#openstreetmap #osm-pbf #osm #topology #file-format #multi-line

bin+lib osm-lump-ways

Group OpenStreetMap lines based on graph topology and OSM tags

47 releases (10 stable)

new 2.0.0-rc2 Nov 7, 2024
1.7.0 Oct 6, 2024
1.0.0 Jul 2, 2024
0.29.0 May 28, 2024
0.1.0 Mar 31, 2023

#12 in Geospatial

Download history 60/week @ 2024-07-25 10/week @ 2024-08-01 227/week @ 2024-08-08 31/week @ 2024-08-15 107/week @ 2024-09-05 194/week @ 2024-09-12 586/week @ 2024-09-19 424/week @ 2024-09-26 201/week @ 2024-10-03 29/week @ 2024-10-10 2/week @ 2024-10-17 45/week @ 2024-10-31

91 downloads per month

MIT/Apache and AGPL-3.0

335KB
6.5K SLoC

osm-lump-ways group OSM ways based on topology & shared tags

Crates.io Number of Downloads Crates.io Latest Version

Answer questions about OSM data like:

  • “What's the longest Main Street?”
  • “How far can I drive on unpaved roads in this region?”
  • “How long is the M1 motorway?”
  • “Are these rivers connected?”
  • “What's the river drainage basins?”

2 similar programmes are included: osm-lump-ways, which ignores the direction of the OSM way, and osm-lump-ways-down, which uses direction of OSM ways to produce data, incl. QA suitable files. Both share similarities.

Background

OSM linear features (eg roads, rivers, walls) are stored as way object. The OSM tagging model often requires one feature to be mapped as many different ways. osm-lump-ways will assemble them all together.

Filtering OSM Data

There are 2 ways to select which OSM ways will be used. All relations are currently ignored.

Tag Filter Rules

  • key / ∃key way has this tag
  • ~key_regex / ~key_regex There is a key, which matches this regex.
  • ∄key way does not has this key
  • key=value way has this key and this value
  • key≠value way either doesn't have this key, or if it does, it's not equal to value
  • key∈value1,value2, way has this key and the value is one of these
  • key∉value1,value2, way either doesn't have this key, or if it does, it's not one of these values
  • key~regex way has this key and the value matches this regex.
  • F1F2F3 logical OR of the other tag filters F1, F2, …
  • F1F2F3 logical AND of the other tag filters F1, F2, …

The popular regex crate is used for matching. Regexes, and string comparison, are case sensitive. Add (?i) at start of regex to switch to case insensitive (e.g. name~(?i).* street) Regexes match the whole value, name~[Ss]treet will match Street or street, but not Main Street North nor Main Street. Use name~.*[Ss]treet.* to match all.

Simple Tag Filtering

The -f/--tag-filter can be specified one or more times, and an OSM object is only included if it matches all defined filter's, i.e. a logical AND of all filters.

  • -f highway: Only ways with a highway tag are included
  • -f highway -f name: Only ways with a highway and name tag are included.
  • -f highway -f ∄name: Only ways with a highway and without a name tag are included.

More complex Tag Filtering Functions

The -F/--tag-filter-func takes a single ordered list of tag filters (separated by ;), and includes (with →T), or excludes (with →F), the OSM object based on the first filter function which matches. A bare T or F applies to all.

Example: We want to include all waterways. But not waterway=canal. But we want a waterway=canal iff it also has a lock=yes tag.

-F "waterway=canal∧lock=yes→T; waterway=canal→F; waterway→T; F

If the argument to -F/--tag-filter-func starts with @, the rest is a filename containing the tag filter func code. e.g. -F @myrules.txt. In this mode, a line include FILENAME; will include the contents of another file there. FILENAME is a path relative to the original filename. It will be expanded recursively.

Comments start with # and continue to the end of the line. Since the ; is special, it cannot be directly used in tag filtering. Use \u{3B} instead. e.g. waterway=put_in\u{3B}egress→F; is a rule to exclude any tag with key waterway and value put_in;egress.

Output

File format

If a filename ends with .geojson, a GeoJSON file (RFC 7946 will be created. For .geojsons, a GeoJSON Text Sequences (RFC 8142), aka GeoJSONSeq, file.

Geometry format

By default each feature is a MultiLineString, representing every way group.

With --split-into-single-paths, each way group will be split into 1 or more LineStrings, controlled by the --split-into-single-paths-by X argument. By default the as-crow-flies distance is used. longest-path uses the longest path in the graph, and this can take days to calculate on large networks.

Input

The input must be an OSM PBF file. Use osmium to convert between OSM file formats.

The input object ids must be positive. OSM import software often uses negative ids. Use osmium sort and then osmium renumber like so, to create an acceptable file.

osmium sort negative-id-file.osm.pbf -o sorted.osm.obf
osmium renumber sorted.osm.pbf -o new-input.osm.pbf

Installation

cargo install osm-lump-ways

This will install the 2 programmes: osm-lump-ways & osm-lump-ways-down. osm-lump-ways ignores the direction of the OSM ways, and produces single output files. It can be used for waterways or watersheds, but also for roads or similar. osm-lump-ways-down uses the direction of the OSM data to produce many similar files.

osm-lump-ways

Usage

Generate river drainage basins

osm-lump-ways -i path/to/region-latest.osm.pbf -o region-rivers.geojson -f waterway=river

To group based on the river's name:

osm-lump-ways -i path/to/region-latest.osm.pbf -o region-rivers.geojson -f waterway=river -g name

To find long streets and assemble them into connected (Multi)LineStrings:

osm-lump-ways -i path/to/region-latest.osm.pbf -o long-streets.geojson -f highway -g name

Full Options

Run with --help to see all options.

