def _to_user_defined(pif_obj):
"""Read the systems in the PIF to populate the user-defined portion"""
res = {}
# make a read view to flatten the hierarchy
rv = ReadView(pif_obj)
# Iterate over the keys in the read view
for k in rv.keys():
name, value = _extract_key_value(rv[k].raw)
# add any objects that can be extracted
if name and value is not None:
res[name] = value
# Grab interesting values not in the ReadView
pif = pif_obj.as_dictionary()
elements = {}
if pif.get("composition"):
for comp in pif["composition"]:
if comp.get("actualAtomicPercent"):
elements[comp["element"]] = float(
comp["actualAtomicPercent"]["value"])
elif comp.get("actualWeightPercent"):
elements[comp["element"]] = float(
comp["actualWeightPercent"]["value"])
if elements:
res["elemental_percent"] = elements
elif pif.get("chemicalFormula"):
symbol = ""
num = ""
# Chemical formulae are comprised of letters, numbers, and potentially characters we don't care about
for char in pif["chemicalFormula"]:
# Uppercase char indicates beginning of new symbol
if char.isupper():
# If there is already a symbol in holding, process it
if symbol:
try:
elements[symbol] = int(num)
# If num is a float, raises ValueError
except ValueError:
elements[symbol] = float(num) if num else 1
symbol = ""
num = ""
symbol += char
# Lowercase chars or digits are continuations of a symbol
elif char.islower():
symbol += char
elif char.isdigit():
num += char
elif char == ".":
num += char
# All other chars are not useful
if elements:
res["elemental_proportion"] = elements
return res
def test_method_software():
"""Testing that method and software names are elevated"""
method = Method(name="spam", software=[Software(name="magic")])
pif = System(properties=[
Property(name="foo", scalars=[Scalar(value="bar")], methods=[method])
])
r = ReadView(pif)
assert r["foo"].scalars[0].value == "bar", "Didn't elevate property key"
assert "spam" in r.keys(), "Expected spam in keys"
assert "magic" in r.keys(), "Expected magic in keys"
def parsePifKey(pif, key):
"""Parse a single pif key for single scalar values;
return nan if no scalar found.
"""
if (key in ReadView(pif).keys()):
if 'scalars' in dir(ReadView(pif)[key]):
try:
return ReadView(pif)[key].scalars[0].value
except IndexError:
return np.nan
else:
return np.nan
else:
return np.nan
def pifs2df(pifs):
"""Converts a collection of PIFs to tabular data
Very simple, purpose-built utility script. Converts an iterable of PIFs
to a dataframe. Returns the superset of all PIF keys as the set of columns.
Non-scalar values are converted to nan.
Usage
df = pifs2df(pifs)
Arguments
pifs = an iterable of PIFs
Returns
df = Pandas DataFrame
examples
import os
from citrination_client import CitrinationClient
from citrination_client import PifSystemReturningQuery, DatasetQuery
from citrination_client import DataQuery, Filter
## Set-up citrination search client
site = "https://citrination.com"
client = CitrinationClient(api_key = os.environ["CITRINATION_API_KEY"], site = site)
search_client = client.search
## Query the Agrawal (2014) dataset
system_query = \
PifSystemReturningQuery(
size = 500,
query = DataQuery(
dataset = DatasetQuery(id = Filter(equal = "150670"))
)
)
query_result = search_client.pif_search(system_query)
pifs = [x.system for x in query_results.hits]
## Rectangularize the pifs
df = pifs2df(pifs)
"""
## Consolidate superset of keys
key_sets = [set(ReadView(pif).keys()) for pif in pifs]
keys_ref = reduce(lambda s1, s2: s1.union(s2), key_sets)
## Rectangularize
## TODO: Append dataframes, rather than using a comprehension
df_data = \
pd.DataFrame(
columns = keys_ref,
data = [
[
