trunk/src/osmparser.c

/***************************************
 $Header: /home/amb/CVS/routino/src/osmparser.c,v 1.5 2009-01-10 11:53:48 amb Exp $

 OSM XML file parser (either JOSM or planet)
 ******************/ /******************
 Written by Andrew M. Bishop

 This file Copyright 2008,2009 Andrew M. Bishop
 It may be distributed under the GNU Public License, version 2, or
 any higher version.  See section COPYING of the GNU Public license
 for conditions under which this file may be redistributed.
 ***************************************/


#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>

#include "nodes.h"
#include "ways.h"
#include "segments.h"
#include "functions.h"


#define BUFFSIZE 64

static char *fgets_realloc(char *buffer,FILE *file);


/*++++++++++++++++++++++++++++++++++++++
  Parse an OSM XML file (from JOSM or planet download).

  int ParseXML Returns 0 if OK or something else in case of an error.

  FILE *file The file to read from.

  NodesMem *OSMNodes The array of nodes to fill in.

  SegmentsMem *OSMSegments The array of segments to fill in.

  WaysMem *OSMWays The arrray of ways to fill in.
  ++++++++++++++++++++++++++++++++++++++*/

int ParseXML(FILE *file,NodesMem *OSMNodes,SegmentsMem *OSMSegments,WaysMem *OSMWays)
{
 char *line=NULL;
 long nlines=0;
 long nnodes=0,nways=0,nrelations=0;
 int isnode=0,isway=0,isrelation=0;
 way_t way_id=0;
 int way_oneway=0,way_roundabout=0;
 float way_maxspeed=0;
 char *way_highway=NULL,*way_name=NULL,*way_ref=NULL,*way_access=NULL,*way_car=NULL;
 node_t *way_nodes=NULL;
 int way_nnodes=0,way_nalloc=0;

 /* Parse the file */

 while((line=fgets_realloc(line,file)))
   {
    char *l=line,*m;

    nlines++;

    while(isspace(*l))
       l++;

    if(!strncmp(l,"<node",5)) /* The start of a node */
      {
       node_t id;
       latlong_t latitude,longitude;

       nnodes++;

       isnode=1; isway=0; isrelation=0;

       m=strstr(l,"id=");  m+=4; if(*m=='"' || *m=='\'') m++; id=atoll(m);
       m=strstr(l,"lat="); m+=5; if(*m=='"' || *m=='\'') m++; latitude=atof(m);
       m=strstr(l,"lon="); m+=4; if(*m=='"' || *m=='\'') m++; longitude=atof(m);

       AppendNode(OSMNodes,id,latitude,longitude);
      }
    else if(!strncmp(l,"</node",6)) /* The end of a node */
      {
       isnode=0; isway=0; isrelation=0;
      }
    else if(!strncmp(l,"<way",4)) /* The start of a way */
      {
       nways++;

       isnode=0; isway=1; isrelation=0;

       m=strstr(l,"id="); m+=3; if(*m=='"' || *m=='\'') m++; way_id=atol(m);

       way_oneway=0;
       way_roundabout=0;
       way_maxspeed=0;
       way_name=NULL; way_ref=NULL;
       way_nnodes=0;
      }
    else if(!strncmp(l,"</way",5)) /* The end of a way */
      {
       double speed=0;
       int i;

       isnode=0; isway=0; isrelation=0;

       if(!way_highway) speed=0;
       else if(way_access && !strcmp(way_access,"private")) speed=0;
       else if(way_car && !strcmp(way_car,"no")) speed=0;
       else if(way_maxspeed) speed=way_maxspeed;
       else if(!strncmp(way_highway,"motorway",8)) speed=80*1.6;
       else if(!strncmp(way_highway,"trunk",5) && way_oneway) speed=75*1.6;
       else if(!strncmp(way_highway,"primary",7) && way_oneway) speed=70*1.6;
       else if(!strncmp(way_highway,"trunk",5)) speed=65*1.6;
       else if(!strncmp(way_highway,"primary",7)) speed=60*1.6;
       else if(!strncmp(way_highway,"secondary",9)) speed=55*1.6;
       else if(!strcmp(way_highway,"tertiary")) speed=50*1.6;
       else if(!strcmp(way_highway,"unclassified") || !strcmp(way_highway,"road") || !strcmp(way_highway,"minor")) speed=40*1.6;
       else if(!strcmp(way_highway,"residential")) speed=30*1.6;
       else if(!strcmp(way_highway,"service")) speed=20*1.6;
       else if(!strcmp(way_highway,"track") || !strcmp(way_highway,"byway") || !strcmp(way_highway,"unsurfaced") || !strcmp(way_highway,"unpaved")) speed=10*1.6;
//      else if(!strcmp(way_highway =~ m%(steps|path|walkway|footway|pedestrian|bridleway|cycleway|living_street)%) { $speed = 5*1.6; }

