trunk/src/planetsplitter.c

/***************************************
 OSM planet file splitter.

 Part of the Routino routing software.
 ******************/ /******************
 This file Copyright 2008-2013 Andrew M. Bishop

 This program is free software: you can redistribute it and/or modify
 it under the terms of the GNU Affero General Public License as published by
 the Free Software Foundation, either version 3 of the License, or
 (at your option) any later version.

 This program is distributed in the hope that it will be useful,
 but WITHOUT ANY WARRANTY; without even the implied warranty of
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 GNU Affero General Public License for more details.

 You should have received a copy of the GNU Affero General Public License
 along with this program.  If not, see <http://www.gnu.org/licenses/>.
 ***************************************/


#include <stdio.h>
#include <stdlib.h>
#include <unistd.h>
#include <string.h>
#include <errno.h>
#include <sys/time.h>

#include "types.h"
#include "ways.h"

#include "typesx.h"
#include "nodesx.h"
#include "segmentsx.h"
#include "waysx.h"
#include "relationsx.h"
#include "superx.h"
#include "prunex.h"

#include "files.h"
#include "logging.h"
#include "errorlogx.h"
#include "functions.h"
#include "osmparser.h"
#include "tagging.h"
#include "uncompress.h"


/* Global variables */

/*+ The name of the temporary directory. +*/
char *option_tmpdirname=NULL;

/*+ The amount of RAM to use for filesorting. +*/
size_t option_filesort_ramsize=0;

/*+ The number of threads to use for filesorting. +*/
int option_filesort_threads=1;


/* Local functions */

static void print_usage(int detail,const char *argerr,const char *err);


/*++++++++++++++++++++++++++++++++++++++
  The main program for the planetsplitter.
  ++++++++++++++++++++++++++++++++++++++*/

int main(int argc,char** argv)
{
 struct timeval start_time;
 NodesX     *OSMNodes;
 SegmentsX  *OSMSegments,*SuperSegments=NULL,*MergedSegments=NULL;
 WaysX      *OSMWays;
 RelationsX *OSMRelations;
 int         iteration=0,quit=0;
 int         max_iterations=5;
 char       *dirname=NULL,*prefix=NULL,*tagging=NULL,*errorlog=NULL;
 int         option_parse_only=0,option_process_only=0;
 int         option_append=0,option_keep=0,option_changes=0;
 int         option_filenames=0;
 int         option_prune_isolated=500,option_prune_short=5,option_prune_straight=3;
 int         arg;

 gettimeofday(&start_time,NULL);

 /* Parse the command line arguments */

 for(arg=1;arg<argc;arg++)
   {
    if(!strcmp(argv[arg],"--help"))
       print_usage(1,NULL,NULL);
    else if(!strncmp(argv[arg],"--dir=",6))
       dirname=&argv[arg][6];
    else if(!strncmp(argv[arg],"--prefix=",9))
       prefix=&argv[arg][9];
    else if(!strncmp(argv[arg],"--sort-ram-size=",16))
       option_filesort_ramsize=atoi(&argv[arg][16]);
#if defined(USE_PTHREADS) && USE_PTHREADS
    else if(!strncmp(argv[arg],"--sort-threads=",15))
       option_filesort_threads=atoi(&argv[arg][15]);
#endif
    else if(!strncmp(argv[arg],"--tmpdir=",9))
       option_tmpdirname=&argv[arg][9];
    else if(!strncmp(argv[arg],"--tagging=",10))
       tagging=&argv[arg][10];
    else if(!strcmp(argv[arg],"--loggable"))
       option_loggable=1;
    else if(!strcmp(argv[arg],"--logtime"))
       option_logtime=1;
    else if(!strcmp(argv[arg],"--errorlog"))
       errorlog="error.log";
    else if(!strncmp(argv[arg],"--errorlog=",11))
       errorlog=&argv[arg][11];
    else if(!strcmp(argv[arg],"--parse-only"))
       option_parse_only=1;
    else if(!strcmp(argv[arg],"--process-only"))
       option_process_only=1;
    else if(!strcmp(argv[arg],"--append"))
       option_append=1;
    else if(!strcmp(argv[arg],"--keep"))
       option_keep=1;
    else if(!strcmp(argv[arg],"--changes"))
       option_changes=1;
    else if(!strncmp(argv[arg],"--max-iterations=",17))
       max_iterations=atoi(&argv[arg][17]);
    else if(!strncmp(argv[arg],"--prune",7))
      {
       if(!strcmp(&argv[arg][7],"-none"))
          option_prune_isolated=option_prune_short=option_prune_straight=0;
       else if(!strncmp(&argv[arg][7],"-isolated=",10))
          option_prune_isolated=atoi(&argv[arg][17]);
       else if(!strncmp(&argv[arg][7],"-short=",7))
          option_prune_short=atoi(&argv[arg][14]);
       else if(!strncmp(&argv[arg][7],"-straight=",10))
          option_prune_straight=atoi(&argv[arg][17]);
       else
          print_usage(0,argv[arg],NULL);
      }
    else if(argv[arg][0]=='-' && argv[arg][1]=='-')
       print_usage(0,argv[arg],NULL);
    else
       option_filenames++;
   }

