{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Utilities to read/write csv files\n",
    "import csv\n",
    "# Utilities to handle character encodings\n",
    "import unicodedata\n",
    "# Ordered Dicts\n",
    "from collections import OrderedDict\n",
    "\n",
    "import json\n",
    "import re\n",
    "\n",
    "\n",
    "# OPZIONAL IMPORTS\n",
    "\n",
    "# For timestamping/simple speed tests\n",
    "from datetime import datetime\n",
    "# Random number generator\n",
    "from random import *\n",
    "# System & command line utilities\n",
    "import sys\n",
    "# Json for the dictionary\n",
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/CSV/ASPO/gettatelli/'\n",
    "export_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/RDF/ASPO/gettatelli/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Custom class to store URIs + related infos for the ontologies/repositories\n",
    "\n",
    "class RDFcoords:\n",
    "    def __init__(self, uri, prefix, code = None):\n",
    "        self.uri = uri\n",
    "        self.prefix = prefix\n",
    "        self.code = code\n",
    "\n",
    "# Repositories\n",
    "datiniCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoGT002/scheda/IT-ASPO-GT002->', 'dt:')\n",
    "personAuthCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-GT002->', 'pa:')\n",
    "# W3/CIDOC Predicates\n",
    "hasTypeCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>', 'tp:')\n",
    "hasNoteCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P3_has_note>', 'no:')\n",
    "hasTypePCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P2_has_type>', 'te:')\n",
    "carriesCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P128_carries>', 'ca:')\n",
    "identifiedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P1_is_identified_by>', 'ib:')\n",
    "labelCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#label>', 'lb:')\n",
    "wasBroughtCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P92i_was_brought_into_existence_by>', 'wb:')\n",
    "wasBornCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P98i_was_born>', 'wbc:')\n",
    "diedCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P100i_died_in>', 'di:')\n",
    "wasPresentCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P12i_was_present_at>', 'wp:')\n",
    "carriedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P14_carried_out_by>', 'cb:')\n",
    "noteCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P3_has_note>', 'nt:')\n",
    "hasTimeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P4_has_time-span>', 'hs:')\n",
    "consistCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P45_consist_of>', 'cf:')\n",
    "hasConditionCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P44_has_condition>', 'hc:')\n",
    "foafCoords = RDFcoords('<http://xmlns.com/foaf/0.1/>', 'foaf:')\n",
    "nsCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#>', 'rdf:')\n",
    "rdfsCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#>', 'rdfs:')\n",
    "aspoCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/>', 'aspo:')\n",
    "\n",
    "# CIDOC Objects\n",
    "manMadeObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E22_Man-Made_Object>', 'mo:', 'E22')\n",
    "informationObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E73_Information_Object>', 'io:', 'E73')\n",
    "titleCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E35_Title>', 'ti:' ,'E35')\n",
    "placeAppellationCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E44_Place_appellation>', 'pa:', 'E44')\n",
    "identifierCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E42_Identifier>', 'id:', 'E42')\n",
    "typeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E55_Type>', 'ty:', 'E55')\n",
    "creationCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E65_Creation>', 'cr:', 'E65')\n",
    "personCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E21_Person>', 'ps:', 'E21')\n",
    "stringCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E62_String>', 'st:', 'E62')\n",
    "birthCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E67_Birth>', 'th:', 'E67')\n",
    "deathCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E69_Death>', 'dh:', 'E69')\n",
    "timeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E52_Time-Span>', 'ts:', 'E52')\n",
    "materialCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E57_Material>', 'mt:', 'E57')\n",
    "conditionCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E3_Condition_State>', 'cs:', 'E3')\n",
    "entityCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E1_Entity>', 'ey:', 'E1')\n",
    "eventCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E5_Event>', 'eve:', 'E5')\n",
    "refersCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P67_refers_to>', 'rt:')\n",
    "refersHasTypeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P67.1_type>', 'rh:')\n",
    "schemaCoords = RDFcoords('<http://schema.org/>', 'schema:')\n",