       if(speed)
         {
          char *refname;

          if(way_ref && way_name)
            {
             refname=(char*)malloc(strlen(way_ref)+strlen(way_name)+4);
             sprintf(refname,"%s (%s)",way_name,way_ref);
            }
          else if(way_ref && !way_name)
             refname=way_ref;
          else if(!way_ref && way_name)
             refname=way_name;
          else if(way_roundabout)
            {
             refname=(char*)malloc(strlen(way_highway)+14);
             sprintf(refname,"%s (roundabout)",way_highway);
            }
          else /* if(!way_ref && !way_name && !way_roundabout) */
             refname=way_highway;

          for(i=1;i<way_nnodes;i++)
            {
             node_t from=way_nodes[i-1];
             node_t to  =way_nodes[i];

             AppendSegment(OSMSegments,from,to,way_id);

             if(!way_oneway)
                AppendSegment(OSMSegments,to,from,way_id);
            }

          AppendWay(OSMWays,way_id,refname,speed);

          if(refname!=way_ref && refname!=way_name && refname!=way_highway)
             free(refname);
         }

       if(way_highway) {free(way_highway); way_highway=NULL;}
       if(way_name)    {free(way_name);    way_name=NULL;}
       if(way_ref)     {free(way_ref);     way_ref=NULL;}
       if(way_access)  {free(way_access);  way_access=NULL;}
       if(way_car)     {free(way_car);     way_car=NULL;}
      }
    else if(!strncmp(l,"<relation",9)) /* The start of a relation */
      {
       nrelations++;

       isnode=0; isway=0; isrelation=1;
      }
    else if(!strncmp(l,"</relation",10)) /* The end of a relation */
      {
       isnode=0; isway=0; isrelation=0;
      }
    else if(isnode) /* The middle of a node */
      {
      }
    else if(isway) /* The middle of a way */
      {
       node_t id;

       if(!strncmp(l,"<nd",3)) /* The start of a node specifier */
         {
          m=strstr(l,"ref="); m+=4; if(*m=='"' || *m=='\'') m++; id=atoll(m);

          if(way_nnodes<=way_nalloc)
             way_nodes=(node_t*)realloc((void*)way_nodes,(way_nalloc+=16)*sizeof(node_t));

          way_nodes[way_nnodes++]=id;
         }

       if(!strncmp(l,"<tag",4)) /* The start of a tag specifier */
         {
          char delimiter,*k="",*v="";

          m=strstr(l,"k="); m+=2; delimiter=*m; m++; k=m;
          while(*m!=delimiter) m++; *m=0; l=m+1;

          m=strstr(l,"v="); m+=2; delimiter=*m; m++; v=m;
          while(*m!=delimiter) m++; *m=0;

          if(!strcmp(k,"oneway") && (!strcmp(v,"true") || !strcmp(v,"yes"))) way_oneway=1;
          if(!strcmp(k,"junction") && !strcmp(v,"roundabout")) {way_oneway=1; way_roundabout=1;}
          if(!strcmp(k,"highway") && !strncmp(v,"motorway",8)) way_oneway=1;

          if(!strcmp(k,"highway"))
             way_highway=strcpy((char*)malloc(strlen(v)+1),v);

          if(!strcmp(k,"name"))
             way_name=strcpy((char*)malloc(strlen(v)+1),v);

          if(!strcmp(k,"ref"))
             way_ref=strcpy((char*)malloc(strlen(v)+1),v);

          if(!strcmp(k,"access"))
             way_access=strcpy((char*)malloc(strlen(v)+1),v);

          if(!strcmp(k,"car"))
             way_car=strcpy((char*)malloc(strlen(v)+1),v);

          if(!strcmp(k,"maxspeed"))
            {
             way_maxspeed=atof(v);
             if(strstr(v,"mph"))
                way_maxspeed*=1.6;
            }
         }
      }
    else if(isrelation) /* The middle of a relation */
      {
      }

    if(!(nlines%10000))
      {
       printf("\rReading: Lines=%ld Nodes=%ld Ways=%ld Relations=%ld",nlines,nnodes,nways,nrelations);
       fflush(stdout);
      }
   }

 printf("\rRead   : Lines=%ld Nodes=%ld Ways=%ld Relations=%ld\n",nlines,nnodes,nways,nrelations);
 fflush(stdout);

 if(way_nalloc)
    free(way_nodes);

 return(0);
}


/*++++++++++++++++++++++++++++++++++++++
  Call fgets and realloc the buffer as needed to get a whole line.

