (*place!=NULL) {
1037         gchar *ps;
1038         ps=g_strstrip(*place);
1039 #ifdef VERBOSE
1040         g_printf("Checking (%d) [%s] in [%s]\n",n->type, n->data->name, ps);
1041 #endif
1042         switch (n->type) {
1043         case NODE_PLACE_CITY:
1044         case NODE_PLACE_TOWN:
1045         case NODE_PLACE_VILLAGE:
1046         case NODE_PLACE_HAMLET:
1047                 t=g_hash_table_lookup(osm_place_region, ps);
1048                 if (t)
1049                         return t->id;
1050                 t=g_hash_table_lookup(osm_place_country, ps);
1051                 if (t)
1052                         return t->id;
1053         break;
1054         case NODE_PLACE_SUBURB:
1055         case NODE_PLACE_LOCALITY:
1056                 t=g_hash_table_lookup(osm_place_city, ps);
1057                 if (t)
1058                         return t->id;
1059         break;
1060         case NODE_PLACE_ISLAND:
1061                 return 0;
1062         break;
1063         default:
1064                 t=g_hash_table_lookup(osm_place_city, ps);
1065                 if (t)
1066                         return t->id;
1067         break;
1068         }
1069         place++;
1070 }
1071
1072 return 0;
1073 }
1074
1075 static guint32
1076 osm_find_way_place(way *w, node_type_t nt)
1077 {
1078 gchar **isin;
1079 gchar **place;
1080
1081 isin=g_hash_table_lookup(osm_way_isin, GINT_TO_POINTER(w->id));
1082 if (!isin)
1083         return 0;
1084
1085 place=isin;
1086 while (*place!=NULL) {
1087         node *t;
1088         gchar *ps;
1089
1090         ps=g_strstrip(*place);
1091
1092 #ifdef VERBOSE
1093         g_printf("Checking (%d) in [%s]\n",w->id, ps);
1094 #endif
1095 switch (nt) {
1096         case NODE_PLACE_CITY:
1097         case NODE_PLACE_TOWN:
1098         case NODE_PLACE_VILLAGE:
1099         case NODE_PLACE_HAMLET:
1100         case NODE_PLACE_LOCALITY:
1101                 t=g_hash_table_lookup(osm_place_city, ps);
1102                 if (t)
1103                         return t->id;
1104         break;
1105         case NODE_PLACE_COUNTRY:
1106                 t=g_hash_table_lookup(osm_place_country, ps);
1107                 if (t)
1108                         return t->id;
1109         break;
1110         default:
1111                 g_assert_not_reached();
1112         break;
1113         }
1114         place++;
1115 }
1116
1117 return 0;
1118 }
1119
1120 /***********************************************************************/
1121
1122 static void
1123 osm_node_save_node(gint key, gpointer value, gpointer user_data)
1124 {
1125 node *n=(node *)value;
1126
1127 dbnode_cnt++;
1128 db_insert_node(n);
1129 if (dbnode_cnt % 26214==0)
1130         g_printf("\rNodes: %f%%\n",((float)dbnode_cnt/(float)node_cnt)*100);
1131 }
1132
1133 /**
1134  * Check node type and insert as POI or Place
1135  * Discard extra data after insert.
