Python test_util.get_test_file_path函数代码示例

def test_handles_empty_updates(test_file_path):
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    empty_site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily_empty.xml' % site_code)

    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=empty_site_data_file, autorepack=False)
    empty_site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)
    assert empty_site_data['00060:00003']['values'] == []

    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file, autorepack=False)

    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=empty_site_data_file, autorepack=False)
    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)

    values = site_data['00060:00003']['values']
    test_values = [
        dict(datetime="1964-01-23T00:00:00", last_checked=None, last_modified=None, qualifiers="A", value='240'),
    ]

    for test_value in test_values:
        assert values.index(test_value) >= 0

def test_site_data_update_site_list_with_multiple_updates(test_file_path):
    first_timestamp = '2013-01-01T01:01:01'
    second_timestamp = '2013-02-02T02:02:02'
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    with test_util.mocked_urls(site_data_file):
        with freezegun.freeze_time(first_timestamp):
            nwis.hdf5.update_site_data(site_code, path=test_file_path,
                    autorepack=False)
    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)

    last_value = site_data['00060:00003']['values'][-1]

    assert first_timestamp == last_value['last_checked'] == last_value['last_modified']

    update_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily_update.xml' % site_code)
    with test_util.mocked_urls(update_data_file):
        with freezegun.freeze_time(second_timestamp):
            nwis.hdf5.update_site_data(site_code, path=test_file_path,
                    autorepack=False)
    updated_site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)

    updated_values = updated_site_data['00060:00003']['values']
    last_value = updated_values[-1]
    assert last_value['last_checked'] != first_timestamp
    assert second_timestamp == last_value['last_checked'] == last_value['last_modified']

    original_timestamp = first_timestamp
    modified_timestamp = second_timestamp

    test_values = [
        dict(datetime="1963-01-23T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="A", value='7'),
        dict(datetime="1964-01-23T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="A", value='1017'),
        dict(datetime="1964-01-24T00:00:00", last_checked=original_timestamp, last_modified=original_timestamp, qualifiers="A", value='191'),
        dict(datetime="1964-08-22T00:00:00", last_checked=original_timestamp, last_modified=original_timestamp, qualifiers="A", value='7.9'),
        dict(datetime="1969-05-26T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="A", value='1080'),
        dict(datetime="2011-12-06T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='222'),
        dict(datetime="2011-12-15T00:00:00", last_checked=original_timestamp, last_modified=original_timestamp, qualifiers="P Eqp", value='-999999'),
        dict(datetime="2012-01-15T00:00:00", last_checked=original_timestamp, last_modified=original_timestamp, qualifiers="P e", value='97'),
        dict(datetime="2012-05-25T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='56'),
        dict(datetime="2012-05-26T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='55'),
        dict(datetime="2012-05-27T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="A", value='52'),
        dict(datetime="2012-05-28T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='48'),
        dict(datetime="2012-05-29T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='1099'),
        dict(datetime="2012-05-30T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='1098'),
        dict(datetime="2012-05-31T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='41'),
        dict(datetime="2012-06-01T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='37'),
        dict(datetime="2012-06-02T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='1097'),
        dict(datetime="2012-06-03T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='69'),
        dict(datetime="2012-06-04T00:00:00", last_checked=modified_timestamp, last_modified=original_timestamp, qualifiers="P", value='81'),
        dict(datetime="2012-06-05T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='1071'),
        dict(datetime="2012-06-06T00:00:00", last_checked=modified_timestamp, last_modified=modified_timestamp, qualifiers="P", value='2071'),
    ]

    for test_value in test_values:
        assert updated_values.index(test_value) >= 0

def test_file_size_doesnt_balloon_with_update_site_list(test_file_path):
    test_file_path += 'test.h5'
    site_list_file = test_util.get_test_file_path('usgs/nwis/RI_daily.xml')
    updated_site_list_file = test_util.get_test_file_path('usgs/nwis/RI_daily.xml')
    nwis.hdf5.update_site_list(path=test_file_path,
        input_file=site_list_file)
    nwis.hdf5.update_site_list(path=test_file_path,
            input_file=updated_site_list_file)
    original_size = os.path.getsize(test_file_path)
    for i in range(3):
        nwis.hdf5.update_site_list(path=test_file_path,
                input_file=updated_site_list_file)
    expected_size = original_size * 1.01
    assert os.path.getsize(test_file_path) <= expected_size

