Parser/Museo/CSV_to_RDF/CSV_to_RDF_mpp_Measurement.py

# Utilities to read/write csv files
import csv
# Utilities to handle character encodings
import unicodedata
# Ordered Dicts
from collections import OrderedDict

import json


# OPZIONAL IMPORTS

# For timestamping/simple speed tests
from datetime import datetime
# Random number generator
from random import *
# System & command line utilities
import sys
# Json for the dictionary
import json

import_dir = '/Users/alessiaspadi/Documents/RESTORE/temp_MPP/tabelle/Ospedale/mod/'
export_dir = '/Users/alessiaspadi/Documents/RESTORE/temp_MPP/tabelle/Ospedale/mod/E54_'

# Custom class to store URIs + related infos for the ontologies/repositories

class RDFcoords:
    def __init__(self, uri, prefix, code = None):
        self.uri = uri
        self.prefix = prefix
        self.code = code


# Repositories
museoCoords = RDFcoords('<http://palazzopretorio.comune.prato.it/it/le-opere/alcuni-capolavori/>', 'mpp:')

cidocCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/>', 'crm:')
aatCoords = RDFcoords('<http://vocab.getty.edu/aat/>', 'aat:')
nsCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#>', 'rdf:')
schemaCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#>', 'rdfs:')


# Basic functions for triples / shortened triples in TTL format

def triple(subject, predicate, object1):
    line = subject + ' ' + predicate + ' ' + object1
    return line

def doublet(predicate, object1):
    line = '    ' + predicate + ' ' + object1
    return line

def singlet(object1):
    line = '        ' + object1
    return line

# Line endings in TTL format
continueLine1 = ' ;\n'
continueLine2 = ' ,\n'
closeLine = ' .\n'

def writeTTLHeader(output):
    output.write('@prefix ' + museoCoords.prefix + ' ' + museoCoords.uri + closeLine)
    output.write('@prefix ' + cidocCoords.prefix + ' ' + cidocCoords.uri + closeLine)
    output.write('@prefix ' + aatCoords.prefix + ' ' + aatCoords.uri + closeLine)
    output.write('@prefix ' + schemaCoords.prefix + ' ' + schemaCoords.uri + closeLine)
    output.write('@prefix ' + nsCoords.prefix + ' ' + nsCoords.uri + closeLine)

    output.write('\n')


filePrefix = 'SR20OA_'
fileType = 'Ospedale'
max_entries = 1000000000

with open(import_dir + filePrefix + fileType + '.csv', newline="") as csv_file, open(
        export_dir + filePrefix + fileType + '.ttl', 'w') as output:
    reader = csv.DictReader(csv_file)
    writeTTLHeader(output)
    first = True
    ii = 0
    for row in reader:
        # The index ii is used to process a limited number of entries for testing purposes
        ii = ii + 1

        # Triplify the 'codice' -- should exist for every entry
        if row['RVEL'] == '' or row['RVEL'] == '0':

            codice = ''
            if (row['NCTR'] != '' and row['NCTN'] != ''):
                codice = row['NCTR'] + row['NCTN']

            unit = ''
            if (row['MISU'] != ''):
                unit = row['MISU']

            valueA = ''
            valueL = ''

            if (row['MISA'] != ''):
                value = row['MISA']
                valueA = value.replace(',', 'v')

            if (row['MISL'] != ''):
                value = row['MISL']
                valueL = value.replace(',', 'v')

            url = row['URL']

            #placeHolders
            datplaceHolder = museoCoords.prefix + url
            e55placeHolder = museoCoords.prefix + url + '_E55'
            e58placeHolder = museoCoords.prefix + url + '_unit_' + unit
            e60placeHolderA = museoCoords.prefix + url + '_value_' + valueA
            e60placeHolderL = museoCoords.prefix + url + '_value_' + valueL

            #Altezza
            #E22 P43 E54
            if (row['MISA'] != ''):
                line = triple(datplaceHolder,
                              cidocCoords.prefix + 'P43_has_dimension',
                              museoCoords.prefix + url + '_Altezza') + closeLine
                output.write(line)
                line = triple(museoCoords.prefix + url + '_Altezza',
                              nsCoords.prefix + 'type',
                              cidocCoords.prefix + 'E54_Dimension') + closeLine
                output.write(line)
                line = triple(museoCoords.prefix + url + '_Altezza',
                              schemaCoords.prefix + 'label',
                              '\"Altezza: ' + row['MISA'] + row['MISU'] +'\"') + closeLine
                output.write(line)