 /* Check the specified command line options */

 if(option_parse_only && option_process_only)
    print_usage(0,NULL,"Cannot use '--parse-only' and '--process-only' at the same time.");

 if(option_append && option_process_only)
    print_usage(0,NULL,"Cannot use '--append' and '--process-only' at the same time.");

 if(option_filenames && option_process_only)
    print_usage(0,NULL,"Cannot use '--process-only' and filenames at the same time.");

 if(!option_filenames && !option_process_only)
    print_usage(0,NULL,"File names must be specified unless using '--process-only'");

 if(!option_filesort_ramsize)
   {
#if SLIM
    option_filesort_ramsize=64*1024*1024;
#else
    option_filesort_ramsize=256*1024*1024;
#endif
   }
 else
    option_filesort_ramsize*=1024*1024;

 if(!option_tmpdirname)
   {
    if(!dirname)
       option_tmpdirname=".";
    else
       option_tmpdirname=dirname;
   }

 if(!option_process_only)
   {
    if(tagging)
      {
       if(!ExistsFile(tagging))
         {
          fprintf(stderr,"Error: The '--tagging' option specifies a file that does not exist.\n");
          return(1);
         }
      }
    else
      {
       if(ExistsFile(FileName(dirname,prefix,"tagging.xml")))
          tagging=FileName(dirname,prefix,"tagging.xml");
       else if(ExistsFile(FileName(DATADIR,NULL,"tagging.xml")))
          tagging=FileName(DATADIR,NULL,"tagging.xml");
       else
         {
          fprintf(stderr,"Error: The '--tagging' option was not used and the default 'tagging.xml' does not exist.\n");
          return(1);
         }
      }

    if(ParseXMLTaggingRules(tagging))
      {
       fprintf(stderr,"Error: Cannot read the tagging rules in the file '%s'.\n",tagging);
       return(1);
      }
   }

 /* Create new node, segment, way and relation variables */

 OSMNodes=NewNodeList(option_append||option_changes,option_process_only);

 OSMWays=NewWayList(option_append||option_changes,option_process_only);

 OSMRelations=NewRelationList(option_append||option_changes,option_process_only);

 /* Create the error log file */

 if(errorlog)
    open_errorlog(FileName(dirname,prefix,errorlog),option_append||option_changes||option_process_only,option_keep);

 /* Parse the file */

if(!option_process_only)
  {
   for(arg=1;arg<argc;arg++)
     {
      int fd;
      char *filename,*p;

      if(argv[arg][0]=='-' && argv[arg][1]=='-')
         continue;

      filename=strcpy(malloc(strlen(argv[arg])+1),argv[arg]);

      fd=ReOpenFile(filename);

      if((p=strstr(filename,".bz2")) && !strcmp(p,".bz2"))
        {
         fd=Uncompress_Bzip2(fd);
         *p=0;
        }

      if((p=strstr(filename,".gz")) && !strcmp(p,".gz"))
        {
         fd=Uncompress_Gzip(fd);
         *p=0;
        }

      if(option_changes)
        {
         printf("\nParse OSC Data [%s]\n==============\n\n",filename);
         fflush(stdout);

         if((p=strstr(filename,".pbf")) && !strcmp(p,".pbf"))
           {
            logassert(0,"Unable to read a PBF file to apply changes (format does not permit this)");
           }
         else if((p=strstr(filename,".o5c")) && !strcmp(p,".o5c"))
           {
            if(ParseO5CFile(fd,OSMNodes,OSMWays,OSMRelations))
               exit(EXIT_FAILURE);
           }
         else
           {
            if(ParseOSCFile(fd,OSMNodes,OSMWays,OSMRelations))
               exit(EXIT_FAILURE);
           }
        }
      else
        {
         printf("\nParse OSM Data [%s]\n==============\n\n",filename);
         fflush(stdout);

         if((p=strstr(filename,".pbf")) && !strcmp(p,".pbf"))
           {
            if(ParsePBFFile(fd,OSMNodes,OSMWays,OSMRelations))
               exit(EXIT_FAILURE);
           }
         else if((p=strstr(filename,".o5m")) && !strcmp(p,".o5m"))
           {
            if(ParseO5MFile(fd,OSMNodes,OSMWays,OSMRelations))
               exit(EXIT_FAILURE);
           }
         else
           {
            if(ParseOSMFile(fd,OSMNodes,OSMWays,OSMRelations))
               exit(EXIT_FAILURE);
           }
        }

      CloseFile(fd);

      free(filename);
     }

   DeleteXMLTaggingRules();
  }

 FinishNodeList(OSMNodes);
 FinishWayList(OSMWays);
 FinishRelationList(OSMRelations);

 if(option_parse_only)
   {
    FreeNodeList(OSMNodes,1);
    FreeWayList(OSMWays,1);
    FreeRelationList(OSMRelations,1);

    return(0);
   }


 /* Sort the data */

 printf("\nSort OSM Data\n=============\n\n");
 fflush(stdout);

 /* Sort the nodes, segments, ways and relations */

 SortNodeList(OSMNodes);

 SortWayList(OSMWays);

 SortRelationList(OSMRelations);

 /* Process the data */

 printf("\nProcess OSM Data\n================\n\n");
 fflush(stdout);

 /* Separate the segments and way names and sort them. */

 OSMSegments=SplitWays(OSMWays,option_keep||option_changes);

 SortWayNames(OSMWays);

 SortSegmentList(OSMSegments);

 /* Remove bad segments (must be after sorting the nodes, segments and ways) */

 RemoveBadSegments(OSMSegments,OSMNodes,OSMWays);

 /* Remove non-highway nodes (must be after removing the bad segments) */

 RemoveNonHighwayNodes(OSMNodes,OSMSegments,option_keep||option_changes);

 /* Process the route relations and first part of turn relations (must be before compacting the ways) */

 ProcessRouteRelations(OSMRelations,OSMWays,option_keep||option_changes);

 ProcessTurnRelations1(OSMRelations,OSMNodes,OSMWays,option_keep||option_changes);

 /* Measure the segments and replace node/way id with index (must be after removing non-highway nodes) */

 MeasureSegments(OSMSegments,OSMNodes,OSMWays);

 /* Index the segments */

 IndexSegments(OSMSegments,OSMNodes,OSMWays);

 /* Convert the turn relations from ways into nodes */

 ProcessTurnRelations2(OSMRelations,OSMNodes,OSMSegments,OSMWays);

