mnalis · February 13, 2024 15:17
diff --git a/count_housenumber_letters.sh b/count_housenumber_letters.sh
 #!/bin/sh
 # first step takes about 1h 45m on my 4core VM and produces about 300MB output file. The second step is about 2.5m 
 # it counts housenumbers ending in a number and a single letter (e.g. '42b') and outputs frequency of those letters a-f
 # for testing https://github.com/streetcomplete/StreetComplete/issues/5479#issuecomment-1937812748

 time pv -ptebar planet-240205.osm.bz2 | pbzip2 -dc | ag -F 'addr:housenumber' | zstd > housenumbers.xml.zstd
 time pv housenumbers.xml.zstd | zstdmt -dc | perl -nE 'if (/^\s*<tag k="addr:housenumber"\sv="(.*?)\s*".*$/) { $n=$1; $a++ if $n=~/\da$/i; $b++ if $n=~/\db$/i; $c++ if $n=~/\dc$/i;  $d++ if $n=~/\dd$/i;  $e++ if $n=~/\de$/i;  $f++ if $n=~/\df$/i;} END { say "a=$a\nb=$b\nc=$c\nd=$d\ne=$e\nf=$f\n" }'

 # result:
 #a=4593803
 #b=1862098
 #c=756913
 #d=435247
 #e=251648
 #f=167140
	#!/bin/sh
	# first step takes about 1h 45m on my 4core VM and produces about 300MB output file. The second step is about 2.5m
	# it counts housenumbers ending in a number and a single letter (e.g. '42b') and outputs frequency of those letters a-f
	# for testing https://github.com/streetcomplete/StreetComplete/issues/5479#issuecomment-1937812748

	time pv -ptebar planet-240205.osm.bz2 \| pbzip2 -dc \| ag -F 'addr:housenumber' \| zstd > housenumbers.xml.zstd
	time pv housenumbers.xml.zstd \| zstdmt -dc \| perl -nE 'if (/^\s<tag k="addr:housenumber"\sv="(.?)\s".$/) { $n=$1; $a++ if $n=~/\da$/i; $b++ if $n=~/\db$/i; $c++ if $n=~/\dc$/i; $d++ if $n=~/\dd$/i; $e++ if $n=~/\de$/i; $f++ if $n=~/\df$/i;} END { say "a=$a\nb=$b\nc=$c\nd=$d\ne=$e\nf=$f\n" }'

	# result:
	#a=4593803
	#b=1862098
	#c=756913
	#d=435247
	#e=251648
	#f=167140