Start extracting a cycle path layer from OSM

README.md

@@ -85,6 +85,7 @@ ATIP can display extra contextual layers:
   - Points of interest, like schools, hospitals, sports centres, etc
   - Roads with bus lanes and bus routes
   - Cycle parking
+  - Existing cycle paths
 - the [Major Road Network](https://www.data.gov.uk/dataset/95f58bfa-13d6-4657-9d6f-020589498cfd/major-road-network)
 - Boundaries
   - Parliament constituency boundaries, from [OS Boundary-Line](https://www.ordnancesurvey.co.uk/products/boundary-line)
@@ -107,7 +108,7 @@ is a single GeoJSON file if it's small enough, or
 To run this:
 
 1.  Get `england-latest.osm.pbf` from Geofabrik. The `split_uk_osm.sh` script above does this.
-2.  Run `cd layers; ./generate_layers.py --osm_input=../england-latest.osm.pbf --schools --hospitals --mrn --parliamentary_constituencies --combined_authorities --local_authority_districts --local_planning_authorities --sports_spaces --railway_stations --bus_routes --cycle_parking`
+2.  Run `cd layers; ./generate_layers.py --osm_input=../england-latest.osm.pbf --schools --hospitals --mrn --parliamentary_constituencies --combined_authorities --local_authority_districts --local_planning_authorities --sports_spaces --railway_stations --bus_routes --cycle_parking --cycle_paths`
 3.  Pick an arbitrary version number, and upload the files: `for x in output/*; do aws s3 cp --dry $x s3://atip.uk/layers/v1/; done`
 
 If you're rerunning the script for the same output, you may need to manually delete the output files from the previous run.
@@ -116,6 +117,8 @@ You can debug a PMTiles file using <https://protomaps.github.io/PMTiles>.
 
 There's a manual step required to generate `--wards`, `--census_output_areas`, and `--imd`. See the comment in the code.
 
+For `--cycle_paths`, you'll need about 20GB of RAM, until we switch to a streaming JSON parser.
+
 ### One-time cloud setup for PMTiles
 
 Currently we're using S3 and Cloudfront to host files generated by this repo.

layers/cycle_paths.py

@@ -0,0 +1,149 @@
+from utils import *
+
+
+def makeCyclePaths(osm_input):
+    if not osm_input:
+        raise Exception("You must specify --osm_input")
+
+    filename = "cycle_paths"
+    tmp = f"tmp_{filename}"
+    ensureEmptyTempDirectoryExists(tmp)
+
+    run(
+        [
+            "osmium",
+            "tags-filter",
+            osm_input,
+            "w/highway",
+            "-o",
+            f"{tmp}/cycle_paths.osm.pbf",
+        ]
+    )
+    gjPath = f"{tmp}/cycle_paths.geojson"
+    convertPbfToGeoJson(
+        f"{tmp}/cycle_paths.osm.pbf", gjPath, "linestring", includeOsmID=True
+    )
+
+    cleanUpGeojson(
+        gjPath, getProps, filterFeatures=lambda f: getProps(f["properties"]) != None
+    )
+
+    convertGeoJsonToPmtiles(f"{tmp}/cycle_paths.geojson", "output/cycle_paths.pmtiles")
+
+
+# If this has some kind of cycle-path, returns a dictionary with:
+#
+# - kind = track | lane | shared_use_segregated | shared_use_unsegregated
+# - osm_id
+# - direction = one-way | two-way | unknown
+#   (For cyclist traffic; unrelated to whether the path is a contraflow on a one-way street)
+# - width = number in meters | unknown
+#
+# If there's no cycle path, returns None
+def getProps(props):
+    if props.get("highway") == "cycleway":
+        kind = "track"
+        # Is it shared-use? Note "foot" may be missing (https://www.openstreetmap.org/way/849651126)
+        if props.get("foot") in ["yes", "designated"] or props.get("segregated"):
+            kind = getSharedUseKind(props)
+
+        return {
+            "kind": kind,
+            "osm_id": props["@id"],
+            "width": getSeparateWayWidth(props),
+            "direction": getSeparateWayDirection(props),
+        }
+    elif hasLane(props):
+        return {
+            "kind": "lane",
+            "osm_id": props["@id"],
+            "width": getLaneWidth(props),
+            "direction": getLaneDirection(props),
+        }
+    elif props["highway"] in ["footway", "pedestrian", "path", "track"]:
+        if props.get("bicycle") not in ["yes", "designated"]:
+            return None
+
+        return {
+            "kind": getSharedUseKind(props),
+            "osm_id": props["@id"],
+            "width": getSeparateWayWidth(props),
+            "direction": getSeparateWayDirection(props),
+        }
+    else:
+        return None
+
+
+def getSeparateWayWidth(props):
+    # TODO Handle suffixes / units
+    if props.get("width"):
+        return props["width"]
+    if props.get("est_width"):
+        return props["est_width"]
+    return "unknown"
+
+
+def getSeparateWayDirection(props):
+    oneway = props.get("oneway")
+    if oneway == "yes":
+        return "one-way"
+    elif oneway == "no":
+        return "two-way"
+    else:
+        return "unknown"
+
+
+def valueIndicatesLane(value):
+    return value and value not in ["no", "separate", "share_busway"]
+
+
+def hasLane(props):
+    # TODO Handle bicycle:lanes
+    for suffix in ["", ":left", ":right", ":both"]:
+        if valueIndicatesLane(props.get("cycleway" + suffix)):
+            return True
+    return False
+
+
+# If there are two lanes, the result could capture one or both of the lanes
+def getLaneWidth(props):
+    for suffix in ["", ":left", ":right", ":both"]:
+        for width in [":width", ":est_width"]:
+            value = props.get("cycleway" + suffix + width)
+            if value:
+                return value
+    return "unknown"
+
+
+def getLaneDirection(props):
+    if valueIndicatesLane(props.get("cycleway:both")):
+        return "two-way"
+    if valueIndicatesLane(props.get("cycleway:left")) and valueIndicatesLane(
+        props.get("cycleway:right")
+    ):
+        return "two-way"
+    if props.get("oneway:bicycle") == "no":
+        # TODO On one-way roads, this might just mean cyclists can travel both
+        # directions, but there's only a dedicated lane in one
+        return "two-way"
+    if (
+        props.get("cycleway:left:oneway") == "no"
+        or props.get("cycleway:right:oneway") == "no"
+    ):
+        return "two-way"
+
+    # TODO opposite_track
+    # TODO cycleway=* + oneway=yes
+    # TODO bicycle:lanes
+    return "unknown"
+
+
+def getSharedUseKind(props):
+    segregated = props.get("segregated")
+    if segregated == "yes":
+        return "shared_use_segregated"
+    elif segregated == "no":
+        return "shared_use_unsegregated"
+    else:
+        # Pessimistically assume unsegregated if unknown
+        return "shared_use_unsegregated"