 /* Compact the ways (must be after turn relations 2) */

 CompactWayList(OSMWays,OSMSegments);

 /* Index the segments */

 IndexSegments(OSMSegments,OSMNodes,OSMWays);

 /* Prune unwanted nodes/segments. */

 if(option_prune_straight || option_prune_isolated || option_prune_short)
   {
    printf("\nPrune Unneeded Data\n===================\n\n");
    fflush(stdout);

    StartPruning(OSMNodes,OSMSegments,OSMWays);

    if(option_prune_isolated)
       PruneIsolatedRegions(OSMNodes,OSMSegments,OSMWays,option_prune_isolated);

    if(option_prune_short)
       PruneShortSegments(OSMNodes,OSMSegments,OSMWays,option_prune_short);

    if(option_prune_straight)
       PruneStraightHighwayNodes(OSMNodes,OSMSegments,OSMWays,option_prune_straight);

    FinishPruning(OSMNodes,OSMSegments,OSMWays);

    /* Remove the pruned nodes and segments and update the indexes */

    RemovePrunedNodes(OSMNodes,OSMSegments);
    RemovePrunedSegments(OSMSegments,OSMWays);
    CompactWayList(OSMWays,OSMSegments);
    RemovePrunedTurnRelations(OSMRelations,OSMNodes);
    IndexSegments(OSMSegments,OSMNodes,OSMWays);
   }

 /* Repeated iteration on Super-Nodes and Super-Segments */

 do
   {
    index_t nsuper;

    printf("\nProcess Super-Data (iteration %d)\n================================%s\n\n",iteration,iteration>9?"=":"");
    fflush(stdout);

    if(iteration==0)
      {
       /* Select the super-nodes */

       ChooseSuperNodes(OSMNodes,OSMSegments,OSMWays);

       /* Select the super-segments */

       SuperSegments=CreateSuperSegments(OSMNodes,OSMSegments,OSMWays);

       nsuper=OSMSegments->number;
      }
    else
      {
       SegmentsX *SuperSegments2;

       /* Select the super-nodes */

       ChooseSuperNodes(OSMNodes,SuperSegments,OSMWays);

       /* Select the super-segments */

       SuperSegments2=CreateSuperSegments(OSMNodes,SuperSegments,OSMWays);

       nsuper=SuperSegments->number;

       FreeSegmentList(SuperSegments);

       SuperSegments=SuperSegments2;
      }

    /* Sort the super-segments and remove duplicates */

    DeduplicateSuperSegments(SuperSegments,OSMWays);

    /* Index the segments */

    IndexSegments(SuperSegments,OSMNodes,OSMWays);

    /* Check for end condition */

    if(SuperSegments->number==nsuper)
       quit=1;

    iteration++;

    if(iteration>max_iterations)
       quit=1;
   }
 while(!quit);

 /* Combine the super-segments */

 printf("\nCombine Segments and Super-Segments\n===================================\n\n");
 fflush(stdout);

 /* Merge the super-segments */

 MergedSegments=MergeSuperSegments(OSMSegments,SuperSegments);

 FreeSegmentList(OSMSegments);

 FreeSegmentList(SuperSegments);

 OSMSegments=MergedSegments;

 /* Re-index the merged segments */

 IndexSegments(OSMSegments,OSMNodes,OSMWays);

 /* Cross reference the nodes and segments */

 printf("\nCross-Reference Nodes and Segments\n==================================\n\n");
 fflush(stdout);

 /* Sort the nodes and segments geographically */

 SortNodeListGeographically(OSMNodes);

 SortSegmentListGeographically(OSMSegments,OSMNodes);

 /* Re-index the segments */

 IndexSegments(OSMSegments,OSMNodes,OSMWays);

 /* Sort the turn relations geographically */

 SortTurnRelationListGeographically(OSMRelations,OSMNodes,OSMSegments);

 /* Output the results */

 printf("\nWrite Out Database Files\n========================\n\n");
 fflush(stdout);

 /* Write out the nodes */

 SaveNodeList(OSMNodes,FileName(dirname,prefix,"nodes.mem"),OSMSegments);

 /* Write out the segments */

 SaveSegmentList(OSMSegments,FileName(dirname,prefix,"segments.mem"));

 /* Write out the ways */

 SaveWayList(OSMWays,FileName(dirname,prefix,"ways.mem"));

 /* Write out the relations */

 SaveRelationList(OSMRelations,FileName(dirname,prefix,"relations.mem"));

 /* Close the error log file and process the data */

 if(errorlog)
   {
    close_errorlog();

    if(option_keep)
      {
       printf("\nCreate Error Log\n================\n\n");
       fflush(stdout);

       ProcessErrorLogs(OSMNodes,OSMWays,OSMRelations);

       SortErrorLogsGeographically();

       SaveErrorLogs(OSMNodes,FileName(dirname,prefix,"errorlogs.mem"));
      }
   }

 /* Free the memory (delete the temporary files) */

 FreeNodeList(OSMNodes,0);
 FreeWayList(OSMWays,0);
 FreeRelationList(OSMRelations,0);

 FreeSegmentList(OSMSegments);

 /* Print the total time */

 if(option_logtime)
   {
    printf("\n");
    fprintf_elapsed_time(stdout,&start_time);
    printf("Complete\n");
    fflush(stdout);
   }

 return(0);
}


/*++++++++++++++++++++++++++++++++++++++
  Print out the usage information.

  int detail The level of detail to use - 0 = low, 1 = high.

  const char *argerr The argument that gave the error (if there is one).