  char *fgets_realloc Returns the modified buffer (NULL at the end of the file).

  char *buffer The current buffer.

  FILE *file The file to read from.
  ++++++++++++++++++++++++++++++++++++++*/

static char *fgets_realloc(char *buffer,FILE *file)
{
 int n=0;
 char *buf;

 if(!buffer)
    buffer=(char*)malloc(BUFFSIZE+1);

 while((buf=fgets(&buffer[n],BUFFSIZE,file)))
   {
    int s=strlen(buf);
    n+=s;

    if(buffer[n-1]=='\n')
       break;
    else
       buffer=(char*)realloc(buffer,n+BUFFSIZE+1);
   }

 if(!buf)
   {free(buffer);buffer=NULL;}

 return(buffer);
}
1	amb	2	/***************************************
2	amb	26	$Header: /home/amb/CVS/routino/src/osmparser.c,v 1.5 2009-01-10 11:53:48 amb Exp $
3	amb	2
4			OSM XML file parser (either JOSM or planet)
5			****************/ /****************
6			Written by Andrew M. Bishop
7
8	amb	5	This file Copyright 2008,2009 Andrew M. Bishop
9	amb	2	It may be distributed under the GNU Public License, version 2, or
10			any higher version. See section COPYING of the GNU Public license
11			for conditions under which this file may be redistributed.
12			***************************************/
13
14
15			#include <stdio.h>
16			#include <stdlib.h>
17			#include <string.h>
18			#include <ctype.h>
19
20	amb	26	#include "nodes.h"
21			#include "ways.h"
22			#include "segments.h"
23	amb	2	#include "functions.h"
24
25	amb	26
26	amb	2	#define BUFFSIZE 64
27
28			static char fgets_realloc(char buffer,FILE *file);
29
30
31			/*++++++++++++++++++++++++++++++++++++++
32			Parse an OSM XML file (from JOSM or planet download).
33
34			int ParseXML Returns 0 if OK or something else in case of an error.
35
36			FILE *file The file to read from.
37	amb	26
38			NodesMem *OSMNodes The array of nodes to fill in.
39
40			SegmentsMem *OSMSegments The array of segments to fill in.
41
42			WaysMem *OSMWays The arrray of ways to fill in.
43	amb	2	++++++++++++++++++++++++++++++++++++++*/
44
45	amb	26	int ParseXML(FILE file,NodesMem OSMNodes,SegmentsMem OSMSegments,WaysMem OSMWays)
46	amb	2	{
47			char *line=NULL;
48			long nlines=0;
49			long nnodes=0,nways=0,nrelations=0;
50			int isnode=0,isway=0,isrelation=0;
51			way_t way_id=0;
52	amb	6	int way_oneway=0,way_roundabout=0;
53	amb	2	float way_maxspeed=0;
54	amb	8	char way_highway=NULL,way_name=NULL,way_ref=NULL,way_access=NULL,*way_car=NULL;
55	amb	2	node_t *way_nodes=NULL;
56			int way_nnodes=0,way_nalloc=0;
57
58			/* Parse the file */
59
60			while((line=fgets_realloc(line,file)))
61			{
62			char l=line,m;
63
64			nlines++;
65
66			while(isspace(*l))
67			l++;
68
69			if(!strncmp(l,"<node",5)) /* The start of a node */
70			{
71			node_t id;
72			latlong_t latitude,longitude;
73
74			nnodes++;
75
76			isnode=1; isway=0; isrelation=0;
77
78			m=strstr(l,"id="); m+=4; if(m=='"' \|\| m=='\'') m++; id=atoll(m);
79			m=strstr(l,"lat="); m+=5; if(m=='"' \|\| m=='\'') m++; latitude=atof(m);
80			m=strstr(l,"lon="); m+=4; if(m=='"' \|\| m=='\'') m++; longitude=atof(m);
81
82	amb	26	AppendNode(OSMNodes,id,latitude,longitude);
83	amb	2	}
84			else if(!strncmp(l,"</node",6)) /* The end of a node */
85			{
