trunk/src/osmparser.c

/***************************************
 $Header: /home/amb/CVS/routino/src/osmparser.c,v 1.2 2009-01-02 11:33:47 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 "types.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.
  ++++++++++++++++++++++++++++++++++++++*/

int ParseXML(FILE *file)
{
 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;
 float way_maxspeed=0;
 char *way_highway=NULL,*way_name=NULL,*way_ref=NULL,*way_access=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(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_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_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 =~ m%(track|byway|unsurfaced|unpaved)%) { $speed = 10*1.6; }
//      else if(!strcmp(way_highway =~ m%(steps|path|walkway|footway|pedestrian|bridleway|cycleway|living_street)%) { $speed = 5*1.6; }
//      else if(!strcmp(way_highway =~ m%(raceway)%) { next; }
//      else { print STDERR "\nhighway=$highway\n"; next; }

       if(speed)
         {
//       if($ref && $name)
//         {$refname="$name ($ref)";}
//       if($ref && !$name)
//         {$refname=$ref;}
//       if(!$ref && $name)
//         {$refname=$name;}
//       if(!$ref && !$name)
//         {$refname="unamed $highway road";}
//
//       print WAYS "$id\t$refname\n";

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

             distance_t distance=SegmentLength(FindNode(from),FindNode(to));
             duration_t duration=hours_to_duration(distance_to_km(distance)/speed);

             AppendSegment(from,to,way_id,distance,duration);

             if(!way_oneway)
                AppendSegment(to,from,way_id,distance,duration);

//          $junctions{$from}++;
//          $junctions{$to}++;
            }
         }

