Update parsers

EAC_to_CSV

@@ -1,77 +0,0 @@
-import xml.etree.ElementTree as Xet
-from typing import Dict, Any
-import pandas as pd
-import os
-import csv
-from xml.dom import minidom
-import sys
-import re
-
-xml_file_name = '/Users/federicaspinelli/Google Drive/OVI:CNR/LAVORO 2020/SELEZIONE CONTENUTI/01_ASPO/XDAMS/export_aspoAuth002--22_09_20_17_34-3se381075198510401.xml'
-tree = Xet.parse(xml_file_name)
-root = tree.getroot()
-schede = root.findall("xw_doc")
-
-eac_data = open('eac_Data.csv', 'w')
-csvwriter = csv.writer(eac_data)
-
-params = ["recordId", "entityType", "nameEntry@normal", "nameEntry@prime",  "biogHist p",
-          "nameEntry@ulterior", "nameEntry@ord", "place", "occupation"]
-
-csvwriter.writerow(params)
-
-for scheda in schede:
-    ul = []
-    identifier = scheda.find("eac-cpf/control/recordId")
-    entityType = scheda.find("eac-cpf/cpfDescription/identity/entityType")
-    nameEntry_N = scheda.find("eac-cpf/cpfDescription/identity/nameEntry/part[@localType='normal']")
-    nameEntry_P = scheda.find("eac-cpf/cpfDescription/identity/nameEntry/part[@localType='prime']")
-    description = scheda.find("eac-cpf/cpfDescription/description/biogHist/p")
-    nameEntry_U = scheda.find("eac-cpf/cpfDescription/identity/nameEntry/part[@localType='ulterior']")
-    nameEntry_O = scheda.find("eac-cpf/cpfDescription/identity/nameEntry/part[@localType='ord']")
-    places = scheda.findall("eac-cpf/cpfDescription/description/places/place/placeEntry")
-    occupation = scheda.find("eac-cpf/cpfDescription/description/occupation/term")
-    ul.append(identifier.text)
-    ul.append(entityType.text)
-    if nameEntry_N is None:
-        ul.append(" ")
-    else:
-        ul.append(nameEntry_N.text)
-
-    if nameEntry_P is None:
-        ul.append(" ")
-    else:
-        ul.append(nameEntry_P.text)
-
-    if description is None:
-       ul.append(" ")
-    else:
-        ul.append(description.text)
-
-    if nameEntry_U is None:
-        ul.append(" ")
-    else:
-        ul.append(nameEntry_U.text)
-
-    if nameEntry_O is None:
-        ul.append(" ")
-    else:
-        ul.append(nameEntry_O.text)
-
-    if places is None:
-        ul.append(" ")
-    else:
-        cell = []
-        for place in places:
-            cell.append(place.text)
-        ul.append(" | ".join(cell))
-    
-    if occupation is None:
-        ul.append(" ")
-    else:
-        ul.append(occupation.text)
-
-    csvwriter.writerow(ul)
-
-
-eac_data.close()

OVI/CSV_to_RDF/CSV_to_RDF_ovi_date.ipynb

@@ -0,0 +1,246 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Utilities to read/write csv files\n",
+    "import csv\n",
+    "import unicodedata\n",
+    "# Ordered Dicts\n",
+    "from collections import OrderedDict\n",
+    "import json\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\n",
+    "import re"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 14,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/CSV/OVI/'\n",
+    "export_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/RDF/OVI/'"
+   ]
+  },
+  {
+   "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",
+    "\n",
+    "# Repositories\n",
+    "datiniCoords = RDFcoords('<http://datini.archiviodistato.prato.it/la-ricerca/scheda/>', 'dt:')\n",
+    "personAuthCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-AU00003->', 'pa:')\n",
+    "# W3 Predicates\n",
+    "hasTypeCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>', 'tp:')\n",
+    "labelCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#label>', 'lb:')\n",
+    "# CIDOC Predicates\n",
+    "identifiedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P1_is_identified_by>', 'ib:')\n",
+    "hasTypePCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P2_has_type>', 'te:')\n",
+    "hasTimeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P4_has_time-span>', 'hs:')\n",
+    "tookPlaceCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P7_took_place_at>', 'tk:')\n",
+    "carriedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P14_carried_out_by>', 'cb:')\n",
+    "movedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P25i_moved_by>', 'mb:')\n",
+    "movedToCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P26_moved_to>', 'mt:')\n",
+    "movedFromCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P27_moved_from>', 'mf:')\n",
+    "wasBroughtCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P92i_was_brought_into_existence_by>', 'wb:')\n",
+    "hasProducedCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P108_has_produced>', 'hp:')\n",