parsePifKey(pif, key) \
for key in keys_ref
] for pif in pifs
]
)
return df_data
def test_ambiguity():
"""Test that ambiguous keys are removed from the top level dict"""
pif = System()
pif.uid = "10245"
pif.properties = [
Property(name="foo", scalars=[Scalar(value=1.0)]),
Property(name="bar", scalars=[Scalar(value=2.0)])
]
pif2 = System(
sub_systems=[
pif,
],
properties=[Property(name="foo", scalars=[Scalar(value=10.0)])])
r = ReadView(pif2)
assert r.properties["foo"].scalars[0].value == 10.0
assert r.sub_systems["10245"].properties["foo"].scalars[0].value == 1.0
assert "foo" not in r.keys()
assert r.sub_systems["10245"]["foo"].scalars[0].value == 1.0
assert r["bar"].scalars[0].value == 2.0
def test_condition_elevation():
"""Test that read views elevate conditions"""
condition = Value(name="spam", scalars=[Scalar(value="eggs")])
pif = System(properties=[
Property(
name="foo", scalars=[Scalar(value="bar")], conditions=[condition])
])
r = ReadView(pif)
assert r["foo"].scalars[0].value == "bar", "Didn't elevate property key"
assert r["spam"].scalars[0].value == "eggs", "Didn't elevate condition key"
def test_unambig():
"""Test that properties are mirrored in a top level dic"""
pif = System()
pif.properties = [
Property(name="foo", scalars=[Scalar(value=1.0)]),
Property(name="bar", scalars=[Scalar(value=2.0)])
]
r = ReadView(pif)
assert r["foo"].scalars[0].value == 1.0
assert r["bar"].scalars[0].value == 2.0
def parsePifKey(pif, key):
"""Parse a single pif key for single scalar values; return nan if no scalar found.
:param pif: PIF to access
:type pif: pif
:param key: key to access data
:type key: string
:returns: scalar value or np.nan
:rtype:
"""
if (key in ReadView(pif).keys()):
if 'scalars' in dir(ReadView(pif)[key]):
try:
return ReadView(pif)[key].scalars[0].value
except IndexError:
return np.nan
else:
return np.nan
else:
return np.nan
def test_read_view():
"""Test that properties are passed through to the readview"""
pif = System()
pif.uid = "10245"
pif.names = ["example", "ex"]
pif.properties = [
Property(name="foo", scalars=[Scalar(value=1.0)]),
Property(name="bar", scalars=[Scalar(value=2.0)])
]
r = ReadView(pif)
assert r.uid == pif.uid
assert r.names == pif.names
assert r.properties["foo"].scalars[0].value == 1.0
assert r.properties["bar"].scalars[0].value == 2.0
def test_nested_read_view():
"""Test that nested Pios (system here) are recursively processed"""
pif = System()
pif.uid = "10245"
pif.properties = [
Property(name="foo", scalars=[Scalar(value=1.0)]),
Property(name="bar", scalars=[Scalar(value=2.0)])
]
pif2 = System(sub_systems=[pif])
r = ReadView(pif2)
assert r.sub_systems["10245"].uid == pif.uid
assert r["10245"].uid == pif.uid
assert r.sub_systems["10245"].properties["foo"].scalars[0].value == 1.0
assert r.sub_systems["10245"].properties["bar"].scalars[0].value == 2.0
assert r["foo"].scalars[0].value == 1.0
assert r["bar"].scalars[0].value == 2.0
def test_multiple_instances():
"""Test that keys that show up in different places with the same value are kept"""
pif = System()
pif.uid = "10245"
pif.properties = [
Property(name="foo", scalars=[Scalar(value=1.0)]),
Property(name="bar", scalars=[Scalar(value=2.0)])
]
pif2 = System(
sub_systems=[
pif,
],
properties=[Property(name="bar", scalars=[Scalar(value=2.0)])])
r = ReadView(pif2)
assert r.properties["bar"].scalars[0].value == 2.0
assert r.sub_systems["10245"].properties["bar"].scalars[0].value == 2.0
assert r["bar"].scalars[0].value == 2.0
def test_key_error_case():
with open("./pypif_sdk/readview/tests/csv_pif.pif") as f:
test_case = load(f)
rv = ReadView(test_case)
assert len(rv.keys()) > 0
