def test_tile_metadata_with_geometry_check(self):
def geometry_check(meta):
if meta['latitude_band'] == 'N':
return False
return True
f = open('tests/samples/tileInfo.json', 'rb')
tile_info = json.loads(f.read().decode(),
object_pairs_hook=OrderedDict)
tile = tile_metadata(tile_info,
metadata_to_dict('tests/samples/metadata.xml'),
geometry_check)
assert isinstance(tile, OrderedDict)
assert tile[
'thumbnail'] == 'http://sentinel-s2-l1c.s3.amazonaws.com/tiles/56/X/NF/2016/3/16/0/preview.jp2'
assert tile[
'tile_name'] == 'S2A_OPER_MSI_L1C_TL_SGS__20160316T054120_A003818_T56XNF_N02.01'
assert tile['utm_zone'] == 56
assert tile['data_coverage_percentage'] == 65.58
assert tile['sensing_orbit_direction'] == 'DESCENDING'
assert len(tile['download_links']['aws_s3']) == 13
# Make sure bands urls are left padded
d_link = tile['download_links']['aws_s3'][0].split('.')[-2].split('/')
assert d_link[-1] == 'B01'
def test_metadata_to_dict_new_format(self):
xml = open('tests/samples/metadata_new.xml')
product = metadata_to_dict(xml)
assert isinstance(product, OrderedDict)
assert 'band_list' in product
assert 'tiles' in product
assert product['spacecraft_name'] == 'Sentinel-2A'
assert len(product['band_list']) == 13
tiles = list(iterkeys(product['tiles']))
assert len(tiles) == 1
def test_metadata_to_dict_new_format(self):
xml = open('tests/samples/metadata_new.xml')
product = metadata_to_dict(xml)
assert isinstance(product, OrderedDict)
assert 'band_list' in product
assert 'tiles' in product
assert product['spacecraft_name'] == 'Sentinel-2A'
assert len(product['band_list']) == 13
tiles = list(iterkeys(product['tiles']))
assert len(tiles) == 1
def test_tile_metadata(self):
f = open('tests/samples/tileInfo.json', 'rb')
tile_info = json.loads(f.read().decode(), object_pairs_hook=OrderedDict)
tile = tile_metadata(tile_info, metadata_to_dict('tests/samples/metadata.xml'))
assert isinstance(tile, OrderedDict)
assert tile['thumbnail'] == 'http://sentinel-s2-l1c.s3.amazonaws.com/tiles/56/X/NF/2016/3/16/0/preview.jp2'
assert tile['tile_name'] == 'S2A_OPER_MSI_L1C_TL_SGS__20160316T054120_A003818_T56XNF_N02.01'
assert tile['utm_zone'] == 56
assert tile['data_coverage_percentage'] == 65.58
assert tile['sensing_orbit_direction'] == 'DESCENDING'
assert len(tile['download_links']['aws_s3']) == 13
# Make sure bands urls are left padded
d_link = tile['download_links']['aws_s3'][0].split('.')[-2].split('/')
assert d_link[-1] == 'B01'
def product_metadata(product, dst_folder, counter=None, writers=[file_writer], geometry_check=None):
""" Extract metadata for a specific product """
if not counter:
counter = {
'products': 0,
'saved_tiles': 0,
'skipped_tiles': 0,
'skipped_tiles_paths': []
}
s3_url = 'http://sentinel-s2-l1c.s3.amazonaws.com'
product_meta_link = '{0}/{1}'.format(s3_url, product['metadata'])
product_info = requests.get(product_meta_link, stream=True)
product_metadata = metadata_to_dict(product_info.raw)
product_metadata['product_meta_link'] = product_meta_link
counter['products'] += 1
for tile in product['tiles']:
tile_info = requests.get('{0}/{1}'.format(s3_url, tile))
try:
metadata = tile_metadata(tile_info.json(), copy(product_metadata), geometry_check)
for w in writers:
w(dst_folder, metadata)
logger.info('Saving to disk: %s' % metadata['tile_name'])
counter['saved_tiles'] += 1
except JSONDecodeError:
logger.warning('Tile: %s was not found and skipped' % tile)
counter['skipped_tiles'] += 1
counter['skipped_tiles_paths'].append(tile)
return counter
def test_tile_metadata(self):
f = open('tests/samples/tileInfo.json', 'rb')
tile_info = json.loads(f.read().decode(),
object_pairs_hook=OrderedDict)
tile = tile_metadata(tile_info,
metadata_to_dict('tests/samples/metadata.xml'))
assert isinstance(tile, OrderedDict)
assert tile[
'thumbnail'] == 'http://sentinel-s2-l1c.s3.amazonaws.com/tiles/56/X/NF/2016/3/16/0/preview.jp2'
assert tile[
'tile_name'] == 'S2A_OPER_MSI_L1C_TL_SGS__20160316T054120_A003818_T56XNF_N02.01'
assert tile['utm_zone'] == 56
assert tile['data_coverage_percentage'] == 65.58
assert tile['sensing_orbit_direction'] == 'DESCENDING'
assert len(tile['download_links']['aws_s3']) == 13
assert tile['tile_origin']['crs']['properties'][
'name'] == 'urn:ogc:def:crs:EPSG:8.9:4326'
# Make sure bands urls are left padded
d_link = tile['download_links']['aws_s3'][0].split('.')[-2].split('/')
assert d_link[-1] == 'B01'