ricgraph 2.8__py3-none-any.whl

ricgraph/ricgraph.py

@@ -0,0 +1,2866 @@
+# ########################################################################
+#
+# Ricgraph - Research in context graph
+#
+# ########################################################################
+#
+# MIT License
+# 
+# Copyright (c) 2023, 2024, 2025 Rik D.T. Janssen
+# 
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+# 
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+# 
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+#
+# ########################################################################
+#
+# This file is the main file for Ricgraph.
+# Original version Rik D.T. Janssen, December 2022.
+# Updated Rik D.T. Janssen, February, March, September to December 2023.
+# Updated Rik D.T. Janssen, February, March 2024.
+# Updated Rik D.T. Janssen, January 2025
+#
+# ########################################################################
+
+
+"""Ricgraph - Research in context graph.
+
+This is Ricgraph - Research in context graph, a graph structure with nodes and edges
+to represent information (as properties in the nodes)
+and their relations (the edges). You can find the code and documentation on
+https://github.com/UtrechtUniversity/ricgraph/.
+
+Ricgraph uses Neo4j as graph database, https://neo4j.com.
+Neo4j only allows directed graphs, so Ricgraph is a directed graph.
+They have several products. Their free products are:
+
+- Neo4j Desktop (https://neo4j.com/download-center/#desktop);
+
+- Neo4j Community Edition (https://neo4j.com/download-center/#community).
+
+Neo4j Desktop includes Bloom, which allows exploring the graph in a more
+intuitive way. Community Edition allows only Cypher queries.
+
+Ricgraph can be extended for other brands of graph databases
+by changing minor bits of the code.
+
+All nodes have the properties in RICGRAPH_PROPERTIES_STANDARD and
+RICGRAPH_PROPERTIES_HIDDEN (see ricgraph.ini):
+
+In RICGRAPH_PROPERTIES_STANDARD:
+
+- name: name of the node, e.g. ISNI, ORCID, DOI, etc.
+
+- category: category of the node,
+  e.g. person, person-root, book, journal article, dataset, etc.
+
+- value: value of the node.
+
+In RICGRAPH_PROPERTIES_HIDDEN:
+
+- _key: key value of the node, not to be modified by the user.
+
+- _source: sorted list of sources a record has been harvested from,
+  not to be modified by the user
+
+- _history: list of history events of the node, not to be modified by the user.
+
+Additional properties for nodes can be added by changing entry RICGRAPH_PROPERTIES_ADDITIONAL
+in ricgraph.ini (make sure 'history_event' is last).
+In the default configuration of Ricgraph, the following properties are included:
+
+- comment: comment for a node, can be anything that is thought to be useful.
+
+- url_main: main URL for a node, pointing to e.g. the corresponding ISNI, ORCID or DOI
+  record on the web.
+
+- url_other: other URL for a node, pointing to e.g. the originating record in a source system.
+
+- year: year of a research output.
+
+- source_event: an event to be added to the _source list.
+
+- history_event: an event to be added to the _history list.
+"""
+
+
+import os
+import sys
+from datetime import datetime
+import numpy
+import pandas
+import requests
+import csv
+import uuid
+import json
+from typing import Union, Tuple
+from collections import defaultdict
+import configparser
+from neo4j import GraphDatabase, Driver, Result
+from neo4j.graph import Node
+
+__author__ = 'Rik D.T. Janssen'
+__copyright__ = 'Copyright (c) 2023, 2024, 2025 Rik D.T. Janssen'
+__email__ = ''
+__license__ = 'MIT License'
+__package__ = 'Ricgraph'
+# __version__ is in __init__.py of this module.
+
+
+# ########################################################################
+# Start of constants section
+# ########################################################################
+RICGRAPH_INI_FILENAME = 'ricgraph.ini'
+RICGRAPH_KEY_SEPARATOR = '|'
+
+# Used for some loop iterations, in case no max iteration for such a loop is specified.
+A_LARGE_NUMBER = 9999999999
+
+# For the REST API, we need to return an HTTP response status code. These are
+# listed on https://www.iana.org/assignments/http-status-codes/http-status-codes.xhtml.
+# For Ricgraph, we only use one HTTP response status code:
+# https://www.rfc-editor.org/rfc/rfc9110.html#name-200-ok.
+HTTP_RESPONSE_OK = 200
+# We use these if an item cannot be found. They should be in the 2xx series
+# since we still return a valid HTTP response. These are in the "unassigned" range,
+# which may be freely used.
+HTTP_RESPONSE_NOTHING_FOUND = 250
+HTTP_RESPONSE_INVALID_SEARCH = 251
+
+# This dict is used as a cache for node id's. If we have a node id, we can
+# do a direct lookup for a node in O(1), instead of a search in O(log n).
+# The dict has the format: [Ricgraph _key]: [Node element_id].
+nodes_cache_key_element_id = {}
+# Cache size. 'nodes_cache_key_element_id' will be emptied if it has this number of elements.
+MAX_NODES_CACHE_KEY_ID = 15000
+
+
+# These counters are used to count the number of accesses to the
+# graph database backend.
+graphdb_nr_creates = 0
+graphdb_nr_reads = 0
+graphdb_nr_updates = 0
+graphdb_nr_deletes = 0
+
+
+# ########################################################################
+# Research output types used in Ricgraph.
+# Harvested sources most often have a type for a research output.
+# In a harvest script for a certain source, you specify a mapping
+# from the names used in the source you harvest, to the values below.
+# This ensures that Ricgraph uses the same wording for types of
+# research outputs harvested from different sources.
+#
+# An example mapping may look like this:
+# mapping = {
+#    'book': ROTYPE_BOOK,
+#    'dataset': ROTYPE_DATASET,
+#    'software': ROTYPE_SOFTWARE
+# }
+#
+# Call the function lookup_resout_type() to do this mapping.
+# If you add a type, also add it to ROTYPE_ALL.
+#
+# This list is inspired by the Strategy Evaluation Protocol 2021-2027
+# https://www.universiteitenvannederland.nl/files/documenten/Domeinen/Onderzoek/SEP_2021-2027.pdf,
+# Appendix E2.
+ROTYPE_ABSTRACT = 'abstract'
+ROTYPE_BOOK = 'book'
+ROTYPE_BOOKCHAPTER = 'book chapter'
+ROTYPE_CONFERENCE_ARTICLE = 'conference article'
+ROTYPE_DATASET = 'data set'
+ROTYPE_DESIGN = 'design'
+ROTYPE_DIGITAL_VISUAL_PRODUCT = 'digital or visual product'
+ROTYPE_EDITORIAL = 'editorial'
+ROTYPE_ENTRY = 'entry for encyclopedia or dictionary'
+ROTYPE_EXHIBITION_PERFORMANCE = 'exhibition or performance'
+ROTYPE_JOURNAL_ARTICLE = 'journal article'
+ROTYPE_LETTER = 'letter to the editor'
+ROTYPE_MEMORANDUM = 'memorandum'
+ROTYPE_METHOD_DESCRIPTION = 'method description'
+ROTYPE_MODEL = 'model'
+ROTYPE_OTHER_CONTRIBUTION = 'other contribution'
+ROTYPE_PATENT = 'patent'
+ROTYPE_PHDTHESIS = 'PhD thesis'
+ROTYPE_POSTER = 'poster'
+ROTYPE_PREPRINT = 'preprint'
+ROTYPE_PRESENTATION = 'presentation'
+ROTYPE_REGISTERED_REPORT = 'registered report'
+ROTYPE_REPORT = 'report'
+ROTYPE_RETRACTION = 'retraction'
+ROTYPE_REVIEW = 'review'
+ROTYPE_SOFTWARE = 'software'
+ROTYPE_THESIS = 'thesis'
+ROTYPE_WEBSITE = 'website or web publication'
+
+ROTYPE_ALL = [ROTYPE_ABSTRACT,
+              ROTYPE_BOOK,
+              ROTYPE_BOOKCHAPTER,
+              ROTYPE_CONFERENCE_ARTICLE,
+              ROTYPE_DATASET,
+              ROTYPE_DESIGN,
+              ROTYPE_DIGITAL_VISUAL_PRODUCT,
+              ROTYPE_EDITORIAL,
+              ROTYPE_ENTRY,
+              ROTYPE_EXHIBITION_PERFORMANCE,
+              ROTYPE_JOURNAL_ARTICLE,
+              ROTYPE_LETTER,
+              ROTYPE_MEMORANDUM,
+              ROTYPE_METHOD_DESCRIPTION,
+              ROTYPE_MODEL,
+              ROTYPE_OTHER_CONTRIBUTION,
+              ROTYPE_PATENT,
+              ROTYPE_PHDTHESIS,
+              ROTYPE_POSTER,
+              ROTYPE_PREPRINT,
+              ROTYPE_PRESENTATION,
+              ROTYPE_REGISTERED_REPORT,
+              ROTYPE_REPORT,
+              ROTYPE_RETRACTION,
+              ROTYPE_REVIEW,
+              ROTYPE_SOFTWARE,
+              ROTYPE_THESIS,
+              ROTYPE_WEBSITE]
+
+# ########################################################################
+# End of constants section
+# ########################################################################
+
+
+global _graph
+
+
+# ##############################################################################
+# Modification for Python module neo4j.
+# ##############################################################################
+
+def node_eq(self, other):
+    """Modified __eq__ method for Python module neo4j.
+
+    The __eq__ method of class Node does not work as expected (as of April 2024).
+    Two Nodes are supposed to be equal if their Node.element_id's are equal.
+
+    I would like to be able to test if a 'node' of type Node is in a list called
+    'mylist' with an 'if' statement like this:
+    'if node in mylist ...' or 'if node not in mylist ...',
+    but it does not work: even if Nodes are in that list, the 'if' fails.
+    I could have made my own class and extended it with an __eq__ method, but
+    I choose this hack. It is called a 'Monkey patch': A Monkey patch is a piece
+    of Python code which extends or modifies other code at runtime
+    (typically at startup).
+
+    :return: True on equality, or False if not.
+    """
+    return isinstance(other, Node) and self.element_id == other.element_id
+
+
+# Modified __eq__ method for Python module neo4j (also see above).
+Node.__eq__ = node_eq
+
+
+# ##############################################################################
+# Cypher functions.
+# ##############################################################################
+# These are also functions with cypher:
+# - read_all_nodes()
+# - read_all_values_of_property()
+# - get_all_neighbor_nodes()
+# ##############################################################################
+# Note the use of WHERE clauses below. Some of them uses the function elementId(),
+# which does a _direct_ lookup for a node with that id. That is the fastest way possible,
+# compared to a WHERE clause on node['_key'], which is a search.
+# To observe this difference, prefix a Cypher query with PROFILE, e.g.:
+# PROFILE MATCH (n) WHERE elementId(n)=[a string here] RETURN *
+#
+# Read: https://neo4j.com/docs/cypher-manual/current/functions/scalar/#functions-elementid
+# https://neo4j.com/docs/cypher-manual/current/planning-and-tuning/operators/operators-detail/#query-plan-node-by-elementid-seek
+#
+# [May 6, 2024] Note however, that memgraph on does not have elementId() but only id():
+# https://memgraph.com/docs/querying/differences-in-cypher-implementations.
+# ##############################################################################
+
+def open_ricgraph() -> Driver:
+    """Open Ricgraph.
+
+    :return: graph that has been opened.
+    """
+    global _graph
+
+    print('Opening ricgraph.\n')
+    try:
+        _graph = GraphDatabase.driver(GRAPHDB_URL,
+                                      auth=(GRAPHDB_USER, GRAPHDB_PASSWORD))
+        _graph.verify_connectivity()
+    except Exception as e:
+        print('open_ricgraph(): An exception occurred. Name: ' + type(e).__name__ + ',')
+        print('  error message: ' + str(e) + '.')
+        print('Error opening graph database backend using these parameters:')
+        print('GRAPHDB_URL: ' + GRAPHDB_URL)
+        print('GRAPHDB_DATABASENAME: ' + ricgraph_databasename())
+        print('GRAPHDB_USER: ' + GRAPHDB_USER)
+        print('GRAPHDB_PASSWORD: [see file ' + get_ricgraph_ini_file() + ']')
+        print('\nAre these parameters correct and did you start your graph database backend "' + GRAPHDB + '"?')
+        print('Exiting.')
+        exit(1)
+
+    return _graph
+
+
+def close_ricgraph() -> None:
+    """Close Ricgraph.
+    """
+    global _graph
+
+    print('Closing ricgraph.\n')
+    _graph.close()
+    return
+
+
+def ricgraph_database() -> str:
+    """Return the name of the Ricgraph database backend.
+    """
+    return GRAPHDB
+
+
+def ricgraph_databasename() -> str:
+    """Return the name of the database in the Ricgraph database backend.
+    """
+    return GRAPHDB_DATABASENAME
+
+
+def empty_ricgraph(answer: str = '') -> None:
+    """Empty Ricgraph and create new indexes. Side effect: indexes are deleted and created.
+
+    :param answer: prefilled answer whether the user wants to empty Ricgraph.
+      'yes': Ricgraph will be emptied, no questions asked;
+      'no': Ricgraph will not be emptied, no questions asked;
+      any other answer: the user will be asked whether to empty Ricgraph.
+    :return: None.
+    """
+    global _graph
+
+    if answer == 'yes':
+        print('Emptying Ricgraph: Ricgraph will be emptied.\n')
+        # Fall through.
+    elif answer == 'no':
+        print('Emptying Ricgraph: Ricgraph will not be emptied.\n')
+        return
+    else:
+        print('Do you want to empty Ricgraph? You have the following options:')
+        print('- "yes": Ricgraph will be emptied.')
+        print('- "no": Ricgraph will not be emptied.')
+        print('- any other answer: Ricgraph will not be emptied, execution of this script will abort.')
+        user_input = input('Please make a choice: ')
+        if user_input.lower() == 'yes':
+            print('\nRicgraph will be emptied.\n')
+            # Fall through.
