targetgeom_config.yaml

target_geom1:
  geom_name: "gb1900"
  gis_file: "../data/input/target_geoms/gb1900/gb1900_gazetteer_complete_july_2018.csv"
  gis_uid_field: "pin_id"
  gis_geocode_field: "final_text"
  gis_lat_long: True
  gis_long_field: "osgb_east"
  gis_lat_field: "osgb_north"
  gis_projection: "EPSG:27700"
  gis_read_params:
    encoding: "utf-16"
    sep: ","
    usecols: ["pin_id", "final_text", "osgb_east", "osgb_north", ]
    # nrows: 20000

  gis_field_to_clean: "final_text"
  gis_convert_non_ascii: True
  gis_standardisation_file: "../configuration/standardisation_files/gb1900_standardisation.json"
  gis_min_len: 5
  cleaned_field_suffix: "_alt"
  dedup: True
  dedup_max_points: 2
  dedup_max_distance_between_points: 1000



  item_per_unit_uid: "street_uid"

  gis_write_params:
    sep: "\t"
    encoding: "utf-8"
    index: False

target_geom2:
  geom_name: "osopenroads"
  gis_file: "../data/input/target_geoms/osopenroads/osopenroads.shp"
  gis_uid_field: "nameTOID"
  gis_convert_non_ascii: True
  gis_standardisation_file: "../configuration/standardisation_files/osopenroads_standardisation.json"
  gis_geocode_field: "name1"
  gis_field_to_clean: "name1"
  cleaned_field_suffix: "_alt"

  gis_read_params:
    engine: "pyogrio"
    columns: ["nameTOID", "name1", ]
    # max_features: 10000

  item_per_unit_uid: "street_uid"

  gis_write_params:
    sep: "\t"
    encoding: "utf-8"
    index: False