  const char *err Other error message (if there is one).
  ++++++++++++++++++++++++++++++++++++++*/

static void print_usage(int detail,const char *argerr,const char *err)
{
 fprintf(stderr,
         "Usage: planetsplitter [--help]\n"
         "                      [--dir=<dirname>] [--prefix=<name>]\n"
#if defined(USE_PTHREADS) && USE_PTHREADS
         "                      [--sort-ram-size=<size>] [--sort-threads=<number>]\n"
#else
         "                      [--sort-ram-size=<size>]\n"
#endif
         "                      [--tmpdir=<dirname>]\n"
         "                      [--tagging=<filename>]\n"
         "                      [--loggable] [--logtime]\n"
         "                      [--errorlog[=<name>]]\n"
         "                      [--parse-only | --process-only]\n"
         "                      [--append] [--keep] [--changes]\n"
         "                      [--max-iterations=<number>]\n"
         "                      [--prune-none]\n"
         "                      [--prune-isolated=<len>]\n"
         "                      [--prune-short=<len>]\n"
         "                      [--prune-straight=<len>]\n"
         "                      [<filename.osm> ... | <filename.osc> ...\n"
         "                       | <filename.pbf> ...\n"
         "                       | <filename.osm> ... | <filename.osc> ..."
#if defined(USE_BZIP2) && USE_BZIP2
         "\n                       | <filename.(osm|osc|o5m|o5c).bz2> ..."
#endif
#if defined(USE_GZIP) && USE_GZIP
         "\n                       | <filename.(osm|osc|o5m|o5c).gz> ..."
#endif
         "]\n");

 if(argerr)
    fprintf(stderr,
            "\n"
            "Error with command line parameter: %s\n",argerr);

 if(err)
    fprintf(stderr,
            "\n"
            "Error: %s\n",err);

 if(detail)
    fprintf(stderr,
            "\n"
            "--help                    Prints this information.\n"
            "\n"
            "--dir=<dirname>           The directory containing the routing database.\n"
            "--prefix=<name>           The filename prefix for the routing database.\n"
            "\n"
            "--sort-ram-size=<size>    The amount of RAM (in MB) to use for data sorting\n"
#if SLIM
            "                          (defaults to 64MB otherwise.)\n"
#else
            "                          (defaults to 256MB otherwise.)\n"
#endif
#if defined(USE_PTHREADS) && USE_PTHREADS
            "--sort-threads=<number>   The number of threads to use for data sorting.\n"
#endif
            "\n"
            "--tmpdir=<dirname>        The directory name for temporary files.\n"
            "                          (defaults to the '--dir' option directory.)\n"
            "\n"
            "--tagging=<filename>      The name of the XML file containing the tagging rules\n"
            "                          (defaults to 'tagging.xml' with '--dir' and\n"
            "                           '--prefix' options or the file installed in\n"
            "                           '" DATADIR "').\n"
            "\n"
            "--loggable                Print progress messages suitable for logging to file.\n"
            "--logtime                 Print the elapsed time for each processing step.\n"
            "--errorlog[=<name>]       Log parsing errors to 'error.log' or the given name\n"
            "                          (the '--dir' and '--prefix' options are applied).\n"
            "\n"
            "--parse-only              Parse the OSM/OSC file(s) and store the results.\n"
            "--process-only            Process the stored results from previous option.\n"
            "--append                  Parse the OSM file(s) and append to existing results.\n"
            "--keep                    Keep the intermediate files after parsing & sorting.\n"
            "--changes                 Parse the data as an OSC file and apply the changes.\n"
            "\n"
            "--max-iterations=<number> The number of iterations for finding super-nodes\n"
            "                          (defaults to 5).\n"
            "\n"
            "--prune-none              Disable the prune options below, they are re-enabled\n"
            "                          by adding them to the command line after this option.\n"
            "--prune-isolated=<len>    Remove access from small disconnected segment groups\n"
            "                          (defaults to removing groups under 500m).\n"
            "--prune-short=<len>       Remove short segments (defaults to removing segments\n"
            "                          up to a maximum length of 5m).\n"
            "--prune-straight=<len>    Remove nodes in almost straight highways (defaults to\n"
            "                          removing nodes up to 3m offset from a straight line).\n"
            "\n"
            "<filename.osm>, <filename.osc>, <filename.pbf>, <filename.o5m>, <filename.o5c>\n"
            "                          The name(s) of the file(s) to read and parse.\n"
            "                          Filenames ending '.pbf' read as PBF, filenames ending\n"
            "                          '.o5m' or '.o5c' read as O5M/O5C, others as XML.\n"
#if defined(USE_BZIP2) && USE_BZIP2
            "                          Filenames ending '.bz2' will be bzip2 uncompressed.\n"
#endif
#if defined(USE_GZIP) && USE_GZIP
            "                          Filenames ending '.gz' will be gzip uncompressed.\n"
#endif
            "\n"
            "<transport> defaults to all but can be set to:\n"
            "%s"
            "\n"
            "<highway> can be selected from:\n"
            "%s"
            "\n"
            "<property> can be selected from:\n"
            "%s",
            TransportList(),HighwayList(),PropertyList());

