csf_utilities/reparsing_2_TO_REVIEW.py

# %%
import json
import csv

# IMPORT MASTER FILE
with open('.dat/man_draft.json', 'r') as in_file:
    data = json.load(in_file)
with open('entities_in.csv', 'r') as csv_in_1:
    reader = csv.DictReader(csv_in_1)
    ent_csv_in = [row for row in reader]
with open('relations_in.csv', 'r') as csv_in_2:
    reader = csv.DictReader(csv_in_2)
    rel_csv_in = [row for row in reader]
# %%
# Consistency check

entity_rels = {ent for rel in data['Relazioni'] for ent in [rel['Entità 1'], rel['Entità 2']]}

entities = set(data['Entità'].keys())
entities.add('#any') # For compatibility

if not entity_rels.issubset(entities):
    print(entity_rels.difference(entities))
# %%
# USE A DIRTY SHORTCUT: paste entity/relation info on a precompiled rdf template file.

# Load template
with open('template_2.rdf', 'r') as in_file:
    raw_rdf = in_file.read()

# Defined rdf snippets; info will replace placeholder tags (in uppercase between '#')
entity_template = '''
    <!-- http://www.h2iosc.it/onto##NAME# -->

    <owl:Class rdf:about="&h2iosc;#NAME#">
        <rdfs:label>#LABEL#</rdfs:label>
        <rdfs:subClassOf>#PARENT#</rdfs:subClassOf>
        <rdfs:isDefinedBy rdf:resource="#URI#"/>
    </owl:Class>
'''
subclass_string = "        <rdfs:subClassOf>#PARENT#</rdfs:subClassOf>\n"
class_defined_string = '        <rdfs:isDefinedBy rdf:resource="#URI#"/>\n'

object_property_template = '''
    <!-- http://www.h2iosc.it/onto##NAME# -->

    <owl:ObjectProperty rdf:about="&h2iosc;#NAME#">
        <rdfs:label>#LABEL#</rdfs:label>
        <rdfs:range rdf:resource="&h2iosc;#RANGE#"/>
        <rdfs:domain rdf:resource="&h2iosc;#DOMAIN#"/>
        <rdfs:isDefinedBy rdf:resource="#URI#"/>
    </owl:ObjectProperty>
'''

object_property_inverse_template = '''
    <!-- http://www.h2iosc.it/onto##NAME# -->

    <owl:ObjectProperty rdf:about="&h2iosc;#NAME#">
        <rdfs:label>#LABEL#</rdfs:label>
        <owl:inverseOf rdf:resource="&h2iosc;#INV#"/>
        <rdfs:range rdf:resource="&h2iosc;#RANGE#"/>
        <rdfs:domain rdf:resource="&h2iosc;#DOMAIN#"/>
        <rdfs:isDefinedBy rdf:resource="#URI#"/>
    </owl:ObjectProperty>
'''
object_defined_string = '        <rdfs:isDefinedBy rdf:resource="#URI#"/>\n'

datatype_property_template = '''
    <!-- http://www.h2iosc.it/onto##NAME# -->

    <owl:DatatypeProperty rdf:about="&h2iosc;#NAME#" rdf:isDefinedBy="#URI#">
        <rdfs:label>#LABEL#</rdfs:label>
        <rdfs:domain rdf:resource="&h2iosc;#DOMAIN#"/>
        <rdfs:isDefinedBy rdf:resource="#URI#"/>
    </owl:DatatypeProperty>
'''
datatype_defined_string = '        <rdfs:isDefinedBy rdf:resource="#URI#"/>\n'

# Define a normalization function for rdf labels for easier portability
def label_to_name(label):
    return label.replace(' ', '_').replace('à', 'a').replace('è', 'e').replace('é', 'e').replace('ì', 'i').replace('ò', 'o').replace('ù', 'u')

# Generic ('propietary') datatypes to std. xsd datatypes mapping
datatype_xsd = {
    "#string": 'string',
    '#uri': '#uri',
    '#number': 'decimal',
    '#date': 'date',
    '#coordinates': '#coordinates'
}

# %%
# Map entity info to dedicated lists
entities_rdf_list = []
entities_csv = []
datatype_properties_rdf_list = []
same_as = list(data['Same_as'].keys())
for label, ent in data['Entità'].items():
    
    entity_name = label_to_name(label)
    entity_rdf = entity_template.replace('#LABEL#', label).replace('#NAME#', entity_name)

    # Try to find entity in extra csv, see if there is CIDOC info and if so, map it
    entity_in_csv = next((ent for ent in ent_csv_in if ent['ENTITÀ']==label), None)
    cidoc_class = None
    if entity_in_csv is not None:
        cidoc_class = entity_in_csv.get('CIDOC-LINK')
    if cidoc_class is not None and cidoc_class!='':
        entity_rdf = entity_rdf.replace('#URI#', cidoc_class)
    else:
        entity_rdf = entity_rdf.replace(class_defined_string, '')