layers/generate_layers.py

@@ -6,6 +6,7 @@
 from utils import *
 import census
 import boundaries
+import cycle_paths
 import osm
 
 
@@ -38,6 +39,7 @@ def main():
         help="Path to the manually downloaded Indices_of_Multiple_Deprivation_(IMD)_2019.geojson",
         type=str,
     )
+    parser.add_argument("--cycle_paths", action="store_true")
     # Inputs required for some outputs
     parser.add_argument(
         "-i", "--osm_input", help="Path to england-latest.osm.pbf file", type=str
@@ -109,6 +111,10 @@ def main():
         made_any = True
         census.makeIMD(args.imd)
 
+    if args.cycle_paths:
+        made_any = True
+        cycle_paths.makeCyclePaths(args.osm_input)
+
     if not made_any:
         print(
             "Didn't create anything. Call with --help to see possible layers that can be created"

layers/osmium_with_ids.cfg

@@ -0,0 +1,17 @@
+{
+    "attributes": {
+        "type": false,
+        "id": true,
+        "version": false,
+        "changeset": false,
+        "timestamp": false,
+        "uid": false,
+        "user": false,
+        "way_nodes": false,
+    },
+    "format_options": {},
+    "linear_tags": true,
+    "area_tags": true,
+    "exclude_tags": [],
+    "include_tags": [],
+}

layers/utils.py

@@ -14,7 +14,14 @@ def ensureEmptyTempDirectoryExists(directoryName):
     os.makedirs(directoryName, exist_ok=True)
 
 
-def convertPbfToGeoJson(pbfPath, geojsonPath, geometryType):
+def convertPbfToGeoJson(pbfPath, geojsonPath, geometryType, includeOsmID=False):
+    config = []
+    if includeOsmID:
+        # Created by `osmium export --print-default-config` and changing `id`
+        # TODO We can do this with a CLI flag with newer osmium, but it's not
+        # easy to install on Ubuntu 20
+        config = ["--config", "osmium_with_ids.cfg"]
+
     run(
         [
             "osmium",
@@ -24,6 +31,7 @@ def convertPbfToGeoJson(pbfPath, geojsonPath, geometryType):
             "-o",
             geojsonPath,
         ]
+        + config
     )