    "owlCoords = RDFcoords('<http://www.w3.org/2002/07/owl#>', 'owl:')\n",
    "fallsCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E89_falls_within>', 'fw:', 'E89')\n",
    "residenceCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P74_has_current_or_former_residence>', 'res:')\n",
    "placeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E53_Place>', 'pl:', 'E53')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Basic functions for triples / shortened triples in TTL format\n",
    "\n",
    "def triple(subject, predicate, object1):\n",
    "    line = subject + ' ' + predicate + ' ' + object1\n",
    "    return line\n",
    "\n",
    "def doublet(predicate, object1):\n",
    "    line = '    ' + predicate + ' ' + object1\n",
    "    return line\n",
    "\n",
    "def singlet(object1):\n",
    "    line = '        ' + object1\n",
    "    return line\n",
    "\n",
    "# Line endings in TTL format\n",
    "continueLine1 = ' ;\\n'\n",
    "continueLine2 = ' ,\\n'\n",
    "closeLine = ' .\\n'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def writeTTLHeader(output):\n",
    "    output.write('@prefix ' + datiniCoords.prefix + ' ' + datiniCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + personAuthCoords.prefix + ' ' + personAuthCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + hasTypeCoords.prefix + ' ' + hasTypeCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + hasTypePCoords.prefix + ' ' + hasTypePCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + manMadeObjectCoords.prefix + ' ' + manMadeObjectCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + carriesCoords.prefix + ' ' + carriesCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + informationObjectCoords.prefix + ' ' + informationObjectCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + identifiedByCoords.prefix + ' ' + identifiedByCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + titleCoords.prefix + ' ' + titleCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + labelCoords.prefix + ' ' + labelCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + identifierCoords.prefix + ' ' + identifierCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + wasBroughtCoords.prefix + ' ' + wasBroughtCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + typeCoords.prefix + ' ' + typeCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + carriedByCoords.prefix + ' ' + carriedByCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + personCoords.prefix + ' ' + personCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + stringCoords.prefix + ' ' + stringCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + noteCoords.prefix + ' ' + noteCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + wasBornCoords.prefix + ' ' + wasBornCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + diedCoords.prefix + ' ' + diedCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + wasPresentCoords.prefix + ' ' + wasPresentCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + birthCoords.prefix + ' ' + birthCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + deathCoords.prefix + ' ' + deathCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + hasTimeSpanCoords.prefix + ' ' + hasTimeSpanCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + timeSpanCoords.prefix + ' ' + timeSpanCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + materialCoords.prefix + ' ' + materialCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + consistCoords.prefix + ' ' + consistCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + conditionCoords.prefix + ' ' + conditionCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + hasConditionCoords.prefix + ' ' + hasConditionCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + refersCoords.prefix + ' ' + refersCoords.uri + closeLine)  \n",
    "    output.write('@prefix ' + entityCoords.prefix + ' ' + entityCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + refersHasTypeCoords.prefix + ' ' + refersHasTypeCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + hasNoteCoords.prefix + ' ' + hasNoteCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + creationCoords.prefix + ' ' + creationCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + foafCoords.prefix + ' ' + foafCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + nsCoords.prefix + ' ' + nsCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + eventCoords.prefix + ' ' + eventCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + schemaCoords.prefix + ' ' + schemaCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + rdfsCoords.prefix + ' ' + rdfsCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + owlCoords.prefix + ' ' + owlCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + aspoCoords.prefix + ' ' + aspoCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + fallsCoords.prefix + ' ' + fallsCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + residenceCoords.prefix + ' ' + residenceCoords.uri + closeLine)\n",
