WIP: link between request and response

python-package/lets_plot/geo_data/gis/response.py

@@ -36,13 +36,13 @@ def __init__(self, geometry: Union[Multipolygon, Polygon, GeoPoint]):
 
 
 class GeocodedFeature:
- def __init__(self, query: str, id: str, name: str,
+ def __init__(self, query: Optional[str], id: str, name: str,
  highlights: Optional[List[str]],
  boundary: Optional[Boundary],
  centroid: Optional[GeoPoint],
  limit: Optional[GeoRect],
  position: Optional[GeoRect]):
- assert_type(query, str)
+ assert_optional_type(query, str)
  assert_type(id, str)
  assert_type(name, str)
  assert_optional_list_type(highlights, str)
@@ -51,7 +51,7 @@ def __init__(self, query: str, id: str, name: str,
  assert_optional_type(limit, GeoRect)
  assert_optional_type(position, GeoRect)
 
- self.query: str = query
+ self.query: Optional[str] = query # optional for requests like find all countries
  self.id: str = id
  self.name: str = name
  self.highlights: Optional[List[str]] = highlights
@@ -78,11 +78,11 @@ def __init__(self, message: str):
  self.message: str = message
 
 class Answer:
- def __init__(self, query: str, features: List[GeocodedFeature]):
- assert_type(query, str)
+ def __init__(self, query: Optional[str], features: List[GeocodedFeature]):
+ assert_optional_type(query, str)
  assert_list_type(features, GeocodedFeature)
 
- self.query: str = query
+ self.query: Optional[str] = query
  self.features: List[GeocodedFeature] = features
 
 

python-package/lets_plot/geo_data/regions.py

@@ -49,19 +49,11 @@ def select_not_empty_name(feature: GeocodedFeature) -> str:
  return feature.name if feature.query is None or feature.query == '' else feature.query
 
 
-def arrange_queries(features: List[GeocodedFeature], queries: List[RegionQuery]) -> List[Optional[RegionQuery]]:
- res = []
- for feature in features:
- request: Optional[RegionQuery] = None
-
- for query in queries:
- if query.request == feature.query:
- request = query
- break
-
- res.append(request)
-
- return res
+def zip_answers(queries, answers):
+ if len(queries) > 0:
+ return zip(queries, answers)
+ elif len(queries) == len(answers):
+ return zip([None] * len(answers), answers)
 
 
 class PlacesDataFrameBuilder:
@@ -72,11 +64,10 @@ def __init__(self):
  self._state: List[Optional[str]] = []
  self._country: List[Optional[str]] = []
 
- def append_row(self, request: str, found_name: str, queries: List[Optional[RegionQuery]], parent_row: int):
+ def append_row(self, request: str, found_name: str, query: Optional[RegionQuery]):
  self._request.append(request)
  self._found_name.append(found_name)
 
- query: Optional[RegionQuery] = queries[parent_row]
  if query is None:
  self._county.append(MapRegion.name_or_none(None))
  self._state.append(MapRegion.name_or_none(None))
@@ -105,14 +96,15 @@ def build_dict(self):
 
 
  @abstractmethod
- def to_data_frame(self, features: List[GeocodedFeature], queries: List[RegionQuery] = []) -> DataFrame:
+ def to_data_frame(self, answers: List[Answer], queries: List[RegionQuery] = []) -> DataFrame:
  raise ValueError('Not implemented')
 
 
 class Regions(CanToDataFrame):
  def __init__(self, level_kind: LevelKind, answers: List[Answer], highlights: bool = False, queries: List[RegionQuery] = []):
  assert_list_type(answers, Answer)
  assert_list_type(queries, RegionQuery)
+ assert len(queries) == 0 or len(answers) == len(queries)
 
  try:
  import geopandas
@@ -140,7 +132,12 @@ def to_map_regions(self):
  return [MapRegion.place(feature.id, feature.query, self._level_kind) for feature in self._geocoded_features]
 
  def as_list(self) -> List['Regions']:
- return [Regions(self._level_kind, [answer], self._highlights) for answer in self._answers]
+ if len(self._queries) == 0:
+ return [Regions(self._level_kind, [answer], self._highlights) for answer in self._answers]
+
+ assert len(self._queries) == len(self._answers)
+ return [Regions(self._level_kind, [answer], self._highlights, [query]) for query, answer in zip(self._queries, self._answers)]
+
 
  def unique_ids(self) -> List[str]:
  seen = set()
@@ -276,11 +273,9 @@ def to_data_frame(self) -> DataFrame:
  data[DF_ID] = [feature.id for feature in self._geocoded_features]
 