86			isnode=0; isway=0; isrelation=0;
87			}
88			else if(!strncmp(l,"<way",4)) /* The start of a way */
89			{
90			nways++;
91
92			isnode=0; isway=1; isrelation=0;
93
94			m=strstr(l,"id="); m+=3; if(m=='"' \|\| m=='\'') m++; way_id=atol(m);
95
96			way_oneway=0;
97	amb	6	way_roundabout=0;
98	amb	2	way_maxspeed=0;
99			way_name=NULL; way_ref=NULL;
100			way_nnodes=0;
101			}
102			else if(!strncmp(l,"</way",5)) /* The end of a way */
103			{
104			double speed=0;
105			int i;
106
107			isnode=0; isway=0; isrelation=0;
108
109			if(!way_highway) speed=0;
110			else if(way_access && !strcmp(way_access,"private")) speed=0;
111	amb	8	else if(way_car && !strcmp(way_car,"no")) speed=0;
112	amb	2	else if(way_maxspeed) speed=way_maxspeed;
113			else if(!strncmp(way_highway,"motorway",8)) speed=80*1.6;
114			else if(!strncmp(way_highway,"trunk",5) && way_oneway) speed=75*1.6;
115			else if(!strncmp(way_highway,"primary",7) && way_oneway) speed=70*1.6;
116			else if(!strncmp(way_highway,"trunk",5)) speed=65*1.6;
117			else if(!strncmp(way_highway,"primary",7)) speed=60*1.6;
118			else if(!strncmp(way_highway,"secondary",9)) speed=55*1.6;
119			else if(!strcmp(way_highway,"tertiary")) speed=50*1.6;
120			else if(!strcmp(way_highway,"unclassified") \|\| !strcmp(way_highway,"road") \|\| !strcmp(way_highway,"minor")) speed=40*1.6;
121			else if(!strcmp(way_highway,"residential")) speed=30*1.6;
122			else if(!strcmp(way_highway,"service")) speed=20*1.6;
123	amb	8	else if(!strcmp(way_highway,"track") \|\| !strcmp(way_highway,"byway") \|\| !strcmp(way_highway,"unsurfaced") \|\| !strcmp(way_highway,"unpaved")) speed=10*1.6;
124	amb	2	// else if(!strcmp(way_highway =~ m%(steps\|path\|walkway\|footway\|pedestrian\|bridleway\|cycleway\|living_street)%) { $speed = 5*1.6; }
125
126			if(speed)
127			{
128	amb	6	char *refname;
129	amb	2
130	amb	6	if(way_ref && way_name)
131			{
132			refname=(char*)malloc(strlen(way_ref)+strlen(way_name)+4);
133			sprintf(refname,"%s (%s)",way_name,way_ref);
134			}
135			else if(way_ref && !way_name)
136			refname=way_ref;
137			else if(!way_ref && way_name)
138			refname=way_name;
139			else if(way_roundabout)
140			{
141	amb	8	refname=(char*)malloc(strlen(way_highway)+14);
142			sprintf(refname,"%s (roundabout)",way_highway);
143	amb	6	}
144			else /* if(!way_ref && !way_name && !way_roundabout) */
145			refname=way_highway;
146
147	amb	2	for(i=1;i<way_nnodes;i++)
148			{
149			node_t from=way_nodes[i-1];
150			node_t to =way_nodes[i];
151
152	amb	26	AppendSegment(OSMSegments,from,to,way_id);
153	amb	2
154			if(!way_oneway)
155	amb	26	AppendSegment(OSMSegments,to,from,way_id);
156	amb	6	}
157	amb	2
158	amb	26	AppendWay(OSMWays,way_id,refname,speed);
159	amb	6
160			if(refname!=way_ref && refname!=way_name && refname!=way_highway)
161			free(refname);
162	amb	2	}
163
164			if(way_highway) {free(way_highway); way_highway=NULL;}
165			if(way_name) {free(way_name); way_name=NULL;}
166			if(way_ref) {free(way_ref); way_ref=NULL;}
167			if(way_access) {free(way_access); way_access=NULL;}
168	amb	8	if(way_car) {free(way_car); way_car=NULL;}
169	amb	2	}