+        elif user_input.lower() == 'no':
+            print('\nRicgraph has not been emptied.\n')
+            return
+        else:
+            print('\nRicgraph has not been emptied, exiting.\n')
+            exit(1)
+
+    graphdb_name = ricgraph_database()
+    graphdb_databasename = ricgraph_databasename()
+
+    # Delete the database and start over again.
+    # Sometimes the following statement fails with Neo4j Desktop graph database backend
+    # if there are many nodes, see e.g.
+    # https://stackoverflow.com/questions/23310114/how-to-reset-clear-delete-neo4j-database.
+    # Apparently, the community edition of Neo4j does not have a
+    # "CREATE OR REPLACE DATABASE customers" command.
+    print('Deleting all nodes and edges in Ricgraph...\n')
+    _graph.execute_query('MATCH (node) DETACH DELETE node', database_=graphdb_databasename)
+
+    if graphdb_name == 'neo4j':
+        # Do not use TEXT indexes, use RANGE indexes, these are much faster during harvesting
+        # (but they do not support a CONTAINS substring search which is being used for a broad search
+        # in read_all_nodes()). The RANGE index is the default in Neo4j. More info:
+        # https://neo4j.com/docs/cypher-manual/current/indexes/search-performance-indexes/managing-indexes
+        # [Jan. 2023] The graph database backend Neo4j can have at most 3 indexes.
+        # [June 5, 2024] This is not true anymore. I use 4 indexes.
+        _graph.execute_query('DROP INDEX KeyIndex IF EXISTS', database_=graphdb_databasename)
+        _graph.execute_query('DROP INDEX NameIndex IF EXISTS', database_=graphdb_databasename)
+        _graph.execute_query('DROP INDEX CategoryIndex IF EXISTS', database_=graphdb_databasename)
+        _graph.execute_query('DROP INDEX ValueIndex IF EXISTS', database_=graphdb_databasename)
+
+        print('Creating indexes...')
+        _graph.execute_query('CREATE INDEX KeyIndex IF NOT EXISTS FOR (node:RicgraphNode) ON (node._key)',
+                             database_=graphdb_databasename)
+        _graph.execute_query('CREATE INDEX NameIndex IF NOT EXISTS FOR (node:RicgraphNode) ON (node.name)',
+                             database_=graphdb_databasename)
+        _graph.execute_query('CREATE INDEX CategoryIndex IF NOT EXISTS FOR (node:RicgraphNode) ON (node.category)',
+                             database_=graphdb_databasename)
+        _graph.execute_query('CREATE INDEX ValueIndex IF NOT EXISTS FOR (node:RicgraphNode) ON (node.value)',
+                             database_=graphdb_databasename)
+
+        print('These indexes have been created:')
+        records, summary, keys = _graph.execute_query('SHOW INDEXES',
+                                                      database_=graphdb_databasename)
+        index_table = pandas.DataFrame(data=records, columns=keys)
+        print(index_table.to_string(index=False))
+        print('')
+    elif graphdb_name == 'memgraph':
+        # For Memgraph, we need to create indexes in the following way, otherwise we get an
+        # 'Index manipulation not allowed in multicommand transactions'
+        # error. We should use implicit (or auto-commit) transactions.
+        # See https://memgraph.com/docs/client-libraries/python#transaction-management.
+        with _graph.session(database=GRAPHDB) as session:
+            # These also work in case they do not exist.
+            session.run('DROP INDEX ON :RicgraphNode;')
+            session.run('DROP INDEX ON :RicgraphNode(_key);')
+            session.run('DROP INDEX ON :RicgraphNode(name);')
+            session.run('DROP INDEX ON :RicgraphNode(category);')
+            session.run('DROP INDEX ON :RicgraphNode(value);')
+
+            print('Creating indexes...')
+            session.run('CREATE INDEX ON :RicgraphNode;')
+            session.run('CREATE INDEX ON :RicgraphNode(_key);')
+            session.run('CREATE INDEX ON :RicgraphNode(name);')
+            session.run('CREATE INDEX ON :RicgraphNode(category);')
+            session.run('CREATE INDEX ON :RicgraphNode(value);')
+
+            print('These indexes have been created:')
+            for index_line in session.run('SHOW INDEX INFO;'):
+                print(index_line)
+            print('')
+
+        # session.close() is done automatically because of 'with'.
+    else:
+        print('empty_ricgraph(): Unknown graph database backend "' + graphdb_name + '".')
+        print('Exiting.')
+        exit(1)
+    return
+
+
+def ricgraph_nr_nodes() -> int:
+    """Count the number of nodes in Ricgraph.
+
+    :return: the number of nodes, or -1 on error.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nricgraph_nr_nodes(): Error: graph has not been initialized or opened.\n\n')
+        return -1
+
+    cypher_query = 'MATCH () RETURN count(*) AS count'
+    result = _graph.execute_query(cypher_query,
+                                  result_transformer_=Result.data,
+                                  database_=ricgraph_databasename())
+    if len(result) == 0:
+        return -1
+    result = result[0]
+    if 'count' not in result:
+        return -1
+    result = int(result['count'])
+    return result
+
+
+def ricgraph_nr_edges() -> int:
+    """Count the number of edges in Ricgraph.
+
+    :return: the number of edges, or -1 on error.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nricgraph_nr_edges(): Error: graph has not been initialized or opened.\n\n')
+        return -1
+
+    cypher_query = 'MATCH ()-[r]->() RETURN count(r) AS count'
+    result = _graph.execute_query(cypher_query,
+                                  result_transformer_=Result.data,
+                                  database_=ricgraph_databasename())
+    if len(result) == 0:
+        return -1
+    result = result[0]
+    if 'count' not in result:
+        return -1
+    result = int(result['count'])
+    return result
+
+
+def ricgraph_nr_edges_of_node(node_element_id: str) -> int:
+    """Count the number of edges in Ricgraph of a given node.
+
+    :param node_element_id: the element_id of the node.
+    :return: the number of nodes, or -1 on error.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nricgraph_nr_edges_of_node(): Error: graph has not been initialized or opened.\n\n')
+        return -1
+
+    cypher_query = 'MATCH (node:RicgraphNode)-[r]->() '
+    if ricgraph_database() == 'neo4j':
+        cypher_query += 'WHERE elementId(node)=$node_element_id '
+    else:
+        cypher_query += 'WHERE id(node)=toInteger($node_element_id) '
+    cypher_query += 'RETURN count(r) AS count'
+
+    result = _graph.execute_query(cypher_query,
+                                  node_element_id=node_element_id,
+                                  result_transformer_=Result.data,
+                                  database_=ricgraph_databasename())
+    if len(result) == 0:
+        return -1
+    result = result[0]
+    if 'count' not in result:
+        return -1
+    result = int(result['count'])
+    return result
+
+
+def cypher_create_node(node_properties: dict) -> Union[Node, None]:
+    """
+    Create a node in the graph database.
+    Create means: the node will be created, regardless if it is already
+    present or not. No test for presence of a node is done.
+    The properties and their values in 'node_properties' will be added
+    as properties & values of the node.
+    The 'name', 'category', and 'value' properties are expected to be
+    in 'node_properties'.
+
+    :param node_properties: the properties of the node.
+    :return: the node updated, or None on error.
+    """
+    global graphdb_nr_creates
+
+    # There are several methods for creating a node in the graph database.
+    # This would be an alternative, but it seems to use more memory than the
+    # one used now, according to 'PROFILE <cypher query>'.
+    # cypher_query = 'CREATE (node:RicgraphNode) SET node=$node_properties ' [etc.]
+    cypher_query = 'CREATE (node:RicgraphNode $node_properties) RETURN node'
+
+    # print('cypher_create_node(): cypher_query: ' + cypher_query)
+    # print('                      node_properties: ' + str(node_properties))
+
+    nodes = _graph.execute_query(cypher_query,
+                                 node_properties=node_properties,
+                                 result_transformer_=Result.value,
+                                 database_=ricgraph_databasename())
+    graphdb_nr_creates += 1
+    if len(nodes) == 0:
+        return None
+    else:
+        return nodes[0]
+
+
+def cypher_delete_node(node_element_id: str) -> None:
+    """Delete a node in the graph database.
+    If it has any edges, these will also be deleted.
+
+    :param node_element_id: the element_id of the node.
+    :return: None.
+    """
+    global graphdb_nr_deletes
+
+    cypher_query = 'MATCH (node:RicgraphNode) '
+    if ricgraph_database() == 'neo4j':
+        cypher_query += 'WHERE elementId(node)=$node_element_id '
+    else:
+        cypher_query += 'WHERE id(node)=toInteger($node_element_id) '
+    cypher_query += 'DETACH DELETE node'
+    _graph.execute_query(cypher_query,
+                         node_element_id=node_element_id,
+                         result_transformer_=Result.value,
+                         database_=ricgraph_databasename())
+
+    nr_edges = ricgraph_nr_edges_of_node(node_element_id=node_element_id)
+    graphdb_nr_deletes += 1 + nr_edges      # '1' for node, add all of its edges.
+    return
+
+
+def cypher_update_node_properties(node_element_id: str, node_properties: dict) -> Union[Node, None]:
+    """
+    Update node properties in a node in the graph database.
+    If the node is not present, nothing will happen.
+    Only the properties and their values in 'node_properties' will be changed (if
+    a property is present in the node) or added (if the property is not present).
+    All other properties (i.e. not specified properties in 'node_properties')
+    will be left as they are.
+
+    :param node_element_id: the element_id of the node.
+    :param node_properties: the properties of the node.
+    :return: the node updated, or None on error.
+    """
+    global graphdb_nr_reads, graphdb_nr_updates
+
+    cypher_query = 'MATCH (node:RicgraphNode) '
+    if ricgraph_database() == 'neo4j':
+        cypher_query += 'WHERE elementId(node)=$node_element_id '
+    else:
+        cypher_query += 'WHERE id(node)=toInteger($node_element_id) '
+    cypher_query += 'SET node+=$node_properties RETURN node'
+
+    # print('cypher_update_node_properties(): node_element_id: ' + str(node_element_id) +
+    #       ', cypher_query: ' + cypher_query)
+    # print('                     node_properties: ' + str(node_properties))
+
+    nodes = _graph.execute_query(cypher_query,
+                                 node_element_id=node_element_id,
+                                 node_properties=node_properties,
+                                 result_transformer_=Result.value,
+                                 database_=ricgraph_databasename())
+    graphdb_nr_reads += 1
+    graphdb_nr_updates += 1
+    if len(nodes) == 0:
+        return None
+    else:
+        return nodes[0]
+
+
+def cypher_merge_nodes(node_merge_from_element_id: str,
+                       node_merge_to_element_id: str,
+                       node_merge_to_properties: dict) -> Union[Node, None]:
+    """Merge two nodes in the graph database.
+    The neighbors of 'node_merge_from' will be merged to node 'node_merge_to'
+    and 'node_merge_from' will be deleted.
+    The node properties in 'node_merge_to' will be set according to
+    'node_merge_to_properties'.
+
+    Only the properties and their values in 'node_properties' will be changed (if
+    a property is present in the node) or added (if the property is not present).
+    All other properties (i.e. not specified properties in 'node_properties')
+    will be left as they are.
+
+    :param node_merge_from_element_id: the element_id of node_merge_from.
+    :param node_merge_to_element_id: the element_id of node_merge_to.
+    :param node_merge_to_properties: the properties of node_merge_to.
+    :return: the node updated, or None on error.
+    """
+    global graphdb_nr_reads, graphdb_nr_updates
+
+    graphdb_name = ricgraph_database()
+
+    if node_merge_from_element_id == node_merge_to_element_id:
+        # We only update node properties.
+        node = cypher_update_node_properties(node_element_id=node_merge_to_element_id,
+                                             node_properties=node_merge_to_properties)
+        return node
+
+    cypher_query = 'MATCH (node_from:RicgraphNode) '
+    if graphdb_name == 'neo4j':
+        cypher_query += 'WHERE elementId(node_from)=$node_merge_from_element_id '
+    else:
+        cypher_query += 'WHERE id(node_from)=toInteger($node_merge_from_element_id) '
+    cypher_query += 'MATCH (node_to:RicgraphNode) '
+    if graphdb_name == 'neo4j':
+        cypher_query += 'WHERE elementId(node_to)=$node_merge_to_element_id '
+    else:
+        cypher_query += 'WHERE id(node_to)=toInteger($node_merge_to_element_id) '
+
+    cypher_query += 'SET node_to+=$node_merge_to_properties '
+    cypher_query += 'WITH node_from, node_to '
+    # Only test for an edge in one direction, that is sufficient.
+    cypher_query += 'MATCH (node_from)-[:LINKS_TO]->(neighbor_node) '
+    # Prevent creating a self-relationship on node_to.
+    cypher_query += 'WHERE node_to <> neighbor_node '
+    cypher_query += 'MERGE (node_to)-[:LINKS_TO]->(neighbor_node) '
+    cypher_query += 'MERGE (node_to)<-[:LINKS_TO]-(neighbor_node) '
+    cypher_query += 'DETACH DELETE node_from '
+    cypher_query += 'RETURN node_to'
+
+    # print('cypher_merge_nodes(): node_merge_from_element_id: ' + str(node_merge_from_element_id))
+    # print('                      node_merge_to_element_id: ' + str(node_merge_to_element_id))
+    # print('                      node_merge_to_properties: ' + str(node_merge_to_properties))
+    # print('                      cypher_query: ' + cypher_query)
+
+    nodes = _graph.execute_query(cypher_query,
+                                 node_merge_from_element_id=node_merge_from_element_id,
+                                 node_merge_to_element_id=node_merge_to_element_id,
+                                 node_merge_to_properties=node_merge_to_properties,
+                                 result_transformer_=Result.value,
+                                 database_=ricgraph_databasename())
+
+    # If merge_node_from does not have neighbors, it will not be deleted.
+    # Do another Cypher query to be sure it is gone.
