Openstreetmap changeset statistics of source tags.

id+source.pl

# wget http://planet.openstreetmap.org/changesets-120229.osm.bz2
# tail -c 1000000  changesets-120229.osm.bz2  >tail.bz2
# bzip2recover tail.bz2 
# bzcat rec000* |perl id+source.pl >all_sources.txt

$id=0;
$source=0;
while(<>){    
    if(/<changeset.* id="([^"]*)"/){
        $id=$1;
        $changesets++;
    }

    if(/tag k="source" v="([^"]*)/){
        $source++;
        print "$id: $1\n"
    }
    
    if ( $changesets % 10000 == 1)
    {
        print STDERR "$changesets/$source: ". $source/$changesets."\n";
    }
}

print STDERR "DONE\n";
print STDERR "$changesets/$source: ". $source/$changesets."\n";

idsource.pl

# Latest means "changesets-120229.osm.bz2"
# wget http://planet.openstreetmap.org/changesets-latest.osm.bz2
# tail -c 1000000  changesets-latest.osm.bz2  >tail.bz2
# bzip2recover tail.bz2 
# bzcat rec000* |perl id+source.pl >all_sources.txt; rm rec000*
# bzcat changesets-latest.osm.bz2 |perl id+source.pl >all_sources_2.txt
$id=0;
$source=0;
while(<>){    
    if(/<changeset.* id="([^"]*).* num_changes="([^"]*)/){
        $id=$1;
        $num_changes=$2;
        $total_changes+=$num_changes;
        $changesets++;
    }

    if(/tag k="source" v="([^"]*)/){
        $source++;
        print "$id: $1: $num_changes\n";
        $source_changes+=$num_changes;
    }
    
    if ( $changesets % 10000 == 1)
    {
        stats();
    }
}

print STDERR "DONE\n";
stats;

sub stats{
print STDERR "Changets: $source/$changesets: ". $source/$changesets." Number of changes: $source_changes/$total_changes:".  $source_changes/$total_changes . "\n";

}