+    "wasProducedCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P108i_was_produced_by>', 'wp:')\n",
+    "carriesCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P128_carries>', 'ca:')\n",
+    "hasAlternativeFormCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P139_has_alternative_form>', 'af:')\n",
+    "onGoingTCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P81_ongoing_throughout>', 'gt:')\n",
+    "nsCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#>', 'rdf:')\n",
+    "schemaCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#>', 'rdfs:')\n",
+    "# CIDOC Objects\n",
+    "moveCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E9_Move>', 'mv:', 'E9')\n",
+    "productionCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E12_Production>', 'pr:', 'E12')\n",
+    "personCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E21_Person>', 'ps:', 'E21')\n",
+    "manMadeObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E22_Man-Made_Object>', 'mo:', 'E22')\n",
+    "titleCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E35_Title>', 'ti:' ,'E35')\n",
+    "identifierCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E42_Identifier>', 'id:', 'E42')\n",
+    "#placeAppellationCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E44_Place_appellation>', 'pa:', 'E44')\n",
+    "timeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E52_Time-Span>', 'ts:', 'E52')\n",
+    "placeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E53_Place>', 'pl:', 'E53')\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",
+    "informationObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E73_Information_Object>', 'io:', 'E73')\n",
+    "# New classes (subclasses of E7 Activity) - Exchange, Sending, Recive Letters\n",
+    "exchangeLettersCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL1_Exchange_Letters>', 'el:', 'EL1')\n",
+    "sendLetterCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL2_Send_Letter>', 'sl:', 'EL2')\n",
+    "receiveLetterCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL3_Receive_Letter>', 'rl:', 'EL3')"
+   ]
+  },
+  {
+   "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 ' + labelCoords.prefix + ' ' + labelCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + identifiedByCoords.prefix + ' ' + identifiedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasTypePCoords.prefix + ' ' + hasTypePCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasTimeSpanCoords.prefix + ' ' + hasTimeSpanCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + tookPlaceCoords.prefix + ' ' + tookPlaceCoords.uri + closeLine)        \n",
+    "    output.write('@prefix ' + carriedByCoords.prefix + ' ' + carriedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedByCoords.prefix + ' ' + movedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedToCoords.prefix + ' ' + movedToCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedFromCoords.prefix + ' ' + movedFromCoords.uri + closeLine)     \n",
+    "    output.write('@prefix ' + wasBroughtCoords.prefix + ' ' + wasBroughtCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasProducedCoords.prefix + ' ' + hasProducedCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + wasProducedCoords.prefix + ' ' + wasProducedCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + carriesCoords.prefix + ' ' + carriesCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasAlternativeFormCoords.prefix + ' ' + hasAlternativeFormCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + moveCoords.prefix + ' ' + moveCoords.uri + closeLine) \n",
+    "    output.write('@prefix ' + productionCoords.prefix + ' ' + productionCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + personCoords.prefix + ' ' + personCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + manMadeObjectCoords.prefix + ' ' + manMadeObjectCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + titleCoords.prefix + ' ' + titleCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + identifierCoords.prefix + ' ' + identifierCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + timeSpanCoords.prefix + ' ' + timeSpanCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + placeCoords.prefix + ' ' + placeCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + typeCoords.prefix + ' ' + typeCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + creationCoords.prefix + ' ' + creationCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + informationObjectCoords.prefix + ' ' + informationObjectCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + exchangeLettersCoords.prefix + ' ' + exchangeLettersCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + sendLetterCoords.prefix + ' ' + sendLetterCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + receiveLetterCoords.prefix + ' ' + receiveLetterCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + onGoingTCoords.prefix + ' ' + onGoingTCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + schemaCoords.prefix + ' ' + schemaCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + nsCoords.prefix + ' ' + nsCoords.uri + closeLine)\n",
