{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 10,
   "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": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "import_dir = '/Users/federicaspinelli/TEAMOVI/Parser/DATA/OVI/CSV/'\n",
    "export_dir = '/Users/federicaspinelli/TEAMOVI/Parser/DATA/OVI/RDF/'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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",
    "subClassOfCoords = RDFcoords('<http://www.w3.org/2000/01/rdf-schema#subClassOf>', 'so:')\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",
    "\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": 13,
   "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": 14,
   "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",
    "    output.write('@prefix ' + subClassOfCoords.prefix + ' ' + subClassOfCoords.uri + closeLine)\n",
    "    output.write('\\n')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "filePrefix = 'merge'\n",
    "fileType = 'OVI-TLIO'\n",
    "max_entries = 10000000000000000\n",
    "\n",
    "def get_merce(lemma, iperlemma):\n",
    "    merci_file = open(\n",
    "        '/Users/federicaspinelli/TEAMOVI/Parser/DATA/OVI/CSV/OVI_lemmi_iperlemmi_OLD - Lemmi con Merce.csv', newline=\"\")\n",
    "    reader = csv.DictReader(merci_file)\n",
    "    for row in reader:\n",
    "        if (row['lemma'] == lemma and row['iperlemma'] == iperlemma):\n",
    "            return [row['merce']]\n",
    "\n",
    "\n",
    "with open(import_dir + filePrefix + fileType + '.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '.ttl', 'w') as output:\n",
    "    reader = csv.DictReader(csv_file)\n",
    "    writeTTLHeader(output)\n",
    "    first = True\n",
    "    ii = 0\n",
    "    e55placeHolder = '<http://www.archiviodistato.prato.it/LEMMA>'\n",
    "    line = triple(e55placeHolder, schemaCoords.prefix +\n",
    "                  'label', '\\\"Lemma\\\"') + closeLine\n",
    "    output.write(line)\n",
    "    line = triple(e55placeHolder, nsCoords.prefix +\n",
    "                  'type', '\\\"Lemma\\\"') + closeLine\n",
    "    output.write(line)\n",
    "    e33POS55placeHolder = '<http://www.archiviodistato.prato.it/' + 'POS>'\n",
    "    line = triple(e33POS55placeHolder, schemaCoords.prefix + 'label', '\\\"Part of Speech\\\"') + closeLine\n",
    "    output.write(line)\n",
    "    line = triple(e33POS55placeHolder, nsCoords.prefix + 'type', '\\\"Part of Speech\\\"') + closeLine\n",
    "    output.write(line)\n",
    "\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",
    "        e33POSplaceHolder = '<http://www.archiviodistato.prato.it/' + row['pos'].replace(\" \", \"\").replace( \".\", \"\").replace(\"/\", \"_\") + '_POS>'\n",
    "        \n",
    "        e28placeHolder = '<http://www.archiviodistato.prato.it/' + row['iperlemma'].replace(\" \", \"\").replace(\" \", \"_\") + '_IP>'\n",
    "        \n",
    "        if (row['link_tlio']!='' and row['link_tlio'] is not None):\n",
    "            e33placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['link_tlio'] + '>'\n",
    "        elif (row['link_tlio']==''):\n",
    "            e33placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") +'>'\n",
    "        else:\n",
    "            e33placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") +'>'\n",
    "        if (row['link_tlio']!='' and row['link_tlio'] is not None):\n",
    "            e28oplaceHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['link_tlio'] + 'E28>'\n",
    "        elif (row['link_tlio']==''):\n",
    "            e28oplaceHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E28>'\n",
    "        else:\n",
    "            e28oplaceHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E28>'\n",
    "        if (row['link_tlio']!='' and row['link_tlio'] is not None and row['link_tlio']!=' ' and row['link_tlio'] != 'None'):\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['link_tlio'] + 'E70>'\n",
    "        elif (row['link_tlio']== ' '):\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E70>'\n",
    "        elif (row['link_tlio']== ''):\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E70>'\n",
    "        else:\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E70>'\n",
    "\n",
    "        lemma = row['lemma']\n",
    "        iperlemma = row['iperlemma']\n",
    "        merci = get_merce(lemma, iperlemma)\n",
    "        if (row['thing'] == 'TRUE'):\n",
    "            if (merci is not None):\n",
    "                if (row['id'] is not None and row['id'] != '' and row['id'] != ' '): \n",
    "                    e73ASPOplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + informationObjectCoords.code + \">\"\n",
    "                    line = triple(e73ASPOplaceHolder, cidocCoords.prefix + 'P67_refers_to', e70placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "            if (row['pos'] != 'antr.' and row['pos'] != 'n.g.'):\n",
    "                if (row['id'] is not None and row['id'] != '' and row['id'] != ' '):\n",
    "                    e73OVIplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + row['id'] + \"/\" + informationObjectCoords.code + \"_OVI>\"\n",
    "                    line = triple(e73OVIplaceHolder, cidocCoords.prefix +\n",