+    cypher_delete_node(node_element_id=node_merge_from_element_id)
+
+    nr_edges = ricgraph_nr_edges_of_node(node_element_id=node_merge_to_element_id)
+    graphdb_nr_reads += 2 * nr_edges        # Approximation: edges are in two directions.
+    graphdb_nr_updates += 2 * nr_edges      # Approximation: edges are in two directions.
+    if len(nodes) == 0:
+        return None
+    else:
+        return nodes[0]
+
+
+def cypher_create_edge_if_not_exists(left_node_element_id: str, right_node_element_id: str) -> None:
+    """Create an edge between two nodes, but only if the edge does not exist.
+
+    :param left_node_element_id: the element_id of the left node.
+    :param right_node_element_id: the element_id of the right node.
+    :return: None.
+    """
+    global graphdb_nr_reads, graphdb_nr_updates
+
+    graphdb_name = ricgraph_database()
+
+    # There are several methods for getting the nodes in the graph database.
+    # This one takes the most time.
+    # It amounts to ?? database hits according to 'PROFILE <cypher query>', using 'AllNodesScan'.
+    # cypher_query = 'MATCH (left_node {ID:' + str(left_node.element_id) + '}) ' [etc.]
+    # The next one takes average time.
+    # It amounts to 23 database hits according to 'PROFILE', using 'AllNodesScan'.
+    # cypher_query = 'MATCH (left_node) WHERE left_node._key="' + str(left_node['_key']) + '" ' [etc.]
+    # The next one is the fastest.
+    # It amounts to 1 database hits according to 'PROFILE' using 'NodeByIdSeek'.
+    cypher_query = 'MATCH (left_node:RicgraphNode) '
+    if graphdb_name == 'neo4j':
+        cypher_query += 'WHERE elementId(left_node)=$left_node_element_id '
+    else:
+        cypher_query += 'WHERE id(left_node)=toInteger($left_node_element_id) '
+    cypher_query += 'MATCH (right_node:RicgraphNode) '
+    if graphdb_name == 'neo4j':
+        cypher_query += 'WHERE elementId(right_node)=$right_node_element_id '
+    else:
+        cypher_query += 'WHERE id(right_node)=toInteger($right_node_element_id) '
+
+    # We could use 'CREATE', but then we may get multiple edges for the same direction.
+    # cypher_query += 'CREATE (left_node)-[:LINKS_TO]->(right_node), ' [etc.]
+    # A MERGE is more expensive since it first checks for presence of the edge.
+    cypher_query += 'MERGE (left_node)-[:LINKS_TO]->(right_node) '
+    cypher_query += 'MERGE (left_node)<-[:LINKS_TO]-(right_node) '
+
+    # print('cypher_create_edge_if_not_exists(): left_node_id: ' + str(left_node_id) + ', right_node_id: '
+    #       + str(right_node_id) + ', cypher_query: ' + cypher_query)
+
+    _graph.execute_query(cypher_query,
+                         left_node_element_id=left_node_element_id,
+                         right_node_element_id=right_node_element_id,
+                         database_=ricgraph_databasename())
+    graphdb_nr_reads += 2             # one for left_node, one for right_node
+    graphdb_nr_updates += 2           # one for ->, one for <-.
+    return
+
+
+# ##############################################################################
+# General Ricgraph functions.
+# ##############################################################################
+def get_ricgraph_ini_file() -> str:
+    """Get the location of the ricgraph ini file.
+
+    :return: the location of the ini file.
+    """
+    # Try to find RICGRAPH_INI_FILENAME in the root of the virtual environment.
+    ricgraph_ini_path = sys.prefix
+    ricgraph_ini = os.path.join(ricgraph_ini_path, RICGRAPH_INI_FILENAME)
+    if os.path.exists(ricgraph_ini):
+        return ricgraph_ini
+
+    # Try to find RICGRAPH_INI_FILENAME in the parent directory of the venv,
+    # which may happen when using a Python IDE.
+    ricgraph_ini_path_parent = os.path.dirname(ricgraph_ini_path)
+    ricgraph_ini = os.path.join(ricgraph_ini_path_parent, RICGRAPH_INI_FILENAME)
+    if os.path.exists(ricgraph_ini):
+        return ricgraph_ini
+
+    print('Ricgraph initialization: error, Ricgraph ini file "' + RICGRAPH_INI_FILENAME + '" not found in')
+    print('   directory "' + ricgraph_ini_path + '", nor in')
+    print('   directory "' + ricgraph_ini_path_parent + '", exiting.')
+    exit(1)
+
+
+def timestamp(seconds: bool = False) -> str:
+    """Get a timestamp only consisting of a time.
+
+    :param seconds: If True, also show seconds in the timestamp.
+    :return: the timestamp.
+    """
+    now = datetime.now()
+    if seconds:
+        time_stamp = now.strftime("%H:%M:%S")
+    else:
+        time_stamp = now.strftime("%H:%M")
+    return time_stamp
+
+
+def datetimestamp(seconds: bool = False) -> str:
+    """Get a timestamp consisting of a date and a time.
+
+    :param seconds: If True, also show seconds in the timestamp.
+    :return: the timestamp.
+    """
+    now = datetime.now()
+    if seconds:
+        datetime_stamp = now.strftime("%Y-%m-%d %H:%M:%S")
+    else:
+        datetime_stamp = now.strftime("%Y-%m-%d %H:%M")
+    return datetime_stamp
+
+
+def graphdb_nr_accesses_reset() -> None:
+    """Reset the counters that are used to count the number of accesses
+    to the graph database backend.
+
+    :return: None.
+    """
+    global graphdb_nr_creates, graphdb_nr_reads
+    global graphdb_nr_updates, graphdb_nr_deletes
+
+    graphdb_nr_creates = 0
+    graphdb_nr_reads = 0
+    graphdb_nr_updates = 0
+    graphdb_nr_deletes = 0
+    return
+
+
+def graphdb_nr_accesses_print() -> None:
+    """Print the values of the counters for the number of accesses to
+    the graph database backend.
+
+    :return: None.
+    """
+    # These counters are used to count the number of accesses to the graph database.
+    global graphdb_nr_creates, graphdb_nr_reads, graphdb_nr_updates, graphdb_nr_deletes
+
+    graphdb_total_accesses = graphdb_nr_creates + graphdb_nr_reads
+    graphdb_total_accesses += (2 * graphdb_nr_updates) + graphdb_nr_deletes
+    nr_nodes = ricgraph_nr_nodes()
+    nr_edges = ricgraph_nr_edges()
+
+    print('\n')
+    print('These are the number of accesses to the graph database backend at ' + datetimestamp() + ':')
+    print('- total number of creates:  {:>10}'.format(graphdb_nr_creates))
+    print('- total number of reads:    {:>10}'.format(graphdb_nr_reads))
+    print('- total number of updates:  {:>10}'.format(graphdb_nr_updates))
+    print('- total number of deletes:  {:>10}'.format(graphdb_nr_deletes))
+    print('- total number of accesses: {:>10}'.format(graphdb_total_accesses))
+    print('Please note that for the total number of accesses, an "update" counts twice: '
+          + 'one for the read and one for the update of the value.')
+    print('\n')
+    print('These are the number of nodes and edges in Ricgraph:')
+    print('- total number of nodes: {:>8}'.format(nr_nodes))
+    print('- total number of edges: {:>8}'.format(nr_edges))
+    print('\n')
+    return
+
+
+def create_ricgraph_key(name: str, value: str) -> str:
+    """Create a key for a node.
+    This function generates a composite key for a node in a graph.
+
+    :param name: name of the node.
+    :param value: value of the node.
+    :return: key generated for the node.
+    """
+    # Check and correct for occurrences of RICGRAPH_KEY_SEPARATOR.
+    name = name.replace(RICGRAPH_KEY_SEPARATOR, '#')
+    value = value.replace(RICGRAPH_KEY_SEPARATOR, '#')
+
+    return value.lower() + RICGRAPH_KEY_SEPARATOR + name.lower()
+
+
+def get_namepart_from_ricgraph_key(key: str) -> str:
+    """Get the 'name' part from the composite key in a node.
+
+    :param key: key of the node.
+    :return: name part of key for the node.
+    """
+    key_list = key.split(sep=RICGRAPH_KEY_SEPARATOR)
+    if len(key_list) != 2:
+        return ''
+    return key_list[1]
+
+
+def get_valuepart_from_ricgraph_key(key: str) -> str:
+    """Get the 'value' part from the composite key in a node.
+
+    :param key: key of the node.
+    :return: value part of key for the node.
+    """
+    key_list = key.split(sep=RICGRAPH_KEY_SEPARATOR)
+    if len(key_list) != 2:
+        return ''
+    return key_list[0]
+
+
+def create_well_known_url(name: str, value: str) -> str:
+    """Create a URL to refer to the source of a 'well known' identifier (i.e. ORCID, ISNI, etc.).
+
+    :param name: an identifier name, e.g. ORCID, ISNI, etc.
+    :param value: the value.
+    :return: URL.
+    """
+    if name == '' or value == '':
+        return ''
+
+    if name == 'DOI':
+        return 'https://doi.org/' + value
+    elif name == 'GITHUB':
+        return 'https://www.github.com/' + value
+    elif name == 'ISNI':
+        return 'https://isni.org/isni/' + value
+    elif name == 'LINKEDIN':
+        return 'https://www.linkedin.com/in/' + value
+    # elif name == 'OPENALEX':
+    #     return 'https://openalex.org/' + value
+    elif name == 'ORCID':
+        return 'https://orcid.org/' + value
+    elif name == 'RESEARCHER_ID':
+        return 'https://www.webofscience.com/wos/author/record/' + value
+    elif name == 'ROR':
+        return 'https://ror.org/' + value
+    elif name == 'SCOPUS_AUTHOR_ID':
+        return 'https://www.scopus.com/authid/detail.uri?authorId=' + value
+    elif name == 'TWITTER':
+        return 'https://www.twitter.com/' + value
+    else:
+        return ''
+
+
+def create_history_line(property_name: str, old_value: str, new_value: str) -> str:
+    """Create a line to be used in the 'history' property of a node.
+    """
+    return 'Updated property "' + property_name + '" from "' + old_value + '" to "' + new_value + '". '
+
+
+def create_name_cache_in_personroot(node: Node, personroot: Node) -> None:
+    """This function caches the value of a node with 'name' 'FULL_NAME'
+    in the 'comment' property of its person-root node 'personroot'.
+    If the cache is not present, it is created.
+
+    :param node: node to add FULL_NAME to cache.
+    :param personroot: personroot node of 'node'. For efficiency, we do not check
+      if this really is the case.
+    :return: No return value.
+    """
+    if node is None:
+        return
+    if personroot is None:
+        return
+    if node['name'] == 'FULL_NAME':
+        time_stamp = datetimestamp()
+        if isinstance(personroot['comment'], str):
+            if personroot['comment'] == '':
+                old_value = str(personroot['comment'])      # Not necessary, for symmetry.
+                node_properties = {'comment': [node['value']]}
+                history_line = create_history_line(property_name='comment',
+                                                   old_value=old_value,
+                                                   new_value=str(node_properties['comment']))
+                node_properties['_history'] = personroot['_history'].copy()
+                node_properties['_history'].append(time_stamp + ': ' + history_line)
+                cypher_update_node_properties(node_element_id=personroot.element_id,
+                                              node_properties=node_properties)
+        elif isinstance(personroot['comment'], list):
+            if node['value'] not in personroot['comment']:
+                old_value = str(personroot['comment'])
+                node_properties = {'comment': personroot['comment'].copy()}
+                node_properties['comment'].append(node['value'])
+                history_line = create_history_line(property_name='comment',
+                                                   old_value=old_value,
+                                                   new_value=str(node_properties['comment']))
+                node_properties['_history'] = personroot['_history'].copy()
+                node_properties['_history'].append(time_stamp + ': ' + history_line)
+                cypher_update_node_properties(node_element_id=personroot.element_id,
+                                              node_properties=node_properties)
+        # In all other cases: it is already in the cache,
+        # or leave it as it is: 'comment' seems to be used for something we don't know.
+    return
+
+
+def recreate_name_cache_in_personroot(personroot: Node) -> None:
+    """This function recreates the cache of all the nodes with 'name' 'FULL_NAME'
+    in the 'comment' property of its person-root node 'personroot'.
+    If the cache is not present, it is created.
+
+    :param personroot: personroot node of 'node'. For efficiency, we do not check
+      if this really is the case.
+    :return: No return value.
+    """
+    if personroot is None:
+        return
+    neighbornodes = get_all_neighbor_nodes(node=personroot,
+                                           name_want='FULL_NAME')
+    name_cache = []
+    for node in neighbornodes:
+        if node['value'] not in name_cache:
+            name_cache.append(node['value'])
+
+    # We need to do this irrespective of the length of name_cache.
+    # If the length is 0, we might be deleting FULL_NAME nodes and that
+    # should also be reflected in the name cache.
+    if (isinstance(personroot['comment'], str) and personroot['comment'] == '') \
+       or isinstance(personroot['comment'], list):
+        time_stamp = datetimestamp()
+        history_line = create_history_line(property_name='comment',
+                                           old_value=str(personroot['comment']),
+                                           new_value=str(name_cache))
+        node_properties = {'comment': name_cache.copy(),
+                           '_history': personroot['_history'].copy()}
+        node_properties['_history'].append(time_stamp + ': Cleaned and recreated name cache. ' + history_line)
+        cypher_update_node_properties(node_element_id=personroot.element_id,
+                                      node_properties=node_properties)
+    # In all other cases: leave it as it is: 'comment' seems to be used for something we don't know.
+    return
+
+
+def create_update_node(name: str, category: str, value: str,
+                       other_properties: dict = None) -> Union[Node, None]:
+    """Create a node, or update its values if is already present.
+    The new/updated node will have values 'name', 'category', and 'value', and
+    the values in other_properties.
+    For now all properties will have a string value, except for '_source'
+    (a sorted list), and '_history' (a list).
