Open Street Map データセット (OpenStreetMap data set)

このページでは、 Open Street Map データセットをダウンロードし、リレーショナルデータベースに格納する手順を説明する． In this Wwb page，Open Street Map dataset are downloaded and stored into relational database.

◆ 作成するSQLite 3 データベース: osmdb

前準備

使用するソフトウェア

R システムのインストールが済んでいること
SQLite 3 のインストール: SQLite 3インストール，データベース作成，テーブル定義，レコード挿入（Windows 上）別ページ »で説明

あらかじめ決めておく事項

このページでは，データベースの作成を行いますので，作成するデータベースのデータベース名を決めておくこと．このページでは，次のように書く．

データベース名: osmdb
データベース名は，自由に決めてよいが，半角文字（つまり英字と英記号）を使い，スペースを含まないこと，

OSM データのダウンロードと CSV ファイル形式への変換 (Download OSM data and Generate CSV Files)

＜要点＞緯度、経度等を次のように設定

130.21688103675842, 33.59656025053064, 500, 500

◆ bash プログラム

 
#
# download OSM data and generate CSV files 
#
cat >/tmp/a.$$.r <<-RCOMMAND
#
  library(stats)
  library(graphics)
  library(datasets)
  library(ggplot2)
  library(xts)
  library(chron)
  library(osmar)
src <- osmsource_api()
bb <- center_bbox(130.21688103675842, 33.59656025053064, 500, 500)
ua <- get_osm(bb, source = src)
#
osm_nodes <- ua\$nodes\$attrs[c("id", "visible", "timestamp", "version", "changeset", "user", "uid", "lat", "lon")]
names(osm_nodes) <- c("node_id", "visible", "timestamp", "version", "changeset", "user", "uid", "lat", "lon")
osm_node_tags <- ua\$nodes\$tags[c("id", "k", "v")]
names(osm_node_tags) <- c("node_id", "key", "value")
# あとでCSVファイルを読み込むときに「,」が邪魔になるので
osm_node_tags\$value <- gsub(",", "", osm_node_tags\$value)
#
osm_ways <- ua\$ways\$attrs[c("id", "visible", "timestamp", "version", "changeset", "user", "uid")]
names(osm_ways) <- c("way_id", "visible", "timestamp", "version", "changeset", "user", "uid")
osm_way_tags <- ua\$ways\$tags[c("id", "k", "v")]
names(osm_way_tags) <- c("way_id", "key", "value")
# あとでCSVファイルを読み込むときに「,」が邪魔になるので
osm_way_tags\$value <- gsub(",", "", osm_way_tags\$value)
#
way_nodes <- ua\$ways\$refs[c("id", "ref")]
names(way_nodes) <- c("way_id", "node_id")
#
osm_relations <- ua\$relations\$attrs[c("id", "visible", "timestamp", "version", "changeset", "user", "uid")]
names(osm_relations) <- c("relation_id", "visible", "timestamp", "version", "changeset", "user", "uid")
osm_relation_tags <- ua\$relations\$tags[c("id", "k", "v")]
names(osm_relation_tags) <- c("relation_id", "key", "value")
# あとでCSVファイルを読み込むときに「,」が邪魔になるので
osm_relation_tags\$value <- gsub(",", "", osm_relation_tags\$value)
relation_way_nodes <- ua\$relations\$refs[c("id", "type", "ref", "role")]
names(relation_way_nodes) <- c("relation_id", "type", "ref", "role")
#
write.csv(osm_nodes, file="/tmp/osm_nodes.csv", row.names=FALSE)
write.csv(osm_node_tags, file="/tmp/osm_node_tags.csv", row.names=FALSE)
write.csv(osm_ways, file="/tmp/osm_ways.csv", row.names=FALSE)
write.csv(osm_way_tags, file="/tmp/osm_way_tags.csv", row.names=FALSE)
write.csv(way_nodes, file="/tmp/osm_way_nodes.csv", row.names=FALSE)
write.csv(osm_relations, file="/tmp/osm_relations.csv", row.names=FALSE)
write.csv(osm_relation_tags, file="/tmp/osm_relation_tags.csv", row.names=FALSE)
write.csv(relation_way_nodes, file="/tmp/osm_relation_way_nodes.csv", row.names=FALSE)
RCOMMAND
cat /tmp/a.$$.r | r

CSV ファイルの表示 (View OSM CSV files)

node の緯度、経度

◆ R プログラム

require(data.table)
require(ggplot2)
nodes <- data.table( read.csv("/tmp/osm_nodes.csv", header=TRUE, as.is=TRUE) )
ggplot(nodes, aes(x=lat, y=lon)) + geom_point()
# または tkplot(x=nodes$lat, y=nodes$lon)

