Create a CSV file of OSM tag changes
Lotsa speed ups & speed improvements



You can also use your local clone with git send-email.


Use CSV tools to see who's mapping what in OpenStreetMap.

Given a OSM history file, it produces a CSV file, where each row refers to a change (addition, removal or modification) to a tag all OSM objects in an OSM data file with history.

Getting data

Planet.OpenStreetMap.org provides a “full history” file, updated every week, where you can download the latest full history file (⚠ 80+ GB! ⚠), although it's quite large.

Geofabrik provides an download service which includes full history files for lots of regions & countries. You must log into that with your OpenStreetMap account. You can also use this tool on regular, non-history, OSM data files.


If you have Rust installed, you can install it with:

cargo install osm-tag-csv-history

You can download prebuild binary released from the Github release page, (e.g. download the v0.1.0 release).


osm-tag-csv-history -i mydata.osm.pbf -o mydata.csv


Many programmes can use CSV files. It's also possible to use hacky unix command line programmes to calculate who's adding fuel stations (amenity=fuel in OSM) in Ireland:

osm-tag-csv-history -i ./ireland-and-northern-ireland-internal.osh.pbf -o - --no-header | grep '^amenity,fuel,' | cut -d, -f9 | sort | uniq -c | sort -n | tail -n 20

Here can find all times someone has upgraded a building from building=yes to something else.

osm-tag-csv-history -i data.osh.pbf -o - --no-header | grep -P '^building,[^,]+,yes,' | cat -n

And with some other command line commands, we can get a list of who's doing the most to make OSM more descriptive by upgrading building=yes.

osm-tag-csv-history -i data.osh.pbf -o - --no-header | grep -P '^building,[^,]+,yes,' | xsv select 8 | sort | uniq -c | sort -n | tail -n 20
Using with osmium getid

The id column (column 4) can be used by osmium-tool to filter an OSM file by object id. This is how you get a file of all the pet shops in OSM in a file:

osm-tag-csv-history -i country-latest.osm.pbf -o - --no-header | grep '^shop,pet,' | xsv select 4 | osmium getid -i - country-latest.osm.pbf -o pets.osm.pbf -r

(For this simple case, osmiums's tag filtering is probably better)

Non-history files

This programme can run on non-history files just fine. The old_value, and old_version will be empty. This can be a way to convert OSM data into CSV format for further processing.

Using on privacy preserving files.

The Geofabrik Public Download Service provides non-history files which do not include some metadata, like usernames, uids or changesetids. This tool can run on them and just give an empty value for username, and 0 for uid & changesetid.

If you have an OSM account, you can get full metada from the internal service.

Output file format

Records are separated by a newline (\n). A header line is included by default, but it can be turned off with --no-header (or forcibly included with --header).

If any string (e.g. tag value, username) has a newline or characters like that, it will be escaped with a backslash (i.e. a newline is written as 2 characters, \ then n).


(in order)

  • key The tag key
  • new_value The current/new version. "" (empty string) if the current version doesn't have this key (i.e. it has been removed from the object)
  • old_value The previous value. "" (empty string) if the previous version didn't have this key
  • id The object type and id. First character is the type (n/w/r), then the id. n123 is node with id 123. This format is used by osmium-tool to filter an OSM file by object id
  • new_version The current/new version number
  • old_version The previous version number. "" (empty string) for the first version of an object
  • Either:
    • datetime Date time (RFC3339 format in UTC) the object was created. Default, or if --timestamp-format dateime was used.
    • epoch_time Date time (Unix epoch time) the object was created. Only if --timestamp-format epoch_time was used.
  • username The username of the user who changes it (remember: in OSM, users can change their username, UIDs remain constant)
  • uid The user id of the user.
  • changeset_id Changeset id where this change was made

Timestamp format

One column contains the timestamp. Be default, the column will be the timestamp in RFC3339 (a subset of ISO 8601 format), i.e. YYYY-MM-DDTHH:MM:SSZ.

