1

我正在尝试将GRASS GIS矢量图层读入GeoPandas Dataframe.

因为我无法在查看相关问题GRASS GIS中启用（只读）驱动程序Fiona

我做了一个“hackish”方法（有效）将GRASS GIS矢量图层读入GeoPandas Dataframe：


import os
from osgeo import ogr
import pandas as pd
from shapely import wkt
import geopandas as gpd


def grass2gpd(layername, grassdb, location_name, mapset):
    datafile = os.path.join(GRASSDB, LOCATION_NAME, MAPSET, 'vector', layername, 'head')
    driver = ogr.GetDriverByName('GRASS')
    # here the data is read in readonly mode, but also the GDAL GRASS driver has no write capabilities
    # I am not sure at the end of the method on how to properly close the datasource (and, actually, if I have to ... )  
    dataSource = driver.Open(datafile, 0)     
    layer = dataSource.GetLayer() 
    srs = layer.GetSpatialRef().ExportToWkt()
    lyrDefn = layer.GetLayerDefn() 
    fieldnames = []
    for i in range( lyrDefn.GetFieldCount() ):  
        fieldnames.append(lyrDefn.GetFieldDefn(i).GetName() )
    # hack to avoid to call `layer.ResetReading()` and iterate again over the features
    #
    # I first build a list of dictionaries
    # each element of the list is of the form:
    # {geom: WKT_Geometry
    #  attr: {field_1: val, field_2: val, ..., field_N: val}}
    # So the attr key is a dictionary itself
    # the nested loop first get the feature then create a dictionary of attributes looping over the list of fields
    # 
    wktgeom = [{'geom':feature.GetGeometryRef().ExportToWkt(), 
                'attr':{i:feature.GetField(i) for i in fieldnames}} for feature in layer] 
    # At this point I should close or unlink the datasource, but I can't find the right method to do it
    #
    # Create a dataframe from the list of dictionaries
    #
    df = pd.DataFrame(wktgeom)
    # convert the WKT string to a shapely WKT object
    # concatenate a Geometry dataframe with an attribute dataframe 
    df_geom = pd.concat([df['geom'].apply(wkt.loads),  
                        pd.DataFrame(list(df['attr'].values))], 
                       axis=1, sort=False)
    # transform the pandas dataframe into a geopandas dataframe
    gdf = gpd.GeoDataFrame(df_geom, geometry='geom', crs=srs)
    return gdf

运行它例如：


GRASSDB="/home/epinux/Data/grassdata"
LOCATION_NAME="lonlat"
MAPSET="PERMANENT"

layername="img_left_filteredBS"

gdf = grass2gpd(layername=layername, 
                grassdb=GRASSDB, 
                location_name=LOCATION_NAME, 
                 mapset=MAPSET)
type(gdf)

# returns
# geopandas.geodataframe.GeoDataFrame

geopandas.geodataframe.GeoDataFrame如我所愿，该方法将返回 a ，但是...

我想知道是否存在一种直接传递到接口GeoPandas读取的 OGR 数据源的方法GDAL-OGR Python。如果没有，您对改进我提出的“hackish 方法”有什么建议吗？我在代码中添加了一些内联注释，试图解释我的担忧。

4

1 回答 1

2

我刚刚收到了 我在 GitHub 上打开的问题的回复，通过启用OGR_GRASS驱动程序解决了问题Fiona，下面的代码工作正常：

import fiona
import geopandas as gpd

fiona.supported_drivers["OGR_GRASS"] = "r" 
gdf = gpd.read_file('/GRASSDB/LOCATION_NAME/MAPSET/vector/layername/head')
于 2019-06-06T15:18:16.200 回答