       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;}
      }
    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;
          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,"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	5	$Header: /home/amb/CVS/routino/src/osmparser.c,v 1.2 2009-01-02 11:33:47 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			#include "types.h"
21			#include "functions.h"
22
23			#define BUFFSIZE 64
24
25			static char fgets_realloc(char buffer,FILE *file);
26
27
28			/*++++++++++++++++++++++++++++++++++++++
29			Parse an OSM XML file (from JOSM or planet download).
30
31			int ParseXML Returns 0 if OK or something else in case of an error.
32
33			FILE *file The file to read from.
34			++++++++++++++++++++++++++++++++++++++*/
35
36			int ParseXML(FILE *file)
37			{
38			char *line=NULL;
39			long nlines=0;
40			long nnodes=0,nways=0,nrelations=0;
41			int isnode=0,isway=0,isrelation=0;
42			way_t way_id=0;
43			int way_oneway=0;
44			float way_maxspeed=0;
45			char way_highway=NULL,way_name=NULL,way_ref=NULL,way_access=NULL;
46			node_t *way_nodes=NULL;
47			int way_nnodes=0,way_nalloc=0;
48
49			/* Parse the file */
50
51			while((line=fgets_realloc(line,file)))
52			{
53			char l=line,m;
54
55			nlines++;
56
57			while(isspace(*l))
58			l++;
59
60			if(!strncmp(l,"<node",5)) /* The start of a node */
61			{
62			node_t id;
63			latlong_t latitude,longitude;
64
65			nnodes++;
66
67			isnode=1; isway=0; isrelation=0;
68
69			m=strstr(l,"id="); m+=4; if(m=='"' \|\| m=='\'') m++; id=atoll(m);
70			m=strstr(l,"lat="); m+=5; if(m=='"' \|\| m=='\'') m++; latitude=atof(m);
71			m=strstr(l,"lon="); m+=4; if(m=='"' \|\| m=='\'') m++; longitude=atof(m);
72
73			AppendNode(id,latitude,longitude);
74			}
75			else if(!strncmp(l,"</node",6)) /* The end of a node */
76			{
77			isnode=0; isway=0; isrelation=0;
78			}
79			else if(!strncmp(l,"<way",4)) /* The start of a way */
80			{
81			nways++;
82
83			isnode=0; isway=1; isrelation=0;
84
85			m=strstr(l,"id="); m+=3; if(m=='"' \|\| m=='\'') m++; way_id=atol(m);
86
87			way_oneway=0;
88			way_maxspeed=0;
89			way_name=NULL; way_ref=NULL;
90			way_nnodes=0;
91			}
92			else if(!strncmp(l,"</way",5)) /* The end of a way */
93			{
94			double speed=0;
95			int i;
96
97			isnode=0; isway=0; isrelation=0;
98
99			if(!way_highway) speed=0;
100			else if(way_access && !strcmp(way_access,"private")) speed=0;
101			else if(way_maxspeed) speed=way_maxspeed;
102			else if(!strncmp(way_highway,"motorway",8)) speed=80*1.6;
103			else if(!strncmp(way_highway,"trunk",5) && way_oneway) speed=75*1.6;
104			else if(!strncmp(way_highway,"primary",7) && way_oneway) speed=70*1.6;
105			else if(!strncmp(way_highway,"trunk",5)) speed=65*1.6;
106			else if(!strncmp(way_highway,"primary",7)) speed=60*1.6;
107			else if(!strncmp(way_highway,"secondary",9)) speed=55*1.6;
108			else if(!strcmp(way_highway,"tertiary")) speed=50*1.6;
109			else if(!strcmp(way_highway,"unclassified") \|\| !strcmp(way_highway,"road") \|\| !strcmp(way_highway,"minor")) speed=40*1.6;
110			else if(!strcmp(way_highway,"residential")) speed=30*1.6;
111			else if(!strcmp(way_highway,"service")) speed=20*1.6;
112			// else if(!strcmp(way_highway =~ m%(track\|byway\|unsurfaced\|unpaved)%) { $speed = 10*1.6; }
113			// else if(!strcmp(way_highway =~ m%(steps\|path\|walkway\|footway\|pedestrian\|bridleway\|cycleway\|living_street)%) { $speed = 5*1.6; }
114			// else if(!strcmp(way_highway =~ m%(raceway)%) { next; }
115			// else { print STDERR "\nhighway=$highway\n"; next; }
116
117			if(speed)
118			{
119			// if($ref && $name)
120			// {$refname="$name ($ref)";}
121			// if($ref && !$name)
122			// {$refname=$ref;}
123			// if(!$ref && $name)
124			// {$refname=$name;}
125			// if(!$ref && !$name)
126			// {$refname="unamed $highway road";}
127			//
128			// print WAYS "$id\t$refname\n";
129
130			for(i=1;i<way_nnodes;i++)
131			{
132			node_t from=way_nodes[i-1];
133			node_t to =way_nodes[i];
134
135			distance_t distance=SegmentLength(FindNode(from),FindNode(to));
136	amb	5	duration_t duration=hours_to_duration(distance_to_km(distance)/speed);
137	amb	2
138			AppendSegment(from,to,way_id,distance,duration);
139
140			if(!way_oneway)
141			AppendSegment(to,from,way_id,distance,duration);
142
143			// $junctions{$from}++;
144			// $junctions{$to}++;
145			}
146			}
147
148			if(way_highway) {free(way_highway); way_highway=NULL;}
149			if(way_name) {free(way_name); way_name=NULL;}
150			if(way_ref) {free(way_ref); way_ref=NULL;}
151			if(way_access) {free(way_access); way_access=NULL;}
152			}
153			else if(!strncmp(l,"<relation",9)) /* The start of a relation */
154			{
155			nrelations++;
156
157			isnode=0; isway=0; isrelation=1;
158			}
159			else if(!strncmp(l,"</relation",10)) /* The end of a relation */
160			{
161			isnode=0; isway=0; isrelation=0;
162			}
163			else if(isnode) /* The middle of a node */
164			{
165			}
166			else if(isway) /* The middle of a way */
167			{
168			node_t id;
169
170			if(!strncmp(l,"<nd",3)) /* The start of a node specifier */
171			{
172			m=strstr(l,"ref="); m+=4; if(m=='"' \|\| m=='\'') m++; id=atoll(m);
173
174			if(way_nnodes<=way_nalloc)
175			way_nodes=(node_t)realloc((void)way_nodes,(way_nalloc+=16)*sizeof(node_t));
176
177			way_nodes[way_nnodes++]=id;
178			}
179
180			if(!strncmp(l,"<tag",4)) /* The start of a tag specifier */
181			{
182			char delimiter,k="",v="";
183
184			m=strstr(l,"k="); m+=2; delimiter=*m; m++; k=m;
185			while(m!=delimiter) m++; m=0; l=m+1;
186
187			m=strstr(l,"v="); m+=2; delimiter=*m; m++; v=m;
188			while(m!=delimiter) m++; m=0;
189
190			if(!strcmp(k,"oneway") && (!strcmp(v,"true") \|\| !strcmp(v,"yes"))) way_oneway=1;
191			if(!strcmp(k,"junction") && !strcmp(v,"roundabout")) way_oneway=1;
192			if(!strcmp(k,"highway") && !strncmp(v,"motorway",8)) way_oneway=1;
193
194			if(!strcmp(k,"highway"))
195			way_highway=strcpy((char*)malloc(strlen(v)+1),v);
196
197			if(!strcmp(k,"name"))
198			way_name=strcpy((char*)malloc(strlen(v)+1),v);
199
200			if(!strcmp(k,"ref"))
201			way_ref=strcpy((char*)malloc(strlen(v)+1),v);
202
203			if(!strcmp(k,"access"))
204			way_access=strcpy((char*)malloc(strlen(v)+1),v);
205
206			if(!strcmp(k,"maxspeed"))
207			{
208			way_maxspeed=atof(v);
209			if(strstr(v,"mph"))
210			way_maxspeed*=1.6;
211			}
212			}
213			}
214			else if(isrelation) /* The middle of a relation */
215			{
216			}
217
218			if(!(nlines%10000))
219			{
220			printf("\rReading: Lines=%ld Nodes=%ld Ways=%ld Relations=%ld",nlines,nnodes,nways,nrelations);
221			fflush(stdout);
222			}
223			}
224
225			printf("\rRead : Lines=%ld Nodes=%ld Ways=%ld Relations=%ld\n",nlines,nnodes,nways,nrelations);
226			fflush(stdout);
227
228			if(way_nalloc)
229			free(way_nodes);
230
231			return(0);
232			}
233
234
235			/*++++++++++++++++++++++++++++++++++++++
236			Call fgets and realloc the buffer as needed to get a whole line.
237
238			char *fgets_realloc Returns the modified buffer (NULL at the end of the file).
239
240			char *buffer The current buffer.
241
242			FILE *file The file to read from.
243			++++++++++++++++++++++++++++++++++++++*/
244
245			static char fgets_realloc(char buffer,FILE *file)
246			{
247			int n=0;
248			char *buf;
249
250			if(!buffer)
251			buffer=(char*)malloc(BUFFSIZE+1);
252
253			while((buf=fgets(&buffer[n],BUFFSIZE,file)))
254			{
255			int s=strlen(buf);
256			n+=s;
257
258			if(buffer[n-1]=='\n')
259			break;
260			else
261			buffer=(char*)realloc(buffer,n+BUFFSIZE+1);
262			}
263
264			if(!buf)
265			{free(buffer);buffer=NULL;}
266
267			return(buffer);
268			}