Strict version of some dependencies and dataframe helper

pyproject.toml

@@ -12,8 +12,8 @@ packages = [{include = "utils"}]
 python = ">=3.9, <3.13"
 
 pnnl-buildingid = {git = "https://github.com/seed-platform/buildingid.git", rev = "master"}
-usaddress = "^0.5.10"
-street-address = "^0.4.0"
+usaddress = "0.5.10"
+street-address = "0.4.0"
 geopandas = "^0.14.3"
 mercantile = "^1.2.1"
 python-dotenv = "^1.0.1"

utils/normalize_address.py

@@ -7,7 +7,7 @@
 import re
 
 import usaddress
-from streetaddress import StreetAddressFormatter
+from streetaddress.streetaddress import StreetAddressFormatter
 
 
 def _normalize_subaddress_type(subaddress_type):

utils/ubid.py

@@ -5,6 +5,7 @@
 """
 
 from buildingid.code import decode, encode
+from geopandas import GeoDataFrame
 from openlocationcode.openlocationcode import PAIR_CODE_LENGTH_
 from shapely.geometry import Point, Polygon
 
@@ -36,3 +37,42 @@ def bounding_box(ubid: str) -> Polygon:
 def centroid(ubid: str) -> Point:
     code_area = decode(ubid)
     return Point(code_area.centroid.longitudeCenter, code_area.centroid.latitudeCenter)
+
+
+def add_ubid_to_geodataframe(
+    gdf: GeoDataFrame,
+    footprint_column: str = "geometry",
+    additional_ubid_columns_to_create: list[str] = ["ubid_centroid", "ubid_bbox"],
+) -> GeoDataFrame:
+    """Add UBID and related fields to a GeoDataFrame
+
+    Args:
+        gdf (GeoDataFrame): Incoming data frame
+        footprint_column (str, optional): Where is the footprint polygon defined. Defaults to "geometry".
+        additional_ubid_columns_to_create (list[str], optional): Which fields to create.
+            Defaults to ['ubid_centroid', 'ubid_bbox'].
+
+    Returns:
+        GeoDataFrame: New data frame with UBID and related fields
+    """
+    # only apply to the fields that have footprints
+    filter_str = gdf[footprint_column].notna()
+
+    # make sure the columns are created
+    if "ubid" not in gdf.columns:
+        gdf["ubid"] = None
+
+    for column in additional_ubid_columns_to_create:
+        if column not in gdf.columns:
+            gdf[column] = None
+
+    # UBID is always calculated and stored in 'ubid'
+    gdf.loc[filter_str, "ubid"] = gdf[filter_str].apply(lambda x: encode_ubid(x[footprint_column]), axis=1)
+
+    if "ubid_centroid" in additional_ubid_columns_to_create:
+        gdf.loc[filter_str, "ubid_centroid"] = gdf[filter_str].apply(lambda x: centroid(x["ubid"]), axis=1)
+
+    if "ubid_bbox" in additional_ubid_columns_to_create:
+        gdf.loc[filter_str, "ubid_bbox"] = gdf[filter_str].apply(lambda x: bounding_box(x["ubid"]), axis=1)
+
+    return gdf