+    "\n",
+    "    output.write('\\n')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "filePrefix = 'Biblio'\n",
+    "fileType = 'Datini'\n",
+    "max_entries = 1000000000\n",
+    "\n",
+    "with open(import_dir + filePrefix + fileType + '_IDAspo.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '_date.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+1\n",
+    "        # Skip the first line as it carries info we don't want to triplify\n",
+    "        if(first):\n",
+    "            first = False\n",
+    "            continue\n",
+    "        \n",
+    "        #Evento send letter\n",
+    "        el2placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + sendLetterCoords.code + \"_OVI>\"\n",
+    "        el3placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + receiveLetterCoords.code + \"_OVI>\"\n",
+    "\n",
+    "        # Data invio\n",
+    "        if(row['data_partenza'] != ''):\n",
+    "            e52PplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + sendLetterCoords.code + \"_\" + timeSpanCoords.code + \"_OVI>\"\n",
+    "            line = triple(el2placeHolder, e52PplaceHolder, '\\\"'+ row['data_partenza'] + '\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "\n",
+    "        if(row['data_arrivo'] != ''):\n",
+    "            e52AplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + receiveLetterCoords.code + \"_\" + timeSpanCoords.code + \"_OVI>\"\n",
+    "            line = triple(el3placeHolder, e52AplaceHolder, '\\\"' + row['data_arrivo'] + '\\\"') + 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": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
+  },
+  "kernelspec": {
+   "display_name": "Python 3.9.0 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
+}

OVI/CSV_to_RDF/CSV_to_RDF_ovi_event_exchange.ipynb

@@ -0,0 +1,280 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 85,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Utilities to read/write csv files\n",
+    "import csv\n",
+    "import unicodedata\n",
+    "# Ordered Dicts\n",
+    "from collections import OrderedDict\n",
+    "import json\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\n",
+    "import re"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 86,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/CSV/OVI/'\n",
+    "export_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/RDF/OVI/'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 87,
+   "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",
+    "\n",
+    "# Repositories\n",
+    "datiniCoords = RDFcoords('<http://datini.archiviodistato.prato.it/la-ricerca/scheda/>', 'dt:')\n",
+    "personAuthCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-AU00003->', 'pa:')\n",
+    "# W3 Predicates\n",
+    "hasTypeCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>', 'tp:')\n",
+    "labelCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#label>', 'lb:')\n",
+    "subClassOfCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#subClassOf>', 'so:')\n",
+    "# CIDOC Predicates\n",
+    "identifiedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P1_is_identified_by>', 'ib:')\n",
+    "hasTypePCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P2_has_type>', 'te:')\n",
+    "hasTimeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P4_has_time-span>', 'hs:')\n",
+    "tookPlaceCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P7_took_place_at>', 'tk:')\n",
+    "carriedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P14_carried_out_by>', 'cb:')\n",
+    "movedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P25i_moved_by>', 'mb:')\n",
+    "movedToCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P26_moved_to>', 'mt:')\n",
+    "movedFromCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P27_moved_from>', 'mf:')\n",
+    "wasBroughtCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P92i_was_brought_into_existence_by>', 'wb:')\n",
+    "hasProducedCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P108_has_produced>', 'hp:')\n",
+    "wasProducedCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P108i_was_produced_by>', 'wp:')\n",