With --timestampformat epoch_time, the column will be called epoch_time, and the timestamp will be in unix epoch time. This is how the data is stored in an OSM PBF file. This makes processing about 15% faster.


Imagine this simple file:

<?xml version='1.0' encoding='UTF-8'?>
<osm version="0.6" generator="osmium/1.7.1">
  <node id="1" version="1" timestamp="2019-01-01T00:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="2">
      <tag k="place" v="city"/>
      <tag k="name" v="Nice City"/>
  <node id="1" version="2" timestamp="2019-03-01T12:30:00Z" lat="0.0" lon="0.0" user="Bob" uid="2" changeset="10">
      <tag k="place" v="city"/>
      <tag k="name" v="Nice City"/>
      <tag k="population" v="1000000"/>
  <node id="2" version="1" timestamp="2019-04-01T00:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="20">
      <tag k="amenity" v="restaurant"/>
      <tag k="name" v="TastyEats"/>
  <node id="2" version="2" timestamp="2019-04-01T02:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="21">
      <tag k="amenity" v="restaurant"/>
      <tag k="name" v="TastyEats"/>
      <tag k="cuisine" v="regional"/>
  <node id="2" version="3" timestamp="2019-04-01T03:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="22">
      <tag k="amenity" v="restaurant"/>
      <tag k="name" v="TastyEats"/>
      <tag k="cuisine" v="burger"/>
  <node id="2" version="4" timestamp="2019-04-01T03:00:00Z" lat="1.0" lon="0.0" user="Alice" uid="12" changeset="22">
      <tag k="amenity" v="restaurant"/>
      <tag k="name" v="TastyEats"/>
      <tag k="cuisine" v="burger"/>
  <node id="3" version="1" timestamp="2019-04-01T00:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="50">
      <tag k="amenity" v="bench"/>
  <node id="3" version="2" timestamp="2019-06-01T00:00:00Z" lat="0.0" lon="0.0" user="Alice" uid="12" changeset="100" visible="false">

NB: This programme cannot read XML files, only PBF. This file was converted to PBF with osmium cat example.osm.xml -o example.osm.pbf.

Running osm-tag-csv-history on it produces this CSV file (formatted here as a table by with csvtomd):

key new_value old_value id new_version old_version datetime username uid changeset_id
name Nice City n1 1 2019-01-01T00:00:00Z Alice 12 2
place city n1 1 2019-01-01T00:00:00Z Alice 12 2
population 1000000 n1 2 1 2019-03-01T12:30:00Z Bob 2 10
amenity restaurant n2 1 2019-04-01T00:00:00Z Alice 12 20
name TastyEats n2 1 2019-04-01T00:00:00Z Alice 12 20
cuisine regional n2 2 1 2019-04-01T02:00:00Z Alice 12 21
cuisine burger regional n2 3 2 2019-04-01T03:00:00Z Alice 12 22
amenity bench n3 1 2019-04-01T00:00:00Z Alice 12 50
amenity bench n3 2 1 2019-06-01T00:00:00Z Alice 12 100

Some things to note:

  • There can be more than one record (line) per version (n1 v1 has 2 lines, one for each tag that was added).
  • If no tags are changed, then there are no lines. There is no line for node 2 v4 because the location, not the tags was changed.
  • An empty value for old_version means there was no previous, or earlier, version.
  • When an object (and hence tag) is deleted, the previous value is in old_value, and the new_value is empty, as for n3 v2.

Possible useful tools

The following other tools might be useful:

  • xsv. a command line tool for slicing & filtering CSV data.
  • osmium a programme to process OSM data. You can use this to filter an OSM history file to a certain area, or time range.
  • datamash, command line CSV statistical tool.


Copyright 2020, GNU Affero General Public Licence (AGPL) v3 or later. See LICENCE.txt. Source code is on Sourcehut, and Github.

The output file should be viewed as a Derived Database of the OpenStreetMap database, and hence under the ODbL 1.0 licence, the same as the OpenStreetMap copyright