リレーショナルデータベースの生成

作成したCSVファイルを、いったん、作業用のテーブルに格納する．作業用のテーブルを使って、最終的に欲しいテーブルを作る．

テーブル定義

OSM データ（元の OSM データを表現するテーブル）

osm_nodes(node_id, visible, timestamp, version, changeset, user, uid, lat, lon)
osm_node_tags(node_id, key, value)
osm_ways(way_id, visible, timestamp, version, changeset, user, uid)
osm_way_tags(node_id, key, value)
osm_way_nodes(seq, way_id, node_id)
osm_relations(relation_id, visible, timestamp, version, changeset, user, uid)
osm_relation_tags(relation_id, key, value)

OSM 派生データ

nodes(node_id, lat, lon, classification, feature, name, degree)
# classification and features are not implemented, see http://wiki.openstreetmap.org/wiki/Map_Features
# 1-to-1 (nodes, osm_nodes)
# 1-to-n (nodes, osm_node_tags)
ways(way_id, is_open, classification, feature, name)
# open polyline or closed polyline or area
# classification and features are not implemented, see http://wiki.openstreetmap.org/wiki/Map_Features
# 1-to-1 (ways, osm_ways)
# 1-to-n (ways, osm_way_tags)
relations(way_id, type)

# relations(relation_id) # relation_node_ways(relation_id, type, ref, role) # n-to-m # 1-to-1 (relations, osm_relations) # 1-to-n (nodes, osm_node_tags)

◆ bash プログラム

#!/bin/bash

rm -f /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table classification; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table classification (
  name           text
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_nodes; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_nodes (
  node_id        int,
  visible        text,
  timestamp      datetime,
  version        integer,
  changeset      integer,
  user           text,
  uid            integer,
  lat            real,
  lon            real
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_node_tags; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_node_tags (
  node_id        integer, 
  key            text, 
  value          text 
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_ways; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_ways (
  way_id        int,
  visible        text,
  timestamp      datetime,
  version        integer,
  changeset      integer,
  user           text,
  uid            integer
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_way_tags; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_way_tags (
  way_id        integer, 
  key            text, 
  value          text 
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_way_nodes; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_way_nodes (
  seq            integer  PRIMARY KEY autoincrement not null, 
  way_id         integer, 
  node_id        integer 
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_relations; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_relations (
  relation_id        int,
  visible        text,
  timestamp      datetime,
  version        integer,
  changeset      integer,
  user           text,
  uid            integer
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table osm_relation_tags; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table osm_relation_tags (
  relation_id        integer, 
  key            text, 
  value          text 
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb





#
cat >/tmp/a.$$.sql <<-SQL
drop table nodes; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table nodes (
  node_id        integer  PRIMARY KEY autoincrement not null, 
  lat            real  not null, 
  lon            real  not null, 
  classification text, 
  feature        text, 
  name           text,  
  degree         integer 
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

#
cat >/tmp/a.$$.sql <<-SQL
drop table ways; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
#
cat >/tmp/a.$$.sql <<-SQL
create table ways (
  way_id         integer  PRIMARY KEY autoincrement not null, 
  classification text, 
  feature        text, 
  name           text, 
  is_open        integer
);  
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

テーブル classification の作成

参考 http://wiki.openstreetmap.org/wiki/Map_Features

cat > /tmp/a.$$.csv <<-CSV
"aerialway"
"aeroway"
"amenity"
"barrier"
"boundary"
"building"
"craft"
"mmergency"
"geological"
"highway"
"historic"
"landuse"
"leisure"
"man made"
"military"
"natural"
"office"
"places"
"power"
"public transport"
"railway"
"route"
"shop"
"sport"
"tourism"
"waterway"
CSV
#
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/a.$$.csv classification 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from classification limit 10;' | sqlite3 /tmp/osmdb

テーブル osm_nodes の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_nodes.csv 
tail -n +2 /tmp/osm_nodes.csv | nkf -w > /tmp/osm_nodes_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_nodes_uft8.csv osm_nodes 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_nodes limit 10;' | sqlite3 /tmp/osmdb

テーブル osm_node_tags の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_node_tags.csv 
tail -n +2 /tmp/osm_node_tags.csv | nkf -w > /tmp/osm_node_tags_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_node_tags_uft8.csv osm_node_tags 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_node_tags limit 10;' | sqlite3 /tmp/osmdb

classficationテーブルのname属性に記載されている値は、osm_node_tags テーブルでは各nodeごとにたかだか１回しか現れないことの確認。

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from classification, osm_node_tags where classification.name = osm_node_tags.key;
select node_id, count(*) from T group by node_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

osm_node_tags テーブルでは, key の値が「"name"」であるような行は、各nodeごとにたかだか１回しか現れないことの確認。

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from osm_node_tags where osm_node_tags.key = '"name"';
select node_id, count(*) from T group by node_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

テーブル osm_ways の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_ways.csv 
tail -n +2 /tmp/osm_ways.csv | nkf -w > /tmp/osm_ways_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_ways_uft8.csv osm_ways 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_ways limit 10;' | sqlite3 /tmp/osmdb