+    "carriesCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P128_carries>', 'ca:')\n",
+    "hasAlternativeFormCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P139_has_alternative_form>', 'af:')\n",
+    "noteCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P3_has_note>', 'nt:')\n",
+    "roleOfCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P14.1_in_the_role_of>', 'ro:')\n",
+    "hasDomainCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P01_has_domain>', 'hd:')\n",
+    "hasRangeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P02_has_range>', 'hr:')\n",
+    "\n",
+    "# CIDOC Objects\n",
+    "moveCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E9_Move>', 'mv:', 'E9')\n",
+    "productionCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E12_Production>', 'pr:', 'E12')\n",
+    "personCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E21_Person>', 'ps:', 'E21')\n",
+    "manMadeObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E22_Man-Made_Object>', 'mo:', 'E22')\n",
+    "titleCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E35_Title>', 'ti:' ,'E35')\n",
+    "identifierCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E42_Identifier>', 'id:', 'E42')\n",
+    "#placeAppellationCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E44_Place_appellation>', 'pa:', 'E44')\n",
+    "timeSpanCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E52_Time-Span>', 'ts:', 'E52')\n",
+    "placeCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E53_Place>', 'pl:', 'E53')\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",
+    "informationObjectCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E73_Information_Object>', 'io:', 'E73')\n",
+    "pcarriedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/PC14_carried_out_by>', 'cy:')\n",
+    "\n",
+    "# New classes (subclasses of E7 Activity) - Exchange, Sending, Recive Letters\n",
+    "exchangeLettersCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL1_Exchange_Letters>', 'el:', 'EL1')\n",
+    "sendLetterCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL2_Send_Letter>', 'sl:', 'EL2')\n",
+    "receiveLetterCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/EL3_Receive_Letter>', 'rl:', 'EL3')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 88,
+   "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": 89,
+   "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 ' + labelCoords.prefix + ' ' + labelCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + identifiedByCoords.prefix + ' ' + identifiedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasTypePCoords.prefix + ' ' + hasTypePCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasTimeSpanCoords.prefix + ' ' + hasTimeSpanCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + tookPlaceCoords.prefix + ' ' + tookPlaceCoords.uri + closeLine)        \n",
+    "    output.write('@prefix ' + carriedByCoords.prefix + ' ' + carriedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedByCoords.prefix + ' ' + movedByCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedToCoords.prefix + ' ' + movedToCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + movedFromCoords.prefix + ' ' + movedFromCoords.uri + closeLine)     \n",
+    "    output.write('@prefix ' + wasBroughtCoords.prefix + ' ' + wasBroughtCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasProducedCoords.prefix + ' ' + hasProducedCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + wasProducedCoords.prefix + ' ' + wasProducedCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + carriesCoords.prefix + ' ' + carriesCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasAlternativeFormCoords.prefix + ' ' + hasAlternativeFormCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + moveCoords.prefix + ' ' + moveCoords.uri + closeLine) \n",
+    "    output.write('@prefix ' + productionCoords.prefix + ' ' + productionCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + personCoords.prefix + ' ' + personCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + manMadeObjectCoords.prefix + ' ' + manMadeObjectCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + titleCoords.prefix + ' ' + titleCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + identifierCoords.prefix + ' ' + identifierCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + timeSpanCoords.prefix + ' ' + timeSpanCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + placeCoords.prefix + ' ' + placeCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + typeCoords.prefix + ' ' + typeCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + creationCoords.prefix + ' ' + creationCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + informationObjectCoords.prefix + ' ' + informationObjectCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + exchangeLettersCoords.prefix + ' ' + exchangeLettersCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + sendLetterCoords.prefix + ' ' + sendLetterCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + receiveLetterCoords.prefix + ' ' + receiveLetterCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + subClassOfCoords.prefix + ' ' + subClassOfCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + noteCoords.prefix + ' ' + noteCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + roleOfCoords.prefix + ' ' + roleOfCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasDomainCoords.prefix + ' ' + hasDomainCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasRangeCoords.prefix + ' ' + hasRangeCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + pcarriedByCoords.prefix + ' ' + pcarriedByCoords.uri + closeLine)\n",
+    "    \n",
+    "    \n",
+    "    \n",
+    "    output.write('\\n')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 90,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "filePrefix = 'Biblio'\n",
+    "fileType = 'Datini'\n",
+    "max_entries = 1000000000\n",
+    "\n",
+    "with open(import_dir + filePrefix + fileType + '_IDAspo.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '_event_exchange.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+1\n",
+    "        # Skip the first line as it carries info we don't want to triplify\n",
+    "        if(first):\n",
+    "            first = False\n",
+    "            continue   \n",
+    "\n",
+    "        el1placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + exchangeLettersCoords.code + \"_OVI>\"\n",
+    "        el2placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + sendLetterCoords.code + \"_OVI>\"\n",
+    "        el3placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + receiveLetterCoords.code + \"_OVI>\"\n",
+    "        #PC14splaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + sendLetterCoords.code + \"_\" + pcarriedByCoords.code + \"_OVI>\"   \n",
+    "        #PC14rplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + receiveLetterCoords.code + \"_\" + pcarriedByCoords.code + \"_OVI>\"   \n",
+    "        #E55splaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + sendLetterCoords.code + \"_\" + typeCoords.code + \"_OVI>\"       \n",
+    "        #E55rplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + receiveLetterCoords.code + \"_\" + typeCoords.code + \"_OVI>\"\n",
+    "    \n",
+    "        if(row['luogo_emissione'] != '' and row['luogo_arrivo'] != ''):\n",
+    "            line = triple(exchangeLettersCoords.prefix, labelCoords.prefix, '\\\"Exchange of letters\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(sendLetterCoords.prefix, labelCoords.prefix, '\\\"Send letter\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(receiveLetterCoords.prefix, labelCoords.prefix, '\\\"Receive letter\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el2placeHolder, subClassOfCoords.prefix, el1placeHolder) + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el3placeHolder, subClassOfCoords.prefix, el1placeHolder) + closeLine\n",
+    "            output.write(line)    \n",
+    "            line = triple(datiniCoords.prefix + row['id'], movedByCoords.prefix, el1placeHolder) + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el1placeHolder, labelCoords.prefix, '\\\"' + row['tipologia'] + ': ' + row['titolo'] + '\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el1placeHolder, hasTypeCoords.prefix, exchangeLettersCoords.prefix) + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el2placeHolder, labelCoords.prefix, '\\\"Invio\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el2placeHolder, hasTypeCoords.prefix, sendLetterCoords.prefix) + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el3placeHolder, labelCoords.prefix, '\\\"Ricezione\\\"') + closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(el3placeHolder, hasTypeCoords.prefix, receiveLetterCoords.prefix) + closeLine\n",
+    "            output.write(line)\n",
+    "            e62placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + exchangeLettersCoords.code + \"_OVI_E62>\" \n",
+    "            line = triple(el1placeHolder, noteCoords.prefix, e62placeHolder) +  closeLine\n",
+    "            output.write(line)\n",