def test_remove_values(test_file_path):
    from datetime import datetime
    site_code = '07335390'
    parameter_code = '00062:00011'
    values_to_remove = {
        parameter_code: ['2012-10-25 06:00', '2012-10-25 23:00',
            '2012-10-30 15:00:00', datetime(2012, 11, 15, 13)]
    }
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_instantaneous.xml' % site_code)
    nwis.hdf5.update_site_data(site_code, period='all',
            path=test_file_path, input_file=site_data_file, autorepack=False)
    nwis.hdf5.remove_values(site_code, values_to_remove, path=test_file_path,
        autorepack=False)

    test_values = [
        dict(datetime="2012-10-25T01:00:00-05:00", last_checked=None, last_modified=None, qualifiers="P", value=None),
        dict(datetime="2012-10-25T18:00:00-05:00", last_checked=None, last_modified=None, qualifiers="P", value=None),
        dict(datetime="2012-10-30T10:00:00-05:00", last_checked=None, last_modified=None, qualifiers="P", value=None),
        dict(datetime="2012-11-15T07:00:00-06:00", last_checked=None, last_modified=None, qualifiers="P", value=None),
    ]

    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)
    site_values = site_data[parameter_code]['values']
    for test_value in test_values:
        assert test_value in site_values

def test_last_refresh_gets_updated(test_file_path):
    test_file_path = test_file_path + "test.h5"

    first_timestamp = '2013-01-01T01:01:01'
    second_timestamp = '2013-02-02T02:02:02'
    forth_timestamp = '2013-03-03T03:03:03'
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)

    with test_util.mocked_urls(site_data_file):
        with freezegun.freeze_time(first_timestamp):
            nwis.hdf5.update_site_data(site_code, path=test_file_path,
                    autorepack=False)
        first_refresh = nwis.hdf5._get_last_refresh(site_code, test_file_path)
        assert first_refresh == first_timestamp

        with freezegun.freeze_time(second_timestamp):
            nwis.hdf5.update_site_data(site_code, path=test_file_path,
                    autorepack=False)
        second_refresh = nwis.hdf5._get_last_refresh(site_code, test_file_path)
        assert second_refresh == second_timestamp

        nwis.hdf5.update_site_data(site_code, path=test_file_path,
                input_file=site_data_file, autorepack=False)
        third_refresh = nwis.hdf5._get_last_refresh(site_code, test_file_path)
        assert third_refresh == None

        with freezegun.freeze_time(forth_timestamp):
            nwis.hdf5.update_site_data(site_code, path=test_file_path,
                    autorepack=False)
        forth_refresh = nwis.hdf5._get_last_refresh(site_code, test_file_path)
        assert forth_refresh is not None
        assert forth_refresh == forth_timestamp

def test_update_site_data_updates_site_list(test_file_path):
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file, autorepack=False)
    site = nwis.hdf5.get_site(site_code, path=test_file_path)

    test_site = {
        'agency': 'USGS',
        'code': '01117800',
        'county': '44009',
        'huc': '01090005',
        'location': {
            'latitude': '41.5739884',
            'longitude': '-71.72062318',
            'srs': 'EPSG:4326'
        },
        'name': 'WOOD RIVER NEAR ARCADIA, RI',
        'network': 'NWIS',
        'site_type': 'ST',
        'state_code': '44',
        'timezone_info': {
            'default_tz': {'abbreviation': 'EST', 'offset': '-05:00'},
            'dst_tz': {'abbreviation': 'EDT', 'offset': '-04:00'},
            'uses_dst': True
        }
    }

    assert site == test_site

def test_get_site(test_file_path):
    site_code = '08068500'
    site_data_file = 'usgs/nwis/site_%s_daily.xml' % site_code
    input_file = test_util.get_test_file_path(site_data_file)
    nwis.hdf5.update_site_list(path=test_file_path,
            input_file=input_file, autorepack=False)

