def values(self, tind, zind=None):
"""
Args:
tind: Value to test for inclusion between time bounds
"""
span = (None, None)
for m in self._members:
if m.period is not None:
# If a period is defined, extract the attribute from the
# pd.Timestamp object before comparison. The min and max
# values are in this period unit already.
tind_copy = getattr(tind, m.period)
else:
# If a period isn't defined, make a new Timestamp object
# to align with the above name 'tind_copy'
tind_copy = tind
# If we are between times
if tind_copy > m.tspan.minv and tind_copy <= m.tspan.maxv:
if not isnan(zind) and not isnan(m.zspan):
# If we are between depths
if zind > m.zspan.minv and zind <= m.zspan.maxv:
span = m.vspan
elif isnan(zind) and isnan(m.zspan):
span = m.vspan
return span
def values(self, tind, zind=None):
span = (None, None)
for m in self._members:
# If we are between times
if tind > m.tspan.minv and tind <= m.tspan.maxv:
if not isnan(zind) and not isnan(m.zspan):
# If we are between depths
if zind > m.zspan.minv and zind <= m.zspan.maxv:
span = m.vspan
elif isnan(zind) and isnan(m.zspan):
span = m.vspan
return span
def check(self, tinp, inp, zinp):
# Start with everything as UNKNOWN (1)
flag_arr = np.ma.empty(inp.size, dtype='uint8')
flag_arr.fill(QartodFlags.UNKNOWN)
# If the value is masked set the flag to MISSING
flag_arr[inp.mask] = QartodFlags.MISSING
# Iterate over each member and apply its spans on the input data.
# Member spans are applied in order and any data points that fall into
# more than one member are flagged by each one.
for m in self._members:
if m.period is not None:
# If a period is defined, extract the attribute from the
# pd.DatetimeIndex object before comparison. The min and max
# values are in this period unit already.
tinp_copy = getattr(tinp, m.period).to_series()
else:
# If a period isn't defined, make a new Timestamp object
# to align with the above name 'tinp_copy'
tinp_copy = tinp
# If a zspan is defined but we don't have z input (zinp), skip this member
# Note: `zinp.count()` can return `np.ma.masked` so we also check using isnan
if not isnan(m.zspan) and (not zinp.count() or isnan(zinp.any())):
continue
# Indexes that align with the T
t_idx = (tinp_copy >= m.tspan.minv) & (tinp_copy <= m.tspan.maxv)
# Indexes that align with the Z
if not isnan(m.zspan):
# Only test non-masked values between the min and max
z_idx = (~zinp.mask) & (zinp >= m.zspan.minv) & (zinp <=
m.zspan.maxv)
else:
# If there is no z data in the config, don't try to filter by depth!
# Set z_idx to all True to prevent filtering
z_idx = np.ones(inp.size, dtype=bool)
# Combine the T and Z indexes
values_idx = (t_idx & z_idx)
# Failed and suspect data for this value span. Combining fail_idx or
# suspect_idx with values_idx represents the subsets of data that should be
# fail and suspect respectively.
if not isnan(m.fspan):
fail_idx = (inp < m.fspan.minv) | (inp > m.fspan.maxv)
else:
fail_idx = np.zeros(inp.size, dtype=bool)
suspect_idx = (inp < m.vspan.minv) | (inp > m.vspan.maxv)
with np.errstate(invalid='ignore'):
flag_arr[(values_idx & fail_idx)] = QartodFlags.FAIL
flag_arr[(values_idx & ~fail_idx
& suspect_idx)] = QartodFlags.SUSPECT
flag_arr[(values_idx & ~fail_idx
& ~suspect_idx)] = QartodFlags.GOOD
return flag_arr