                # E54 P90 E60
                line = triple(museoCoords.prefix + url + '_Altezza',
                              cidocCoords.prefix + 'P90_has_value',
                              e60placeHolderA) + closeLine
                output.write(line)
                line = triple(e60placeHolderA,
                              nsCoords.prefix + 'type',
                              cidocCoords.prefix + 'E60') + closeLine
                output.write(line)
                line = triple(e60placeHolderA,
                              schemaCoords.prefix + 'label',
                              '\"' + row['MISA'] + '\"') + closeLine
                output.write(line)

                #E54 P2 E55

                line = triple(museoCoords.prefix + url + '_Altezza',
                              cidocCoords.prefix + 'P2_has_type',
                              aatCoords.prefix + '300055644') + closeLine
                output.write(line)
                line = triple(aatCoords.prefix + '300055644',
                              schemaCoords.prefix + 'label',
                              '\"altezza\"') + closeLine
                output.write(line)

                # E54 P91 E58
                if (row['MISU'] != ''):
                    line = triple(museoCoords.prefix + url + '_Altezza',
                                  cidocCoords.prefix + 'P91_has_unit',
                                  e58placeHolder) + closeLine
                    output.write(line)
                    line = triple(e58placeHolder,
                                  nsCoords.prefix + 'type',
                                  cidocCoords.prefix + 'E58_Measurement_Unit') + closeLine
                    output.write(line)
                    line = triple(e58placeHolder,
                                  schemaCoords.prefix + 'label',
                                  '\"' + row['MISU'] + '\"') + closeLine
                    output.write(line)

            # Larghezza
            # E22 P43 E54
            if (row['MISL'] != ''):
                line = triple(datplaceHolder,
                              cidocCoords.prefix + 'P43_has_dimension',
                              museoCoords.prefix + url + '_Larghezza') + closeLine
                output.write(line)
                line = triple(museoCoords.prefix + url + '_Larghezza',
                              nsCoords.prefix + 'type',
                              cidocCoords.prefix + 'E54_Dimension') + closeLine
                output.write(line)
                line = triple(museoCoords.prefix + url + '_Larghezza',
                              schemaCoords.prefix + 'label',
                              '\"Larghezza: ' + row['MISL'] + row['MISU'] + '\"') + closeLine
                output.write(line)

                # E54 P90 E60
                line = triple(museoCoords.prefix + url + '_Larghezza',
                              cidocCoords.prefix + 'P90_has_value',
                              e60placeHolderL) + closeLine
                output.write(line)
                line = triple(e60placeHolderL,
                              nsCoords.prefix + 'type',
                              cidocCoords.prefix + 'E60') + closeLine
                output.write(line)
                line = triple(e60placeHolderL,
                              schemaCoords.prefix + 'label',
                              '\"' + row['MISL'] + '\"') + closeLine
                output.write(line)

                # E54 P2 E55

                line = triple(museoCoords.prefix + url + '_Larghezza',
                              cidocCoords.prefix + 'P2_has_type',
                              aatCoords.prefix + '300055647') + closeLine
                output.write(line)
                line = triple(aatCoords.prefix + '300055647',
                              schemaCoords.prefix + 'label',
                              '\"larghezza\"') + closeLine
                output.write(line)

                # E54 P91 E58
                if (row['MISU'] != ''):
                    line = triple(museoCoords.prefix + url + '_Larghezza',
                                  cidocCoords.prefix + 'P91_has_unit',
                                  e58placeHolder) + closeLine
                    output.write(line)
                    line = triple(e58placeHolder,
                                  nsCoords.prefix + 'type',
                                  cidocCoords.prefix + 'E58_Measurement_Unit') + closeLine
                    output.write(line)
                    line = triple(e58placeHolder,
                                  schemaCoords.prefix + 'label',
                                  '\"' + row['MISU'] + '\"') + closeLine
                    output.write(line)


            output.write('\n')
        #
        #
        # Limit number of entries processed (if desired)
        if (ii > max_entries):
            break