{ "cells": [ { "cell_type": "code", "execution_count": 7, "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", "\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": 8, "metadata": {}, "outputs": [], "source": [ "import_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/CSV/ASPO/marcovaldi/'\n", "export_dir = '/Users/federicaspinelli/Google Drive/OVI-CNR/RDF/ASPO/marcovaldi/'" ] }, { "cell_type": "code", "execution_count": 9, "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('', 'dt:')\n", "personAuthCoords = RDFcoords('', 'pa:')\n", "# W3/CIDOC Predicates\n", "hasTypeCoords = RDFcoords('', 'tp:')\n", "hasTypePCoords = RDFcoords('', 'te:')\n", "carriesCoords = RDFcoords('', 'ca:')\n", "identifiedByCoords = RDFcoords('', 'ib:')\n", "labelCoords = RDFcoords('', 'lb:')\n", "wasBroughtCoords = RDFcoords('', 'wb:')\n", "carriedByCoords = RDFcoords('', 'cb:')\n", "movedByCoords = RDFcoords('', 'mb:')\n", "subClassOfCoords = RDFcoords('', 'so:')\n", "tookPlaceCoords = RDFcoords('', 'tk:')\n", "carriedByCoords = RDFcoords('', 'cb:')\n", "roleOfCoords = RDFcoords('', 'ro:')\n", "hasDomainCoords = RDFcoords('', 'hd:')\n", "hasRangeCoords = RDFcoords('', 'hr:')\n", "\n", "# CIDOC Objects\n", "manMadeObjectCoords = RDFcoords('', 'mo:', 'E22')\n", "informationObjectCoords = RDFcoords('', 'io:', 'E73')\n", "titleCoords = RDFcoords('', 'ti:' ,'E35')\n", "placeAppellationCoords = RDFcoords('', 'pa:', 'E44')\n", "identifierCoords = RDFcoords('', 'id:', 'E42')\n", "typeCoords = RDFcoords('', 'ty:', 'E55')\n", "creationCoords = RDFcoords('', 'cr:', 'E65')\n", "personCoords = RDFcoords('', 'ps:', 'E21')\n", "pcarriedByCoords = RDFcoords('', 'cy:', 'PC14')\n", "\n", "# New classes (subclasses of E7 Activity) - Exchange, Sending, Recive Letters\n", "exchangeLettersCoords = RDFcoords('', 'el:', 'EL1')\n", "sendLetterCoords = RDFcoords('', 'sl:', 'EL2')\n", "receiveLetterCoords = RDFcoords('', 'rl:', 'EL3')" ] }, { "cell_type": "code", "execution_count": 10, "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": 11, "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 ' + 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 ' + movedByCoords.prefix + ' ' + movedByCoords.uri + closeLine)\n", " output.write('@prefix ' + roleOfCoords.prefix + ' ' + roleOfCoords.uri + closeLine)\n", " output.write('@prefix ' + hasDomainCoords.prefix + ' ' + hasDomainCoords.uri + closeLine)\n", " output.write('@prefix ' + pcarriedByCoords.prefix + ' ' + pcarriedByCoords.uri + closeLine)\n", " output.write('@prefix ' + hasRangeCoords.prefix + ' ' + hasRangeCoords.uri + closeLine)\n", " \n", " output.write('\\n')\n" ] }, { "cell_type": "code", "execution_count": 12, "metadata": {}, "outputs": [], "source": [ "filePrefix = 'data_'\n", "fileType = 'item'\n", "max_entries = 1000000000\n", "\n", "with open(import_dir + filePrefix + fileType + '.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 = \"\"\n", " el2placeHolder = \"\"\n", " el3placeHolder = \"\"\n", " PC14splaceHolder = \"\" \n", " PC14rplaceHolder = \"\" \n", " E55splaceHolder = \"\" \n", " E55rplaceHolder = \"\"\n", " \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(\"\", movedByCoords.prefix, el1placeHolder) + closeLine\n", " output.write(line)\n", " line = triple(el1placeHolder, labelCoords.prefix, '\\\"' + row['tipologia'] + ': ' + row['titolo_aspo'] + '\\\"') + 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", "\n", " # If the 'persname' property is not empty for the given entry, write down triples\n", " if(row['persona_mittente'] != ''):\n", " mittente = row['persona_mittente'].replace('{\"nome\":', '').replace(',', '').replace('\"authID\":', '')\n", " name_mittente = re.sub('IT-ASPO-AU00003-[0-9].*}', '', mittente).replace('\"', '').replace('}', '').strip()\n", " auth = row['persona_mittente']\n", " authcode = re.sub('{\"nome\": .* ', '', auth)\n", " authcodeprefix= authcode.replace('IT-ASPO-AU00003-', '').replace('\"', '').replace('}', '').strip()\n", " actorplaceHolder = personAuthCoords.prefix + authcodeprefix \n", " line = triple(el2placeHolder, hasDomainCoords.prefix, PC14splaceHolder) + closeLine\n", " output.write(line)\n", " line = triple(PC14splaceHolder, hasTypeCoords.prefix, pcarriedByCoords.prefix) + closeLine\n", " output.write(line)\n", " line = triple(PC14splaceHolder, labelCoords.prefix, \"\\\"\" + name_mittente + \" nel ruolo di mittente\" + \"\\\"\" ) + closeLine\n", " output.write(line)\n", " line = triple(PC14splaceHolder, roleOfCoords.prefix, E55splaceHolder) + closeLine\n", " output.write(line)\n", " line = triple(E55splaceHolder, labelCoords.prefix, \"\\\"Mittente\\\"\" ) + closeLine\n", " output.write(line)\n", " line = triple(PC14splaceHolder, hasRangeCoords.prefix, actorplaceHolder) + closeLine\n", " output.write(line)\n", "\n", " # If the 'persname' property is not empty for the given entry, write down triples\n", " if(row['persona_destinatario'] != ''):\n", " destinatario = row['persona_destinatario'].replace('{\"nome\":', '').replace(',', '').replace('\"authID\":', '')\n", " name_destinatario = re.sub('IT-ASPO-AU00003-[0-9].*}', '', destinatario).replace('\"', '').replace('}', '').strip()\n", " auth = row['persona_destinatario']\n", " authcode = re.sub('{\"nome\": .* ', '', auth)\n", " authcodeprefix= authcode.replace('IT-ASPO-AU00003-', '').replace('\"', '').replace('}', '').strip()\n", " actorplaceHolder = personAuthCoords.prefix + authcodeprefix \n", " line = triple(el3placeHolder, hasDomainCoords.prefix, PC14rplaceHolder) + closeLine\n", " output.write(line)\n", " line = triple(PC14rplaceHolder, hasTypeCoords.prefix, pcarriedByCoords.prefix) + closeLine\n", " output.write(line)\n", " line = triple(PC14rplaceHolder, labelCoords.prefix, \"\\\"\" + name_destinatario + \" nel ruolo di destinatario\" + \"\\\"\" ) + closeLine\n", " output.write(line)\n", " line = triple(PC14rplaceHolder, roleOfCoords.prefix, E55rplaceHolder) + closeLine\n", " output.write(line)\n", " line = triple(E55rplaceHolder, labelCoords.prefix, \"\\\"Destinatario\\\"\" ) + closeLine\n", " output.write(line)\n", " line = triple(PC14rplaceHolder, hasRangeCoords.prefix, actorplaceHolder) + 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", " " ] }, { "cell_type": "code", "execution_count": null, "metadata": {}, "outputs": [], "source": [] } ], "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 }