基盤地図情報の PackDLMap.zip を全ファイル展開し、種別ごとにマージするやつ

layer_spec.json

{
	"common_fields" : [
		["gml_id", "String"],
		["fid", "String"],
		["lfSpanFr", "Date"],
		["lfSpanTo", "Date"],
		["devDate", "Date"],
		["orgGILvl", "String"],
		["orgMDId", "String"],
		["vis", "String"]
	],
	"AdmArea" : {
		"geom_type" : "Polygon",
		"fields" : [
			["type", "String"],
			["name", "String"],
			["admCode", "String"]
		]
	},
	"AdmBdry" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"]
		]
	},
	"AdmPt" : {
		"geom_type" : "Point",
		"fields" : [
			["type", "String"],
			["name", "String"],
			["admCode", "String"]
		]
	},
	"BldA" : {
		"geom_type" : "Polygon",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"BldL" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"Cntr" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["alti", "Real"]
		]
	},
	"CommBdry" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"]
		]
	},
	"CommPt" : {
		"geom_type" : "Point",
		"fields" : [
			["type", "String"],
			["name", "String"],
			["admCode", "String"]
		]
	},
	"Cstline" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"ElevPt" : {
		"geom_type" : "Point",
		"fields" : [
			["type", "String"],
			["alti", "Real"]
		]
	},
	"GCP" : {
		"geom_type" : "Point",
		"fields" : [
			["advNo", "String"],
			["orgName", "String"],
			["type", "String"],
			["gcpClass", "String"],
			["gcpCode", "String"],
			["name", "String"],
			["B", "Real"],
			["L", "Real"],
			["alti", "Real"],
			["altiAcc", "Integer"]
		]
	},
	"RailCL" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"RdCompt" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"],
			["admOffice", "String"]
		]
	},
	"RdEdg" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"],
			["admOffice", "String"]
		]
	},
	"WA" : {
		"geom_type" : "Polygon",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"WL" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"WStrA" : {
		"geom_type" : "Polygon",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	},
	"WStrL" : {
		"geom_type" : "LineString",
		"fields" : [
			["type", "String"],
			["name", "String"]
		]
	}
}

merge_gml.py

import glob
import json
import zipfile
import ogr
import osr

GEOM_TYPE = {
    'Point'      : ogr.wkbPoint,
    'LineString' : ogr.wkbLineString,
    'Polygon'    : ogr.wkbPolygon
}
FIELD_TYPE = {
    'Integer'    : ogr.OFTInteger,
    'Integer64'  : ogr.OFTInteger64,
    'Real'       : ogr.OFTReal,
    'String'     : ogr.OFTString,
    'Date'       : ogr.OFTDate,
    'Time'       : ogr.OFTTime,
    'DateTime'   : ogr.OFTDateTime
}



def main():
    with open('layer_spec.json', 'r') as f:
        layer_spec = json.load(f)

    out_filename = 'merged.gpkg'
    out_driver   = ogr.GetDriverByName('GPKG')
    out_srs      = osr.SpatialReference()
    out_srs.ImportFromEPSG(6668)
    driver = ogr.GetDriverByName('GML')

    extract_zips()

    # XML ファイルのリストを作り、 layername ごとに整理する
    filelist = {}
    for filename in glob.glob('FG-GML-*.xml'):
        meshcode, layername, pubdate, num = filename[7:-4].split('-')
        if layername not in filelist:
            filelist[layername] = []
        filelist[layername].append((layername, meshcode, filename))

    # 出力用のファイルを作成する
    out_ds = out_driver.CreateDataSource(out_filename)

    # layername ごとにデータのマージ処理を行う
    for layername in sorted(list(filelist.keys())):
        if layername not in layer_spec:
            print("Warning: undefined layer spec: %s" % layername)
            continue

        # 新しいレイヤを作成する
        out_layer = out_ds.CreateLayer(layername, out_srs,
                GEOM_TYPE[layer_spec[layername]['geom_type']],
                options=["FID=ID"])

        # 新しいレイヤの属性の設定
        for i in range(len(layer_spec['common_fields'])):
            f_name, f_type = layer_spec['common_fields'][i]
            defn = ogr.FieldDefn(f_name, FIELD_TYPE[f_type])
            out_layer.CreateField(defn)
        for i in range(len(layer_spec[layername]['fields'])):
            f_name, f_type = layer_spec[layername]['fields'][i]
            defn = ogr.FieldDefn(f_name, FIELD_TYPE[f_type])
            out_layer.CreateField(defn)
        out_layer.CreateField(ogr.FieldDefn('filename', ogr.OFTString))
        out_layer.CreateField(ogr.FieldDefn('meshcode', ogr.OFTString))

        out_layer_defn = out_layer.GetLayerDefn()

        # XML ファイルごとにデータをマージする
        for f in sorted(filelist[layername], key=lambda v: v[1]):
            _, meshcode, filename = f

            print("data appending... %s" % filename)
            ds = driver.Open(filename, 0)
            layer = ds.GetLayer()

            for feature in layer:
                out_feature = ogr.Feature(out_layer_defn)
                out_feature.SetGeometry(feature.GetGeometryRef().Clone())
                for i in range(out_layer_defn.GetFieldCount()):
                    defn = out_layer_defn.GetFieldDefn(i)
                    field_name = defn.GetName()
                    if field_name == 'filename':
                        field_value = filename
                    elif field_name == 'meshcode':
                        field_value = meshcode
                    else:
                        field_value = feature.GetField(field_name)
                    out_feature.SetField(defn.GetNameRef(), field_value)
                out_layer.CreateFeature(out_feature)
                out_feature = None
            out_layer.SyncToDisk()


def extract_zips():
    for f in glob.glob('PackDLMap*.zip'):
        with zipfile.ZipFile(f) as z:
            for info in z.infolist():
                if info.filename[-4:] == '.zip':
                    z.extract(info)

    for f in glob.glob('FG-GML-*.zip'):
        with zipfile.ZipFile(f) as z:
            for info in z.infolist():
                z.extract(info)


main()