+    Changing the 'name' or 'value' properties (and, subsequently, the '_key' value),
+    cannot be done with this function, since in that case we would not be able
+    to find it.
+
+    :param name: 'name' property of node to create or update.
+    :param category: 'category' property of node.
+    :param value: 'value' property of node.
+    :param other_properties: a dictionary of all the other properties.
+    :return: the node created, or None if this was not possible
+    """
+    if other_properties is None:
+        other_properties = {}
+    global _graph
+
+    if _graph is None:
+        print('\ncreate_update_node(): Error: graph has not been initialized or opened.\n\n')
+        return None
+
+    if not isinstance(name, str) \
+       or not isinstance(category, str) \
+       or not isinstance(value, str):
+        return None
+
+    if other_properties is None:
+        other_properties = {}
+
+    lname = str(name)
+    lcategory = str(category)
+    lvalue = str(value)
+
+    if lname == '' or lcategory == '' or lvalue == '':
+        return None
+
+    node_properties = {}
+    time_stamp = datetimestamp()
+    node = read_node(name=lname, value=lvalue)
+    if node is None:
+        # Create a node.
+        # First do the properties in RICGRAPH_PROPERTIES_STANDARD.
+        node_properties['name'] = lname
+        node_properties['category'] = lcategory
+        node_properties['value'] = lvalue
+
+        # Then do the properties in RICGRAPH_PROPERTIES_ADDITIONAL.
+        for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+            if prop_name in other_properties:
+                node_properties[prop_name] = str(other_properties[prop_name])
+            else:
+                node_properties[prop_name] = ''
+
+        # If no url_main has been passed, insert an ISNI, DOI, etc. url if category is ISNI, DOI, etc.
+        url = create_well_known_url(name=lname, value=lvalue)
+        if node_properties['url_main'] == '' and url != '':
+            node_properties['url_main'] = url
+
+        # Then do the properties in RICGRAPH_PROPERTIES_HIDDEN.
+        node_properties['_key'] = create_ricgraph_key(name=lname, value=lvalue)
+        if node_properties['source_event'] == '':
+            node_properties['_source'] = []
+        else:
+            node_properties['_source'] = [node_properties['source_event']]
+            node_properties['source_event'] = ''
+
+        if node_properties['history_event'] == '':
+            node_properties['_history'] = [time_stamp + ': Created. ']
+        else:
+            node_properties['_history'] = [time_stamp + ': Created. ' + node_properties['history_event']]
+            node_properties['history_event'] = ''
+
+        new_node = cypher_create_node(node_properties=node_properties)
+        return new_node
+
+    if RICGRAPH_NODEADD_MODE == 'strict' and node['name'] == 'FULL_NAME':
+        # We only get here if we want to connect some other node A to this FULL_NAME node B.
+        # This FULL_NAME node B already exists. Most probably it is connected to a person-root C
+        # and that person-root C to some other nodes D and E.
+        # If we would continue, A would be connected to B, and also to C, D and E.
+        # However, connecting based on a name is not a very good idea, since two
+        # different persons may have the same name.
+        # So don't do this if we are in NODEADD_MODE 'strict'.
+        return None
+
+    # Update a node.
+    history_line = ''
+
+    # First do the properties in RICGRAPH_PROPERTIES_STANDARD.
+    # It is not possible to update 'name' or 'value'.
+    if node['category'] != lcategory:
+        history_line += create_history_line(property_name='category',
+                                            old_value=node['category'],
+                                            new_value=lcategory)
+        node_properties['category'] = lcategory
+
+    # Then do the properties in RICGRAPH_PROPERTIES_ADDITIONAL.
+    for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+        if prop_name not in other_properties:
+            continue
+        if prop_name == 'history_event' or prop_name == 'source_event':
+            # Do not add these to node_properties, and don't add them to _history.
+            continue
+        # Only in case a property is in other_properties, its value may change.
+        present_val = str(node[prop_name])
+        if present_val != str(other_properties[prop_name]):
+            # But its value is only changed if it is different then the old value.
+            node_properties[prop_name] = str(other_properties[prop_name])
+            history_line += create_history_line(property_name=prop_name,
+                                                old_value=present_val,
+                                                new_value=str(other_properties[prop_name]))
+
+    # Finally do the properties in RICGRAPH_PROPERTIES_HIDDEN.
+    # It is not possible to update '_key'.
+    if 'source_event' in other_properties:
+        if other_properties['source_event'] != '':
+            if other_properties['source_event'] not in node['_source']:
+                node_properties['_source'] = node['_source'].copy()
+                node_properties['_source'].append(other_properties['source_event'])
+                node_properties['_source'].sort()
+                history_line += create_history_line(property_name='_source',
+                                                    old_value=str(node['_source']),
+                                                    new_value=str(node_properties['_source']))
+
+    if history_line == '':
+        # No changes.
+        return node
+
+    node_properties['_history'] = node['_history'].copy()
+    node_properties['_history'].append(time_stamp + ': Updated. ' + history_line)
+    updated_node = cypher_update_node_properties(node_element_id=node.element_id,
+                                                 node_properties=node_properties)
+    return updated_node
+
+
+def read_node(name: str = '', value: str = '') -> Union[Node, None]:
+    """Read a node based on name and value.
+    Since all nodes are supposed to be unique if both are
+    specified, return the first one found.
+
+    :param name: 'name' property of node.
+    :param value: 'value' property of node.
+    :return: the node read, or None if no node was found.
+    """
+    global _graph
+    global nodes_cache_key_element_id
+    global graphdb_nr_reads
+
+    if _graph is None:
+        print('\nread_node(): Error: graph has not been initialized or opened.\n\n')
+        return None
+
+    if not isinstance(name, str) or not isinstance(value, str):
+        return None
+
+    if name == '' or value == '':
+        return None
+
+    graphdb_nr_reads += 1
+    key = create_ricgraph_key(name=name, value=value)
+    if key in nodes_cache_key_element_id:
+        node_element_id = nodes_cache_key_element_id[key]
+        cypher_query = 'MATCH (node:RicgraphNode) WHERE '
+        if ricgraph_database() == 'neo4j':
+            cypher_query += 'elementId(node)=$node_element_id '
+        else:
+            cypher_query += 'id(node)=toInteger($node_element_id) '
+        cypher_query += 'RETURN node'
+        nodes = _graph.execute_query(cypher_query,
+                                     node_element_id=node_element_id,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+        if len(nodes) == 0:
+            return None
+        else:
+            return nodes[0]
+    else:
+        if len(nodes_cache_key_element_id) > MAX_NODES_CACHE_KEY_ID:
+            # Empty cache.
+            nodes_cache_key_element_id = {}
+
+        cypher_query = 'MATCH (node:RicgraphNode) WHERE (node.name=$node_name) AND (node.value=$node_value) '
+        cypher_query += 'RETURN node'
+        nodes = _graph.execute_query(cypher_query,
+                                     node_name=name,
+                                     node_value=value,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+        if len(nodes) == 0:
+            return None
+        else:
+            node = nodes[0]
+            nodes_cache_key_element_id[key] = node.element_id
+            return node
+
+
+def read_all_nodes(name: str = '', category: str = '', value: str = '',
+                   key: str = '',
+                   name_is_exact_match: bool = True,
+                   value_is_exact_match: bool = True,
+                   max_nr_nodes: int = 0) -> list:
+    """Read a number of nodes based on name, category or value.
+    Any of these parameters can be specified.
+    It is also possible to read a number of nodes based on key.
+    But not name, category, value, and key at the same time.
+    If key has been specified, the values of name_is_exact_match and
+    value_is_exact_match are not used.
+
+    :param name: 'name' property of node.
+    :param category: idem.
+    :param value: idem.
+    :param key: idem.
+    :param name_is_exact_match: if True, then do an exact match search
+      on field 'name', if False, then do a case-insensitive match.
+      Note that a case-insensitive match is more expensive.
+    :param value_is_exact_match: if True, then do an exact match search
+      on field 'value', if False, then do a case-insensitive match.
+      Note that a case-insensitive match is more expensive.
+    :param max_nr_nodes: return at most this number of nodes, 0 = all nodes.
+    :return: list of nodes read, or empty list if nothing found.
+    """
+    global _graph
+    global graphdb_nr_reads
+
+    if _graph is None:
+        print('\nread_all_nodes(): Error: graph has not been initialized or opened.\n\n')
+        return []
+
+    if not isinstance(name, str) \
+       or not isinstance(category, str) \
+       or not isinstance(value, str) \
+       or not isinstance(key, str):
+        return []
+
+    # Special case, for speed.
+    if value_is_exact_match and name != '' and category == '' and value != '':
+        # Ignore 'max_nr_nodes'.
+        cypher_query = 'MATCH (node:RicgraphNode) WHERE (node.name=$node_name) AND (node.value=$node_value) '
+        cypher_query += 'RETURN node'
+        nodes = _graph.execute_query(cypher_query,
+                                     node_name=name,
+                                     node_value=value,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+        nr_nodes = len(nodes)
+        # Unsure what to count here, this seems reasonable. '+ 1' for first node.
+        graphdb_nr_reads += nr_nodes + 1
+        if nr_nodes == 0:
+            return []
+        else:
+            return nodes
+
+    # Now all other cases.
+    # Don't allow a search for everything.
+    if name == '' and category == '' and value == '' and key == '':
+        return []
+
+    if key != '' and name == '' and category == '' and value == '':
+        # Ignore 'max_nr_nodes'.
+        cypher_query = 'MATCH (node:RicgraphNode) WHERE (node._key=$node_key) '
+        cypher_query += 'RETURN node'
+        nodes = _graph.execute_query(cypher_query,
+                                     node_key=key,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+        nr_nodes = len(nodes)
+        # Unsure what to count here, this seems reasonable. '+ 1' for first node.
+        graphdb_nr_reads += nr_nodes + 1
+        if nr_nodes == 0:
+            return []
+        else:
+            return nodes
+
+    cypher_query = 'MATCH (node:RicgraphNode) WHERE '
+    if name != '':
+        if name_is_exact_match:
+            cypher_query += '(node.name="' + name + '") AND '
+        else:
+            cypher_query += '(toLower(node.name) CONTAINS toLower("' + name + '")) AND '
+    if category != '':
+        cypher_query += '(node.category="' + category + '") AND '
+    if value != '':
+        # Value may contain special characters.
+        if value_is_exact_match:
+            # Exact match search.
+            cypher_query += '(node.value=$node_value) AND '
+        else:
+            # Case-insensitive search.
+            cypher_query += '(toLower(node.value) CONTAINS toLower($node_value)) AND '
+
+    # Remove last 'AND ', is of length -4.
+    cypher_query = cypher_query[:-4]
+    cypher_query += 'RETURN node '
+
+    if max_nr_nodes > 0:
+        cypher_query += 'LIMIT ' + str(max_nr_nodes)
+    # print(cypher_query)
+
+    if value != '':
+        # Value may contain special characters.
+        nodes = _graph.execute_query(cypher_query,
+                                     node_value=value,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+    else:
+        nodes = _graph.execute_query(cypher_query,
+                                     result_transformer_=Result.value,
+                                     database_=ricgraph_databasename())
+
+    nr_nodes = len(nodes)
+    # Unsure what to count here, this seems reasonable. '+ 1' for first node.
+    graphdb_nr_reads += nr_nodes + 1
+    if nr_nodes == 0:
+        return []
+    else:
+        return nodes
+
+
+def read_all_values_of_property(node_property: str = '') -> list:
+    """Read all the values of a certain property.
+
+    :param node_property: the property to find all values:
+       'name': all different name properties.
+       'category': all different category properties.
+       '_source': all different _source properties.
+       'name_personal': all different name properties that are personal identifiers.
+    :return: a sorted list with all the values, or empty list on error.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nread_all_values_of_property(): Error: graph has not been initialized or opened.')
+        return []
+
+    if ricgraph_nr_nodes() == 0:
+        print('\nread_all_values_of_property(): Warning: graph is empty.')
+        return []
+
+    if node_property != 'name' \
+       and node_property != 'name_personal' \
+       and node_property != 'category' \
+       and node_property != '_source':
+        print('\nread_all_values_of_property(): Error: function does not work for property "'
+              + node_property + '".\n\n')
+        return []
+
+    if node_property == 'name_personal':
+        # Note that the comment property of 'person-root' contains FULL_NAMEs,
+        # so it is also a name property that is a personal identifier.
+        cypher_query = 'MATCH (node:RicgraphNode) '
+        cypher_query += 'WHERE node.category = "person" '
+        cypher_query += 'RETURN COLLECT (DISTINCT node.name) AS entries'
+    else:
+        cypher_query = 'MATCH (node:RicgraphNode) '
+        cypher_query += 'RETURN COLLECT (DISTINCT node.' + node_property + ') AS entries'
+    result = _graph.execute_query(cypher_query,
+                                  result_transformer_=Result.data,
+                                  database_=ricgraph_databasename())
+    if len(result) == 0:
+        return []
+    result = result[0]
+    if 'entries' not in result:
+        return []
+    result_list = result['entries']
+    if len(result_list) == 0:
+        return []
+    if node_property == '_source':
+        # This is a special case, we have a list of lists that we need to untangle.
+        list_of_lists = result_list
+        result_list = []
+        for item in list_of_lists:
+            if len(item) == 0:
+                continue
+            for encapsulated_item in item:
+                if encapsulated_item not in result_list:
+                    result_list.append(encapsulated_item)
+
+    # We need 'key' to do a case insenstive search.
+    result_list_sorted = sorted(result_list, key=str.lower)
+    return result_list_sorted
+
+
+def update_node_value(name: str, old_value: str, new_value: str) -> Union[Node, None]:
+    """Update a node, change the value property.
+    This is a special case because we change the key.
+    Use carefully, because other property that contain 'old_value' (such
+    as 'url_main' or 'url_other') are not being updated, so they will
+    point to the wrong URL.
+
+    :param name: 'name' property of node.
+    :param old_value: old 'value' property of node.