    "    output.write('@prefix ' + placeCoords.prefix + ' ' + placeCoords.uri + closeLine)\n",
    "\n",
    "    output.write('\\n')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "filePrefix = 'data_'\n",
    "fileType = 'item'\n",
    "max_entries = 1000000000\n",
    "\n",
    "with open(import_dir + filePrefix + fileType + '_persone_newdataset.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '_persone_newdataset.ttl', 'w') as output:\n",
    "    reader = csv.DictReader(csv_file)\n",
    "    writeTTLHeader(output)\n",
    "    first = True\n",
    "    ii = 0\n",
    "    for row in reader:\n",
    "        # The index ii is used to process a limited number of entries for testing purposes\n",
    "        ii = ii+0\n",
    "        if (row['EAD_EAC'] == 'FALSE'):    \n",
    "            e21placeHolder = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] + '>'\n",
    "            #e22placeHolder = \"<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoGT002/scheda/\" + row['ID_persona'] + \">\"\n",
    "            e65placeHolder = \"<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoGT002/scheda/\" + row['ID_persona'] + \"/\" + creationCoords.code + \">\"\n",
    "\n",
    "            #line = triple(e22placeHolder, refersCoords.prefix, e21placeHolder) + closeLine\n",
    "            #output.write(line)\n",
    "            if (row['gettatello'] == 'TRUE'):\n",
    "                e55placeHolder = '<http://www.archiviodistato.prato.it/gettatello' + \">\"\n",
    "                line = triple(e21placeHolder, hasTypePCoords.prefix, e55placeHolder) + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e55placeHolder, labelCoords.prefix, '\\\"Gettatello\\\"') + closeLine\n",
    "                output.write(line)\n",
    "            \n",
    "            # nome rilevato\n",
    "            if(row['nome rilevato'] != ''):\n",
    "                line = triple(e21placeHolder, hasTypeCoords.prefix, personCoords.prefix) + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e21placeHolder, foafCoords.prefix + 'name', '\\\"' + row['nome rilevato'] + '\\\"') + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e21placeHolder, labelCoords.prefix, '\\\"' + row['nome rilevato'] + '\\\"') +  closeLine\n",
    "                output.write(line)\n",
    "                e62placeHolder = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] +'/E62>'\n",
    "                line = triple(e21placeHolder, noteCoords.prefix, e62placeHolder) +  closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e62placeHolder, hasTypeCoords.prefix, stringCoords.prefix) +  closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e62placeHolder, labelCoords.prefix, '\\\"Fonte: Archivio di Stato di Prato - Fondo Ospedale della Misericordia e Dolce\\\"') +  closeLine\n",
    "                output.write(line)\n",
    "\n",
    "            # codice gettatello\n",
    "            e42placeHolder = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] + '/E42>'\n",
    "            if(row['gettatello'] == 'TRUE' and row['codice gettatello numero'] != ''):\n",
    "                line = triple(e21placeHolder, identifiedByCoords.prefix, e42placeHolder) + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e42placeHolder, hasTypeCoords.prefix, identifierCoords.prefix) + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e42placeHolder, labelCoords.prefix, '\\\"Matricola: ' + row['codice gettatello numero'] + '\\\"') + closeLine\n",
    "                output.write(line)\n",
    "            \n",
    "            # soprannome\n",
    "            if row['soprannome'] != '' and row['soprannome'] != ' ':\n",
    "                    #Remove all white-space characters:\n",
    "                    txt = row['soprannome']\n",
    "                    x = re.sub(\"\\n\", \" \", txt)\n",
    "                    y = re.sub(\"\\s\\s\", \"\", x)\n",
    "                    line = triple(e21placeHolder,\n",
    "                                schemaCoords.prefix + 'alternateName',\n",
    "                                '\\\"' + y + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "            \n",
    "            if row['nome proprio'] != '':\n",
    "                    #Remove all white-space characters:\n",
    "                    txt = row['nome proprio']\n",
    "                    x = re.sub(\" \\n\", \"\", txt)\n",
    "                    y = re.sub(\"\\s\\s\", \"\", x)\n",
    "                    name = re.sub(\"\\n\", \"\", y)\n",
    "                    line = triple(e21placeHolder,\n",
    "                                foafCoords.prefix + 'givenName',\n",
    "                                '\\\"' + name + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "\n",
    "            if row['cognome'] != '':\n",
    "                    #Remove all white-space characters:\n",
    "                    txt = row['cognome']\n",
    "                    x = re.sub(\"\\n\", \" \", txt)\n",
    "                    y = re.sub(\"\\s\\s\", \"\", x)\n",
