71 lines
3.2 KiB
Python
71 lines
3.2 KiB
Python
import uuid # For UUID creation
|
|
from initDb import initDb # For database initialization
|
|
from wisski.api import Api, Pathbuilder, Entity # For WissKI API
|
|
import os # For environment variable loading
|
|
from dotenv import load_dotenv # For environment variable loading
|
|
import pandas as pd # For dataframe handling
|
|
|
|
def importArtistAssignment(api, engine):
|
|
print('Importing artist assignment...')
|
|
|
|
tableName = "c__ob30_bez_kuenstler"
|
|
bundleId = 'bc8826cc7d9c9373ce71cfc0251c2a4f'
|
|
|
|
try:
|
|
processedRows = pd.read_csv(f'./logs/{tableName}.csv')
|
|
except FileNotFoundError:
|
|
processedRows = pd.DataFrame(columns=['id', 'uuid', 'uri'])
|
|
|
|
# Load sources table
|
|
artistRelationsTable = pd.read_sql_table(tableName, con=engine)
|
|
|
|
# Create artistRelations
|
|
for index, row in artistRelationsTable.iterrows():
|
|
# For every row in table...
|
|
if index < len(processedRows) and artistRelationsTable.loc[index, 'id'] == processedRows.loc[index, 'id']:
|
|
# skip if already processed
|
|
print(f'Skipping already processed artistAssignment {artistRelationsTable.loc[index, "id"]}')
|
|
continue
|
|
# Create Entity property dicts
|
|
artistRelationValues = {}
|
|
for key, value in row.items():
|
|
# For every column in row...
|
|
if (value is None) or (value == ''):
|
|
# skip if cell has no value
|
|
continue
|
|
# Properties of an entity have to be an array, so...
|
|
value = str(value).replace('&###{{new_line}}###'.format(), '&')
|
|
value = str(value).replace('###{{new_line}}###', '&')
|
|
value = str(value).replace(' & ', '&')
|
|
if '&' in str(value):
|
|
# ...Explode "&"-separated values to array items
|
|
value = [x.strip() for x in str(value).split('&')]
|
|
else:
|
|
# ...Or parse to array
|
|
value = [value]
|
|
# Map columns to fields. We use assignments for reification.
|
|
match key:
|
|
case 'id':
|
|
docId = value[0]
|
|
case 'f__uuid':
|
|
artistRelationValues['fc150259d31fea8a3f992e7beb901fa4'] = value # UUID
|
|
case 'f__3100_name':
|
|
artistRelationValues['ff5bf58133f9351d03e2ee92b6f8bb7e'] = value # Artist Name
|
|
case 'f__3475_ber__funkt_':
|
|
artistRelationValues['fc0c7d8c6b736489210bc42ef0f1406a'] = value # Occupation
|
|
case 'f__ob30_bez_kuenstler':
|
|
artistRelationValues['f575d4f2c8ea5d37618cea708c2a7c5e'] = value # Relation
|
|
case _:
|
|
print(f'{key} is not a valid field, skipping.')
|
|
|
|
|
|
artistRelation = Entity(api=api, fields=artistRelationValues, bundle_id=bundleId)
|
|
api.save(artistRelation)
|
|
|
|
print(f'Created artist assignment {index}: {artistRelation.uri} of {len(artistRelationsTable)}')
|
|
|
|
# Write log
|
|
processedRows = processedRows._append({'id': row['id'], 'uuid': artistRelationValues['fc150259d31fea8a3f992e7beb901fa4'][0], 'uri': artistRelation.uri}, ignore_index=True)
|
|
processedRows.to_csv(f'./logs/{tableName}.csv', index=False)
|
|
|
|
print('finished importing artist assignment')
|