 exit(!detail);
}
1	/***************************************
2	OSM planet file splitter.
3
4	Part of the Routino routing software.
5	****************/ /****************
6	This file Copyright 2008-2013 Andrew M. Bishop
7
8	This program is free software: you can redistribute it and/or modify
9	it under the terms of the GNU Affero General Public License as published by
10	the Free Software Foundation, either version 3 of the License, or
11	(at your option) any later version.
12
13	This program is distributed in the hope that it will be useful,
14	but WITHOUT ANY WARRANTY; without even the implied warranty of
15	MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16	GNU Affero General Public License for more details.
17
18	You should have received a copy of the GNU Affero General Public License
19	along with this program. If not, see <http://www.gnu.org/licenses/>.
20	***************************************/
21
22
23	#include <stdio.h>
24	#include <stdlib.h>
25	#include <unistd.h>
26	#include <string.h>
27	#include <errno.h>
28	#include <sys/time.h>
29
30	#include "types.h"
31	#include "ways.h"
32
33	#include "typesx.h"
34	#include "nodesx.h"
35	#include "segmentsx.h"
36	#include "waysx.h"
37	#include "relationsx.h"
38	#include "superx.h"
39	#include "prunex.h"
40
41	#include "files.h"
42	#include "logging.h"
43	#include "errorlogx.h"
44	#include "functions.h"
45	#include "osmparser.h"
46	#include "tagging.h"
47	#include "uncompress.h"
48
49
50	/* Global variables */
51
52	/+ The name of the temporary directory. +/
53	char *option_tmpdirname=NULL;
54
55	/+ The amount of RAM to use for filesorting. +/
56	size_t option_filesort_ramsize=0;
57
58	/+ The number of threads to use for filesorting. +/
59	int option_filesort_threads=1;
60
61
62	/* Local functions */
63
64	static void print_usage(int detail,const char argerr,const char err);
65
66
67	/*++++++++++++++++++++++++++++++++++++++
68	The main program for the planetsplitter.
69	++++++++++++++++++++++++++++++++++++++*/
70
71	int main(int argc,char** argv)
72	{
73	struct timeval start_time;
74	NodesX *OSMNodes;
75	SegmentsX OSMSegments,SuperSegments=NULL,*MergedSegments=NULL;
76	WaysX *OSMWays;
77	RelationsX *OSMRelations;
78	int iteration=0,quit=0;
79	int max_iterations=5;
80	char dirname=NULL,prefix=NULL,tagging=NULL,errorlog=NULL;
81	int option_parse_only=0,option_process_only=0;
82	int option_append=0,option_keep=0,option_changes=0;
83	int option_filenames=0;
84	int option_prune_isolated=500,option_prune_short=5,option_prune_straight=3;
85	int arg;
86
87	gettimeofday(&start_time,NULL);
88
89	/* Parse the command line arguments */
90
91	for(arg=1;arg<argc;arg++)
92	{
93	if(!strcmp(argv[arg],"--help"))
94	print_usage(1,NULL,NULL);
95	else if(!strncmp(argv[arg],"--dir=",6))
96	dirname=&argv[arg][6];
97	else if(!strncmp(argv[arg],"--prefix=",9))
98	prefix=&argv[arg][9];
99	else if(!strncmp(argv[arg],"--sort-ram-size=",16))
100	option_filesort_ramsize=atoi(&argv[arg][16]);
101	#if defined(USE_PTHREADS) && USE_PTHREADS
102	else if(!strncmp(argv[arg],"--sort-threads=",15))
103	option_filesort_threads=atoi(&argv[arg][15]);
104	#endif
105	else if(!strncmp(argv[arg],"--tmpdir=",9))
106	option_tmpdirname=&argv[arg][9];
107	else if(!strncmp(argv[arg],"--tagging=",10))
108	tagging=&argv[arg][10];
109	else if(!strcmp(argv[arg],"--loggable"))
110	option_loggable=1;
111	else if(!strcmp(argv[arg],"--logtime"))
112	option_logtime=1;
113	else if(!strcmp(argv[arg],"--errorlog"))
114	errorlog="error.log";
115	else if(!strncmp(argv[arg],"--errorlog=",11))
116	errorlog=&argv[arg][11];
117	else if(!strcmp(argv[arg],"--parse-only"))
118	option_parse_only=1;
119	else if(!strcmp(argv[arg],"--process-only"))
120	option_process_only=1;
121	else if(!strcmp(argv[arg],"--append"))
122	option_append=1;
123	else if(!strcmp(argv[arg],"--keep"))
124	option_keep=1;
125	else if(!strcmp(argv[arg],"--changes"))
126	option_changes=1;
127	else if(!strncmp(argv[arg],"--max-iterations=",17))
128	max_iterations=atoi(&argv[arg][17]);
129	else if(!strncmp(argv[arg],"--prune",7))
130	{
131	if(!strcmp(&argv[arg][7],"-none"))
132	option_prune_isolated=option_prune_short=option_prune_straight=0;
133	else if(!strncmp(&argv[arg][7],"-isolated=",10))
134	option_prune_isolated=atoi(&argv[arg][17]);
135	else if(!strncmp(&argv[arg][7],"-short=",7))
136	option_prune_short=atoi(&argv[arg][14]);
137	else if(!strncmp(&argv[arg][7],"-straight=",10))
138	option_prune_straight=atoi(&argv[arg][17]);
139	else
140	print_usage(0,argv[arg],NULL);
141	}
142	else if(argv[arg][0]=='-' && argv[arg][1]=='-')
143	print_usage(0,argv[arg],NULL);
144	else
145	option_filenames++;
146	}
147
148	/* Check the specified command line options */
149
150	if(option_parse_only && option_process_only)
151	print_usage(0,NULL,"Cannot use '--parse-only' and '--process-only' at the same time.");
152
153	if(option_append && option_process_only)
154	print_usage(0,NULL,"Cannot use '--append' and '--process-only' at the same time.");