1136  */
1137 static gboolean
1138 osm_node_save_poi(node *n, gpointer user_data)
1139 {
1140 if (!n) {
1141         g_printerr("ERROR: null poi\n");
1142         return FALSE;
1143 }
1144
1145 if (!n->data) {
1146         g_printerr("POI node with no data ?\n");
1147         return FALSE;
1148 }
1149
1150 n->data->isin_p=osm_find_node_place(n);
1151 n->data->isin_c=0;
1152
1153 if (n->type>NODE_POI_START && n->type<NODE_POI_END) {
1154         print_node(n);
1155         db_insert_poi(n);
1156         osm_free_node_data(n);
1157 } else if (n->type>NODE_PLACE_START && n->type<NODE_PLACE_END) {
1158         print_node(n);
1159         db_insert_place(n);
1160 } else {
1161         osm_free_node_data(n);
1162         return FALSE;
1163 }
1164
1165 return TRUE;
1166 }
1167
1168 static gboolean
1169 osm_planet_poi_clear_nodes(void)
1170 {
1171 g_print("Removing old OSM POIs...\n");
1172 db_transaction_begin(db);
1173 sqlite3_step(sql.delete_osm_poi);
1174 sqlite3_step(sql.delete_place);
1175 return db_transaction_commit(db);
1176 }
1177
1178 static gboolean
1179 osm_planet_poi_save_nodes(void)
1180 {
1181 g_print("Storing new POIs...\n");
1182 db_transaction_begin(db);
1183 g_slist_foreach(osm_poi, osm_node_save_poi, NULL);
1184 g_slist_free(osm_poi);
1185 return db_transaction_commit(db);
1186 }
1187
1188 /*********************************************************************/
1189
1190 static void
1191 osm_planet_clear_nodes(void)
1192 {
1193 g_print("Clearing old nodes\n");
1194 sqlite3_step(sql.delete_nodes);
1195 }
1196
1197 static gboolean
1198 osm_planet_save_nodes(void)
1199 {
1200 g_print("Storing nodes...\n");
1201 db_transaction_begin(db);
1202 g_hash_table_foreach(osm_nodes, osm_node_save_node, NULL);
1203 return db_transaction_commit(db);
1204 }
1205
1206 /*********************************************************************/
1207
1208 static void
1209 osm_way_save(way *value, gpointer user_data)
1210 {
1211 dbway_cnt++;
1212 db_insert_way(value);
1213 if (dbway_cnt % 16384==0 && dbway_cnt>0) {
1214                 g_printf("\rWays: %f%%\n",(((float)dbway_cnt/(float)way_cnt)*100));
1215                 print_way(value);
1216 }
1217 }
1218
1219 static void
1220 osm_planet_clear_ways(void)
1221 {
1222 g_print("Clearing old data\n");
1223 sqlite3_step(sql.delete_way);
1224 sqlite3_step(sql.delete_way_name);
1225 sqlite3_step(sql.delete_way_ref);
1226 sqlite3_step(sql.delete_way_n2n);
1227 }
1228
1229 static gboolean
1230 osm_planet_save_ways(void)
1231 {
1232 g_print("Inserting new ways\n");
1233 db_transaction_begin(db);
1234 g_slist_foreach(osm_ways, osm_way_save, NULL);
1235 return db_transaction_commit(db);
1236 }
1237
1238 /*********************************************************************/
1239
1240 static void
1241 osm_planet_save_all_nodes(void)
1242 {
1243 g_printf("Saving planet nodes to database:\n");
1244
1245 osm_planet_poi_clear_nodes();
1246 osm_planet_poi_save_nodes();
1247
1248 if (!is_update) {
1249         osm_planet_clear_nodes();
1250         osm_planet_clear_ways();
1251 }
1252 osm_planet_save_nodes();
1253 }
1254
1255 static void
1256 osm_planet_save_all_ways(void)
1257 {
1258 g_printf("Saving planet way to database:\n");
1259
1260 osm_planet_save_ways();
1261 }
1262
1263 /***********************************************************************/
1264
1265 static void
1266 _osm_tag_start(void *userData, const char *name, const char **atts)
1267 {
1268 tag_state_t t;
1269 gchar *k, *v;
1270 guint32 id, ndref;
1271 gdouble nlat, nlon;
1272
1273 t=check_tag(name);
1274 switch (t) {
1275         case IN_OSM_TAG:
1276                 g_printf("Starting...\n");
1277         break;
1278         case IN_NODE_TAG:
1279                 tag_parent=IS_NODE;
1280                 node_cnt++;
1281
1282                 id=atoi(get_attr_key_value(atts, "id"));
1283                 nlat=atof(get_attr_key_value(atts, "lat"));
1284                 nlon=atof(get_attr_key_value(atts, "lon"));
1285
1286                 cnode=osm_new_node(id, nlat, nlon);
1287                 osm_node_tags=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1288         break;
1289         case IN_WAY_TAG:
1290                 tag_parent=IS_WAY;
1291                 way_cnt++;
1292                 id=atoi(get_attr_key_value(atts, "id"));
1293                 cway=osm_new_way(id);
1294                 osm_way_tags=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1295         break;
1296         case IN_WNODE_TAG:
1297                 ndref=atoi(get_attr_key_value(atts, "ref"));
1298                 if (use_bbox==TRUE) {
1299                         if (osm_find_node(ndref)==NULL) {
1300                                 cway->id=0;
1301                                 return;
1302                         }
1303                 }
1304                 osm_way_new_node(cway, ndref);
1305         break;
1306         case IN_KEY_TAG:
1307                 k=get_attr_key_value(atts, "k");
1308                 if (strcmp(k,"created_by")==0)
1309                         return;
1310                 if (strcmp(k,"source")==0)
1311                         return;
1312
1313                 v=get_attr_key_value(atts, "v");
1314 #ifdef VERBOSE_KEYS
1315                 g_printf("TAG: K=[%s] V=[%s]\n", k, v);
1316 #endif
1317
1318                 switch (tag_parent) {
1319                 case IS_NONE:
1320                         g_printf("Tag key/value pair but unknown owner\n");
1321                 break;
1322                 case IS_NODE:
1323                 {
1324                         if (!osm_node_tags)
1325                                 return;
1326
1327                         /* Insert key/value pairs into hash table */
1328                         if (cnode==NULL) {
1329                                 g_printerr("In node tags but node is NULL!\n");
1330                                 return;
1331                         }
1332                         g_hash_table_insert(osm_node_tags, g_strdup(k), g_strdup(v));
1333                 }
1334                 break;
1335                 case IS_WAY: 
1336                 {
1337                         if (cway==NULL) {
1338                                 g_printerr("In way tags but way is NULL!\n");
1339                                 return;
1340                         }
1341                         g_hash_table_insert(osm_way_tags, g_strdup(k), g_strdup(v));
1342                         osm_new_way_data(cway);
1343                 }
1344                 break;
1345                 case IS_RELATION:
1346
1347                 break;
1348                 }
1349         break;
1350         case IN_BOUND_TAG:
1351                 /* Ignore for now */
1352                 g_printf("Ignoring bound tag\n");
1353         break;
1354         case IN_RELATION_TAG:
1355                 tag_parent=IS_RELATION;
1356
1357         break;
1358         case IN_MEMBER_TAG:
1359
1360         break;
1361         default:
1362                 tag_parent=IS_NONE;
1363                 g_printf("Unknown tag: %s\n", name);
1364         break;
1365 }
1366 }
1367
1368 static void
1369 _osm_tag_end(void *userData, const char *name)
1370 {
1371 tag_state_t t;
1372 gchar *v;
1373 guint i;
1374 t=check_tag(name);
1375 switch (t) {
1376         case IN_NODE_TAG:
1377
1378                 if (node_cnt % 262140==0) {
1379                         g_printf("Nodes: %d of %d, POIs: %d, Outside box: %d\n", node_cnt-node_skip_cnt, node_cnt, noded_cnt, node_skip_cnt);
1380                 }
1381
1382                 if (!osm_node_tags)
1383                         return;
1384
1385                 osm_new_node_data(cnode);
1386
1387                 for (i=0; nodeinfo[i].k; i++) {
1388                         v=g_hash_table_lookup(osm_node_tags, nodeinfo[i].k);
1389                         if (!v)
1390                                 continue;
1391                         if (strcasecmp (v, nodeinfo[i].v)==0) {
1392                                 cnode->type=nodeinfo[i].type;
1393                                 break;
1394                         }
1395                 }
1396
1397                 /* Check if node is inside bounding box, if not skip it. 
1398                  * But keep it if it's something we might need for other nodes:
1399                  * - Places (for is_in)
1400                  * - ...