    "                    line = triple(e21placeHolder,\n",
    "                                foafCoords.prefix + 'familyName',\n",
    "                                '\\\"' + y + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "\n",
    "            # nome_alternativo\n",
    "            if row['nome_alternativo'] != '': \n",
    "                line = triple(e21placeHolder,\n",
    "                owlCoords.prefix + 'sameAs',\n",
    "                aspoCoords.prefix + row['nome_alternativo'].replace(' ', '_')) + closeLine\n",
    "                output.write(line)\n",
    "                line = triple(aspoCoords.prefix + row['nome_alternativo'].replace(' ', '_'),\n",
    "                rdfsCoords.prefix + 'label',\n",
    "                '\\\"' + row['nome_alternativo'] + '\\\"') + closeLine\n",
    "                output.write(line)\n",
    "\n",
    "            # genere\n",
    "            if row['m/f'] != '':\n",
    "                    #Remove all white-space characters:\n",
    "                    txt = row['m/f']\n",
    "                    x = re.sub(\"\\n\", \" \", txt)\n",
    "                    y = re.sub(\"\\s\\s\", \"\", x)\n",
    "                    line = triple(e21placeHolder,\n",
    "                                foafCoords.prefix + 'gender',\n",
    "                                '\\\"' + y + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "            \n",
    "            # patronimico\n",
    "            if row['patronimico'] != '':\n",
    "                    #Remove all white-space characters:\n",
    "                    txt = row['patronimico']\n",
    "                    x = re.sub(\"\\n\", \" \", txt)\n",
    "                    y = re.sub(\"\\s\\s\", \"\", x)\n",
    "                    line = triple(e21placeHolder,\n",
    "                                personCoords.prefix + 'patronymicName',\n",
    "                                '\\\"' + y + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "            \n",
    "            # professione     \n",
    "            if row['professione'] != '' and row['professione'] != ' ' :\n",
    "                        occupationPlaceHolder = '<http://www.archiviodistato.prato.it/' + row['professione'].replace(\" \",\"_\") + '>'\n",
    "                        line = triple(e21placeHolder,\n",
    "                                    schemaCoords.prefix + 'hasOccupation',\n",
    "                                    occupationPlaceHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(occupationPlaceHolder,\n",
    "                                    nsCoords.prefix + 'type',\n",
    "                                    schemaCoords.prefix + 'Occupation') + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(occupationPlaceHolder,\n",
    "                                    rdfsCoords.prefix + 'label',\n",
    "                                    '\\\"' + row['professione'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "            # avo 1\n",
    "            if row['avo 1'] != '':\n",
    "                    avo1 = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] + \"/avo1>\"\n",
    "                    line = triple(e21placeHolder,\n",
    "                                schemaCoords.prefix + 'relatedTo',\n",
    "                                avo1) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(avo1,\n",
    "                                nsCoords.prefix + 'type',\n",
    "                                foafCoords.prefix + 'Person') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(avo1,\n",
    "                                rdfsCoords.prefix + 'label',\n",
    "                                '\\\"' + row['avo 1'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "\n",
    "            if row['avo 2'] != '':\n",
    "                    avo2 = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] + \"/avo2>\"\n",
    "                    line = triple(e21placeHolder,\n",
    "                                schemaCoords.prefix + 'relatedTo',\n",
    "                                avo2) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(avo2,\n",
    "                                nsCoords.prefix + 'type',\n",
    "                                foafCoords.prefix + 'Person') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(avo2,\n",
    "                                rdfsCoords.prefix + 'label',\n",
    "                                '\\\"' + row['avo 2'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "            \n",
    "            # titolo \n",
    "            if row['titolo'] != '':\n",
    "                txt = row['titolo']\n",
    "                x = re.sub(\"\\n\", \" \", txt)\n",
    "                y = re.sub(\"\\s\\s\", \" \", x)\n",
    "                line = triple(e21placeHolder, schemaCoords.prefix + 'honorificPrefix', '\\\"' + y + '\\\"') + closeLine\n",
    "                output.write(line)\n",
    "            if row['titolo 2'] != '':\n",
    "                txt = row['titolo 2']\n",