155
156	if(option_filenames && option_process_only)
157	print_usage(0,NULL,"Cannot use '--process-only' and filenames at the same time.");
158
159	if(!option_filenames && !option_process_only)
160	print_usage(0,NULL,"File names must be specified unless using '--process-only'");
161
162	if(!option_filesort_ramsize)
163	{
164	#if SLIM
165	option_filesort_ramsize=6410241024;
166	#else
167	option_filesort_ramsize=25610241024;
168	#endif
169	}
170	else
171	option_filesort_ramsize=10241024;
172
173	if(!option_tmpdirname)
174	{
175	if(!dirname)
176	option_tmpdirname=".";
177	else
178	option_tmpdirname=dirname;
179	}
180
181	if(!option_process_only)
182	{
183	if(tagging)
184	{
185	if(!ExistsFile(tagging))
186	{
187	fprintf(stderr,"Error: The '--tagging' option specifies a file that does not exist.\n");
188	return(1);
189	}
190	}
191	else
192	{
193	if(ExistsFile(FileName(dirname,prefix,"tagging.xml")))
194	tagging=FileName(dirname,prefix,"tagging.xml");
195	else if(ExistsFile(FileName(DATADIR,NULL,"tagging.xml")))
196	tagging=FileName(DATADIR,NULL,"tagging.xml");
197	else
198	{
199	fprintf(stderr,"Error: The '--tagging' option was not used and the default 'tagging.xml' does not exist.\n");
200	return(1);
201	}
202	}
203
204	if(ParseXMLTaggingRules(tagging))
205	{
206	fprintf(stderr,"Error: Cannot read the tagging rules in the file '%s'.\n",tagging);
207	return(1);
208	}
209	}
210
211	/* Create new node, segment, way and relation variables */
212
213	OSMNodes=NewNodeList(option_append\|\|option_changes,option_process_only);
214
215	OSMWays=NewWayList(option_append\|\|option_changes,option_process_only);
216
217	OSMRelations=NewRelationList(option_append\|\|option_changes,option_process_only);
218
219	/* Create the error log file */
220
221	if(errorlog)
222	open_errorlog(FileName(dirname,prefix,errorlog),option_append\|\|option_changes\|\|option_process_only,option_keep);
223
224	/* Parse the file */
225
226	if(!option_process_only)
227	{
228	for(arg=1;arg<argc;arg++)
229	{
230	int fd;
231	char filename,p;
232
233	if(argv[arg][0]=='-' && argv[arg][1]=='-')
234	continue;
235
236	filename=strcpy(malloc(strlen(argv[arg])+1),argv[arg]);
237
238	fd=ReOpenFile(filename);
239
240	if((p=strstr(filename,".bz2")) && !strcmp(p,".bz2"))
241	{
242	fd=Uncompress_Bzip2(fd);
243	*p=0;
244	}
245
246	if((p=strstr(filename,".gz")) && !strcmp(p,".gz"))
247	{
248	fd=Uncompress_Gzip(fd);
249	*p=0;
250	}
251
252	if(option_changes)
253	{
254	printf("\nParse OSC Data [%s]\n==============\n\n",filename);
255	fflush(stdout);
256
257	if((p=strstr(filename,".pbf")) && !strcmp(p,".pbf"))
258	{
259	logassert(0,"Unable to read a PBF file to apply changes (format does not permit this)");
260	}
261	else if((p=strstr(filename,".o5c")) && !strcmp(p,".o5c"))
262	{
263	if(ParseO5CFile(fd,OSMNodes,OSMWays,OSMRelations))
264	exit(EXIT_FAILURE);
265	}
266	else
267	{
268	if(ParseOSCFile(fd,OSMNodes,OSMWays,OSMRelations))
269	exit(EXIT_FAILURE);
270	}
271	}
272	else
273	{
274	printf("\nParse OSM Data [%s]\n==============\n\n",filename);
275	fflush(stdout);
276
277	if((p=strstr(filename,".pbf")) && !strcmp(p,".pbf"))
278	{
279	if(ParsePBFFile(fd,OSMNodes,OSMWays,OSMRelations))
280	exit(EXIT_FAILURE);
281	}
282	else if((p=strstr(filename,".o5m")) && !strcmp(p,".o5m"))
283	{
284	if(ParseO5MFile(fd,OSMNodes,OSMWays,OSMRelations))
285	exit(EXIT_FAILURE);
286	}
287	else
288	{
289	if(ParseOSMFile(fd,OSMNodes,OSMWays,OSMRelations))
290	exit(EXIT_FAILURE);
291	}
292	}
293
294	CloseFile(fd);
295
296	free(filename);
297	}
298
299	DeleteXMLTaggingRules();
300	}
301
302	FinishNodeList(OSMNodes);
303	FinishWayList(OSMWays);
304	FinishRelationList(OSMRelations);
305
306	if(option_parse_only)
307	{
308	FreeNodeList(OSMNodes,1);
309	FreeWayList(OSMWays,1);
310	FreeRelationList(OSMRelations,1);
311
312	return(0);
313	}
314
315
316	/* Sort the data */
317
318	printf("\nSort OSM Data\n=============\n\n");
319	fflush(stdout);
320
321	/* Sort the nodes, segments, ways and relations */
322
323	SortNodeList(OSMNodes);
324
325	SortWayList(OSMWays);
326
327	SortRelationList(OSMRelations);
328
329	/* Process the data */
330
331	printf("\nProcess OSM Data\n================\n\n");
332	fflush(stdout);
333
334	/* Separate the segments and way names and sort them. */
335
336	OSMSegments=SplitWays(OSMWays,option_keep\|\|option_changes);
337
338	SortWayNames(OSMWays);
339
340	SortSegmentList(OSMSegments);
341
342	/* Remove bad segments (must be after sorting the nodes, segments and ways) */
343
344	RemoveBadSegments(OSMSegments,OSMNodes,OSMWays);
345
346	/* Remove non-highway nodes (must be after removing the bad segments) */
347
348	RemoveNonHighwayNodes(OSMNodes,OSMSegments,option_keep\|\|option_changes);
349
350	/* Process the route relations and first part of turn relations (must be before compacting the ways) */
351
352	ProcessRouteRelations(OSMRelations,OSMWays,option_keep\|\|option_changes);
353