  # for us-48 queries doesnt' count
- queries: List[Optional[RegionQuery]] = arrange_queries(self._geocoded_features, self._queries)
-
- for i in range(len(self._geocoded_features)):
- feature = self._geocoded_features[i]
- places.append_row(select_not_empty_name(feature), feature.name, queries, i)
+ for query, answer in zip_answers(self._queries, self._answers):
+ for feature in answer.features:
+ places.append_row(select_not_empty_name(feature), feature.name, query)
 
  data = {**data, **places.build_dict()}
 
@@ -303,7 +298,7 @@ def _execute(self, request_builder: RequestBuilder, df_converter):
 
  self._join_payload(features)
 
- return df_converter.to_data_frame(self._geocoded_features, self._queries)
+ return df_converter.to_data_frame(self._answers, self._queries)
 
  def _request_builder(self, payload_kind: PayloadKind) -> RequestBuilder:
  assert_type(payload_kind, PayloadKind)

python-package/lets_plot/geo_data/to_geo_data_frame.py

@@ -5,8 +5,8 @@
 from pandas import DataFrame
 from shapely.geometry import box
 
-from lets_plot.geo_data import PlacesDataFrameBuilder, arrange_queries, select_not_empty_name, DF_REQUEST, DF_FOUND_NAME, abstractmethod
-from lets_plot.geo_data.gis.response import GeocodedFeature, GeoRect, Boundary, Multipolygon, Polygon, GeoPoint
+from lets_plot.geo_data import PlacesDataFrameBuilder, zip_answers, select_not_empty_name, DF_REQUEST, DF_FOUND_NAME, abstractmethod
+from lets_plot.geo_data.gis.response import Answer, GeocodedFeature, GeoRect, Boundary, Multipolygon, Polygon, GeoPoint
 from lets_plot.geo_data.gis.request import RegionQuery
 
 ShapelyPoint = shapely.geometry.Point
@@ -40,21 +40,22 @@ def __init__(self):
  self._lonmax: List[float] = []
  self._latmax: List[float] = []
 
- def to_data_frame(self, features: List[GeocodedFeature], queries: List[RegionQuery] = []) -> DataFrame:
+ def to_data_frame(self, answers: List[Answer], queries: List[RegionQuery] = []) -> DataFrame:
  places = PlacesDataFrameBuilder()
- queries: List[Optional[RegionQuery]] = arrange_queries(features, queries)
-
- for i in range(len(features)):
- feature = features[i]
- rects: List[GeoRect] = self._read_rect(feature)
- for rect in rects:
- places.append_row(request=select_not_empty_name(feature), found_name=feature.name, queries=queries, parent_row=i)
- self._lonmin.append(rect.min_lon)
- self._latmin.append(rect.min_lat)
- self._lonmax.append(rect.max_lon)
- self._latmax.append(rect.max_lat)
+
+
+ for query, answer in zip_answers(queries, answers):
+ for feature in answer.features:
+ rects: List[GeoRect] = self._read_rect(feature)
+ for rect in rects:
+ places.append_row(request=select_not_empty_name(feature), found_name=feature.name, query=query)
+ self._lonmin.append(rect.min_lon)
+ self._latmin.append(rect.min_lat)
+ self._lonmax.append(rect.max_lon)
+ self._latmax.append(rect.max_lat)
+
  geometry = [RectGeoDataFrame.limit2geometry(lmt[0], lmt[1], lmt[2], lmt[3]) for lmt in
- zip(self._lonmin, self._latmin, self._lonmax, self._latmax)]
+  zip(self._lonmin, self._latmin, self._lonmax, self._latmax)]
  return _create_geo_data_frame(places.build_dict(), geometry=geometry)
 