1401                  */
1402                 if ((osm_node_check_box(cnode->lat, cnode->lon)==FALSE) && 
1403                                 (cnode->type<NODE_PLACE_START)) {
1404                         osm_free_node_data(cnode);
1405                         osm_free_node(cnode);
1406                         g_hash_table_destroy(osm_node_tags);
1407                         node_skip_cnt++;
1408                         return;
1409                 }
1410
1411                 g_hash_table_insert(osm_nodes, GINT_TO_POINTER(cnode->id), cnode);
1412
1413                 if (cnode->type!=NODE_PLAIN) {
1414                         cnode->data->name=NULL;
1415                         v=g_hash_table_lookup(osm_node_tags, "name");
1416                         if (v)
1417                                 cnode->data->name=g_strstrip(g_utf8_normalize(v, -1, G_NORMALIZE_ALL_COMPOSE));
1418                         v=g_hash_table_lookup(osm_node_tags, "note");
1419                         if (v)
1420                                 cnode->data->desc=g_strstrip(g_strdup(v));
1421                         v=g_hash_table_lookup(osm_node_tags, "postal_code");
1422                         if (v)
1423                                 cnode->data->postal_code=g_strstrip(g_strdup(v));
1424
1425                         /* Links */
1426                         v=g_hash_table_lookup(osm_node_tags, "url");
1427                         if (v) {
1428                                 cnode->data->url=g_strstrip(g_strdup(v));
1429                         } else {
1430                                 v=g_hash_table_lookup(osm_node_tags, "wikipedia");
1431                                 if (v && strncmp(v,"http:", 5)==0) 
1432                                         cnode->data->url=g_strstrip(g_strdup(v));
1433                         }
1434                 }
1435
1436                 cnode->data->isin_c=0;
1437                 cnode->data->isin_p=0;
1438                 v=g_hash_table_lookup(osm_node_tags, "is_in");
1439                 if (v) {
1440                         gchar **isin;                           
1441                         isin=g_strsplit(v, ",", 10);
1442                         g_hash_table_insert(osm_node_isin, GINT_TO_POINTER(cnode->id), isin);
1443                 }
1444
1445                 if (cnode->type==NODE_PLAIN) {
1446                         osm_free_node_data(cnode);
1447                 } else {
1448                         osm_poi=g_slist_prepend(osm_poi, cnode);
1449                         if (cnode->data->name) {
1450                                 switch (cnode->type) {
1451                                 case NODE_PLACE_COUNTRY:
1452                                         g_hash_table_insert(osm_place_country, cnode->data->name, cnode);
1453                                 break;
1454                                 case NODE_PLACE_CITY:
1455                                 case NODE_PLACE_TOWN:
1456                                         g_hash_table_insert(osm_place_city, cnode->data->name, cnode);
1457                                 break;
1458                                 case NODE_PLACE_SUBURB:
1459                                         g_hash_table_insert(osm_place_suburb, cnode->data->name, cnode);
1460                                 break;
1461                                 case NODE_PLACE_VILLAGE:
1462                                 case NODE_PLACE_HAMLET:
1463                                 case NODE_PLACE_LOCALITY:
1464                                         g_hash_table_insert(osm_place_village, cnode->data->name, cnode);
1465                                 break;
1466                                 case NODE_PLACE_ISLAND:
1467                                         /* Ignore for now */
1468                                 break;
1469                                 default:;
1470                                 }
1471                         }
1472                 }
1473                 g_hash_table_destroy(osm_node_tags);
1474                 cnode=NULL;
1475         break;
1476         case IN_WAY_TAG:
1477                 if (way_cnt % 1024==0) {
1478                         g_printf("\rWays: %d\n", way_cnt);
1479                 }
1480
1481                 cway->nodes=g_slist_reverse(cway->nodes);