354	ProcessTurnRelations1(OSMRelations,OSMNodes,OSMWays,option_keep\|\|option_changes);
355
356	/* Measure the segments and replace node/way id with index (must be after removing non-highway nodes) */
357
358	MeasureSegments(OSMSegments,OSMNodes,OSMWays);
359
360	/* Index the segments */
361
362	IndexSegments(OSMSegments,OSMNodes,OSMWays);
363
364	/* Convert the turn relations from ways into nodes */
365
366	ProcessTurnRelations2(OSMRelations,OSMNodes,OSMSegments,OSMWays);
367
368	/* Compact the ways (must be after turn relations 2) */
369
370	CompactWayList(OSMWays,OSMSegments);
371
372	/* Index the segments */
373
374	IndexSegments(OSMSegments,OSMNodes,OSMWays);
375
376	/* Prune unwanted nodes/segments. */
377
378	if(option_prune_straight \|\| option_prune_isolated \|\| option_prune_short)
379	{
380	printf("\nPrune Unneeded Data\n===================\n\n");
381	fflush(stdout);
382
383	StartPruning(OSMNodes,OSMSegments,OSMWays);
384
385	if(option_prune_isolated)
386	PruneIsolatedRegions(OSMNodes,OSMSegments,OSMWays,option_prune_isolated);
387
388	if(option_prune_short)
389	PruneShortSegments(OSMNodes,OSMSegments,OSMWays,option_prune_short);
390
391	if(option_prune_straight)
392	PruneStraightHighwayNodes(OSMNodes,OSMSegments,OSMWays,option_prune_straight);
393
394	FinishPruning(OSMNodes,OSMSegments,OSMWays);
395
396	/* Remove the pruned nodes and segments and update the indexes */
397
398	RemovePrunedNodes(OSMNodes,OSMSegments);
399	RemovePrunedSegments(OSMSegments,OSMWays);
400	CompactWayList(OSMWays,OSMSegments);
401	RemovePrunedTurnRelations(OSMRelations,OSMNodes);
402	IndexSegments(OSMSegments,OSMNodes,OSMWays);
403	}
404
405	/* Repeated iteration on Super-Nodes and Super-Segments */
406
407	do
408	{
409	index_t nsuper;
410
411	printf("\nProcess Super-Data (iteration %d)\n================================%s\n\n",iteration,iteration>9?"=":"");
412	fflush(stdout);
413
414	if(iteration==0)
415	{
416	/* Select the super-nodes */
417
418	ChooseSuperNodes(OSMNodes,OSMSegments,OSMWays);
419
420	/* Select the super-segments */
421
422	SuperSegments=CreateSuperSegments(OSMNodes,OSMSegments,OSMWays);
423
424	nsuper=OSMSegments->number;
425	}
426	else
427	{
428	SegmentsX *SuperSegments2;
429
430	/* Select the super-nodes */
431
432	ChooseSuperNodes(OSMNodes,SuperSegments,OSMWays);
433
434	/* Select the super-segments */
435
436	SuperSegments2=CreateSuperSegments(OSMNodes,SuperSegments,OSMWays);
437
438	nsuper=SuperSegments->number;
439
440	FreeSegmentList(SuperSegments);
441
442	SuperSegments=SuperSegments2;
443	}
444
445	/* Sort the super-segments and remove duplicates */
446
447	DeduplicateSuperSegments(SuperSegments,OSMWays);
448
449	/* Index the segments */
450
451	IndexSegments(SuperSegments,OSMNodes,OSMWays);
452
453	/* Check for end condition */
454
455	if(SuperSegments->number==nsuper)
456	quit=1;
457
458	iteration++;
459
460	if(iteration>max_iterations)
461	quit=1;
462	}
463	while(!quit);
464
465	/* Combine the super-segments */
466
467	printf("\nCombine Segments and Super-Segments\n===================================\n\n");
468	fflush(stdout);
469
470	/* Merge the super-segments */
471
472	MergedSegments=MergeSuperSegments(OSMSegments,SuperSegments);
473
474	FreeSegmentList(OSMSegments);
475
476	FreeSegmentList(SuperSegments);
477
478	OSMSegments=MergedSegments;
479
480	/* Re-index the merged segments */
481
482	IndexSegments(OSMSegments,OSMNodes,OSMWays);
483
484	/* Cross reference the nodes and segments */
485
486	printf("\nCross-Reference Nodes and Segments\n==================================\n\n");
487	fflush(stdout);
488
489	/* Sort the nodes and segments geographically */
490
491	SortNodeListGeographically(OSMNodes);
492
493	SortSegmentListGeographically(OSMSegments,OSMNodes);
494
495	/* Re-index the segments */
496
497	IndexSegments(OSMSegments,OSMNodes,OSMWays);
498
499	/* Sort the turn relations geographically */
500
501	SortTurnRelationListGeographically(OSMRelations,OSMNodes,OSMSegments);
502
503	/* Output the results */
504
505	printf("\nWrite Out Database Files\n========================\n\n");
506	fflush(stdout);
507
508	/* Write out the nodes */
509
510	SaveNodeList(OSMNodes,FileName(dirname,prefix,"nodes.mem"),OSMSegments);
511
512	/* Write out the segments */
513
514	SaveSegmentList(OSMSegments,FileName(dirname,prefix,"segments.mem"));
515
516	/* Write out the ways */
517
518	SaveWayList(OSMWays,FileName(dirname,prefix,"ways.mem"));
519
520	/* Write out the relations */
521
522	SaveRelationList(OSMRelations,FileName(dirname,prefix,"relations.mem"));
523
524	/* Close the error log file and process the data */
525
526	if(errorlog)
527	{
528	close_errorlog();
529
530	if(option_keep)
531	{
532	printf("\nCreate Error Log\n================\n\n");
533	fflush(stdout);
534
535	ProcessErrorLogs(OSMNodes,OSMWays,OSMRelations);
536
537	SortErrorLogsGeographically();
538
539	SaveErrorLogs(OSMNodes,FileName(dirname,prefix,"errorlogs.mem"));
540	}
541	}
542
543	/* Free the memory (delete the temporary files) */
