include csv in package

MANIFEST.in

+include rules/gem_occupancy/data/*.csv
+include LICENSE

rabotnikobm/instance.py

@@ -24,7 +24,7 @@ import argparse
 from rabotnik import Rabotnik, Assembly
 from rabotnik.bus import MessageBus
 
-from rabotnikobm.rules.gem_occupancy.mapping import OccupancyMapper
+from rabotnikobm.rules.gem_occupancy.get_building_occupancy import GetBuildingOccupancy
 
 from .rules import GetBuilding
 from .rules import GetFloorspace
@@ -62,15 +62,10 @@ async def start_rabotnik_obm(message_bus):
     await message_bus.subscribe("building", rules.run)
 
 
-async def start_rabotnik_gem_occupancy(
-    message_bus, storage_consumer, storage_contributor, n_processes_max
-):
-
-    fn_mapping = "building_and_PoIs_tags.csv"
-    occupancy_mapper = OccupancyMapper.from_csv(fn=fn_mapping)
+async def start_rabotnik_gem_occupancy(message_bus, storage_consumer, _, n_processes_max):
 
     rules = [
-        GetBuildingOccupancy(storage_consumer, occupancy_mapper=occupancy_mapper),
+        GetBuildingOccupancy(storage_consumer),
     ]
     rules = Assembly(rules=rules, n_processes_max=n_processes_max)
 
@@ -82,9 +77,13 @@ async def start_rabotnik(args):
 
     message_bus = await connected_message_bus(args.config_message_bus)
 
-    # await start_rabotnik_obm(
-    #     message_bus, storage_consumer, storage_contributor, args.n_processes_max
-    # )
+    rabotnik_args = (message_bus, storage_consumer, storage_contributor, args.n_processes_max)
+
+    if args.target in ("all", "obm"):
+        asyncio.create_task(start_rabotnik_obm(*rabotnik_args))
+
+    if args.target in ("all", "gem"):
+        asyncio.create_task(start_rabotnik_gem_occupancy(*rabotnik_args))
 
 
 def main():
@@ -108,6 +107,15 @@ def main():
         "--start-celery-worker",
         action="store_true",
     )
+
+    parser.add_argument(
+        "-t",
+        "--target",
+        default="all",
+        type=str,
+        help="Process to start (defaults to 'all')",
+    )
+
     args = parser.parse_args()
 
     default_log_level = logging.WARNING

rabotnikobm/rules/gem_occupancy/get_building_occupancy.py

@@ -109,12 +109,12 @@ class RulesOneUniqueSubGroup:
 class GetBuildingOccupancy(Rule):
     """A rule to map OSM tags to building occupancies."""
 
-    def __init__(self, storage: StorageBase, occupancy_mapper: OccupancyMapper):
+    def __init__(self, storage: StorageBase):
         self.storage = storage
 
         self.mappers = [
-            occupancy_mapper.landuse_mapper(),
-            occupancy_mapper.building_poi_mapper(),
+            OccupancyMapper.landuse_mapper(),
+            OccupancyMapper.building_poi_mapper(),
         ]
 
         self.candidates = [
@@ -130,14 +130,15 @@ class GetBuildingOccupancy(Rule):
             f"SELECT tags FROM osm_building_relations WHERE osm_id={building_id} AND index=0"
         )
 
-        # Mapping to GEM taxonomies
+        # Mapping to GEM taxonomy strings
         occupancies = []
         for mapper in self.mappers:
             occupancies.extend(mapper.apply(tags))
 
+        # Convert derived occupancies to GEMTags
         occupancies = TagStatistics.from_strings(occupancies)
 
-        # Apply rules to find ultimate taxonomy
+        # Apply rules to find final taxonomy
         occupancy = apply_rules(self.candidates, occupancies)
 
         logger.debug("occupancies %s: %s", building_id, occupancy)

setup.py

@@ -33,7 +33,8 @@ setup(
         "tests": tests_require,
         "linters": linters_require,
     },
-    packages=find_packages(),
     entry_points={"console_scripts": ["rabotnikobm = rabotnikobm.instance:main"]},
     python_requires=">=3.6",
+    packages=find_packages(),
+    include_package_data=True,
 )