+    "            line = triple(e62placeHolder, labelCoords.prefix, '\\\"Fonte: Istituto Opera del Vocavolario Italiano\\\"') + 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": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
+  },
+  "kernelspec": {
+   "display_name": "Python 3.9.0 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
+}

OVI/CSV_to_RDF/CSV_to_RDF_ovi_trascr.ipynb

@@ -0,0 +1,294 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Utilities to read/write csv files\n",
+    "import csv\n",
+    "import unicodedata\n",
+    "# Ordered Dicts\n",
+    "from collections import OrderedDict\n",
+    "import json\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/OVI/'\n",
+    "export_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/RDF/OVI/'"
+   ]
+  },
+  {
+   "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",
+    "museoCoords = RDFcoords('<http://palazzopretorio.comune.prato.it/it/le-opere/alcuni-capolavori/>', 'mpp:')\n",
+    "autCoords = RDFcoords('<http://palazzopretorio.comune.prato.it/it/opere/autori/>', 'aut:')\n",
+    "cidocCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/>', 'crm:')\n",
+    "aatCoords = RDFcoords('<http://vocab.getty.edu/aat/>', 'aat:')\n",
+    "nsCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#>', 'rdf:')\n",
+    "schemaCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#>', 'rdfs:')\n",
+    "\n",
+    "# Repositories\n",
+    "datiniCoords = RDFcoords('<http://datini.archiviodistato.prato.it/la-ricerca/scheda/>', 'dt:')\n",
+    "personAuthCoords = RDFcoords('<http://www.archiviodistato.prato.it/accedi-e-consulta/aspoMV001/scheda/IT-ASPO-AU00003->', 'pa:')\n",
+    "# W3/CIDOC Predicates\n",
+    "hasTypeCoords = RDFcoords('<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>', 'tp:')\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",
+    "carriedByCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P14_carried_out_by>', 'cb:')\n",
+    "hasAlternativeFormCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P139_has_alternative_form>', 'af:')\n",
+    "hasNoteCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P3_has_note>', 'no:')\n",
+    "hasTypeNCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P3.1_has_type>', 'tn:')\n",
+    "hasLanguageCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P72_has_language>', 'hl:')\n",
+    "documentsCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P70_documents>', 'ds:')\n",
+    "hasComponentCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/P148_has_component>', 'hc:')\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>', 'sr:', 'E62')\n",
+    "linguisticObjCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E33_Linguistic_Object>', 'lj:', 'E33')\n",
+    "languageCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E56_Language>', 'ln:', 'E56')\n",
+    "appellationCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E41_appellation>', 'ap:', 'E41')\n",
+    "propositionalObjCoords = RDFcoords('<http://www.cidoc-crm.org/cidoc-crm/E89_Propositional_Object>', 'pj:', 'E89')\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 ' + hasAlternativeFormCoords.prefix + ' ' + hasAlternativeFormCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasNoteCoords.prefix + ' ' + hasNoteCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasTypeNCoords.prefix + ' ' + hasTypeNCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + stringCoords.prefix + ' ' + stringCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + linguisticObjCoords.prefix + ' ' + linguisticObjCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + languageCoords.prefix + ' ' + languageCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasLanguageCoords.prefix + ' ' + hasLanguageCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + documentsCoords.prefix + ' ' + documentsCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + appellationCoords.prefix + ' ' + appellationCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + propositionalObjCoords.prefix + ' ' + propositionalObjCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + hasComponentCoords.prefix + ' ' + hasComponentCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + museoCoords.prefix + ' ' + museoCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + cidocCoords.prefix + ' ' + cidocCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + aatCoords.prefix + ' ' + aatCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + schemaCoords.prefix + ' ' + schemaCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + nsCoords.prefix + ' ' + nsCoords.uri + closeLine)\n",
