Blogpost: https://lucahammer.com/2019/02/11/network-visualization-of-every-registered-company-in-germany/

offeneregister_network.py

import json_lines

import pprint
pp = pprint.PrettyPrinter(indent=1)

def get_companies(line_number=0,lines=1):
    """
    Returns as many companies as you want as a list.
    The higher the starting line, the longer it takes.
    """
    companies = []
    with open('local_data/de_companies_ocdata.jsonl', 'rb') as f:
        current_line = 0
        max_line = line_number + lines
        for company in json_lines.reader(f):
            if current_line >= max_line:
                return(companies)
            if current_line >= line_number:
                companies.append(company)
            current_line += 1

def get_companies_and_officers():
    '''
    Returns lists of all currently registered
    companies and their officers.
    '''
    data = {'allofficers': {}, #[{name: ['company 1', 'company 2']}]
            'allcompanies': {}}
    with open('local_data/de_companies_ocdata.jsonl', 'rb') as f:
        for company in json_lines.reader(f):
            if 'current_status' in company:
                if company['current_status'] == 'currently registered':
                    data['allcompanies'][company['company_number']] = company['name']
                    if 'officers' in company:
                        for officer in company['officers']:
                            if 'end_date' not in officer:
                                if officer['name'] in data['allofficers']:
                                    data['allofficers'][officer['name']].append(company['company_number'])
                                else:
                                    data['allofficers'][officer['name']] = [company['company_number']]
    return(data)

def create_companies_network(data):
    '''
    Writes a .gdf with companies
    connected by officers.
    '''
    with open('local_data/offeneregister-companies.gdf', 'w', encoding='utf-8') as output:
        output.write('nodedef>name VARCHAR,label VARCHAR\n')
        for company,name in data['allcompanies'].items():
            output.write('{0},{1}\n'.format(company,name.replace(',','COMMA').replace('\n', 'NEWLINE')))
        output.write('edgedef>node1 VARCHAR,node2 VARCHAR\n')
        for officer,companies in data['allofficers'].items():
            for i, company in enumerate(companies):
                for y in range(i+1, len(companies)):
                    if companies[i] != companies[y]:
                        output.write('{0},{1}\n'.format(companies[i],companies[y]))
        print('Companies network created.')
        
def create_officers_network():
    '''
    Writes a .gdf with officers
    connected by their companies.
    '''
    with open('local_data/de_companies_ocdata.jsonl', 'rb') as f:
        with open('local_data/offeneregister-officers.gdf', 'w', encoding='utf-8') as output:
            output.write('nodedef>name VARCHAR,label VARCHAR\n')
            output.write('edgedef>node1 VARCHAR,node2 VARCHAR\n')
            for company in json_lines.reader(f):
                if 'current_status' in company:
                    if company['current_status'] == 'currently registered':
                        if 'officers' in company:
                            for officer_a in company['officers']:
                                if 'end_date' not in officer_a:
                                    for officer_b in company['officers']:
                                        if 'end_date' not in officer_b:
                                            if officer_a['name'] != officer_b['name']:
                                                output.write('{0},{1}\n'.format(
                                                    officer_a['name'].replace(',', 'COMMA').replace('\n', 'NEWLINE'), 
                                                    officer_b['name'].replace(',', 'COMMA').replace('\n', 'NEWLINE')))
        print('Officers network created.')
        
def create_officers_companies_network(registrar='München'):
    '''
    Writes a .gdf with companies and 
    officers connected by each other.
    '''
    nodes = []
    edges = []
    with open('local_data/de_companies_ocdata.jsonl', 'rb') as f:
        for company in json_lines.reader(f):
            if company['all_attributes']['registrar'] == registrar:
                if 'current_status' in company:
                    if company['current_status'] == 'currently registered':
                        nodes.append('{0},{1}\n'.format(
                        company['company_number'],
                        company['name'].replace(',', 'COMMA').replace('\n', 'NEWLINE')))
                        if 'officers' in company:
                            for officer in company['officers']:
                                if 'end_date' not in officer:
                                    edges.append('{0},{1},TRUE\n'.format(
                                        officer['name'].replace(',', 'COMMA').replace('\n', 'NEWLINE'), 
                                        company['company_number']))
    with open('local_data/offeneregister-combined-{0}.gdf'.format(registrar), 'w', encoding='utf-8') as output:
        output.write('nodedef>name VARCHAR,label VARCHAR\n')
        for node in nodes:
            output.write(node)
        output.write('edgedef>node1 VARCHAR,node2 VARCHAR,directed BOOLEAN\n')
        for edge in edges:
            output.write(edge)
    print('Combined network created.')

#pp.pprint(get_companies(0,2))
#data = get_companies_and_officers()
#create_officers_network()
#create_companies_network(data)
create_officers_companies_network('Köln')