+    :param new_value: old 'value' property of node.
+    :return: the node updated, or None if this was not possible.
+    """
+    # A lot of the code below is copied from create_update_node().
+    global _graph
+
+    if _graph is None:
+        print('\nupdate_node_value(): Error: graph has not been initialized or opened.\n\n')
+        return None
+
+    if not isinstance(name, str) \
+       or not isinstance(old_value, str) \
+       or not isinstance(new_value, str):
+        return None
+
+    lname = str(name)
+    loldvalue = str(old_value)
+    lnewvalue = str(new_value)
+
+    if lname == '' or loldvalue == '' or lnewvalue == '':
+        return None
+
+    time_stamp = datetimestamp()
+    node = read_node(name=lname, value=loldvalue)
+    if node is None:
+        # Node we want to change does not exist.
+        return None
+
+    newnode = read_node(name=lname, value=lnewvalue)
+    if newnode is not None:
+        # The node we want to change to does already exist.
+        # So this happens to be a merge of two nodes.
+        merged_node = merge_two_nodes(node_merge_from=node,
+                                      node_merge_to=newnode)
+        return merged_node
+
+    node_properties = {'value': lnewvalue}
+    oldkey = node['_key']
+    newkey = create_ricgraph_key(name=lname, value=lnewvalue)
+    node_properties['_key'] = newkey
+    history_line = create_history_line(property_name='value', old_value=loldvalue, new_value=lnewvalue)
+    history_line += create_history_line(property_name='_key', old_value=oldkey, new_value=newkey)
+    node_properties['_history'] = node['_history'].copy()
+    node_properties['_history'].append(time_stamp + ': Updated. ' + history_line)
+    updated_node = cypher_update_node_properties(node_element_id=node.element_id,
+                                                 node_properties=node_properties)
+
+    if updated_node['name'] == 'FULL_NAME':
+        personroot = get_or_create_personroot_node(person_node=updated_node)
+        recreate_name_cache_in_personroot(personroot=personroot)
+    return node
+
+
+# Note the similarity with create_nodepairs_and_edges_df().
+def update_nodes_df(nodes: pandas.DataFrame) -> None:
+    """Update the values in a node, but only if values have been changed.
+    This is done for all the nodes passed in the DataFrame.
+
+    The column names specify the property in a node (i.e. 'name', 'category', 'value').
+    The row value of that column specify the property value of that property.
+
+    :param nodes: the nodes in a DataFrame.
+    :return: None.
+    """
+    nodes_clean = nodes.copy(deep=True)
+    # Ensure that all '' values are NaN, so that those rows can be easily removed with dropna()
+    nodes_clean.replace('', numpy.nan, inplace=True)
+    nodes_clean.dropna(axis=0, how='any', inplace=True)
+    nodes_clean.drop_duplicates(keep='first', inplace=True, ignore_index=True)
+
+    print('There are ' + str(len(nodes_clean)) + ' nodes ('
+          + timestamp() + '), updating node: 0  ', end='')
+    count = 0
+    columns = nodes_clean.columns
+    for row in nodes_clean.itertuples():
+        count += 1
+        if count % 250 == 0:
+            print(count, ' ', end='', flush=True)
+        if count % 2500 == 0:
+            print('(' + timestamp() + ')\n', end='', flush=True)
+
+        node_properties = {}
+        for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+            for other_name in columns:
+                if prop_name == other_name:
+                    node_properties[prop_name] = getattr(row, other_name)
+
+        create_update_node(name=row.name, category=row.category, value=row.value,
+                           other_properties=node_properties)
+
+    print(count, '(' + timestamp() + ')\n', end='', flush=True)
+    return
+
+
+def delete_node(node: Node) -> None:
+    """Delete a node in the graph database.
+    If it has any edges, these will also be deleted.
+
+    :param node: the node to delete.
+    :return: None.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\ndelete_node(): Error: graph has not been initialized or opened.\n\n')
+        return
+
+    if node is None:
+        print('delete_node(): Error: node is None - cannot be found.')
+        return
+
+    if node['name'] == 'FULL_NAME':
+        # We need to recreate the name cache in the person-root node.
+        personroot = get_personroot_node(node=node)
+    else:
+        personroot = None
+
+    cypher_delete_node(node_element_id=node.element_id)
+
+    if personroot is not None:
+        recreate_name_cache_in_personroot(personroot=personroot)
+
+    return
+
+
+def get_or_create_personroot_node(person_node: Node) -> Union[Node, None]:
+    """Get a 'person-root' node for a given 'person' node, if that node has
+    a 'person-root'. If not, create the 'person-root' node.
+
+    :param person_node: the node.
+    :return: the 'person-root' node, or None on error.
+    """
+    global _graph
+
+    if person_node is None:
+        print('get_or_create_personroot_node(): Error: node is None.')
+        return
+
+    if person_node['category'] != 'person':
+        print('get_or_create_personroot_node(): node category is not person.')
+        return None
+
+    if person_node['name'] == 'person-root':
+        return person_node
+
+    personroot_nodes = get_all_personroot_nodes(node=person_node)
+    if len(personroot_nodes) == 0:
+        # Create the 'person-root' node with a unique value.
+        value = str(uuid.uuid4())
+        personroot = create_update_node(name='person-root', category='person', value=value)
+        cypher_create_edge_if_not_exists(left_node_element_id=person_node.element_id,
+                                         right_node_element_id=personroot.element_id)
+        return personroot
+
+    if len(personroot_nodes) > 1:
+        print('get_or_create_personroot_node(): not anticipated: person_node "'
+              + person_node['_key'] + '" has more than one person-root nodes: ')
+        for node in personroot_nodes:
+            print('"' + node['_key'] + '" ', end='')
+        print('\nReturn the first.')
+        # And fall through.
+
+    personroot = personroot_nodes[0]
+    return personroot
+
+
+def connect_person_and_non_person_node(person_node: Node,
+                                       non_person_node: Node) -> None:
+    """Connect a 'person' and a non 'person' node.
+
+    :param person_node: the left node, being of category 'person'.
+    :param non_person_node: the right node, being of a different category than 'person'.
+    :return: None.
+    """
+    global _graph
+
+    if person_node is None or non_person_node is None:
+        print('connect_persons_and_non_person_node(): Error: (one of the) nodes is None.')
+        return
+
+    if person_node['category'] != 'person' or non_person_node['category'] == 'person':
+        print('connect_person_and_non_person_node(): (one of the) nodes have wrong category.')
+        return
+
+    if person_node['name'] == 'person-root':
+        personroot = person_node
+    else:
+        personroot = get_or_create_personroot_node(person_node=person_node)
+        if personroot is None:
+            return
+
+    cypher_create_edge_if_not_exists(left_node_element_id=non_person_node.element_id,
+                                     right_node_element_id=personroot.element_id)
+    return
+
+
+def merge_two_nodes(node_merge_from: Node, node_merge_to: Node) -> Union[Node, None]:
+    """Merge two nodes.
+    The neighbors of 'node_merge_from' will be merged to node 'node_merge_to'
+    and 'node_merge_from' will be deleted.
+
+    :param node_merge_from: the node to merge from.
+    :param node_merge_to: the node to merge to.
+    :return: the merged node, or None if this was not possible.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nmerge_two_nodes(): Error: graph has not been initialized or opened.\n\n')
+        return None
+
+    if node_merge_from is None and node_merge_to is not None:
+        # Done.
+        return node_merge_to
+
+    if node_merge_from is None or node_merge_to is None:
+        print('merge_two_nodes(): Error: one or both of the nodes is None - cannot be found.')
+        return None
+
+    if node_merge_from['name'] != node_merge_to['name'] \
+       or node_merge_from['category'] != node_merge_to['category']:
+        print('merge_two_nodes(): nodes "' + node_merge_from['_key']
+              + '" and "' + node_merge_to['_key']
+              + '" have a different "name" or "category" field: not supported.')
+        return None
+
+    if node_merge_from == node_merge_to:
+        # They are the same, done.
+        return node_merge_to
+
+    node_merge_to_properties = {'_history': node_merge_to['_history'].copy()}
+    node_merge_to_properties['_source'] = list(set(node_merge_from['_source'] + node_merge_to['_source']))
+    node_merge_to_properties['_source'].sort()
+
+    time_stamp = datetimestamp(seconds=True)
+    count = 0
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += 'Merged node "'
+    what_happened += node_merge_from['_key'] + '" to this node '
+    what_happened += 'and then deleted it.'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += '----- Start of deleted node history and neighbors -----'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += 'Start of history of the deleted node.'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    for history in node_merge_from['_history']:
+        count += 1
+        what_happened = time_stamp + '-' + format(count, '02d') + ': '
+        what_happened += history
+        node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += 'End of history of the deleted node.'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += 'These were the neighbors of the deleted node, '
+    what_happened += 'now merged with the neighbors of this node:'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    neighbornodes = get_all_neighbor_nodes(node=node_merge_from)
+    if len(neighbornodes) == 0:
+        count += 1
+        what_happened = time_stamp + '-' + format(count, '02d') + ': '
+        what_happened += '[the node to be deleted has no neighbors]'
+        node_merge_to_properties['_history'].append(what_happened)
+    else:
+        for node in neighbornodes:
+            count += 1
+            what_happened = time_stamp + '-' + format(count, '02d') + ': '
+            what_happened += '"' + str(node['_key']) + '" '
+            node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += 'End of list of neighbors of the deleted node.'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    count += 1
+    what_happened = time_stamp + '-' + format(count, '02d') + ': '
+    what_happened += '----- End of deleted node history and neighbors -----'
+    node_merge_to_properties['_history'].append(what_happened)
+
+    # Note only properties '_history' and '_source' are changed.
+    merged_node = cypher_merge_nodes(node_merge_from_element_id=node_merge_from.element_id,
+                                     node_merge_to_element_id=node_merge_to.element_id,
+                                     node_merge_to_properties=node_merge_to_properties)
+    if merged_node is None:
+        return None
+    if merged_node['name'] == 'person-root':
+        # Property 'comment' is also updated.
+        recreate_name_cache_in_personroot(personroot=merged_node)
+
+    return merged_node
+
+
+def merge_two_nodes_name_value(node_merge_from_name: str, node_merge_from_value: str,
+                               node_merge_to_name: str, node_merge_to_value: str) -> None:
+    """Merge two nodes, by specifying their 'name' and 'value' fields.
+    The neighbors of 'node_merge_from' will be merged to node 'node_merge_to'
+    and 'node_merge_from' will be deleted.
+
+    :param node_merge_from_name: 'name' property of node_merge_from.
+    :param node_merge_from_value: 'value' property of node_merge_from.
+    :param node_merge_to_name: 'name' property of node_merge_to.
+    :param node_merge_to_value: 'value' property of node_merge_to.
+    :return: None.
+    """
+    global _graph
+
+    if _graph is None:
+        print('\nmerge_two_nodes_name_value(): Error: graph has not been initialized or opened.\n\n')
+        return
+
+    node_merge_from = read_node(name=node_merge_from_name,
+                                value=node_merge_from_value)
+    node_merge_to = read_node(name=node_merge_to_name,
+                              value=node_merge_to_value)
+    merge_two_nodes(node_merge_from=node_merge_from,
+                    node_merge_to=node_merge_to)
+    return
+
+
+def connect_person_and_person_node(left_node: Node, right_node: Node) -> None:
+    """Connect two person nodes.
+
+    :param left_node: the left node.
+    :param right_node: the right node.
+    :return: None.
+    """
+    global _graph
+
+    if left_node is None or right_node is None:
+        print('connect_person_and_person_node(): (one of the) nodes is None.')
+        return
+
+    if left_node['category'] != 'person' or right_node['category'] != 'person':
+        print('connect_person_and_person_node(): (one of the) nodes have wrong category.')
+        return
+
+    if left_node['name'] == 'person-root' or right_node['name'] == 'person-root':
+        print('connect_person_and_person_node(): not anticipated: (one of the) person_nodes '
+              + 'are "person-root" nodes.')
+        return
+
+    nr_edges_left_node = ricgraph_nr_edges_of_node(node_element_id=left_node.element_id)
+    nr_edges_right_node = ricgraph_nr_edges_of_node(node_element_id=right_node.element_id)
+    if nr_edges_left_node == 0 and nr_edges_right_node == 0:
+        # None of the nodes have a 'person-root' node, create one and connect.
+        personroot = get_or_create_personroot_node(person_node=left_node)
+        if personroot is None:
+            return
+        cypher_create_edge_if_not_exists(left_node_element_id=right_node.element_id,
+                                         right_node_element_id=personroot.element_id)
+        return
+
+    if nr_edges_left_node > 0 and nr_edges_right_node == 0:
+        # 'left_node' already has a 'person-root' node.
+        personroot = get_or_create_personroot_node(person_node=left_node)
+        if personroot is None:
+            return
+        cypher_create_edge_if_not_exists(left_node_element_id=right_node.element_id,
+                                         right_node_element_id=personroot.element_id)
+        return
+
+    if nr_edges_left_node == 0 and nr_edges_right_node > 0:
+        # 'right_node' already has a 'person-root' node.
+        personroot = get_or_create_personroot_node(person_node=right_node)
+        if personroot is None:
+            return
+        cypher_create_edge_if_not_exists(left_node_element_id=left_node.element_id,
+                                         right_node_element_id=personroot.element_id)
+        return
+
+    left_personroot_node = get_or_create_personroot_node(person_node=left_node)
+    right_personroot_node = get_or_create_personroot_node(person_node=right_node)
+    if left_personroot_node is None or right_personroot_node is None:
+        return
+
+    if left_personroot_node == right_personroot_node:
+        # Already connected, nothing to do.
+        return
+
+    # Only continue depending on RICGRAPH_NODEADD_MODE.
+    if RICGRAPH_NODEADD_MODE == 'strict':
+        # For more explanation, see file docs/ricgraph_install_configure.md,
+        # section RICGRAPH_NODEADD_MODE.
+        return
+
+    # Connect crosswise.