170			else if(!strncmp(l,"<relation",9)) /* The start of a relation */
171			{
172			nrelations++;
173
174			isnode=0; isway=0; isrelation=1;
175			}
176			else if(!strncmp(l,"</relation",10)) /* The end of a relation */
177			{
178			isnode=0; isway=0; isrelation=0;
179			}
180			else if(isnode) /* The middle of a node */
181			{
182			}
183			else if(isway) /* The middle of a way */
184			{
185			node_t id;
186
187			if(!strncmp(l,"<nd",3)) /* The start of a node specifier */
188			{
189			m=strstr(l,"ref="); m+=4; if(m=='"' \|\| m=='\'') m++; id=atoll(m);
190
191			if(way_nnodes<=way_nalloc)
192			way_nodes=(node_t)realloc((void)way_nodes,(way_nalloc+=16)*sizeof(node_t));
193
194			way_nodes[way_nnodes++]=id;
195			}
196
197			if(!strncmp(l,"<tag",4)) /* The start of a tag specifier */
198			{
199			char delimiter,k="",v="";
200
201			m=strstr(l,"k="); m+=2; delimiter=*m; m++; k=m;
202			while(m!=delimiter) m++; m=0; l=m+1;
203
204			m=strstr(l,"v="); m+=2; delimiter=*m; m++; v=m;
205			while(m!=delimiter) m++; m=0;
206
207			if(!strcmp(k,"oneway") && (!strcmp(v,"true") \|\| !strcmp(v,"yes"))) way_oneway=1;
208	amb	6	if(!strcmp(k,"junction") && !strcmp(v,"roundabout")) {way_oneway=1; way_roundabout=1;}
209	amb	2	if(!strcmp(k,"highway") && !strncmp(v,"motorway",8)) way_oneway=1;
210
211			if(!strcmp(k,"highway"))
212			way_highway=strcpy((char*)malloc(strlen(v)+1),v);
213
214			if(!strcmp(k,"name"))
215			way_name=strcpy((char*)malloc(strlen(v)+1),v);
216
217			if(!strcmp(k,"ref"))
218			way_ref=strcpy((char*)malloc(strlen(v)+1),v);
219
220			if(!strcmp(k,"access"))
221			way_access=strcpy((char*)malloc(strlen(v)+1),v);
222
223	amb	8	if(!strcmp(k,"car"))
224			way_car=strcpy((char*)malloc(strlen(v)+1),v);
225
226	amb	2	if(!strcmp(k,"maxspeed"))
227			{
228			way_maxspeed=atof(v);
229			if(strstr(v,"mph"))
230			way_maxspeed*=1.6;
231			}
232			}
233			}
234			else if(isrelation) /* The middle of a relation */
235			{
236			}
237
238			if(!(nlines%10000))
239			{
240			printf("\rReading: Lines=%ld Nodes=%ld Ways=%ld Relations=%ld",nlines,nnodes,nways,nrelations);
241			fflush(stdout);
242			}
243			}
244
245			printf("\rRead : Lines=%ld Nodes=%ld Ways=%ld Relations=%ld\n",nlines,nnodes,nways,nrelations);
246			fflush(stdout);
247
248			if(way_nalloc)
249			free(way_nodes);
250
251			return(0);
252			}
253
254
255			/*++++++++++++++++++++++++++++++++++++++
256			Call fgets and realloc the buffer as needed to get a whole line.
257
258			char *fgets_realloc Returns the modified buffer (NULL at the end of the file).
259
260			char *buffer The current buffer.
261
262			FILE *file The file to read from.
263			++++++++++++++++++++++++++++++++++++++*/
264
265			static char fgets_realloc(char buffer,FILE *file)
266			{
267			int n=0;
268			char *buf;
269
270			if(!buffer)
271			buffer=(char*)malloc(BUFFSIZE+1);
272
273			while((buf=fgets(&buffer[n],BUFFSIZE,file)))
274			{
275			int s=strlen(buf);
276			n+=s;
277
278			if(buffer[n-1]=='\n')
279			break;
280			else
281			buffer=(char*)realloc(buffer,n+BUFFSIZE+1);
282			}
283
284			if(!buf)
285			{free(buffer);buffer=NULL;}
286
287			return(buffer);
288			}