Frames

Here, a “frame” of a grouping is a shortest path through 2 points in the grouped together ways. This can be useful for waterways to find where a big group is connected.

--output-frames FILENAME.geojsons will save these features to a file. GeoJSONSeq output format only.

Examples of usage

osm-lump-ways-down

It reads & groups an OSM PBF file, like osm-lump-ways, but it uses the direction of the OSM way, to produce many different output files.

Output files

It can output different files, depending on the options:

--loops FILENAME

Cycles in the network. Each is a strongly connected component, and MultiLineString.

Feature Properties

  • areas: Array of Strings. The cycle is geocoded using country-borders. The code for every matching area is included in this array, sorted with longest strings first. e.g.: ["GB-ENG","GB"],
  • areas_s: String. areas, but the array has been converted to a string, and comma separated, with a , prefix & suffix. e.g. ",GB-ENG,GB,". The prefix & suffix ensure you can always search for ,GB, in that property and get a match.
  • area_N: String. One for every element in areas, with a numeric suffix and the item itself. e.g. "areas_s":",GB-ENG,GB," then there will be "area_0":"GB-ENG","area_1":"GB". These properties are for software which doesn't support JSON arrays.
  • length_m: Float. Total length of this cycle, in metres.
  • nodes: String. All the node ids in this cycle, sorted, and deduplicated, and saved e.g. "n318688661,n5354970286,n1016277960.
  • root_node_id: Integer. Node id of the lowest numbered node in this cycle.
  • num_nodes: Integer. Number of nodes in this cycle. incl. duplicates.

Loops stats CSV

With --loops-csv-stats-file FILENAME.csv, a CSV file with statistics of the loops, along with a breakdown per region is created. If the file is already there, the data is appended.

See the doc comment src/bin/osm-lump-ways-down/cli_args.rs, or run osm-lump-ways-down --help for the format.

--upstreams FILENAME

Each way segment (a 2 point LineString) with the upstream data. With --upstream-output-ends-full, each segement get information about all the end point(s) that the segment eventually flows into.

Assigning to an end point

With --upstream-outpout-biggest-end, each segment will be assigned to the end point that it flows into which has the largest total outflow value.

With --upstream-assign-end-by-tag TAG, it will attempt to follow the same OSM tag upstream from an end point. When used with the name tag, this usually produces maps that are more like what people expect.

--grouped\_ends FILENAME

Ways grouped by downhill and the end point they flow into. The upstream value of each segment isn't included, and it attemptes to generate long LineStrings, grouped by the end point that it flows into.

See the --upstream-outpout-biggest-end & --upstream-assign-end-by-tag TAG arguments.

--ends FILENAME

Points where waterways end.

Ends membership of tagged ways

--ends-membership TAGFILTER

With the --ends-membership TAGFILTER arg every end will have a boolean property called is_in:TAGFILTER which is only true if this end node is a member of a way with this tag filter. This argument can be applied many times. An addditional property is_in_count is an integer of how many of these properties are true.

e.g. --ends-membership natural=coastline will cause each end point to have a JSON property is_in:natural=coastline which is true iff this node is also a member of a way with the natural=coastline tag, false otherwise.

--ends-tag TAG

Every end point will have a JSON property tag:X which is the tag

e.g. --ends-tag name will cause every end point to have a tag:name JSON property, with the name tag from the way which flows into this end point.

Unlike --ends-membership this only uses the OSM ways which are included by the tag filters

ends-full-upstreams

Only with --ends-upstreams argument. File of MultiLineStrings showing, for each end, where the upstreams are. Useful to find why there's a big upstream end somewhere.

--ends-upstreams-min-upstream-m 1e6 --ends-upstreams-max-nodes 1000 is a good tradeoff for calculation speed & file size, which still shows the relevant upstreams

Ends stats CSV

With --ends-csv-file FILENAME.csv, the end points are also written to a CSV file. If the file is already there, new data is appended. Values from --ends-tag (in order) are included in the CSV. Changing the arguments (or order) will make a possibly invalid CSV file.

Only end points with an upstream_m greater than --ends-csv-min-length-m X will be included, and only the largest N from --ends-csv-only-largest-n N

See the doc comment src/bin/osm-lump-ways-down/cli_args.rs, or run osm-lump-ways-down --help for the format.

Loop removal

After the loops are detected, all the edges (way segments) in the loops are contracted together, producing a new graph which is loop-free.

TODOs

This software isn't finished. Lots is being motivated for WaterwayMap.org, a topological analysis of waterways in OpenStreetMap.

External Mentions

TBC

Copyright & Licence

Copyright 2023, 2024, MIT/Apache2.0. Source code is on Github (osm-lump-ways).

The output data file(s) are a Derived Database of the OpenStreetMap database, and hence under the ODbL 1.0 licence, the same as the OpenStreetMap copyright, and contains data © OpenStreetMap contributors

Dependencies

~53MB
~864K SLoC