How to create a Feature Collection from a GeoJSON - gis

Question

I have a Feature Collection of Polygons and MultiPolygons and I have to first write it in a temporary file to then load it with geopandas.GeoDataFrame.from_file(tmp_json_file), I'm looking for a way to do it without the temporary file. I've tried to use geopandas.GeoDataFrame.from_feature(), it works pretty well for Feature Collection of simple Polygon but i can't make it work for Feature Collection of Polygons and MultiPolygons, I was thinking about doing something like below, but it's not working yet.
features_collection = []
for feature in json_data['features']:
tmp_properties = {'id': feature['properties']['id']}
if is_multipolygon (feature):
tmp = Feature(geometry=MultiPolygon((feature['geometry']['coordinates'])), properties=tmp_properties)
else:
Feature(geometry=Polygon((feature['geometry']['coordinates'])), properties=tmp_properties)
features_collection.append(tmp)
collection = FeatureCollection(features_collection)
return geopandas.GeoDataFrame.from_features(collection['features'])
The GeoJSON is taken from an API, returning territory (some territory are modelized by a single polygon, other by a set of polygons (formatted as a MultiPolygon).
The GeoJSON are structured as follow : http://pastebin.com/PPdMUGkY
I'm getting the following error from the function above :
Traceback (most recent call last):
File "overlap.py", line 210, in <module>
print bdv_json_to_geodf(contours_bdv)
File "overlap.py", line 148, in json_to_geodf
return geopandas.GeoDataFrame.from_features(collection['features'])
File "/Library/Python/2.7/site-packages/geopandas/geodataframe.py", line 179, in from_features
d = {'geometry': shape(f['geometry'])}
File "/Library/Frameworks/GEOS.framework/Versions/3/Python/2.7/site-packages/shapely/geometry/geo.py", line 40, in shape
return MultiPolygon(ob["coordinates"], context_type='geojson')
File "/Library/Frameworks/GEOS.framework/Versions/3/Python/2.7/site-packages/shapely/geometry/multipolygon.py", line 64, in __init__
self._geom, self._ndim = geos_multipolygon_from_py(polygons)
File "/Library/Frameworks/GEOS.framework/Versions/3/Python/2.7/site-packages/shapely/geometry/multipolygon.py", line 138, in geos_multipolygon_from_py
N = len(ob[0][0][0])
TypeError: object of type 'float' has no len()

Answer 1

For me this works if I just feed the json_data features to GeoDataFrame.from_features:
In [17]: gdf = geopandas.GeoDataFrame.from_features(json_data['features'])
In [18]: gdf.head()
Out[18]:
geometry id
0 (POLYGON ((-0.58570861816406 44.810461337462, ... 2
1 (POLYGON ((-0.5851936340332 44.816550206151, -... 1
2 POLYGON ((-0.58805465698242 44.824018340447, -... 5
3 POLYGON ((-0.59412002563477 44.821664359038, -... 9
4 (POLYGON ((-0.58502197265625 44.817159057661, ... 12
The resulting GeoDataFrame has a mixture of Polygons and MultiPolygons like in the input data:
In [19]: gdf.geom_type.head()
Out[19]:
0 MultiPolygon
1 MultiPolygon
2 Polygon
3 Polygon
4 MultiPolygon
dtype: object
I tried this with GeoPandas 0.2, shapely 1.5.15, pandas 0.18.1 on Windows.