def test_emptyens():
"""Check that we can initialize an empty ensemble"""
ens = ScratchEnsemble("emptyens")
assert not ens
if "__file__" in globals():
# Easen up copying test code into interactive sessions
testdir = os.path.dirname(os.path.abspath(__file__))
else:
testdir = os.path.abspath(".")
emptydf = ens.get_smry()
assert isinstance(emptydf, pd.DataFrame)
assert emptydf.empty
emptydatelist = ens.get_smry_dates()
assert isinstance(emptydatelist, list)
assert not emptydatelist
emptykeys = ens.get_smrykeys()
assert isinstance(emptykeys, list)
assert not emptykeys
emptyrates = ens.get_volumetric_rates()
assert isinstance(emptyrates, pd.DataFrame)
assert emptyrates.empty
emptystats = ens.get_smry_stats()
assert isinstance(emptystats, pd.DataFrame)
assert emptystats.empty
emptywells = ens.get_wellnames()
assert isinstance(emptywells, list)
assert not emptywells
emptygroups = ens.get_groupnames()
assert isinstance(emptygroups, list)
assert not emptygroups
emptymeta = ens.get_smry_meta()
assert isinstance(emptymeta, dict)
assert not emptymeta
emptymeta = ens.get_smry_meta("*")
assert isinstance(emptymeta, dict)
assert not emptymeta
emptymeta = ens.get_smry_meta("FOPT")
assert isinstance(emptymeta, dict)
assert not emptymeta
emptymeta = ens.get_smry_meta(["FOPT"])
assert isinstance(emptymeta, dict)
assert not emptymeta
# Add a realization manually:
ens.add_realizations(
testdir + "/data/testensemble-reek001/" + "realization-0/iter-0"
)
assert len(ens) == 1
def test_get_smry_meta(tmpdir):
"""Test the conservation of smry meta-data in virtual ensembles"""
if "__file__" in globals():
# Easen up copying test code into interactive sessions
testdir = os.path.dirname(os.path.abspath(__file__))
else:
testdir = os.path.abspath(".")
reekensemble = ScratchEnsemble(
"reekmetatest",
testdir + "/data/testensemble-reek001/" + "realization-*/iter-0")
# If no smry loaded before virtualization, nothing should be there:
assert "__smry_metadata" not in reekensemble.to_virtual().keys()
reekensemble.load_smry(time_index="yearly", column_keys=["F*"])
origmeta = reekensemble.get_smry_meta()
vens = reekensemble.to_virtual()
assert "__smry_metadata" in vens.keys()
meta = vens.get_df("__smry_metadata")
# Internally it is stored as a DataFrame, we check that
# since it is possible to get it using get_df(), and thereby
# almost part of the API
assert isinstance(meta, pd.DataFrame)
# But rather users should use get_smry_meta() to obtain
# stuff from the internal frame __smry_metadata:
metadict = vens.get_smry_meta()
assert isinstance(metadict, dict)
assert len(metadict) + 2 == len(
vens.get_smry(time_index="yearly", column_keys="*").columns)
# (the vens only knows of F* columns)
assert len(metadict) + 2 == len(
vens.get_smry(time_index="yearly", column_keys="F*").columns)
assert origmeta["FOPT"] == metadict["FOPT"]
assert origmeta["FWPTH"] == metadict["FWPTH"]
assert not vens.get_smry_meta([])
assert vens.get_smry_meta(column_keys="FOPT")["FOPT"] == origmeta["FOPT"]
assert not vens.get_smry_meta(column_keys="WOPT:NOTEXISTING")
# Test that it is retrievable after dumping to disk:
vens_disk_path = str(tmpdir.join("vens_dumped"))
vens.to_disk(vens_disk_path)
disk_vens = VirtualEnsemble(fromdisk=vens_disk_path)
metadict = disk_vens.get_smry_meta()
assert isinstance(metadict, dict)
assert len(metadict) + 2 == len(
vens.get_smry(time_index="yearly", column_keys="*").columns)
# (the vens only knows of F* columns)
assert len(metadict) + 2 == len(
vens.get_smry(time_index="yearly", column_keys="F*").columns)
assert origmeta["FOPT"] == metadict["FOPT"]
assert origmeta["FWPTH"] == metadict["FWPTH"]
def test_ensemble_ecl():
"""Eclipse specific functionality"""
if "__file__" in globals():
# Easen up copying test code into interactive sessions
testdir = os.path.dirname(os.path.abspath(__file__))
else:
testdir = os.path.abspath(".")