1482
1483                 for (i=0; wayinfo[i].k; i++) {
1484                         v=g_hash_table_lookup(osm_way_tags, wayinfo[i].k);
1485                         if (!v)
1486                                 continue;
1487                         if (strcasecmp (v, wayinfo[i].v)==0) {
1488                                 if (wayinfo[i].link==TRUE)
1489                                         cway->flags|=W_LINK;
1490                                 if (wayinfo[i].area==TRUE)
1491                                         cway->flags|=W_AREA;
1492                                 if (wayinfo[i].oneway==TRUE)
1493                                         cway->flags|=W_ONEWAY;
1494                                 cway->type=wayinfo[i].type;
1495                                 if (cway->data->speed==0)
1496                                         cway->data->speed=wayinfo[i].defspeed;
1497                                 break;
1498                         }
1499                 }
1500
1501                 v=g_hash_table_lookup(osm_way_tags, "name");
1502                 if (v) {
1503                         cway->data->name=g_utf8_normalize(v, -1, G_NORMALIZE_ALL_COMPOSE);
1504                         /* Try to find other language names */
1505                         cway->data->names=g_hash_table_new_full(g_str_hash, g_str_equal, g_free, g_free);
1506                         g_hash_table_foreach(osm_way_tags, find_nls_names, cway->data->names);
1507                         if (g_hash_table_size(cway->data->names)==0) {
1508                                 g_hash_table_destroy(cway->data->names);
1509                                 cway->data->names=NULL;
1510                         }
1511                 }
1512
1513                 v=g_hash_table_lookup(osm_way_tags, "ref");
1514                 if (v)
1515                         cway->data->ref=g_strdup(v);
1516                 v=g_hash_table_lookup(osm_way_tags, "int_ref");
1517                 if (v)
1518                         cway->data->int_ref=g_strdup(v);
1519
1520                 v=g_hash_table_lookup(osm_way_tags, "postal_code");
1521                 if (v)
1522                         cway->data->postal_code=g_strdup(v);
1523
1524                 /* XXX: somehow handle the silly -1 'reversed' oneway */
1525                 v=g_hash_table_lookup(osm_way_tags, "oneway");
1526                 if (v)
1527                         cway->flags|=W_ONEWAY;
1528
1529                 v=g_hash_table_lookup(osm_way_tags, "noexit");
1530                 if (v)
1531                         cway->flags|=W_NOEXIT;
1532                 
1533                 v=g_hash_table_lookup(osm_way_tags, "speedlimit");
1534                 if (v)
1535                         cway->data->speed=atoi(v);
1536                 v=g_hash_table_lookup(osm_way_tags, "maxspeed");
1537                 if (v)
1538                         cway->data->speed=atoi(v);
1539
1540 #if 0
1541                 v=g_hash_table_lookup(osm_way_tags, "layer");
1542                 if (v)
1543                         cway->data->layer=atoi(v);
1544 #endif
1545
1546                 v=g_hash_table_lookup(osm_way_tags, "junction");
1547                 if (v && strcasecmp(v,"roundabout")==0) {
1548                         cway->flags|=W_ROUNDABOUT;
1549                         cway->flags|=W_ONEWAY;
1550                 } else if (v && strcasecmp(v,"mini_roundabout")==0) {
1551                         cway->flags|=W_ROUNDABOUT;
1552                         cway->flags|=W_ONEWAY;
1553                 }
1554
1555                 /* XXX: Should check keys */
1556                 v=g_hash_table_lookup(osm_way_tags, "access");
1557                 if (v && (strcasecmp(v, "private")==0)) {
1558                         cway->flags|=W_NOACCESS;
1559                 }
1560
1561                 print_way(cway);
1562
1563                 v=g_hash_table_lookup(osm_way_tags, "is_in");
1564                 if (v) {
1565                         gchar **isin;                           
1566                         isin=g_strsplit(v, ",", 10);
1567                         g_hash_table_insert(osm_way_isin, GINT_TO_POINTER(cway->id), isin);
1568                 }
1569
