Making Maps Navigable - Programming GPS and OpenStreetMap Applications with Java - page 107

nodes, and it works much better to add this step at the end to deflate the
file significantly.
At the end of the country batch, Germany has been split into separate
les for
networks: highways railways waterways
carto: forest farm residential industrial wood water
admin: cities borders
rest of: Germany natural landuse
Map statistics. From the map algebra, using the file sizes, one can easily
detect overlaps. When extracting layers for the first time, the map engineer
is curious to know what is actually left in the *.rest.osm files. To find this
out, another switch -stats is hacked to the OSMparser . 3 The output file is
directed to a dummy.osm file, while the statistics are written to the command
line to be directed to a log file: quick and dirty, yet spontaneous and useful.
With the following command line, the statistics will be logged to a file (see
statistics2010.log ):
../products/GER>statistics statistics.log
Statistical tools can only indicate error sources, yet they are very useful
for regression analysis and attributes deviating more than 5% from a prior
production can indicate a problem. To avoid getting lost in the details,
the reader should make his own analysis by copying the statistics to a
spreadsheet and ordering them by occurrences. The clean.bat script should
give an idea on how to shrink the file sizes step by step. For reasonable sizes,
the script runs quickly and can be configured to remove all temporary files
at the end. In the process of creating cleaning strategies for dedicated files,
the map engineer can collect unnecessary attributes for resulting products
and possibly move the cleaning up the hierarchy.
Going further. Naturally, there are endless ways to compile a country, and
the batch file might provide some inspiration. Yet, we are still quite some
way from a final PSF. Osmosis has six relevant switches -nk -nkv -wk -wkv
-tf -un to process OSM les, while the OSMparser has the \regEx" window
to get the desired output.
Using the layer files, you should experiment with regular expressions to
do the following:
ˆ reduce all city tags to place , name , isin , and population ;
ˆ reduce the cities to those with population > 500,000 ;
3 The implementation will become transparent in Section 8.3.2. Note that the private
OSMparser $ Occurrences.class has to be copied with the OSMparser.class .