reekensemble = ScratchEnsemble(
"reektest",
testdir + "/data/testensemble-reek001/" + "realization-*/iter-0")
# Eclipse summary keys:
assert len(reekensemble.get_smrykeys("FOPT")) == 1
assert len(reekensemble.get_smrykeys("F*")) == 49
assert len(reekensemble.get_smrykeys(["F*", "W*"])) == 49 + 280
assert not reekensemble.get_smrykeys("BOGUS")
# reading ensemble dataframe
monthly = reekensemble.load_smry(time_index="monthly")
monthly = reekensemble.load_smry(column_keys=["F*"], time_index="monthly")
assert monthly.columns[0] == "REAL" # Enforce order of columns.
assert monthly.columns[1] == "DATE"
assert len(monthly) == 190
# Check that the result was cached in memory, not necessarily on disk..
assert isinstance(reekensemble.get_df("unsmry--monthly.csv"), pd.DataFrame)
assert len(reekensemble.keys()) == 4
# When asking the ensemble for FOPR, we also get REAL as a column
# in return. Note that the internal stored version will be
# overwritten by each load_smry()
assert len(reekensemble.load_smry(column_keys=["FOPR"]).columns) == 3
assert len(reekensemble.load_smry(column_keys=["FOP*"]).columns) == 11
assert len(
reekensemble.load_smry(column_keys=["FGPR", "FOP*"]).columns) == 12
# Check that there is now a cached version with raw dates:
assert isinstance(reekensemble.get_df("unsmry--raw.csv"), pd.DataFrame)
# The columns are not similar, this is allowed!'
# If you get 3205 here, it means that you are using the union of
# raw dates from all realizations, which is not correct
assert len(
reekensemble.load_smry(column_keys=["FGPR", "FOP*"]).index) == 1700
# Date list handling:
assert len(reekensemble.get_smry_dates(freq="report")) == 641
assert len(reekensemble.get_smry_dates(freq="raw")) == 641
assert len(reekensemble.get_smry_dates(freq="yearly")) == 5
assert len(reekensemble.get_smry_dates(freq="monthly")) == 38
assert len(reekensemble.get_smry_dates(freq="daily")) == 1098
assert len(reekensemble.get_smry_dates(freq="first")) == 1
assert len(reekensemble.get_smry_dates(freq="last")) == 1
assert reekensemble.get_smry_dates(
freq="first") == reekensemble.get_smry_dates(freq="first",
start_date="1900-01-01",
end_date="2050-02-01")
assert reekensemble.get_smry_dates(
freq="last") == reekensemble.get_smry_dates(freq="last",
start_date="1900-01-01",
end_date="2050-02-01")
assert str(reekensemble.get_smry_dates(
freq="report")[-1]) == "2003-01-02 00:00:00"
assert str(
reekensemble.get_smry_dates(freq="raw")[-1]) == "2003-01-02 00:00:00"
assert str(reekensemble.get_smry_dates(freq="yearly")[-1]) == "2004-01-01"
assert str(reekensemble.get_smry_dates(freq="monthly")[-1]) == "2003-02-01"
assert str(reekensemble.get_smry_dates(freq="daily")[-1]) == "2003-01-02"
assert str(reekensemble.get_smry_dates(freq="first")[-1]) == "2000-01-01"
assert str(reekensemble.get_smry_dates(freq="last")[-1]) == "2003-01-02"
assert (str(
reekensemble.get_smry_dates(
freq="daily", end_date="2002-03-03")[-1]) == "2002-03-03")
assert (str(
reekensemble.get_smry_dates(
freq="daily", start_date="2002-03-03")[0]) == "2002-03-03")
# Start and end outside of orig data and on the "wrong side"
dates = reekensemble.get_smry_dates(end_date="1999-03-03")
assert len(dates) == 1
assert str(dates[0]) == "1999-03-03"
dates = reekensemble.get_smry_dates(start_date="2099-03-03")
assert len(dates) == 1
assert str(dates[0]) == "2099-03-03"
# Time interpolated dataframes with summary data:
yearly = reekensemble.get_smry_dates(freq="yearly")