+    "    output.write('@prefix ' + autCoords.prefix + ' ' + autCoords.uri + closeLine)\n",
+    "\n",
+    "    output.write('\\n')\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {
+    "tags": []
+   },
+   "outputs": [],
+   "source": [
+    "filePrefix = 'Biblio'\n",
+    "fileType = 'Datini'\n",
+    "max_entries = 10000000000000000\n",
+    "\n",
+    "def get_text(sigla):\n",
+    "    text_file = open('/Users/federicaspinelli/Google Drive/OVI-CNR/CSV/OVI/testi_ovi.csv', newline=\"\")\n",
+    "    reader = csv.DictReader(text_file)\n",
+    "    for row in reader:\n",
+    "        if row['sigla'] == sigla:\n",
+    "            return [row['testo'], row['testo_lemmatizzato']]\n",
+    "\n",
+    "with open(import_dir + filePrefix + fileType + '_IDAspo.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '_trascr.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",
+    "        # Skip the first line as it carries info we don't want to triplify\n",
+    "        if(first):\n",
+    "            first = False\n",
+    "            continue\n",
+    "        \n",
+    "        if(row['num_ovi'] != '2'):\n",
+    "            E73placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + informationObjectCoords.code + \"_OVI\"\n",
+    "            E33placeHolder = '<http://datini.archiviodistato.prato.it/la-ricerca/scheda/' + row['id'] + '/E33_OVI>'\n",
+    "\n",
+    "        else: \n",
+    "            E73placeHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + informationObjectCoords.code + \"_OVI_2\"\n",
+    "            E33placeHolder = '<http://datini.archiviodistato.prato.it/la-ricerca/scheda/' + row['id'] + '/E33_OVI_2>'\n",
+    "\n",
+    "        datplaceHolder = '<http://datini.archiviodistato.prato.it/la-ricerca/scheda/' + row['id'] + '>'\n",
+    "        sigla = row['sigla']\n",
+    "\n",
+    "        #E22 P128 E73\n",
+    "        line = triple(datplaceHolder,\n",
+    "                      cidocCoords.prefix + 'P128_carries',\n",
+    "                      E73placeHolder + '>') + closeLine\n",
+    "        output.write(line)\n",
+    "        line = triple(E73placeHolder + '>',\n",
+    "                      nsCoords.prefix + 'type',\n",
+    "                      cidocCoords.prefix + 'E73_Information_Object') + closeLine\n",
+    "        output.write(line)\n",
+    "        #line = triple(E73placeHolder,\n",
+    "        #              schemaCoords.prefix + 'label',\n",
+    "        #              '\\\"' + row['titolo'] + '\\\"') + closeLine\n",
+    "        #output.write(line)\n",
+    "        #line = triple(E73placeHolder,\n",
+    "        #              cidocCoords.prefix + 'P2_has_type',\n",
+    "        #              '\\\"Testo Lemmatizzato\\\"') + closeLine\n",
+    "        output.write(line)\n",
+    "\n",
+    "        #E22 P128 E33\n",
+    "        line = triple(datplaceHolder,\n",
+    "                      cidocCoords.prefix + 'P128_carries',\n",
+    "                      E33placeHolder + '>') + closeLine\n",
+    "        output.write(line)\n",
+    "        line = triple(E33placeHolder + '>',\n",
+    "                      nsCoords.prefix + 'type',\n",
+    "                      cidocCoords.prefix + 'E33_Linguistic_Object') + closeLine\n",
+    "        output.write(line)\n",
+    "        line = triple(E33placeHolder + '>',\n",
+    "                      schemaCoords.prefix + 'label',\n",
+    "                      '\\\"Trascrizione di ' + row['titolo'] + '\\\"') + closeLine\n",
+    "        output.write(line)\n",
+    "        line = triple(E33placeHolder + '>',\n",
+    "                      cidocCoords.prefix + 'P2_has_type',\n",
+    "                      '\\\"Trascrizione\\\"') + closeLine\n",
+    "        output.write(line)\n",
+    "\n",
+    "        testo = get_text(sigla)\n",
+    "        testo_lem = testo[1].replace('\"', \"'\")\n",
+    "        line = triple(E73placeHolder + '>',\n",
+    "                      cidocCoords.prefix + 'P190_has_symbolic_content',\n",
+    "                      '\\\"' + testo_lem + '\\\"') + closeLine\n",
+    "        output.write(line)\n",
+    "\n",
+    "        trascrizione = testo[0].replace('\"', \"'\")\n",
+    "        line = triple(E33placeHolder + '>',\n",
+    "                      cidocCoords.prefix + 'P190_has_symbolic_content',\n",
+    "                      '\\\"' + trascrizione + '\\\"') + 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": "aee8b7b246df8f9039afb4144a1f6fd8d2ca17a180786b69acc140d282b71a49"
+  },
+  "kernelspec": {
+   "display_name": "Python 3.9.0 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
+}