Read GeoJSON object in a CSV file column and transform it to geometry

Question

I have a column ('geo_shape') that contains GeoJSON object type of linestring geometry and, so far, I didn't manage to read and transform it to a GeoPandas/shapefile type of geometry.

Here is a way to do it with QGIS Python module: Reading GeoJSON-String from CSV as Geometry in QGIS

I have tried this method also : CSV containing GeoJSON format geometries to DataFrame

All I get is this JSONDecodeError: Extra data: line 1 column 12 (char 11) Which I didn't manage to solve yet.

Here how the data look like (it's a type 'object' not a proper dictionary):

{"coordinates": [[5.807033273966051, 43.19284592374117], [5.807314867862657, 43.192757356667016]], "type": "LineString"}

When trying this method: CSV containing GeoJSON format geometries to DataFrame

import ast

def parse_geom(geom_str):
    try:
        return shape(json.loads(geom_str))
    except (TypeError, AttributeError):  # Handle NaN and empty strings
        return None


df["geo_shape"] = df["geo_shape"].apply(parse_geom)
gdf = gpd.GeoDataFrame(df, geometry="geo_shape")
gdf.head()

All I get is a None value :

I also tried this method: https://stackoverflow.com/questions/68820085/how-to-convert-geojson-to-shapely-polygon

geo: dict = df["geo_shape"] linestring: LineString = shape(geo)

And this too: How can I use GeoPandas to read a string containing GeoJSON content into a GeoDataFrame?

gdf = gpd.read_file(df["geo_shape"], driver='GeoJSON') gdf.head()

The result is this: TypeError: expected string or bytes-like object

Does anyone has an idea how to deal with it with GeoPandas and GeoJSON?

Answer 1

from shapely import Point
import json

def make_point(json_feature):
    dict_json = json.loads(json_feature)
    pnt = Point(dict_json['coordinates'][0][0],
                dict_json['coordinates'][0][1])

   return pnt

infra_gaz['new_pnt_geom'] = infra_gaz['geo_shape'].apply(
                   make_point, axis=1)

Can make shapely point with custom function and use apply to create new point column.

Answer 2

shapely.geometry.shape function can create geometries from strings formatted like yours:

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

data = '{"coordinates": [[5.807033273966051, 43.19284592374117], [5.807314867862657, 43.192757356667016]], "type": "LineString"}'
df = gpd.pd.DataFrame(data=[data], columns=["col1"])
df["geometry"] = df.apply(lambda x: shape(ast.literal_eval(x.col1)), axis=1)
df = gpd.GeoDataFrame(data=df, geometry=df["geometry"], crs=4326)

df.geometry
Out[100]: 
0    LINESTRING (5.80703 43.19285, 5.80731 43.19276)
Name: geometry, dtype: geometry