    "                x = re.sub(\"\\n\", \" \", txt)\n",
    "                y = re.sub(\"\\s\\s\", \" \", x)\n",
    "                line = triple(e21placeHolder, schemaCoords.prefix + 'honorificPrefix', '\\\"' + y + '\\\"') + closeLine\n",
    "                output.write(line)\n",
    "\n",
    "            # provenienza microtoponimo\n",
    "            if(row['PROVENIENZA_microtoponimo'] != ''):\n",
    "                if(row['IDASPO_GEO_PROVENIENZA_microtoponimo'] != ''):\n",
    "                    e53placeHolder = \"<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-GEO0001-\" + row['IDASPO_GEO_PROVENIENZA_microtoponimo'].replace('IT-ASPO-GEO0001-', '') + \">\"\n",
    "                    line = triple(e21placeHolder, residenceCoords.prefix, e53placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                else:\n",
    "                    e53placeHolder = '<http://www.dev.restore.ovi.cnr.it/vocabularies/places/' + row['ID_VOCABOLARIO_PROVENIENZA_microtoponimo'] + '>'\n",
    "                    line = triple(e53placeHolder, labelCoords.prefix, '\\\"' + row['PROVENIENZA_microtoponimo'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e53placeHolder, hasTypeCoords.prefix, placeCoords.prefix) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e21placeHolder, residenceCoords.prefix, e53placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "            # provenienza macrotoponimo \n",
    "            if(row['PROVENIENZA_macrotoponimo'] != '' and row['PROVENIENZA_microtoponimo'] != ''):\n",
    "                        if(row['IDASPO_GEO_PROVENIENZA_macrotoponimo'] != ''):\n",
    "                            e53mplaceHolder = \"<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-GEO0001-\" + row['IDASPO_GEO_PROVENIENZA_macrotoponimo'].replace('IT-ASPO-GEO0001-', '') + \">\"\n",
    "                            line = triple(e53placeHolder, fallsCoords.prefix, e53mplaceHolder) + closeLine\n",
    "                            output.write(line)\n",
    "                        else:\n",
    "                            e53mplaceHolder = '<http://www.dev.restore.ovi.cnr.it/vocabularies/places/' + row['ID_VOCABOLARIO_PROVENIENZA_macrotoponimo'] + '>'\n",
    "                            line = triple(e53mplaceHolder, labelCoords.prefix, '\\\"' + row['PROVENIENZA_macrotoponimo'] + '\\\"') + closeLine\n",
    "                            output.write(line)\n",
    "                            line = triple(e53mplaceHolder, hasTypeCoords.prefix, placeCoords.prefix) + closeLine\n",
    "                            output.write(line)\n",
    "                            line = triple(e53placeHolder, fallsCoords.prefix, e53mplaceHolder) + closeLine\n",
    "                            output.write(line)\n",
    "            elif (row['PROVENIENZA_macrotoponimo'] != '' and row['PROVENIENZA_microtoponimo'] == ''):\n",
    "                if(row['IDASPO_GEO_PROVENIENZA_macrotoponimo'] != ''):\n",
    "                    e53placeHolder = \"<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-GEO0001-\" + row['IDASPO_GEO_PROVENIENZA_macrotoponimo'].replace('IT-ASPO-GEO0001-', '') + \">\"\n",
    "                    line = triple(e21placeHolder, residenceCoords.prefix, e53placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                else:\n",
    "                    e53placeHolder = '<http://www.dev.restore.ovi.cnr.it/vocabularies/places/'+ row['ID_VOCABOLARIO_PROVENIENZA_macrotoponimo'] + '>'\n",
    "                    line = triple(e53placeHolder, labelCoords.prefix, '\\\"' + row['PROVENIENZA_macrotoponimo'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e53placeHolder, hasTypeCoords.prefix, placeCoords.prefix) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e21placeHolder, residenceCoords.prefix, e53placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "\n",
    "            # note\n",
    "            if(row['note'] != ''):\n",
    "                e62placeHolder = '<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/' + row['ID_persona'] +'/E62P>'\n",
    "                line = triple(e21placeHolder, noteCoords.prefix, e62placeHolder) +  closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e62placeHolder, hasTypeCoords.prefix, stringCoords.prefix) +  closeLine\n",
    "                output.write(line)\n",
    "                line = triple(e62placeHolder, labelCoords.prefix, '\\\"'+ row['note'] + '\\\"') +  closeLine\n",
    "                output.write(line)\n",
    "\n",
    "            output.write('\\n')\n",
    "        #\n",
    "        #\n",
    "        # Limit number of entries processed (if desired)\n",
    "        if(ii>max_entries):\n",
    "            break\n",
    "        "
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
  },
  "kernelspec": {
   "display_name": "Python 3.7.3 64-bit",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.0"
  },
  "metadata": {
   "interpreter": {
    "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}