def language_string(s, lang="en"):
"""
Function to encode a literal as being in a specific language.
Parameters
----------
s : string
lang : string
ISO character code, default="en"
Returns
-------
s : string
triple quoted Turtle literal with language encoding
Example
-------
>>> print(language_string("Canada goose"))
\"""Canada goose\"""@en
"""
return(
"\"\"\"{0}\"\"\"@{1}".format(
return_string(
s,
[
'"'
],
[
"'"
]
),
lang
)
)
def create_label(input_string):
"""
Clean up a string and create a corresponding (shortened) label.
Parameters
----------
input_string : string
arbitrary string
Returns
-------
output_string : string
stripped input_string
label_string : string
alphanumeric characters of input_string
"""
from mhdb.spreadsheet_io import return_string
from mhdb.spreadsheet_io import convert_string_to_label
if input_string:
if isinstance(input_string, str):
output_string = return_string(input_string,
replace=['"', '\n'],
replace_with=['', ''])
if output_string:
label_string = convert_string_to_label(output_string)
return output_string, label_string
else:
return '', ''
else:
raise Exception('input_string is not a string!')
else:
raise Exception('input_string is None!')
def build_rdf(uri_stem,
rdf_type,
label,
comment=None,
index=None,
worksheet=None,
worksheet2=None,
equivalent_class_uri=None,
subclassof_uri=None,
property_domain=None,
property_range=None,
exclude=[],
conceptualizations={}): #, no_nan=True):
"""
Build a generic RDF text document (with \" to escape for some strings).
Parameters
----------
uri_stem : string
class URI stem
rdf_type : string
rdf:type, such as:
':Disorder',
'owl:Class',
'owl:ObjectProperty',
'owl:DatatypeProperty',
'owl:FunctionalProperty'
label : string
label
comment : string
comment
index : integer
index to row of worksheet
worksheet : pandas dataframe
spreadsheet worksheet containing properties
worksheet2 : pandas dataframe
second worksheet containing references
equivalent_class_uri : string
equivalentClass URI (override worksheet)
subclassof_uri : string
subClassOf URI (override worksheet)
property_domain : string
property domain (override worksheet)
property_range : string
property range (override worksheet)
exclude : list
exclusions
conceptualizations : dictionary
conceptualizaiton scheme (i.e., OWL or SKOS) for a given prefix
#no_nan : Boolean
# return None if NaN?
Returns
-------
rdf_string : string
RDF triples
"""
from mhdb.spreadsheet_io import return_string, get_cells #, get_cell
# Get worksheet contents:
class_uri, subclass_uri, prop_domain, prop_range, \
definition, definition_ref, definition_uri = get_cells(worksheet, index,
worksheet2, exclude,
True)
#try:
# coding_system = get_cell(worksheet, "health-lifesci:codingSystem",
# index, exclude=[], no_nan=True)
#except:
# coding_system = None
# If arguments not provided, get from worksheet:
if comment in exclude:
comment = definition
if equivalent_class_uri in exclude:
equivalent_class_uri = class_uri
if subclassof_uri in exclude:
subclassof_uri = subclass_uri
if property_domain in exclude:
property_domain = prop_domain
if property_range in exclude:
property_range = prop_range
l_con = owl_or_skos(uri_stem, conceptualizations)
if ":" in uri_stem:
rdf_string = """
### {0}
{1} rdf:type {2} """.format(label, uri_stem, rdf_type)
else:
rdf_string = """
### {0}
:{1} rdf:type {2} """.format(label, uri_stem, rdf_type)
if label not in exclude:
rdf_string += """;
rdfs:label \"\"\"{0}\"\"\"^^rdfs:Literal """.format(
label if not label[-1] == "\"" else "".join([label[:-1], "\\\""]))
if comment not in exclude:
if definition_ref in exclude:
refstring = ""
else:
refstring = " [from: {0}]".format(
return_string(definition_ref, ['"'], ["'"]))
rdf_string += """;
rdfs:comment \"\"\"{0}{1}\"\"\"^^rdfs:Literal """.\
format(return_string(comment, ['"'], ["'"]), refstring)
if definition_uri not in exclude:
rdf_string += """;
rdfs:isDefinedBy "{0}"^^rdfs:Literal """.format(
return_string(definition_uri))
if equivalent_class_uri not in exclude:
rel = owl_or_skos_prop(l_con, equivalent_class_uri, conceptualizations,
"equivalence")
if rdf_type == 'owl:ObjectProperty':
rdf_string += """;
owl:equivalentProperty {0} """.format(
return_string(equivalent_class_uri))
else:
rdf_string += """;
{0} {1} """.format(rel, return_string(equivalent_class_uri))
if subclassof_uri not in exclude:
rel = owl_or_skos_prop(l_con, subclassof_uri, conceptualizations,
"subtype")
if not subclassof_uri.startswith(':') and "//" in subclassof_uri:
subclassof_uri = "{0}".format(return_string(subclassof_uri))
if rdf_type == 'owl:ObjectProperty':
rdf_string += """;
rdfs:subPropertyOf {0} """.format(return_string(subclassof_uri))
else:
rdf_string += """;
{0} {1} """.format(rel, return_string(subclassof_uri))
if property_domain not in exclude:
rdf_string += """;
rdfs:domain :{0} """.format(return_string(property_domain))
if property_range not in exclude:
rdf_string += """;
rdfs:range :{0} """.format(return_string(property_range))
rdf_string += """.
"""
return rdf_string