diff --git a/main.py b/main.py index e7d8e1c..cf2ba66 100644 --- a/main.py +++ b/main.py @@ -3,16 +3,20 @@ import argparse from src.librecatastro.scrapping.scrapper_html import ScrapperHTML from src.librecatastro.scrapping.scrapper_xml import ScrapperXML +from src.settings import config if __name__ == "__main__": parser = argparse.ArgumentParser(description='Runs the Cadastro Parser') parser.add_argument('--coords', dest='coords', default=False, action='store_true') parser.add_argument('--filenames', action='store', nargs='+', dest='filenames', default=[]) parser.add_argument('--provinces', action='store', nargs='+', dest='provinces', default=[]) + parser.add_argument('--sleep', action='store', dest='sleep', type=int, default=5) args = parser.parse_args(sys.argv[1:]) + + config['sleep_time'] = args.sleep + if args.coords: - print(args.filenames) ScrapperHTML.scrap_all_coordinates_files(args.filenames) else: ScrapperXML.scrap_all_addresses(args.provinces) diff --git a/src/librecatastro/domain/geometry/geo_bounding_box.py b/src/librecatastro/domain/geometry/geo_bounding_box.py deleted file mode 100644 index 54464f8..0000000 --- a/src/librecatastro/domain/geometry/geo_bounding_box.py +++ /dev/null @@ -1,32 +0,0 @@ -import json -from collections import namedtuple - -from src.settings import config -from src.utils.cadastro_logger import CadastroLogger - -'''Logger''' -logger = CadastroLogger(__name__).logger - - -class GeoBoundingBox: - def __init__(self, data): - self.data = json.loads(data, object_hook=lambda d: namedtuple('X', d.keys())(*d.values())) - - def get_coordinates_tuple(self): - return GeoBoundingBox.get_bb_from_file_static(self.data) - - @staticmethod - def get_bb_from_file_static(data): - location = data.geo_bounding_box.location - return int(location.top_left.lon * config['scale']), int(location.bottom_right.lon * config['scale']), int(location.bottom_right.lat * config['scale']), int(location.top_left.lat * config['scale']) - - @staticmethod - def get_bb_from_file(file): - f = open(file, "r") - content = f.read() - try: - data = json.loads(content, object_hook=lambda d: namedtuple('X', d.keys())(*d.values())) - return GeoBoundingBox.get_bb_from_file_static(data) - except: - logger.error("{} is not formatted properly. Please take a look at the examples.".format(file)) - return None diff --git a/src/tests/scrapper_html_tests.py b/src/tests/scrapper_html_tests.py index 18ae5fb..085db46 100644 --- a/src/tests/scrapper_html_tests.py +++ b/src/tests/scrapper_html_tests.py @@ -1,11 +1,8 @@ import os import unittest -from shapely.geometry import Point - from src.librecatastro.domain.geometry.geo_polygon import GeoPolygon from src.librecatastro.scrapping.scrapper_html import ScrapperHTML -from src.librecatastro.domain.geometry.geo_bounding_box import GeoBoundingBox from src.settings import config from src.utils.elasticsearch_utils import ElasticSearchUtils