    "                                  'P67_refers_to', e70placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e70placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  cidocCoords.prefix + 'E70_Thing') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e70placeHolder,\n",
    "                                  schemaCoords.prefix + 'label',\n",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e70placeHolder, cidocCoords.prefix +\n",
    "                                  'P128_carries', e33placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    # E33\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",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e33placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  e55placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    if (row['iperlemma'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      subClassOfCoords.prefix,\n",
    "                                      e28placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['pos'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P2_has_type',\n",
    "                                      e33POSplaceHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder, schemaCoords.prefix + 'label', '\\\"' + row['pos'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder,\n",
    "                                      nsCoords.prefix + 'type',\n",
    "                                      e33POS55placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['commento'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P3_has_note',\n",
    "                                      '\\\"' + row['commento'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                elif (row['id'] is None or row['id'] == ''):\n",
    "                    line = triple(e70placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  cidocCoords.prefix + 'E70_Thing') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e70placeHolder,\n",
    "                                  schemaCoords.prefix + 'label',\n",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e70placeHolder, cidocCoords.prefix +\n",
    "                                  'P128_carries', e33placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    # E33\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",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e33placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  e55placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    if (row['iperlemma'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      subClassOfCoords.prefix,\n",
    "                                      e28placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['pos'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P2_has_type',\n",
    "                                      e33POSplaceHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder, schemaCoords.prefix + 'label', '\\\"' + row['pos'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder,\n",
    "                                      nsCoords.prefix + 'type',\n",
    "                                      e33POS55placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['commento'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P3_has_note',\n",
    "                                      '\\\"' + row['commento'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "        if (row['thing'] == 'FALSE'):\n",
    "            if (row['pos'] != 'antr.' and row['pos'] != 'n.g.'):\n",
    "                if (row['id'] is not None and row['id'] != ''):\n",
    "                    e73OVIplaceHolder = \"<http://datini.archiviodistato.prato.it/la-ricerca/scheda/\" + \\\n",
    "                        row['id'] + \"/\" + informationObjectCoords.code + \"_OVI>\"\n",
    "                    line = triple(e73OVIplaceHolder, cidocCoords.prefix +\n",
    "                                  'P67_refers_to', e28oplaceHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e28oplaceHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  cidocCoords.prefix + 'E28_Conceptual_Object') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e28oplaceHolder,\n",
    "                                  schemaCoords.prefix + 'label',\n",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e28oplaceHolder, cidocCoords.prefix +\n",
    "                                  'P128_carries', e33placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    # E33\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",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e33placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  e55placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    if (row['iperlemma'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      subClassOfCoords.prefix,\n",
    "                                      e28placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['pos'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P2_has_type',\n",
    "                                      e33POSplaceHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder, schemaCoords.prefix + 'label', '\\\"' + row['pos'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder,\n",
    "                                      nsCoords.prefix + 'type',\n",