    # Subclasses
    if 'Sottoclasse di' in ent.keys():
        parent = ent['Sottoclasse di']
        data['Relazioni'].append({"Entità 1": label,
            "Entità 2": parent,
            "Etichetta": "is_subclass_of", "Inversa": "is_superclass_of"})
        entity_rdf = entity_rdf.replace('#PARENT#', label_to_name(parent))
    else:
        entity_rdf = entity_rdf.replace(subclass_string, '')

    entities_rdf_list.append(entity_rdf)
    #
    if label in same_as:
        entities_csv.append( [label, "", ', '.join(data['Same_as'][label])] )
    else:
        entities_csv.append([label, "", ""])
    for datatype_label, datatype_val in ent.items():
        if not isinstance(datatype_val, str) or not datatype_val.startswith('#'):
            continue
        entities_csv.append(["", datatype_label, ""])
        datatype_name = label_to_name(datatype_label)
        datatype_rdf = datatype_property_template.replace('#LABEL#', datatype_label).replace('#NAME#', datatype_name).replace('#DOMAIN#', entity_name)
        # Try to find entity in extra csv, see if there is CIDOC info and if so, map it
        datatype_in_csv = next((ent for ent in ent_csv_in if ent['ATTRIBUTO (LITERAL)']==datatype_label), None)
        cidoc_class = None
        if datatype_in_csv is not None:
            cidoc_class = datatype_in_csv.get('CIDOC-LINK')
        if cidoc_class is not None and cidoc_class!='':
            datatype_rdf = datatype_rdf.replace('#URI#', cidoc_class)
        else:
            datatype_rdf = datatype_rdf.replace(datatype_defined_string, '')

        datatype_properties_rdf_list.append(datatype_rdf)


# Map relation info to dedicated lists
relations_rdf_list = []
relations_csv = []
for rel in data['Relazioni']:
    label = rel['Etichetta']
    inverse_label = rel['Inversa']
    domain = label_to_name(rel['Entità 1'])
    range1 = label_to_name(rel['Entità 2'])
    relations_csv.append([rel['Entità 1'], rel['Entità 2'], rel['Etichetta'], rel['Inversa']])
    name = domain + '_' + label_to_name(label) + '_' + range1
    inverse_name = range1 + '_' + label_to_name(inverse_label) + '_' + domain

    # Try to find entity in extra csv, see if there is CIDOC info and if so, map it
    relation_in_csv = next((rel_csv for rel_csv in rel_csv_in if (rel_csv['ENTITÀ 1']==rel['Entità 1'] and rel_csv['ENTITÀ 2']==rel['Entità 2']) ), None)
    cidoc_class = None
    #
    relation_rdf = object_property_template.replace('#NAME#', name).replace('#LABEL#', label).replace('#DOMAIN#', domain).replace('#RANGE#', range1)
    #
    if relation_in_csv is not None:
        cidoc_class = relation_in_csv.get('CIDOC-LINK')
    if cidoc_class is not None and cidoc_class!='':
        relation_rdf = relation_rdf.replace('#URI#', cidoc_class)
    else:
        relation_rdf = relation_rdf.replace(object_defined_string, '')
    #
    relation_inverse_rdf = object_property_inverse_template.replace('#NAME#', inverse_name).replace('#LABEL#', inverse_label).replace('#DOMAIN#', range1).replace('#RANGE#', domain).replace('#INV#', name)
    #
    if cidoc_class is not None and cidoc_class!='':
        relation_inverse_rdf = relation_inverse_rdf.replace('#URI#', cidoc_class)
    else:
        relation_inverse_rdf = relation_inverse_rdf.replace('<rdfs:isDefinedBy rdf:resource="#URI#"/>', '')
    #
    relation_full_rdf = relation_rdf + '\n\n\n' + relation_inverse_rdf
    relations_rdf_list.append(relation_full_rdf)


# Write info to template and export it to output file
with open('draft.rdf', 'w') as out_file:
    to_out = raw_rdf.replace(entity_template, '\n\n\n'.join(entities_rdf_list)).replace(datatype_property_template, '\n\n\n'.join(datatype_properties_rdf_list)).replace(object_property_inverse_template, '\n\n\n'.join(relations_rdf_list))
    out_file.write(to_out)
# %%
# Write info to two csv files (one for Entities, one for Relations) for extra human readability
with open('entities.csv', 'w') as out_csv:
    writer = csv.writer(out_csv)
    writer.writerow(['ENTITÀ', 'ATTRIBUTO (LITERAL)', 'SAME AS'])
    writer.writerows(entities_csv)
with open('relations.csv', 'w') as out_csv:
    writer = csv.writer(out_csv)
    writer.writerow(['ENTITÀ 1', 'ENTITÀ 2', 'NOME RELAZIONE', 'NOME RELAZIONE INVERSA'])
    writer.writerows(relations_csv)
# %%
print(raw_rdf)
# %%
entity_template in raw_rdf
# %%
entity_template
# %%