I have a CSV file I open as a DataFrame. Among the columns is one named geom 1 :

df.loc[:, 'geom 1']
0       {'type': 'Polygon', 'coordinates': [[[2.826589...
1       {'type': 'Polygon', 'coordinates': [[[2.225689...
2       {'type': 'Polygon', 'coordinates': [[[2.225689...
3       {'type': 'MultiPolygon', 'coordinates': [[[[5....
4       {'type': 'Polygon', 'coordinates': [[[3.933055...
2998                                                 None
2999                                                  NaN
3000    {'type': 'Polygon', 'coordinates': [[[5.014937...
3001    {'type': 'Polygon', 'coordinates': [[[4.912995...
3002    {'type': 'Polygon', 'coordinates': [[[4.739631...
Name: geom 1, Length: 3003, dtype: object

#Conditionning the dataframe :
df.replace('None', np.nan, inplace=True)

I would like to convert these strings to geometries in a GeoDataFrame.

After a few resarches I did not find a working solution :

TypeError: Input geometry column must contain valid geometry objects :

df.loc[:, 'geom 1'] = df.loc[:, 'geom 1'].apply(wkt.loads) 
>>> ParseException: Unknown type: '{'TYPE':'
WKTReadingError: Could not create geometry because of errors while reading input.

Convert GeoJSON to GeoPandas GeoDataframe :

df.replace('None', np.nan, inplace=True)
geom = [shape(i) for i in df.loc[:, 'geom 1'].dropna()]
>>> AttributeError: 'str' object has no attribute 'get'

What I succeded was to sent the DataFrame to PostGIS, convert the type of the column geom 1 from text to geometry and use the function ST_FromGeoJson to access the geometry. Since I have many columns geom I want to do this under python in order to merge all the columns geom into a single one named all_geoms.


The shape function from shapely needs a dictionary, so you need to parse the strings to dictionaries before calling shape. If you have these imports

import geopandas as gpd
import json
from shapely.geometry import shape

and you define this function

def parse_geom(geom_str):
    if not geom_str:
        return None
    return shape(json.loads(geom_str))

Then this should give you a GeoDataframe

df["geom 1"] = df["geom 1"].apply(parse_geom)
gdf = gpd.GeoDataFrame(df, geometry="geom 1")

EDIT: Fixed wrong spelling of variable parse_geom

| improve this answer | |
  • I get this Error : JSONDecodeError: Expecting property name enclosed in double quotes: line 1 column 2 (char 1) – Basile Apr 23 at 16:46
  • @Basile: It is because of the single quotes around property names in the json. You can try to replace the single quotes. The last line in the parse_geom function becomes return shape(json.loads(geom_str.replace("'", '"'))). This can be a bit dangerous if the values of the json object contains double quotes. I don't think it is a problem in your case. Another option is to use the ast module. First import it with import ast. The last line in parse_geom is then: return shape(ast.literal_eval(geom_str)) – Dataform Apr 24 at 6:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.