pbf file missing 'osmosis_replication_sequence_number' in header

Question

I downloaded the latest pbf file(dated 27-Aug-2020 23:37) from a mirror and read its header using a python wrapper to osmium (v3.0.1)

import osmium

f = osmium.io.Reader("path/to/pbf_file.osm.pbf")
header = f.header()
seqnum = header.get("osmosis_replication_sequence_number", "")
timestamp = header.get("osmosis_replication_timestamp", "")
print(f'{seqnum!r},{timestamp!r}')
# output: '', '2020-08-23T23:59:50Z'

The sequence number was missing. Is this expected behaviour?

In the documentation, all the replication fields in OSMHeader seem to be set as optional. What pbf files are expected to have these values set?

Accepted Answer

The sequence field in the headers are currently set only on extracts from Geofabrik. pyosmium-up-to-date also sets them when you update a PBF file.

There is a good reason that the sequence number is not set on the official planet file: there is no exact sequence number that corresponds to the state of a planet file. Planet files are created independently of the change files. So the content of the planet might correspond to something like sequence number "4164164 3/4".

When you want to use updates with a downloaded planet file, the usual way is to look at the creation time and find a change sequence number that is far enough in the past that you get all new changes that are not yet contained in the planet. You don't have to do the math yourself, pyosmium-get-changes, which is included in the python osmium package, can do that for you. Just run

pyosmium-get-changes -O planet-latest.osm.pbf

and it prints a single number. This is the recommended sequence number where you should start with the updates.

See the section on updating OSM data in the pyosmium manual for more information.

pbf file missing 'osmosis_replication_sequence_number' in header

Follow this question

Related questions