Converts in a few minutes the big, complex and hard to read XML Dutch addresses database (BAG or Basisregistratie Adressen en Gebouwen) into a user-friendly, file based, blazingly fast SQLite database by running a single Python script. No need to install any dependencies or a database server.
Additional scripts will convert this SQLite database to other formats, like CSV.
If you don't want to run the script yourself, download the latest BAG in SQLite or CSV format from our releases section.
The Dutch public addresses and buildings database (BAG or Basisregistratie Adressen en Gebouwen) is freely downloadable from the Dutch cadastre agency named Kadaster. Hooray 🙂.
The bad news is: The original BAG comes in a complex and hard to read XML format using thousands of zipped XML files, which will quickly temper your initial enthusiasm to quite low levels. It also does not include municipalities or provinces and provides coordinates using a system that non-experts won't expect named Rijksdriehoekscoördinaten😲.
This Python utility parses the BAG database and converts it into a clean, easy to read & use SQLite database. Municipalities (gemeenten) and provinces (provincies) are added. Rijksdriehoekscoördinaten coordinates are converted to standard WGS84 latitude and longitude coordinates. A few BAG bugs are fixed. Year of construction, floor area and intended use of buildings are also provided. Several tables (nummers, verblijfsobjecten, panden, ligplaatsen and standplaatsen) are merged into a general 'adressen' table. The SQLite database can be used directly, or as a source to generate a *.csv file or update your own addresses databases. There are a couple of options available in the config.py.
- Python 3.11. Older Python versions may work, but are not tested and certainly slower.
- Download, or better (because it allows easy updates) use git to download the BAG parser. Install parser:
git clone https://github.com/digitaldutch/BAG_parser
Update parser:git pull https://github.com/digitaldutch/BAG_parser - Download the BAG (3 GB) from kadaster.nl
or directly from pdok.nl
and save the file as
bag.zipin theinputfolder. - The gemeenten.csv file is already included in the
inputfolder, but you can download the latest version from the CBS website. Save it asgemeenten.csvin the input folder. - Set your options in config.py
- Run
import_bag.py - Drink a cup of coffee for a few minutes ☕😎 while watching the progress bar.
- Open the SQLite database with your favorite tool. I like DBeaver. Here's an example query on SQLite database to get information about postcode 2514GL, huisnummer 78 (Paleis Noordeinde):
SELECT
a.postcode,
a.huisnummer,
a.huisletter || a.toevoeging AS toevoeging,
o.naam AS straat,
g.naam AS gemeente,
w.naam AS woonplaats,
p.naam AS provincie,
a.bouwjaar,
a.latitude,
a.longitude,
a.rd_x,
a.rd_y,
a.oppervlakte AS vloeroppervlakte,
a.gebruiksdoel,
a.hoofd_nummer_id
FROM adressen a
LEFT JOIN openbare_ruimten o ON a.openbare_ruimte_id = o.id
LEFT JOIN gemeenten g ON a.gemeente_id = g.id
LEFT JOIN woonplaatsen w ON a.woonplaats_id = w.woonplaats_id
LEFT JOIN provincies p ON g.provincie_id = p.id
WHERE postcode = '2514GL'
AND huisnummer = 68;- When done parsing, use the export_to_csv_postcodes.py or export_to_csv.py to create a *.csv file. These functions are easy to customize. I myself use one (not on GitHub yet, as it is customized) to pump the SQLite data into a live Firebird database.
Parses the original BAG file and transforms it into a SQLite database. Takes 9 minutes to complete
on an AMD 7700X PC, or a few minutes more if you switch on the parse_geometries option in the config.py.
Exports the addresses in SQLite database to a *.csv file with address info only. Takes about 15 seconds.
Exports the addresses in SQLite database to a *.csv file including additional information, like year of construction, latitude, longitude, floor area and intended use of buildings. Takes about 40 seconds.
Exports statistics of postal code groups (4, 5 or all 6 characters) in SQLite database to a *.csv file including number of addresses and average lat/lon of addresses in that zone. Takes several seconds.
Checks de SQLite database for info and errors. import_bag.py also performs these tests after parsing.
Reduces the SQlite database size by removing BAG tables (nummers, verblijfsobjecten, panden, ligplaatsen and standplaatsen)
that are no longer needed due to the new 'adressen' table.
The parser also does this as a final step if delete_no_longer_needed_bag_tables is set to True in config.py.
An adres is a nevenadres if the hoofd_nummer_id field is set. It points to the nummer_id of the hoofdadres.
- The WGS84 coordinates are calculated using approximation equations by F.H. Schreutelkamp and G.L. Strang van Hees. This conversion has an error of a few decimeters. Don't use the WGS84 coordinates if you need higher accuracy.
- verblijfsobjecten table:
Some gebruiksdoel, pand_id and nevenadressen fields contain multiple values that are separated by commas. Be careful if you do queries with joins on those fields. - Adressen table:
- Some gebruiksdoel and pand_id fields contain multiple values that are separated by commas.
- The bouwjaar and geometry field only contain the data of one pand, even if an address has multiple panden.
- There are probably several more things missing that I don't know about. Feel free to file a GitHub issue.
The Kadaster has an online BAG viewer where you can search any address or other info in the official database.
This tool does not parse all data. If you need more data or professional support, buy it from nlextract, who have a more complex, but also complete parser.
Bert hubert has written a parser in C++, bagconv, which is quite similar to this one.
This software is made available under the MIT license.