+    cypher_create_edge_if_not_exists(left_node_element_id=left_node.element_id,
+                                     right_node_element_id=right_personroot_node.element_id)
+    cypher_create_edge_if_not_exists(left_node_element_id=right_node.element_id,
+                                     right_node_element_id=left_personroot_node.element_id)
+
+    time_stamp = datetimestamp()
+    message = 'The node pair "'
+    message += left_node['_key'] + '" and "' + right_node['_key']
+    message += '" caused this node to have more than one neighbor. '
+    message += 'These are its person-root nodes: "'
+    message += left_personroot_node['_key'] + '" and "'
+    message += right_personroot_node['_key'] + '". '
+    message += 'This might be caused by a mislabeling in a harvested system.'
+    timestamped_message = time_stamp + ': ' + message
+
+    print('\nconnect_person_and_person_node(): ' + message)
+    node_property = {'_history': left_node['_history'].copy()}
+    node_property['_history'].append(timestamped_message)
+    cypher_update_node_properties(node_element_id=left_node.element_id,
+                                  node_properties=node_property)
+    node_property = {'_history': right_node['_history'].copy()}
+    node_property['_history'].append(timestamped_message)
+    cypher_update_node_properties(node_element_id=right_node.element_id,
+                                  node_properties=node_property)
+    return
+
+
+def connect_two_nodes(left_node: Node, right_node: Node) -> None:
+    """Connect two nodes with two directed edges.
+
+    In case a node is to contain a personal ID (e.g. ORCID, ISNI, etc.) (then it has
+    category = 'person'), a 'person-root' node will be created if it does not exist.
+    This 'person-root' node is to be seen as the 'representative' for a person.
+    All research outputs will be linked to the 'person-root' node, not to the personal ID node.
+
+    :param left_node: the left node.
+    :param right_node: the right node.
+    :return: None.
+    """
+    global _graph
+
+    if left_node is None or right_node is None:
+        return
+
+    if left_node['category'] != 'person' and right_node['category'] != 'person':
+        # This is not a person to person link, link directly
+        cypher_create_edge_if_not_exists(left_node_element_id=left_node.element_id,
+                                         right_node_element_id=right_node.element_id)
+        return
+
+    # At least one of the nodes is a 'person' link. These should be linked via their 'person-root' node.
+    if left_node['category'] == 'person' and right_node['category'] != 'person':
+        connect_person_and_non_person_node(person_node=left_node,
+                                           non_person_node=right_node)
+        return
+
+    if left_node['category'] != 'person' and right_node['category'] == 'person':
+        connect_person_and_non_person_node(person_node=right_node,
+                                           non_person_node=left_node)
+        return
+
+    connect_person_and_person_node(left_node=left_node,
+                                   right_node=right_node)
+    return
+
+
+def create_two_nodes_and_edge(name1: str, category1: str, value1: str,
+                              name2: str, category2: str, value2: str,
+                              **other_properties: Union[dict, str]) -> None:
+    """Create two nodes (if they do not exist) and two directed edges between those two nodes.
+    The nodes are specified by the properties passed.
+    All 'left' nodes end with '1', all right nodes end with '2'.
+
+    :param name1: 'name' property of left node.
+    :param category1: idem.
+    :param value1: idem.
+    :param name2: 'name' property of right node.
+    :param category2: idem.
+    :param value2: idem
+    :param other_properties: a dictionary of all the other properties.
+                             Names ending with '1' belong to a 'left' node,
+                             names ending with '2' to a 'right' node.
+    :return: None.
+    """
+    node_properties1 = {}
+    node_properties2 = {}
+    for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+        for other_name, other_value in other_properties.items():
+            if prop_name + '1' == other_name:
+                node_properties1.update({prop_name: other_value})
+            if prop_name + '2' == other_name:
+                node_properties2.update({prop_name: other_value})
+
+    node1 = create_update_node(name=name1, category=category1, value=value1,
+                               other_properties=node_properties1)
+    node2 = create_update_node(name=name2, category=category2, value=value2,
+                               other_properties=node_properties2)
+    if node1 is None or node2 is None:
+        return
+
+    connect_two_nodes(left_node=node1, right_node=node2)
+    if node1['name'] == 'FULL_NAME':
+        personroot = get_or_create_personroot_node(person_node=node1)
+        create_name_cache_in_personroot(node=node1, personroot=personroot)
+    if node2['name'] == 'FULL_NAME':
+        personroot = get_or_create_personroot_node(person_node=node2)
+        create_name_cache_in_personroot(node=node2, personroot=personroot)
+    return
+
+
+def create_nodepairs_and_edges_df(left_and_right_nodepairs: pandas.DataFrame) -> None:
+    """Create two nodes (called 'node pairs') and two directed edges between those nodes.
+    This is done for all the nodes passed. The nodes are specified in a DataFrame.
+    One row in the DataFrame contains two nodes which are to be connected.
+
+    The column names specify the property in a node (i.e. 'name1', 'category1', 'value1').
+    The row value of that column specify the property value of that property.
+    All 'left' nodes end with '1', all right nodes end with '2'.
+
+    :param left_and_right_nodepairs: the node pairs in a DataFrame.
+    :return: None.
+    """
+    print('There are ' + str(len(left_and_right_nodepairs)) + ' rows ('
+          + timestamp() + '), creating nodes and edges for row: 0  ', end='')
+    count = 0
+    columns = left_and_right_nodepairs.columns
+    for row in left_and_right_nodepairs.itertuples():
+        count += 1
+        if count % 250 == 0:
+            print(count, ' ', end='', flush=True)
+        if count % 2500 == 0:
+            print('(' + timestamp() + ')\n', end='', flush=True)
+
+        node_properties = {}
+        for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+            for other_name in columns:
+                if prop_name + '1' == other_name:
+                    node_properties[prop_name + '1'] = getattr(row, other_name)
+            for other_name in columns:
+                if prop_name + '2' == other_name:
+                    node_properties[prop_name + '2'] = getattr(row, other_name)
+
+        create_two_nodes_and_edge(name1=row.name1, category1=row.category1, value1=row.value1,
+                                  name2=row.name2, category2=row.category2, value2=row.value2,
+                                  **node_properties)
+    print(count, '(' + timestamp() + ')\n', end='', flush=True)
+    return
+
+
+def create_nodepairs_and_edges_params(name1: Union[dict, str], category1: Union[dict, str], value1: Union[dict, str],
+                                      name2: Union[dict, str], category2: Union[dict, str], value2: Union[dict, str],
+                                      **other_properties: Union[dict, str]) -> None:
+    """Create two nodes (called 'node pairs') and two directed edges between those nodes.
+    This is done for all the nodes passed. The nodes are specified by values passed to this function.
+    Each value can be either a string or a dict containing strings.
+    All 'left' nodes end with '1', all right nodes end with '2'.
+
+    If it is a string (like 'ORCID') then that value will be used for every
+    property value for that property. That is, every node pair will have the same value for that property.
+
+    If it is a dict containing strings, then every node pair will have a different value for that property.
+
+    :param name1: 'name' property of left node.
+    :param category1: idem.
+    :param value1: idem.
+    :param name2: 'name' property of right node.
+    :param category2: idem.
+    :param value2: idem.
+    :param other_properties: a dictionary of all the other properties.
+                             Names ending with '1' belong to a 'left' node,
+                             names ending with '2' to a 'right' node.
+    :return: None.
+    """
+    if isinstance(value1, str):
+        left_and_right_nodepairs = pandas.DataFrame(data={'value1': value1},
+                                                    index=[0])
+    else:
+        left_and_right_nodepairs = pandas.DataFrame(data=value1).copy(deep=True)
+        left_and_right_nodepairs.rename(columns={value1.name: 'value1'}, inplace=True)
+
+    # This makes programming more efficient
+    allowed_properties = RICGRAPH_PROPERTIES_ADDITIONAL + RICGRAPH_PROPERTIES_STANDARD
+    ext_other_properties = other_properties.copy()
+    ext_other_properties['name1'] = name1
+    ext_other_properties['name2'] = name2
+    ext_other_properties['category1'] = category1
+    ext_other_properties['category2'] = category2
+    # 'value1' has been included above
+    ext_other_properties['value2'] = value2
+
+    for prop_name in allowed_properties:
+        for other_name in ext_other_properties:
+            other_value = ext_other_properties[other_name]
+            if prop_name + '1' == other_name:
+                if isinstance(other_value, str):
+                    left_and_right_nodepairs[prop_name + '1'] = other_value
+                else:
+                    left_and_right_nodepairs = pandas.concat([left_and_right_nodepairs, other_value], axis='columns')
+                    left_and_right_nodepairs.rename(columns={other_value.name: prop_name + '1'}, inplace=True)
+
+            if prop_name + '2' == other_name:
+                if isinstance(other_value, str):
+                    left_and_right_nodepairs[prop_name + '2'] = other_value
+                else:
+                    left_and_right_nodepairs = pandas.concat([left_and_right_nodepairs, other_value], axis='columns')
+                    left_and_right_nodepairs.rename(columns={other_value.name: prop_name + '2'}, inplace=True)
+
+    # Reorder a bit, starting with the last, only the name-category-value property, since they will surely exist
+    left_and_right_nodepairs.insert(0, 'value2', left_and_right_nodepairs.pop('value2'))
+    left_and_right_nodepairs.insert(0, 'category2', left_and_right_nodepairs.pop('category2'))
+    left_and_right_nodepairs.insert(0, 'name2', left_and_right_nodepairs.pop('name2'))
+    left_and_right_nodepairs.insert(0, 'value1', left_and_right_nodepairs.pop('value1'))
+    left_and_right_nodepairs.insert(0, 'category1', left_and_right_nodepairs.pop('category1'))
+    left_and_right_nodepairs.insert(0, 'name1', left_and_right_nodepairs.pop('name1'))
+
+    create_nodepairs_and_edges_df(left_and_right_nodepairs=left_and_right_nodepairs)
+    return
+
+
+def unify_personal_identifiers(personal_identifiers: pandas.DataFrame,
+                               source_event: Union[dict, str], history_event: Union[dict, str]) -> None:
+    """Unify a collection of personal identifiers (e.g. ORCID, ISNI, etc.).
+    That means that every column is unified to every other column.
+    The identifiers are specified as columns in a DataFrame.
+
+    :param personal_identifiers: DataFrame containing personal identifiers.
+    :param source_event: an event to add to the _source list.
+    :param history_event: a history event to add.
+    :return: None.
+    """
+    # Use i & j to make sure not to unify twice
+    i = 0
+    for column1 in personal_identifiers.columns:
+        i += 1
+        j = 0
+        for column2 in personal_identifiers.columns:
+            j += 1
+            if i >= j:
+                continue
+            print('\nUnifying personal identifiers ' + column1 + ' and ' + column2
+                  + ' at ' + timestamp() + '...')
+            identifiers = personal_identifiers[[column1, column2]].copy(deep=True)
+
+            # Ensure that all '' values are NaN, so that those rows can be easily removed with dropna()
+            identifiers.replace('', numpy.nan, inplace=True)
+            # dropna(how='any'): drop row if one or more row values contain NaN
+            identifiers.dropna(axis=0, how='any', inplace=True)
+            identifiers.drop_duplicates(keep='first', inplace=True, ignore_index=True)
+
+            if RICGRAPH_NODEADD_MODE == 'strict':
+                # Remove rows which have a cell value that occurs more than once.
+                # E.g. if we have a DataFrame:
+                # FULL_NAME     ORCID
+                # Name1         10
+                # Name2         11
+                # Name3         11
+                # then we remove both rows with ORCID 11.
+                # For more explanation, see file docs/ricgraph_install_configure.md,
+                # section RICGRAPH_NODEADD_MODE.
+                identifiers.drop_duplicates(subset=column1, keep=False, inplace=True, ignore_index=True)
+                identifiers.drop_duplicates(subset=column2, keep=False, inplace=True, ignore_index=True)
+
+            create_nodepairs_and_edges_params(name1=column1, category1='person', value1=identifiers[column1],
+                                              name2=column2, category2='person', value2=identifiers[column2],
+                                              source_event1=source_event, source_event2=source_event,
+                                              history_event1=history_event, history_event2=history_event)
+    print('\nDone at ' + timestamp() + '.\n')
+    return
+
+
+def print_node_values(node: Node) -> None:
+    """Print the values of all properties in a node.
+
+    :param node: the node.
+    :return: None.
+    """
+    print('Node values:')
+    print('_key:     ' + node['_key'])
+    print('name:     ' + node['name'])
+    print('category: ' + node['category'])
+    print('value:    ' + node['value'])
+    for prop_name in RICGRAPH_PROPERTIES_ADDITIONAL:
+        print(prop_name + ': ' + node[prop_name])
+    print('source:')
+    for source in node['_source']:
+        print('- ' + source)
+    print('history:')
+    for history in node['_history']:
+        print('- ' + history)
+    print('')
+    return
+
+
+# ##############################################################################
+# Neighbor node and edge functions.
+# These start with "get_", not with "read_" as some above.
+# The difference is that "get_" functions work in constant time (starting from
+# a node) while the "read_" functions read (aka search, find) the whole graph.
+# ##############################################################################
+
+
+# ######
+# This function does not seem necessary. Use get_all_neighbor_nodes() or
+# ricgraph_nr_edges_of_node().
+# Note that this function returns a list of edges, but you cannot get the next node
+# directly from an edge in that list.
+# ######
+# def get_edges(node: Node) -> list:
+#     """Get the edges attached to a node.
+#
+#     :param node: the node.
+#     :return: a list of the edges, or empty list if no edges are connected.
+#     """
+#     global _graph
+#
+#     cypher_query = 'MATCH (node)-[edge]->() WHERE elementId(node)=$node_element_id RETURN edge'
+#     edges = _graph.execute_query(cypher_query,
+#                                  node_element_id=node.element_id,
+#                                  result_transformer_=Result.value,
+#                                  database_=ricgraph_databasename())
+#     if len(edges) == 0:
+#         return []
+#     else:
+#         return edges
+# ######
+
+
+def get_personroot_node(node: Node) -> Union[Node, None]:
+    """Get the 'person-root' node for any type of node.