1570                 if (cway->data && cway->data->name==NULL && cway->data->ref==NULL &&
1571                         cway->data->int_ref==NULL && cway->data->layer==0 && cway->data->speed==0)
1572                         osm_free_way_data(cway);
1573
1574                 if (cway->id!=0)
1575                         osm_way_add_to_list(cway);
1576                 else 
1577                         osm_free_way(cway);
1578
1579                 cway=NULL;
1580                 g_hash_table_destroy(osm_way_tags);
1581         break;
1582         case IN_BOUND_TAG:
1583                 /* */
1584         break;
1585         case IN_OSM_TAG:
1586                 g_printf("\nPlanet loaded.\n");
1587         break;
1588         default:;
1589 }
1590 }
1591
1592 /************************************************************************/
1593
1594 static void
1595 storage_init(void)
1596 {
1597 osm_nodes=g_hash_table_new(g_direct_hash, g_direct_equal);
1598
1599 osm_place_country=g_hash_table_new(g_str_hash, g_str_equal);
1600 osm_place_city=g_hash_table_new(g_str_hash, g_str_equal);
1601 osm_place_suburb=g_hash_table_new(g_str_hash, g_str_equal);
1602 osm_place_village=g_hash_table_new(g_str_hash, g_str_equal);
1603 osm_place_region=g_hash_table_new(g_str_hash, g_str_equal);
1604 osm_node_isin=g_hash_table_new(g_direct_hash, g_direct_equal);
1605 osm_way_isin=g_hash_table_new(g_direct_hash, g_direct_equal);
1606 }
1607
1608 static void
1609 storage_free(void)
1610 {
1611 g_hash_table_destroy(osm_nodes);
1612
1613 g_hash_table_destroy(osm_place_country);
1614 g_hash_table_destroy(osm_place_city);
1615 g_hash_table_destroy(osm_place_suburb);
1616 g_hash_table_destroy(osm_place_village);
1617 g_hash_table_destroy(osm_place_region);
1618 g_hash_table_destroy(osm_node_isin);
1619 }
1620
1621 /************************************************************************/
1622
1623 static gint
1624 print_fail(const gchar *msg, gint ret)
1625 {
1626 g_printerr("ERROR: %s\n", msg);
1627 return ret;
1628 }
1629
1630 /************************************************************************/
1631
1632 static void
1633 print_memory_usage(void)
1634 {
1635 g_print("Memory usage per item:\n");
1636 g_printf("Node  size: %d\n", (gint)sizeof(node));
1637 g_printf("NodeD size: %d\n", (gint)sizeof(node_data));
1638 g_printf("Way   size: %d\n", (gint)sizeof(way));
1639 g_printf("WayD  size: %d\n", (gint)sizeof(way_data));
1640 }
1641
1642 /************************************************************************
1643  * Public inteface
1644  ************************************************************************/
1645
1646 void
1647 osm_planet_parser_init(void)
1648 {
1649 xp=XML_ParserCreate(NULL);
1650 XML_SetElementHandler(xp, _osm_tag_start, _osm_tag_end);
1651 storage_init();
1652 }
1653
1654 void
1655 osm_planet_parser_deinit(void)
1656 {
1657 XML_ParserFree(xp);
1658 storage_free();
1659 }
1660
1661 gboolean
1662 osm_planet_parse_buffer(gchar *buffer, size_t r)
1663 {
1664 if (XML_Parse(xp, buffer, r, r>0 ? 0:1) == XML_STATUS_ERROR) {
1665         g_printerr("Parse error at line %d:\n%s\n",
1666                 (gint)XML_GetCurrentLineNumber(xp),
1667                 XML_ErrorString(XML_GetErrorCode(xp)));
1668         return FALSE;
1669 }
1670 return TRUE;
1671 }
1672
1673 gboolean 
1674 osm_planet_parse_file(gchar *pfile)
1675 {
1676 FILE *f;
1677 BZFILE *b;
1678 int bzerror;
1679 int r;
1680 gchar buffer[FILE_BUFFER];
1681 gboolean res=TRUE;
1682
1683 f=fopen(pfile, "r");
1684 if (!f) {
1685         perror("fopen failed\n");
1686         return FALSE;
1687 }
1688
1689 b=BZ2_bzReadOpen(&bzerror, f, 0, 0, NULL, 0);
1690 if (bzerror != BZ_OK) {
1691         g_printf("BZ2_bzReadOpen failed\n");
1692         BZ2_bzReadClose(&bzerror, b);
1693         return FALSE;
1694 }
1695
1696 do {
1697         r=BZ2_bzRead(&bzerror, b, buffer, FILE_BUFFER);
1698         if ((bzerror!=BZ_STREAM_END) && (bzerror!=BZ_OK)) {
1699                 res=FALSE;
1700                 break;
1701         }
1702         if (!osm_planet_parse_buffer(buffer, r)) {
1703                 res=FALSE;
1704                 break;
1705         }
1706 } while (bzerror==BZ_OK);
1707
1708 BZ2_bzReadClose(&bzerror, b);
1709 fclose(f);
1710 return res;
1711 }
1712
1713 /**
1714  * Set up bounding box for import.