    site = nwis.hdf5.get_site(site_code, path=test_file_path)
    assert site == {
        'agency': 'USGS',
        'code': '08068500',
        'county': '48339',
        'huc': '12040102',
        'location': {
            'latitude': '30.11049517',
            'longitude': '-95.4363275',
            'srs': 'EPSG:4326'
        },
        'name': 'Spring Ck nr Spring, TX',
        'network': 'NWIS',
        'site_type': 'ST',
        'state_code': '48',
        'timezone_info': {
            'default_tz': {'abbreviation': 'CST', 'offset': '-06:00'},
            'dst_tz': {'abbreviation': 'CDT', 'offset': '-05:00'},
            'uses_dst': True
        },
    }

def test_parse_site_values():
    query_isodate = '2000-01-01'
    value_file = test_util.get_test_file_path(
            'usgs/nwis/site_07335390_instantaneous.xml')
    with open(value_file, 'rb') as content_io:
        values = ulmo.waterml.v1_1.parse_site_values(content_io, query_isodate)

    assert len(values['00062:00011']['values']) > 1000

def test_non_usgs_site(test_file_path):
    site_code = '07335390'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_instantaneous.xml' % site_code)
    nwis.hdf5.update_site_data(site_code, period='all',
            path=test_file_path, input_file=site_data_file, autorepack=False)

    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)
    assert len(site_data['00062:00011']['values']) > 1000

def test_get_site_for_missing_raises_lookup(test_file_path):
    site_code = '08068500'
    site_data_file = 'usgs/nwis/site_%s_daily.xml' % site_code
    input_file = test_util.get_test_file_path(site_data_file)
    nwis.hdf5.update_site_list(path=test_file_path,
        input_file=input_file, autorepack=False)

    with pytest.raises(LookupError):
        missing_code = '98068500'
        nwis.hdf5.get_site(missing_code, path=test_file_path)

def test_file_size_doesnt_balloon_with_update_site_data(test_file_path):
    test_file_path += 'test.h5'
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    update_site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily_update.xml' % site_code)

    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file)
    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=update_site_data_file)
    original_size = os.path.getsize(test_file_path)
    for i in range(20):
        nwis.hdf5.update_site_data(site_code, path=test_file_path,
                input_file=update_site_data_file)

    expected_size = original_size * 1.01
    assert os.path.getsize(test_file_path) <= expected_size

def test_site_data_is_sorted(test_file_path):
    site_code = '01117800'
    site_data_file = test_util.get_test_file_path(os.path.join('usgs','nwis', 'site_%s_daily.xml' % site_code))
    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file, autorepack=False)
    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)

    values = site_data['00060:00003']['values']
    assert all(
        values[i]['datetime'] < values[i+1]['datetime']
        for i in range(len(values) - 1))

def test_site_data_filter_by_one_parameter_code(test_file_path):
    site_code = '08068500'
    parameter_code = '00065:00003'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file, autorepack=False)
    all_site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path)
    site_data = nwis.hdf5.get_site_data(site_code, parameter_code=parameter_code, path=test_file_path)

    assert site_data[parameter_code] == all_site_data[parameter_code]

def _test_use_file(elements, by_state):
    if test_util.use_test_files():
        if isinstance(elements, basestring):
            div = "st" if by_state else "dv"
            path = 'ncdc/cirs/climdiv-{elements}{div}-v1.0.0-20140304'.format(
                elements=elements, div=div)
        else:
            path = 'ncdc/cirs/'
        return test_util.get_test_file_path(path)
    else:
        return None

def test_site_data_filter_by_date_single_param(test_file_path):
    site_code = '08068500'
    parameter_code = '00065:00003'
    date_str = '2000-01-01'
    site_data_file = test_util.get_test_file_path(
        'usgs/nwis/site_%s_daily.xml' % site_code)
    nwis.hdf5.update_site_data(site_code, path=test_file_path,
            input_file=site_data_file, autorepack=False)
    site_data = nwis.hdf5.get_site_data(site_code, path=test_file_path, start=date_str)
    first_value = site_data[parameter_code]['values'][0]
    assert datetime.datetime.strptime(first_value["datetime"], '%Y-%m-%dT%H:%M:%S') >= datetime.datetime.strptime(date_str, '%Y-%m-%d')