+    If 'node' is already a 'person-root' node, return 'node'.
+    If there is more than one person-root node (which should not happen
+    if the category is 'person'), return the first.
+
+    :param node: the node.
+    :return: the person-root node.
+    """
+    personroot_nodes = get_all_personroot_nodes(node)
+    if len(personroot_nodes) == 0:
+        return None
+    else:
+        return personroot_nodes[0]
+
+
+def get_all_personroot_nodes(node: Node) -> list:
+    """Get the 'person-root' node(s) for any type of node.
+    If 'node' is already a 'person-root' node, return 'node'.
+    If there is more than one person-root node (which can happen if
+    node is e.g. a research output, and which should not happen
+    if the category is 'person'), all will be returned in a list.
+
+    :param node: the node.
+    :return: a list of all the person-root nodes found.
+    """
+    if node is None:
+        return []
+
+    if node['name'] == 'person-root':
+        return [node]
+
+    personroot_nodes = get_all_neighbor_nodes(node=node,
+                                              name_want='person-root')
+    return personroot_nodes
+
+
+def get_all_neighbor_nodes(node: Node,
+                           name_want: Union[str, list] = '',
+                           name_dontwant: Union[str, list] = '',
+                           category_want: Union[str, list] = '',
+                           category_dontwant: Union[str, list] = '',
+                           max_nr_neighbor_nodes: int = 0) -> list:
+    """Get all the neighbors of 'node' in a list.
+    You can restrict the nodes returned by specifying one or more of the
+    other parameters. If more than one is specified, the result is an AND.
+
+    :param node: the node we need neighbors from.
+    :param name_want: either a string which indicates that we only want neighbor
+      nodes where the property 'name' is equal to 'name_want'
+      (e.g. 'ORCID'),
+      or a list containing several node names, indicating
+      that we want all neighbor nodes where the property 'name' equals
+      one of the names in the list 'name_want'
+      (e.g. ['ORCID', 'ISNI', 'FULL_NAME']).
+      If empty (empty string), return all nodes.
+    :param name_dontwant: similar, but for property 'name' and nodes we don't want.
+      If empty (empty string), all nodes are 'wanted'.
+    :param category_want: similar to 'name_want', but now for the property 'category'.
+    :param category_dontwant: similar, but for property 'category' and nodes we don't want.
+    :param max_nr_neighbor_nodes: return at most this number of nodes, 0 = all nodes.
+    :return: the list of neighboring nodes satisfying all these criteria, or
+      empty list if nothing found.
+    """
+    global _graph
+    global graphdb_nr_reads
+
+    if _graph is None:
+        print('\nget_all_neighbor_nodes(): Error: graph has not been initialized or opened.\n\n')
+        return []
+
+    if node is None:
+        return []
+
+    if name_want is None:
+        name_want = []
+    if name_dontwant is None:
+        name_dontwant = []
+    if category_want is None:
+        category_want = []
+    if category_dontwant is None:
+        category_dontwant = []
+
+    cypher_query = 'MATCH (node:RicgraphNode)-[]->(neighbor:RicgraphNode) '
+    if ricgraph_database() == 'neo4j':
+        cypher_query += 'WHERE (elementId(node)=$node_element_id) '
+    else:
+        cypher_query += 'WHERE (id(node)=toInteger($node_element_id)) '
+
+    nr_of_not_clauses = 0
+    if isinstance(name_want, str):
+        if name_want != '':
+            cypher_query += 'AND (neighbor.name IN ["' + name_want + '"]) '
+    else:
+        if len(name_want) > 0:
+            string_list = '["' + '", "'.join(str(item) for item in name_want) + '"]'
+            cypher_query += 'AND (neighbor.name IN ' + string_list + ') '
+
+    if isinstance(name_dontwant, str):
+        if name_dontwant != '':
+            cypher_query += 'AND (NOT neighbor.name IN ["' + name_dontwant + '"]) '
+            nr_of_not_clauses += 1
+    else:
+        if len(name_dontwant) > 0:
+            string_list = '["' + '", "'.join(str(item) for item in name_dontwant) + '"]'
+            cypher_query += 'AND (NOT neighbor.name IN ' + string_list + ') '
+            nr_of_not_clauses += 1
+
+    if isinstance(category_want, str):
+        if category_want != '':
+            cypher_query += 'AND (neighbor.category IN ["' + category_want + '"]) '
+    else:
+        if len(category_want) > 0:
+            string_list = '["' + '", "'.join(str(item) for item in category_want) + '"]'
+            cypher_query += 'AND (neighbor.category IN ' + string_list + ') '
+
+    if isinstance(category_dontwant, str):
+        if category_dontwant != '':
+            cypher_query += 'AND (NOT neighbor.category IN ["' + category_dontwant + '"]) '
+            nr_of_not_clauses += 1
+    else:
+        if len(category_dontwant) > 0:
+            string_list = '["' + '", "'.join(str(item) for item in category_dontwant) + '"]'
+            cypher_query += 'AND (NOT neighbor.category IN ' + string_list + ') '
+            nr_of_not_clauses += 1
+
+    if nr_of_not_clauses >= 2:
+        # This is a special case in which the Cypher query produces unexpected results.
+        neighbor_nodes = get_all_neighbor_nodes_loop(node=node,
+                                                     name_want=name_want,
+                                                     name_dontwant=name_dontwant,
+                                                     category_want=category_want,
+                                                     category_dontwant=category_dontwant,
+                                                     max_nr_neighbor_nodes=max_nr_neighbor_nodes)
+        return neighbor_nodes
+
+    cypher_query += 'RETURN DISTINCT neighbor '
+    if max_nr_neighbor_nodes > 0:
+        cypher_query += 'LIMIT ' + str(max_nr_neighbor_nodes)
+    # print(cypher_query)
+
+    neighbor_nodes = _graph.execute_query(cypher_query,
+                                          node_element_id=node.element_id,
+                                          result_transformer_=Result.value,
+                                          database_=ricgraph_databasename())
+    nr_neighbors = len(neighbor_nodes)
+    # Unsure what to count here, this seems reasonable. '+ 1' for 'node'.
+    graphdb_nr_reads += nr_neighbors + 1
+    if nr_neighbors == 0:
+        return []
+    else:
+        return neighbor_nodes
+
+
+def get_all_neighbor_nodes_loop(node: Node,
+                                name_want: Union[str, list] = '',
+                                name_dontwant: Union[str, list] = '',
+                                category_want: Union[str, list] = '',
+                                category_dontwant: Union[str, list] = '',
+                                max_nr_neighbor_nodes: int = 0) -> list:
+    """Get all the neighbors of 'node' in a list.
+    You can restrict the nodes returned by specifying one or more of the
+    other parameters. If more than one is specified, the result is an AND.
+
+    This is the same function as get_all_neighbor_nodes() but it is much slower
+    since it uses a loop. We need this in case
+    len(name_dontwant_list) > 0 and len(category_dontwant_list) > 0.
+
+    :param node: the node we need neighbors from.
+    :param name_want: either a string which indicates that we only want neighbor
+      nodes where the property 'name' is equal to 'name_want'
+      (e.g. 'ORCID'),
+      or a list containing several node names, indicating
+      that we want all neighbor nodes where the property 'name' equals
+      one of the names in the list 'name_want'
+      (e.g. ['ORCID', 'ISNI', 'FULL_NAME']).
+      If empty (empty string), return all nodes.
+    :param name_dontwant: similar, but for property 'name' and nodes we don't want.
+      If empty (empty string), all nodes are 'wanted'.
+    :param category_want: similar to 'name_want', but now for the property 'category'.
+    :param category_dontwant: similar, but for property 'category' and nodes we don't want.
+    :param max_nr_neighbor_nodes: return at most this number of nodes, 0 = all nodes.
+    :return: the list of neighboring nodes satisfying all these criteria.
+    """
+    global _graph
+    global graphdb_nr_reads
+
+    if _graph is None:
+        print('\nget_all_neighbor_nodes_loop(): Error: graph has not been initialized or opened.\n\n')
+        return []
+
+    if node is None:
+        return []
+
+    candidate_neighbor_nodes = get_all_neighbor_nodes(node=node)
+    if len(candidate_neighbor_nodes) == 0:
+        return []
+
+    # If 'name_want' etc. are strings, convert them to lists.
+    if isinstance(name_want, str):
+        if name_want == '':
+            name_want_list = []
+        else:
+            name_want_list = [name_want]
+    else:
+        name_want_list = name_want
+
+    if isinstance(name_dontwant, str):
+        if name_dontwant == '':
+            name_dontwant_list = []
+        else:
+            name_dontwant_list = [name_dontwant]
+    else:
+        name_dontwant_list = name_dontwant
+
+    if isinstance(category_want, str):
+        if category_want == '':
+            category_want_list = []
+        else:
+            category_want_list = [category_want]
+    else:
+        category_want_list = category_want
+
+    if isinstance(category_dontwant, str):
+        if category_dontwant == '':
+            category_dontwant_list = []
+        else:
+            category_dontwant_list = [category_dontwant]
+    else:
+        category_dontwant_list = category_dontwant
+
+    count = 0
+    all_nodes = A_LARGE_NUMBER
+    if max_nr_neighbor_nodes == 0:
+        max_nr_neighbor_nodes = all_nodes
+    neighbor_nodes = []
+    for neighbor in candidate_neighbor_nodes:
+        if count >= max_nr_neighbor_nodes:
+            break
+        if node == neighbor:
+            continue
+        if neighbor['name'] in name_dontwant_list:
+            continue
+        if neighbor['category'] in category_dontwant_list:
+            continue
+        if len(name_want) == 0 and len(category_want) == 0:
+            neighbor_nodes.append(neighbor)
+            count += 1
+            continue
+        if neighbor['name'] in name_want_list \
+           and neighbor['category'] in category_want_list:
+            neighbor_nodes.append(neighbor)
+            count += 1
+            continue
+        if neighbor['name'] in name_want_list and len(category_want_list) == 0:
+            neighbor_nodes.append(neighbor)
+            count += 1
+            continue
+        if len(name_want_list) == 0 and neighbor['category'] in category_want_list:
+            neighbor_nodes.append(neighbor)
+            count += 1
+            continue
+        # Any other next_node we do not want.
+
+    # Unsure what to count here, this seems reasonable. '+ 1' for 'node'.
+    graphdb_nr_reads += len(neighbor_nodes) + 1
+    return neighbor_nodes
+
+
+# ##############################################################################
+# Utility functions
+# ##############################################################################
+
+def create_multidimensional_dict(dimension: int, dict_type):
+    """Create a multimensional dict.
+    Non-existing keys will be added automatically.
+    From https://stackoverflow.com/questions/29348345/declaring-a-multi-dimensional-dictionary-in-python,
+    the second answer.
+
+    Example use:
+    new_dict = create_multidimensional_dict(2, int)
+    new_dict['key1']['key2'] += 5
+
+    :param dimension: dimension of the dict.
+    :param dict_type: type of the dict.
+    :return: the dict.
+    """
+    if dimension == 0:
+        return None
+    if dimension == 1:
+        return defaultdict(dict_type)
+    else:
+        return defaultdict(lambda: create_multidimensional_dict(dimension=dimension - 1,
+                                                                dict_type=dict_type))
+
+
+def lookup_resout_type(research_output_type: str,
+                       research_output_mapping: dict) -> str:
+    """Convert a research output type from a harvested system
+    to a shorter and easier Ricgraph research output type, according to a certain mapping.
+    The reason for doing this is to ensure a constant naming of research output
+    types for objects harvested from different sources.
+    For more explanation, see the text at 'Research output types used in Ricgraph' at
+    the start of this file.
+
+    :param research_output_type: A research output type from a source system.
+    :param research_output_mapping: The mapping from the source system to Ricgraph
+    research output types.
+    :return: The result, in a few words.
+    """
+    if research_output_type == '':
+        print('lookup_resout_type(): no research output type specified .')
+        return 'empty'
+
+    if research_output_type not in research_output_mapping:
+        print('lookup_resout_type(): unknown research output type: "' + research_output_type + '".')
+        return 'unknown'
+
+    return research_output_mapping[research_output_type]
+
+
+# #####
+# Note:
+# This function was written using the JSON harvested by the following python files:
+# - GET: harvest_openalex_to_ricgraph.py.
+# - POST: harvest_pure_to_ricgraph.py.
+#
+# Possibly for other harvests some changes should be made. Please also test the result
+# with the above-mentioned files, and add the filename of the new harvest script.
+# #####
+def harvest_json(url: str, headers: dict, body: dict = None, max_recs_to_harvest: int = 0, chunksize: int = 0) -> list:
+    """Harvest json data from a file.
+
+    :param url: URL to harvest.
+    :param headers: headers required.
+    :param body: the body of a POST request, or [] for a GET request.
+    :param max_recs_to_harvest: maximum records to harvest.
+    :param chunksize: chunk size to use (i.e. the number of records harvested in one call to 'url').
+    :return: list of records in json format, or empty list if nothing found.
+    """
+    if body is None:
+        body = []
+
+    print('Harvesting json data from ' + url + '.')
+    print('Getting data at ' + datetimestamp() + '...')
+
+    all_records = A_LARGE_NUMBER
+    if max_recs_to_harvest == 0:
+        max_recs_to_harvest = all_records
+    if chunksize == 0:
+        chunksize = 1
+    if len(body) == 0:
+        # GET http request
+        request_type = 'get'
+        url += '&per_page=' + str(chunksize)
+    else:
+        # POST http request
+        request_type = 'post'
+        body['size'] = chunksize
+
+    # Do a first harvest to determine the number of records to harvest.
