|
| 1 | +"""Normalizer for the metadata of ESA CCI datasets""" |
| 2 | + |
| 3 | +import re |
| 4 | +from datetime import datetime |
| 5 | + |
| 6 | +from dateutil.tz import tzutc |
| 7 | +from dateutil.relativedelta import relativedelta |
| 8 | + |
| 9 | +import metanorm.utils as utils |
| 10 | +from .base import GeoSPaaSMetadataNormalizer |
| 11 | + |
| 12 | + |
| 13 | +class MarineHeatWavesMetadataNormalizer(GeoSPaaSMetadataNormalizer): |
| 14 | + """Generate the properties of a GeoSPaaS Dataset for a NOAA marine |
| 15 | + heatwaves dataset |
| 16 | + """ |
| 17 | + |
| 18 | + def check(self, raw_metadata): |
| 19 | + """Checks that the URL starts with the right prefix""" |
| 20 | + return raw_metadata.get('url', '').startswith( |
| 21 | + 'https://www.star.nesdis.noaa.gov/pub/socd/mecb/crw/data/marine_heatwave/' |
| 22 | + 'v1.0.1/category/nc') |
| 23 | + |
| 24 | + def get_entry_title(self, raw_metadata): |
| 25 | + return 'NOAA marine heatwaves' |
| 26 | + |
| 27 | + @utils.raises((KeyError, AttributeError)) |
| 28 | + def get_entry_id(self, raw_metadata): |
| 29 | + return utils.NC_H5_FILENAME_MATCHER.search(raw_metadata['url']).group(1) |
| 30 | + |
| 31 | + def get_summary(self, raw_metadata): |
| 32 | + return '' |
| 33 | + |
| 34 | + time_patterns = ( |
| 35 | + ( |
| 36 | + re.compile(r'/noaa-crw_mhw.*' + utils.YEARMONTHDAY_REGEX + r'\.nc$'), |
| 37 | + utils.create_datetime, |
| 38 | + lambda time: (time, time + relativedelta(days=1)) |
| 39 | + ), |
| 40 | + ) |
| 41 | + |
| 42 | + @utils.raises(KeyError) |
| 43 | + def get_time_coverage_start(self, raw_metadata): |
| 44 | + return utils.find_time_coverage(self.time_patterns, raw_metadata['url'])[0] |
| 45 | + |
| 46 | + @utils.raises(KeyError) |
| 47 | + def get_time_coverage_end(self, raw_metadata): |
| 48 | + return utils.find_time_coverage(self.time_patterns, raw_metadata['url'])[1] |
| 49 | + |
| 50 | + def get_platform(self, raw_metadata): |
| 51 | + return utils.get_gcmd_platform('Earth Observation Satellites') |
| 52 | + |
| 53 | + def get_instrument(self, raw_metadata): |
| 54 | + return utils.get_gcmd_instrument('Imaging Spectrometers/Radiometers') |
| 55 | + |
| 56 | + def get_location_geometry(self, raw_metadata): |
| 57 | + return utils.WORLD_WIDE_COVERAGE_WKT |
| 58 | + |
| 59 | + def get_provider(self, raw_metadata): |
| 60 | + return utils.get_gcmd_provider(['DOC/NOAA']) |
| 61 | + |
| 62 | + def get_dataset_parameters(self, raw_metadata): |
| 63 | + return [] |
0 commit comments