 
@@ -79,15 +80,14 @@ def __init__(self):
  self._lons: List[float] = []
  self._lats: List[float] = []
 
- def to_data_frame(self, features: List[GeocodedFeature], queries: List[RegionQuery] = []) -> DataFrame:
+ def to_data_frame(self, answers: List[Answer], queries: List[RegionQuery] = []) -> DataFrame:
  places = PlacesDataFrameBuilder()
- queries: List[Optional[RegionQuery]] = arrange_queries(features, queries)
 
- for i in range(len(features)):
- feature = features[i]
- places.append_row(request=select_not_empty_name(feature), found_name=feature.name, queries=queries, parent_row=i)
- self._lons.append(feature.centroid.lon)
- self._lats.append(feature.centroid.lat)
+ for query, answer in zip_answers(queries, answers):
+ for feature in answer.features:
+  places.append_row(request=select_not_empty_name(feature), found_name=feature.name, query=query)
+  self._lons.append(feature.centroid.lon)
+  self._lats.append(feature.centroid.lat)
 
  geometry = [ShapelyPoint(pnt[0], pnt[1]) for pnt in zip(self._lons, self._lats)]
  return _create_geo_data_frame(places.build_dict(), geometry)
@@ -97,15 +97,14 @@ class BoundariesGeoDataFrame:
  def __init__(self):
  super().__init__()
 
- def to_data_frame(self, features: List[GeocodedFeature], queries: List[RegionQuery] = []) -> DataFrame:
+ def to_data_frame(self, answers: List[Answer], queries: List[RegionQuery] = []) -> DataFrame:
  places = PlacesDataFrameBuilder()
- queries: List[Optional[RegionQuery]] = arrange_queries(features, queries)
 
  geometry = []
- for i in range(len(features)):
- feature = features[i]
- places.append_row(request=select_not_empty_name(feature), found_name=feature.name, queries=queries, parent_row=i)
- geometry.append(self._geo_parse_geometry(feature.boundary))
+ for query, answer in zip_answers(queries, answers):
+ for feature in answer.features:
+  places.append_row(request=select_not_empty_name(feature), found_name=feature.name, query=query)
+  geometry.append(self._geo_parse_geometry(feature.boundary))
 
  return _create_geo_data_frame(places.build_dict(), geometry=geometry)
 

python-package/test/geo_data/test_georect.py

@@ -4,7 +4,7 @@
 import pytest
 from pandas import DataFrame
 
-from lets_plot.geo_data.gis.response import GeoRect, FeatureBuilder
+from lets_plot.geo_data.gis.response import Answer, GeoRect, FeatureBuilder
 from lets_plot.geo_data import DF_FOUND_NAME, DF_ID, DF_REQUEST
 from lets_plot.geo_data.to_geo_data_frame import LimitsGeoDataFrame
 
@@ -47,12 +47,15 @@ def make_rect(lon_min: float, lon_max: float) -> GeoRect:
 def data_frame(r: GeoRect):
  return LimitsGeoDataFrame().to_data_frame(
  [
- FeatureBuilder() \
- .set_id(ID) \
- .set_query(NAME) \
- .set_name(FOUND_NAME) \
- .set_limit(r) \
- .build_geocoded()
+ Answer(NAME, [
+ FeatureBuilder() \
+ .set_id(ID) \
+ .set_query(NAME) \
+ .set_name(FOUND_NAME) \
+ .set_limit(r) \
+ .build_geocoded()
+ ]
+ )
  ]
  )
 

python-package/test/geo_data/test_integration_new_api.py

@@ -113,5 +113,5 @@ def test_drop_not_found_with_namesakes():
  request=['jefferson', 'jefferson'],
  found_name=['Jefferson County', 'Jefferson County'],
  state=['alabama', 'arkansas'],
- country=['usa', 'usa', 'usa']
+ country=['usa', 'usa']
  )