assert len(reekensemble.load_smry(column_keys=["FOPT"],
time_index=yearly)) == 25
# NB: This is cached in unsmry-custom.csv, not unsmry--yearly!
# This usage is discouraged. Use 'yearly' in such cases.
# Check that we can shortcut get_smry_dates:
assert len(
reekensemble.load_smry(column_keys=["FOPT"],
time_index="yearly")) == 25
assert len(reekensemble.load_smry(column_keys=["FOPR"],
time_index="first")) == 5
assert isinstance(reekensemble.get_df("unsmry--first.csv"), pd.DataFrame)
assert len(reekensemble.load_smry(column_keys=["FOPR"],
time_index="last")) == 5
assert isinstance(reekensemble.get_df("unsmry--last.csv"), pd.DataFrame)
# Check that time_index=None and time_index="raw" behaves like default
raw = reekensemble.load_smry(column_keys=["F*PT"], time_index="raw")
print(raw)
assert reekensemble.load_smry(column_keys=["F*PT"]).iloc[3,
2] == raw.iloc[3,
2]
assert (reekensemble.load_smry(column_keys=["F*PT"],
time_index=None).iloc[3, 3] == raw.iloc[3,
3])
# Give ISO-dates directly:
assert (len(
reekensemble.get_smry(column_keys=["FOPR"],
time_index="2001-01-02")) == 5)
# Summary metadata:
meta = reekensemble.get_smry_meta()
assert len(meta) == len(reekensemble.get_smrykeys())
assert "FOPT" in meta
assert not meta["FOPT"]["is_rate"]
assert meta["FOPT"]["is_total"]
meta = reekensemble.get_smry_meta("FOPT")
assert meta["FOPT"]["is_total"]
meta = reekensemble.get_smry_meta("*")
assert meta["FOPT"]["is_total"]
meta = reekensemble.get_smry_meta(["*"])
assert meta["FOPT"]["is_total"]
meta = reekensemble.get_smry_meta(["FOPT", "BOGUS"])
assert meta["FOPT"]["is_total"]
assert "BOGUS" not in meta
# Eclipse well names list
assert len(reekensemble.get_wellnames("OP*")) == 5
assert len(reekensemble.get_wellnames(None)) == 8
assert len(reekensemble.get_wellnames()) == 8
assert not reekensemble.get_wellnames("")
assert len(reekensemble.get_wellnames(["OP*", "WI*"])) == 8
# eclipse well groups list
assert len(reekensemble.get_groupnames()) == 3
# delta between two ensembles
diff = reekensemble - reekensemble
assert len(
diff.get_smry(column_keys=["FOPR", "FGPR", "FWCT"]).columns) == 5
# eclipse summary vector statistics for a given ensemble
df_stats = reekensemble.get_smry_stats(column_keys=["FOPR", "FGPR"],
time_index="monthly")
assert isinstance(df_stats, pd.DataFrame)
assert len(df_stats.columns) == 2
assert isinstance(df_stats["FOPR"]["mean"], pd.Series)
assert len(df_stats["FOPR"]["mean"].index) == 38
# check if wild cards also work for get_smry_stats
df_stats = reekensemble.get_smry_stats(column_keys=["FOP*", "FGP*"],
time_index="monthly")
assert len(df_stats.columns) == len(
reekensemble.get_smrykeys(["FOP*", "FGP*"]))
# Check webviz requirements for dataframe
stats = df_stats.index.levels[0]
assert "minimum" in stats
assert "maximum" in stats
assert "p10" in stats
assert "p90" in stats
assert "mean" in stats
assert df_stats["FOPR"]["minimum"].iloc[-2] < df_stats["FOPR"][
"maximum"].iloc[-2]
# Check user supplied quantiles
df_stats = reekensemble.get_smry_stats(column_keys=["FOPT"],
time_index="yearly",
quantiles=[0, 15, 50, 85, 100])
statistics = df_stats.index.levels[0]
assert "p0" in statistics
assert "p15" in statistics
assert "p50" in statistics
assert "p85" in statistics
assert "p100" in statistics
# For oil industry, p15 on FOPT should yield a larger value than p85.
# But the quantiles we get out follows the rest of the world
# so we check for the opposite.
assert df_stats["FOPT"]["p85"][-1] > df_stats["FOPT"]["p15"][-1]
with pytest.raises(ValueError):
reekensemble.get_smry_stats(column_keys=["FOPT"],
time_index="yearly",
quantiles=["foobar"])
noquantiles = reekensemble.get_smry_stats(column_keys=["FOPT"],
time_index="yearly",
quantiles=[])
assert len(noquantiles.index.levels[0]) == 3