テーブル osm_way_tags の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_way_tags.csv 
tail -n +2 /tmp/osm_way_tags.csv | nkf -w > /tmp/osm_way_tags_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_way_tags_uft8.csv osm_way_tags 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_way_tags limit 10;' | sqlite3 /tmp/osmdb

classficationテーブルのname属性に記載されている値は、osm_way_tags テーブルでは各wayごとにたかだか１回しか現れないかの確認。実は２回現れるものがある．

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from classification, osm_way_tags where classification.name = osm_way_tags.key;
select way_id, count(*) from T group by way_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

osm_way_tags テーブルでは, key の値が「"name"」であるような行は、各wayごとにたかだか１回しか現れないことの確認。

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from osm_way_tags where osm_way_tags.key = '"name"';
select way_id, count(*) from T group by way_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

テーブル osm_way_nodes の作成

◆ bash プログラム

一度、テーブルに読み込んで、その後通し番号 seq を生成．

#!/bin/bash

# /tmp/osm_way_nodes.csv 
tail -n +2 /tmp/osm_way_nodes.csv | nkf -w > /tmp/osm_way_nodes_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
create table T (
  way_id         integer, 
  node_id        integer 
);  
.mode csv
.import /tmp/osm_way_nodes_uft8.csv T 
insert into osm_way_nodes(way_id, node_id) select way_id, node_id from T;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_way_nodes limit 10;' | sqlite3 /tmp/osmdb

テーブル osm_relations の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_relations.csv 
tail -n +2 /tmp/osm_relations.csv | nkf -w > /tmp/osm_relations_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_relations_uft8.csv osm_relations 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_relations limit 10;' | sqlite3 /tmp/osmdb

テーブル osm_relation_tags の作成

◆ bash プログラム

#!/bin/bash

# /tmp/osm_relation_tags.csv 
tail -n +2 /tmp/osm_relation_tags.csv | nkf -w > /tmp/osm_relation_tags_uft8.csv
cat >/tmp/a.$$.sql <<-SQL
.mode csv
.import /tmp/osm_relation_tags_uft8.csv osm_relation_tags 
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from osm_relation_tags limit 10;' | sqlite3 /tmp/osmdb

classficationテーブルのname属性に記載されている値は、osm_relation_tags テーブルでは各relationごとにたかだか１回しか現れないかの確認。実は２回現れるものがある．

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from classification, osm_relation_tags where classification.name = osm_relation_tags.key;
select relation_id, count(*) from T group by relation_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

osm_relation_tags テーブルでは, key の値が「"name"」であるような行は、各relationごとにたかだか１回しか現れないことの確認。

cat >/tmp/a.$$.sql <<-SQL
create table T as select * from osm_relation_tags where osm_relation_tags.key = '"name"';
select relation_id, count(*) from T group by relation_id having count(*) > 1;
drop table T;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb

テーブル nodes の作成

◆ bash プログラム

#!/bin/bash
cat >/tmp/a.$$.sql <<-SQL
create table R as select osm_node_tags.node_id as node_id, osm_node_tags.key as key, osm_node_tags.value as value from classification, osm_node_tags where classification.name = osm_node_tags.key;
create table S as select osm_node_tags.node_id as node_id, osm_node_tags.key as key, osm_node_tags.value as value from osm_node_tags where osm_node_tags.key = '"name"';
create table T as select osm_nodes.node_id as node_id, osm_nodes.lat as lat, osm_nodes.lon as lon, R.key as classification, R.value as feature  from osm_nodes left outer join R on osm_nodes.node_id = R.node_id;
create table U as select T.node_id as node_id, T.lat as lat, T.lon as lon, T.classification as classification, T.feature as feature, S.value as name from T left outer join S on T.node_id = S.node_id; 
insert into nodes(node_id, lat, lon, classification, feature, name) select node_id, lat, lon, classification, feature, name from U; 
drop table R;
drop table S;
drop table T;
drop table U;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from nodes limit 10;' | sqlite3 /tmp/osmdb

テーブル ways の作成

◆ bash プログラム

#!/bin/bash
cat >/tmp/a.$$.sql <<-SQL
create table R as select osm_way_tags.way_id as way_id, osm_way_tags.key as key, osm_way_tags.value as value from classification, osm_way_tags where classification.name = osm_way_tags.key;
create table S as select osm_way_tags.way_id as way_id, osm_way_tags.key as key, osm_way_tags.value as value from osm_way_tags where osm_way_tags.key = '"name"';
create table T as select osm_ways.way_id as way_id, osm_ways.lat as lat, osm_ways.lon as lon, R.key as classification, R.value as feature  from osm_ways left outer join R on osm_ways.way_id = R.way_id;
create table U as select T.way_id as way_id, T.lat as lat, T.lon as lon, T.classification as classification, T.feature as feature, S.value as name from T left outer join S on T.way_id = S.way_id; 
insert into ways(way_id, lat, lon, classification, feature, name) select way_id, lat, lon, classification, feature, name from U; 
drop table R;
drop table S;
drop table T;
drop table U;
vacuum; 
SQL
cat /tmp/a.$$.sql | sqlite3 /tmp/osmdb
# 
echo 'select * from ways limit 10;' | sqlite3 /tmp/osmdb