+    if request_type == 'get':
+        response = requests.get(url=url, headers=headers)
+    else:
+        response = requests.post(url=url, headers=headers, json=body)
+
+    if response.status_code != requests.codes.ok:
+        print('harvest_json(): error during harvest, possibly '
+              + 'a missing API-key or mixed up POST body?')
+        print('Status code: ' + str(response.status_code))
+        print('Url: ' + response.url)
+        print('Error: ' + response.text)
+        exit(1)
+
+    chunk_json_data = response.json()
+    total_records = 0
+    if request_type == 'get':
+        if 'meta' in chunk_json_data:
+            if 'count' in chunk_json_data['meta']:
+                total_records = chunk_json_data['meta']['count']
+    else:
+        if 'count' in chunk_json_data:
+            total_records = chunk_json_data['count']
+
+    if total_records == 0:
+        print('harvest_json(): Warning: malformed json, "count" is missing.')
+
+    if max_recs_to_harvest == all_records:
+        print('There are ' + str(total_records)
+              + ' records, harvesting in chunks of ' + str(chunksize)
+              + ' items.')
+    else:
+        print('There are ' + str(total_records)
+              + ' records, harvesting in chunks of ' + str(chunksize)
+              + ', at most ' + str(max_recs_to_harvest) + ' items.')
+
+    print('Harvesting record: ', end='', flush=True)
+
+    json_data = []
+    records_harvested = 0
+    page_nr = 1
+    # And now start the real harvesting.
+    while records_harvested <= max_recs_to_harvest:
+        if request_type == 'get':
+            url_page = url + '&page=' + str(page_nr)
+            response = requests.get(url=url_page, headers=headers)
+        else:
+            if max_recs_to_harvest - records_harvested <= chunksize:
+                # We have to harvest the last few (< chunksize).
+                body['size'] = max_recs_to_harvest - records_harvested
+            body['offset'] = records_harvested
+            response = requests.post(url=url, headers=headers, json=body)
+
+        if response.status_code != requests.codes.ok:
+            print('harvest_json(): error during harvest, possibly '
+                  + 'a missing API-key or mixed up POST body?')
+            print('Status code: ' + str(response.status_code))
+            print('Url: ' + response.url)
+            print('Error: ' + response.text)
+            exit(1)
+
+        chunk_json_data = response.json()
+        if request_type == 'get':
+            if 'results' not in chunk_json_data:
+                print('harvest_json(): Error: malformed json, "results" is missing.')
+                return []
+            if len(chunk_json_data['results']) == 0:
+                break
+            json_items = chunk_json_data['results']
+        else:
+            if 'items' not in chunk_json_data:
+                print('harvest_json(): Error: malformed json, "items" is missing.')
+                return []
+            if len(chunk_json_data['items']) == 0:
+                break
+            json_items = chunk_json_data['items']
+
+        json_data += json_items
+        print(records_harvested, ' ', end='', flush=True)
+        if page_nr % 10 == 0:
+            if page_nr != 0:
+                print('(' + timestamp() + ')\n', end='', flush=True)
+        records_harvested += chunksize
+        page_nr += 1
+
+    print(' Done at ' + timestamp() + '.\n')
+    return json_data
+
+
+def harvest_json_and_write_to_file(filename: str, url: str, headers: dict, body: dict = None,
+                                   max_recs_to_harvest: int = 0, chunksize: int = 0) -> list:
+    """Harvest json data and write the data found to a file.
+    This data is a list of records in json format. If no records are harvested, nothing is written.
+
+    :param filename: filename of the file to use for writing.
+    :param url: URL to harvest.
+    :param headers: headers required.
+    :param body: the body of a POST request, or '' for a GET request.
+    :param max_recs_to_harvest: maximum records to harvest.
+    :param chunksize: chunk size to use (i.e. the number of records harvested in one call to 'url').
+    :return: list of records in json format, or empty list if nothing found.
+    """
+    if body is None:
+        body = []
+
+    json_data = harvest_json(url=url,
+                             headers=headers,
+                             body=body,
+                             max_recs_to_harvest=max_recs_to_harvest,
+                             chunksize=chunksize)
+    if len(json_data) == 0:
+        return []
+    write_json_to_file(json_data=json_data, filename=filename)
+    return json_data
+
+
+def write_json_to_file(json_data: list, filename: str) -> None:
+    """Write json data to a file.
+    If no records are harvested, nothing is written.
+
+    :param json_data: a list of records in json format.
+    :param filename: filename of the file to use for writing.
+    :return: None.
+    """
+    print('Saving json data to ' + filename + '... ', end='', flush=True)
+    with open(filename, 'w') as fd:
+        json.dump(obj=json_data, fp=fd, ensure_ascii=False, indent=2)
+    print('Done.')
+    return
+
+
+def read_json_from_file(filename: str) -> list:
+    """Read json data from a file.
+
+    :param filename: filename of the file to use for writing.
+    :return: list of records in json format, or empty list if nothing found.
+    """
+    print('Reading json data from ' + filename + '... ', end='', flush=True)
+    with open(filename) as fd:
+        json_data = json.load(fp=fd)
+    print('Done.')
+    if json_data is None:
+        return []
+    return json_data
+
+
+def read_dataframe_from_csv(filename: str, columns: dict = None,
+                            nr_rows: int = None, datatype=None) -> pandas.DataFrame:
+    """Reads a datastructure (DataFrame) from file
+
+    :param filename: csv file to read.
+    :param columns: which columns to read.
+    :param nr_rows: how many rows to read (default: all).
+    :param datatype: type of (some) columns to read.
+    :return: dataframe read.
+    """
+    try:
+        # Some inputfiles have problems reading in utf-8
+        csv_data = pandas.read_csv(filename,
+                                   sep=',',
+                                   usecols=columns,
+                                   dtype=datatype,
+                                   engine='python',
+                                   nrows=nr_rows,
+                                   parse_dates=False,
+                                   iterator=False,
+                                   quotechar='"',
+                                   encoding='utf-8')
+    except BaseException:
+        print('read_dataframe_from_csv(): error reading in utf-8 format, reading in latin-1 format.')
+        csv_data = pandas.read_csv(filename,
+                                   sep=',',
+                                   usecols=columns,
+                                   dtype=datatype,
+                                   engine='python',
+                                   nrows=nr_rows,
+                                   parse_dates=False,
+                                   iterator=False,
+                                   quotechar='"',
+                                   encoding='latin-1')
+    return csv_data
+
+
+def write_dataframe_to_csv(filename: str, df: pandas.DataFrame) -> None:
+    """Write a datastructure (DataFrame) to file.
+
+    :param filename: csv file to write.
+    :param df: dataframe to write.
+    :return: None.
+    """
+    print('\nWriting to csv file: ' + filename + '... ', end='', flush=True)
+    df.to_csv(filename,
+              sep=',',
+              quotechar='"',
+              quoting=csv.QUOTE_ALL,
+              encoding='utf-8',
+              index=False)
+    print('Done.')
+    return
+
+
+def convert_nodes_to_list_of_dict(nodes_list: list,
+                                  max_nr_items: str = '0') -> list:
+    """Convert a list of nodes to a list of dict.
+
+    :param nodes_list: The list of nodes.
+    :param max_nr_items: The maximum number of items to return.
+    :return: A list of dicts
+    """
+    if max_nr_items == '0':
+        max_nr_nodes = A_LARGE_NUMBER
+    else:
+        max_nr_nodes = int(max_nr_items)
+
+    result_list = []
+    count = 0
+    for node in nodes_list:
+        count += 1
+        if count > max_nr_nodes:
+            break
+        result = {}
+        for item in node:
+            result[item] = node[item]
+        result_list.append(result)
+    return result_list
+
+
+def create_http_response(result_list: list = None,
+                         message: str = '',
+                         http_status: int = HTTP_RESPONSE_OK) -> Tuple[dict, int]:
+    """Create an HTTP response.
+
+    :param result_list: A list of dicts, to be put in the 'result' section of
+      the response.
+    :param message: An optional message to be put in the 'meta' section of
+      the response.
+    :param http_status: The HTTP status code to be put in the 'meta' section of
+      the response.
+    :return: An HTTP response (as dict, to be translated to json)
+      and an HTTP response code.
+    """
+    if result_list is None:
+        result_list = []
+
+    meta = {'count': len(result_list),
+            'page': 1,                      # More pages not implemented yet.
+            'per_page': len(result_list),   # Should be page length, not implemented yet.
+            'status': http_status}
+    if message != '':
+        meta['message'] = message
+
+    response = {'meta': meta,
+                'results': result_list}
+    return response, http_status
+
+
+def print_commandline_arguments(argument_list: list) -> None:
+    """Print the script name and all command line arguments.
+
+    :param argument_list: the argument list.
+    :return: None.
+    """
+    if len(argument_list) == 0:
+        print('print_commandline_arguments(): Error, empty argument_list passed, exiting.')
+        exit(1)
+
+    print('\nScript name: ' + argument_list[0])
+    if len(argument_list) == 1:
+        print('Command line arguments: [none]')
+    else:
+        print('Command line arguments: '
+              + ' '.join([str(arg) for arg in argument_list[1:]])
+              + '\n')
+    return
+
+
+def get_commandline_argument(argument: str, argument_list: list) -> str:
+    """Get the value of a command line argument
+
+    :param argument: the command line argument name.
+    :param argument_list: the argument list.
+    :return: the value that belongs to 'argument'.
+    """
+    length = len(argument_list)
+    if length == 0:
+        print('get_commandline_argument(): Error, empty argument_list passed, exiting.')
+        exit(1)
+
+    if str(argument) == '':
+        # No argument passed.
+        return ''
+
+    if length == 1:
+        # The argument list contains the script name only.
+        return ''
+
+    for i in range(1, length - 1):
+        if str(argument_list[i]) == str(argument):
+            if i + 1 <= length:
+                # Only get the next index if we are still in the array bounds.
+                return str(argument_list[i + 1])
+    return ''
+
+
+# ############################################
+# ################### main ###################
+# ############################################
+# This will be executed on module initialization
+config = configparser.ConfigParser()
+config.read(get_ricgraph_ini_file())
+try:
+    RICGRAPH_PROPERTIES_STANDARD = tuple(config['Ricgraph']['ricgraph_properties_standard'].split(','))
+    RICGRAPH_PROPERTIES_HIDDEN = tuple(config['Ricgraph']['ricgraph_properties_hidden'].split(','))
+    RICGRAPH_PROPERTIES_ADDITIONAL = tuple(config['Ricgraph']['ricgraph_properties_additional'].split(','))
+    if len(RICGRAPH_PROPERTIES_STANDARD) == 0 \
+       or len(RICGRAPH_PROPERTIES_HIDDEN) == 0 \
+       or len(RICGRAPH_PROPERTIES_ADDITIONAL) == 0:
+        print('Ricgraph initialization: error, ricgraph_properties_standard and/or')
+        print('  ricgraph_properties_hidden and/or ricgraph_properties_additional')
+        print('  are empty in Ricgraph ini file, exiting.')
+        exit(1)
+
+    if RICGRAPH_PROPERTIES_STANDARD[0] == '' \
+       or RICGRAPH_PROPERTIES_HIDDEN[0] == '' \
+       or RICGRAPH_PROPERTIES_ADDITIONAL[0] == '':
+        print('Ricgraph initialization: error, ricgraph_properties_standard and/or')
+        print('  ricgraph_properties_hidden and/or ricgraph_properties_additional')
+        print('  are empty in Ricgraph ini file, exiting.')
+        exit(1)
+
+    # For more explanation, see file docs/ricgraph_install_configure.md,
+    # section RICGRAPH_NODEADD_MODE.
+    RICGRAPH_NODEADD_MODE = config['Ricgraph']['ricgraph_nodeadd_mode']
+    if RICGRAPH_NODEADD_MODE != 'strict' and RICGRAPH_NODEADD_MODE != 'lenient':
+        print('Ricgraph initialization: error, unknown value "' + RICGRAPH_NODEADD_MODE
+              + '" for ricgraph_nodeadd_mode in Ricgraph ini file, exiting.')
+        exit(1)
+    else:
+        print('Ricgraph is using "' + RICGRAPH_NODEADD_MODE + '" for ricgraph_nodeadd_mode.')
+except KeyError:
+    print('Ricgraph initialization: error, ricgraph_properties_standard and/or')
+    print('  ricgraph_properties_hidden and/or ricgraph_properties_additional and/or ricgraph_nodeadd_mode')
+    print('  not found in Ricgraph ini file, exiting.')
+    exit(1)
+
+try:
+    GRAPHDB = config['GraphDB']['graphdb']
+    GRAPHDB_HOSTNAME = config['GraphDB']['graphdb_hostname']
+    GRAPHDB_DATABASENAME = config['GraphDB']['graphdb_databasename']
+    GRAPHDB_USER = config['GraphDB']['graphdb_user']
+    GRAPHDB_PASSWORD = config['GraphDB']['graphdb_password']
+    GRAPHDB_SCHEME = config['GraphDB']['graphdb_scheme']
+    GRAPHDB_PORT = config['GraphDB']['graphdb_port']
+    if GRAPHDB == '' or GRAPHDB_HOSTNAME == '' or GRAPHDB_DATABASENAME == '' \
+       or GRAPHDB_USER == '' \
+       or GRAPHDB_PASSWORD == '' or GRAPHDB_SCHEME == '' or GRAPHDB_PORT == '':
+        print('Ricgraph initialization: error, one or more of the GraphDB parameters '
+              + 'empty in Ricgraph ini file, exiting.')
+        exit(1)
+    GRAPHDB_URL = '{scheme}://{hostname}:{port}'.format(scheme=GRAPHDB_SCHEME,
+                                                        hostname=GRAPHDB_HOSTNAME,
+                                                        port=GRAPHDB_PORT)
+except KeyError:
+    print('Ricgraph initialization: error, one or more of GraphDB parameters '
+          + 'not found in Ricgraph ini file, exiting.')
+    exit(1)
+
+# Make sure DataFrames are printed with all columns on full width
+pandas.set_option('display.max_columns', None)
+pandas.set_option('display.width', 500)