544
545	FreeNodeList(OSMNodes,0);
546	FreeWayList(OSMWays,0);
547	FreeRelationList(OSMRelations,0);
548
549	FreeSegmentList(OSMSegments);
550
551	/* Print the total time */
552
553	if(option_logtime)
554	{
555	printf("\n");
556	fprintf_elapsed_time(stdout,&start_time);
557	printf("Complete\n");
558	fflush(stdout);
559	}
560
561	return(0);
562	}
563
564
565	/*++++++++++++++++++++++++++++++++++++++
566	Print out the usage information.
567
568	int detail The level of detail to use - 0 = low, 1 = high.
569
570	const char *argerr The argument that gave the error (if there is one).
571
572	const char *err Other error message (if there is one).
573	++++++++++++++++++++++++++++++++++++++*/
574
575	static void print_usage(int detail,const char argerr,const char err)
576	{
577	fprintf(stderr,
578	"Usage: planetsplitter [--help]\n"
579	" [--dir=<dirname>] [--prefix=<name>]\n"
580	#if defined(USE_PTHREADS) && USE_PTHREADS
581	" [--sort-ram-size=<size>] [--sort-threads=<number>]\n"
582	#else
583	" [--sort-ram-size=<size>]\n"
584	#endif
585	" [--tmpdir=<dirname>]\n"
586	" [--tagging=<filename>]\n"
587	" [--loggable] [--logtime]\n"
588	" [--errorlog[=<name>]]\n"
589	" [--parse-only \| --process-only]\n"
590	" [--append] [--keep] [--changes]\n"
591	" [--max-iterations=<number>]\n"
592	" [--prune-none]\n"
593	" [--prune-isolated=<len>]\n"
594	" [--prune-short=<len>]\n"
595	" [--prune-straight=<len>]\n"
596	" [<filename.osm> ... \| <filename.osc> ...\n"
597	" \| <filename.pbf> ...\n"
598	" \| <filename.osm> ... \| <filename.osc> ..."
599	#if defined(USE_BZIP2) && USE_BZIP2
600	"\n \| <filename.(osm\|osc\|o5m\|o5c).bz2> ..."
601	#endif
602	#if defined(USE_GZIP) && USE_GZIP
603	"\n \| <filename.(osm\|osc\|o5m\|o5c).gz> ..."
604	#endif
605	"]\n");
606
607	if(argerr)
608	fprintf(stderr,
609	"\n"
610	"Error with command line parameter: %s\n",argerr);
611
612	if(err)
613	fprintf(stderr,
614	"\n"
615	"Error: %s\n",err);
616
617	if(detail)
618	fprintf(stderr,
619	"\n"
620	"--help Prints this information.\n"
621	"\n"
622	"--dir=<dirname> The directory containing the routing database.\n"
623	"--prefix=<name> The filename prefix for the routing database.\n"
624	"\n"
625	"--sort-ram-size=<size> The amount of RAM (in MB) to use for data sorting\n"
626	#if SLIM
627	" (defaults to 64MB otherwise.)\n"
628	#else
629	" (defaults to 256MB otherwise.)\n"
630	#endif
631	#if defined(USE_PTHREADS) && USE_PTHREADS
632	"--sort-threads=<number> The number of threads to use for data sorting.\n"
633	#endif
634	"\n"
635	"--tmpdir=<dirname> The directory name for temporary files.\n"
636	" (defaults to the '--dir' option directory.)\n"
637	"\n"
638	"--tagging=<filename> The name of the XML file containing the tagging rules\n"
639	" (defaults to 'tagging.xml' with '--dir' and\n"
640	" '--prefix' options or the file installed in\n"
641	" '" DATADIR "').\n"
642	"\n"
643	"--loggable Print progress messages suitable for logging to file.\n"
644	"--logtime Print the elapsed time for each processing step.\n"
645	"--errorlog[=<name>] Log parsing errors to 'error.log' or the given name\n"
646	" (the '--dir' and '--prefix' options are applied).\n"
647	"\n"
648	"--parse-only Parse the OSM/OSC file(s) and store the results.\n"
649	"--process-only Process the stored results from previous option.\n"
650	"--append Parse the OSM file(s) and append to existing results.\n"
651	"--keep Keep the intermediate files after parsing & sorting.\n"
652	"--changes Parse the data as an OSC file and apply the changes.\n"
653	"\n"
654	"--max-iterations=<number> The number of iterations for finding super-nodes\n"
655	" (defaults to 5).\n"
656	"\n"
657	"--prune-none Disable the prune options below, they are re-enabled\n"
658	" by adding them to the command line after this option.\n"
659	"--prune-isolated=<len> Remove access from small disconnected segment groups\n"
660	" (defaults to removing groups under 500m).\n"
661	"--prune-short=<len> Remove short segments (defaults to removing segments\n"
662	" up to a maximum length of 5m).\n"
663	"--prune-straight=<len> Remove nodes in almost straight highways (defaults to\n"
664	" removing nodes up to 3m offset from a straight line).\n"
665	"\n"
666	"<filename.osm>, <filename.osc>, <filename.pbf>, <filename.o5m>, <filename.o5c>\n"
667	" The name(s) of the file(s) to read and parse.\n"
668	" Filenames ending '.pbf' read as PBF, filenames ending\n"
669	" '.o5m' or '.o5c' read as O5M/O5C, others as XML.\n"
670	#if defined(USE_BZIP2) && USE_BZIP2
671	" Filenames ending '.bz2' will be bzip2 uncompressed.\n"
672	#endif
673	#if defined(USE_GZIP) && USE_GZIP
674	" Filenames ending '.gz' will be gzip uncompressed.\n"
675	#endif
676	"\n"
677	"<transport> defaults to all but can be set to:\n"
678	"%s"
679	"\n"
680	"<highway> can be selected from:\n"
681	"%s"
682	"\n"
683	"<property> can be selected from:\n"
684	"%s",
685	TransportList(),HighwayList(),PropertyList());
686
687	exit(!detail);
688	}