    "                                      e33POS55placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['commento'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P3_has_note',\n",
    "                                      '\\\"' + row['commento'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                elif (row['id'] is None or row['id'] == ''):\n",
    "                    line = triple(e28oplaceHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  cidocCoords.prefix + 'E28_Conceptual_Object') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e28oplaceHolder,\n",
    "                                  schemaCoords.prefix + 'label',\n",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e28oplaceHolder, cidocCoords.prefix +\n",
    "                                  'P128_carries', e33placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    # E33\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",
    "                                  '\\\"' + row['lemma'] + '\\\"') + closeLine\n",
    "                    output.write(line)\n",
    "                    line = triple(e33placeHolder,\n",
    "                                  nsCoords.prefix + 'type',\n",
    "                                  e55placeHolder) + closeLine\n",
    "                    output.write(line)\n",
    "                    if (row['iperlemma'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      subClassOfCoords.prefix,\n",
    "                                      e28placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['pos'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P2_has_type',\n",
    "                                      e33POSplaceHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder, schemaCoords.prefix + 'label', '\\\"' + row['pos'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "                        line = triple(e33POSplaceHolder,\n",
    "                                      nsCoords.prefix + 'type',\n",
    "                                      e33POS55placeHolder) + closeLine\n",
    "                        output.write(line)\n",
    "                    if (row['commento'] != ''):\n",
    "                        line = triple(e33placeHolder,\n",
    "                                      cidocCoords.prefix + 'P3_has_note',\n",
    "                                      '\\\"' + row['commento'] + '\\\"') + closeLine\n",
    "                        output.write(line)\n",
    "\n",
    "        output.write('\\n')\n",
    "        # Limit number of entries processed (if desired)\n",
    "        if(ii > max_entries):\n",
    "            break\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<>:32: SyntaxWarning: \"is not\" with a literal. Did you mean \"!=\"?\n",
      "<>:32: SyntaxWarning: \"is not\" with a literal. Did you mean \"!=\"?\n",
      "/var/folders/_n/1ldwyw1s547dcpvn3485fr7r0000gn/T/ipykernel_7536/1610792216.py:32: SyntaxWarning: \"is not\" with a literal. Did you mean \"!=\"?\n",
      "  if (row['link_tlio']!= '' and row['link_tlio']!= '' is not None):\n"
     ]
    }
   ],
   "source": [
    "filePrefix = 'merge'\n",
    "fileType = 'OVI-TLIO'\n",
    "max_entries = 10000000000000000\n",
    "\n",
    "def get_merce(lemma, iperlemma):\n",
    "    merci_file = open('/Users/federicaspinelli/TEAMOVI/Parser/DATA/OVI/CSV/OVI_lemmi_iperlemmi_OLD - Lemmi con Merce.csv', newline=\"\")\n",
    "    reader = csv.DictReader(merci_file)\n",
    "    for row in reader:\n",
    "        if (row['lemma'] == lemma and row['iperlemma'] == iperlemma):\n",
    "            return [row['merce']]\n",
    "\n",
    "with open(import_dir + filePrefix + fileType + '.csv', newline=\"\") as csv_file, open(export_dir + filePrefix + fileType + '_merci.ttl', 'w') as output:\n",
    "    reader = csv.DictReader(csv_file)\n",
    "    writeTTLHeader(output)\n",
    "    first = True\n",
    "    ii = 0\n",
    "    e55placeHolder = '<http://www.archiviodistato.prato.it/merce_E55>'\n",
    "    line = triple(e55placeHolder, hasTypeCoords.prefix, typeCoords.prefix) + closeLine\n",
    "    output.write(line)\n",
    "    line = triple(e55placeHolder, labelCoords.prefix, '\\\"Merce\\\"') + closeLine\n",
    "    output.write(line)\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",
    "        lemma = row['lemma']\n",
    "        iperlemma = row['iperlemma']\n",
    "        merci = get_merce(lemma, iperlemma)\n",
    "        if (row['link_tlio']!= '' and row['link_tlio']!= '' is not None):\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['link_tlio'] + 'E70>'\n",
    "        elif (row['link_tlio']== ''):\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E70>'\n",
    "        else:\n",
    "            e70placeHolder = '<http://tlio.ovi.cnr.it/TLIO/index.php?vox=' + row['iperlemma'].replace(\" \", \"\") + row['lemma'].replace(\" \", \"\") + 'E70>'\n",
    "\n",
    "        if (row['thing'] == 'TRUE'):\n",
    "            if (merci is not None):\n",
    "                line = triple(e70placeHolder, hasTypePCoords.prefix, e55placeHolder) + closeLine\n",
    "                output.write(line)\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": "3.10.4"
  },
  "metadata": {
   "interpreter": {
    "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}