1715  *
1716  */
1717 void
1718 osm_import_set_bbox(gboolean use_bb, gdouble latmin, gdouble lonmin, gdouble latmax, gdouble lonmax)
1719 {
1720 use_bbox=use_bb;
1721 bbox.lat_min=latmin;
1722 bbox.lon_min=lonmin;
1723 bbox.lat_max=latmax;
1724 bbox.lon_max=lonmax;
1725 g_printf("Skipping data outside of box: %f,%f - %f,%f\n",
1726         bbox.lat_min, bbox.lon_min,     bbox.lat_max, bbox.lon_max);
1727 }
1728
1729 static void
1730 osm_print_import_stats(void)
1731 {
1732 g_printf("Total nodes %d, POIs: %d and Ways %d.\n",     node_cnt, noded_cnt, way_cnt);
1733 g_printf("Cities/Towns: %d\n", g_hash_table_size(osm_place_city));
1734 g_printf("Villages/Hamlets: %d\n", g_hash_table_size(osm_place_village));
1735 g_printf("Suburbs: %d\n", g_hash_table_size(osm_place_suburb));
1736 g_printf("Nodes: %d\n", g_hash_table_size(osm_nodes));
1737 }
1738
1739 /**
1740  * Simple helper to do all preparations and importing from planet -> database
1741  *
1742  */
1743 gboolean
1744 osm_import(const gchar *planet, const gchar *database)
1745 {
1746 if (db_connect(&db, database)!=TRUE) {
1747         g_printerr("Database open failed: %s", database);
1748         return FALSE;
1749 }
1750
1751 db_create_tables(db);
1752 db_create_indexes(db);
1753 db_prepare(db);
1754 osm_planet_parser_init();
1755
1756 if (osm_planet_parse_file(planet)==FALSE) {
1757         g_printerr("Failed to parse file: %s\n", planet);
1758         return FALSE;
1759 }
1760
1761 osm_print_import_stats();
1762
1763 osm_planet_save_all_nodes();
1764 osm_planet_save_all_ways();
1765 osm_planet_parser_deinit();
1766 db_finalize();
1767 db_close(&db);
1768 g_printf("All done.\n");
1769 return TRUE;
1770 }
1771
1772 static gpointer 
1773 osm_import_thread(gpointer user_data)
1774 {
1775 gboolean r;
1776 osm_import_data_req *req=(osm_import_data_req *)user_data;
1777
1778 g_assert(req);
1779 g_assert(req->planet);
1780 g_assert(req->db);
1781
1782 osm_import_progress_cb=req->progress_cb!=NULL ? req->progress_cb : NULL;
1783
1784 r=osm_import(req->planet, req->db);
1785 g_debug("OSM import result: %d", r);
1786
1787 g_free(req->planet);
1788 g_free(req->db);
1789
1790 if (req->done_cb!=NULL)
1791         g_idle_add(req->done_cb, GINT_TO_POINTER(r==TRUE ? 0 : 1));
1792
1793 return r==TRUE ? 0 : 1;
1794 }
1795
1796 /**
1797  * Helper to start an import in the background using a thread.
1798  *
1799  * Two callback can be given, one for progress feedback and one when the operation is done.
1800  * Done callback must call the join function.
1801  * Only one import thread can run at a time.
1802  *
1803  */
1804 gboolean 
1805 osm_import_bg(const gchar *planet, const gchar *database, GSourceFunc progress_cb, GSourceFunc done_cb)
1806 {
1807 GError *error=NULL;
1808
1809 g_return_val_if_fail(import_thread==NULL, FALSE);
1810
1811 osm_import_req.planet=g_strdup(planet);
1812 osm_import_req.db=g_strdup(database);
1813 osm_import_req.progress_cb=progress_cb;
1814 osm_import_req.done_cb=done_cb;
1815
1816 import_thread=g_thread_create(osm_import_thread, &osm_import_req, TRUE, &error);
1817 if (import_thread==NULL) {
1818         g_free(osm_import_req.planet);
1819         g_free(osm_import_req.db);
1820         g_printerr("Import thread creation failed.\n");
1821         return FALSE;
1822 }
1823 if (osm_import_progress_cb!=NULL)
1824         import_sid=g_timeout_add(1000, osm_import_progress_cb, NULL);
1825 return TRUE;
1826 }
1827
1828 gint
1829 osm_import_join_bg(void)
1830 {
1831 gint ret;
1832 g_assert(import_thread!=NULL);
1833
1834 if (import_sid!=0)
1835         g_source_remove(import_sid);
1836 ret=g_thread_join(import_thread);
1837 import_thread=NULL;
1838 return ret;
1839 }