import networkx as nx

def parse_synthea_reference(ref):
    if not ref.startswith('#'):
        #print("reference: ", ref)
        if '?' in ref and '|' in ref:
            parsed_ref = ref.split('|')[1]
#        elif '/' in ref:
#            parsed_ref = ref.split('/')[1]
        else:
            parsed_ref = ref.split(':')[2]
    else:
        parsed_ref = 'mock'
    return(parsed_ref)

def process_references(graph):
    
    isSynthea = False
    
    nodes_with_reference = [[n, attr['reference']] for n, attr in graph.nodes(data=True) if 'reference' in attr]
    
    directly_referenced_nodes = []
    
    indirectly_referenced_nodes = []

    dummy_references = []
    
    if isSynthea:
    
        nodes_with_mock_reference = []

        for i in range(len(nodes_with_reference)):
            reference = nodes_with_reference[i][1]
            parsed_reference = parse_synthea_reference(reference)

            if parsed_reference != 'mock':
                nodes_with_reference[i].append(parsed_reference)
            else:
                nodes_with_mock_reference.append(i)

        for i in sorted(nodes_with_mock_reference, reverse=True):
            del nodes_with_reference[i]

        id_to_node = {data["id"]: node for node, data in graph.nodes(data=True) if "id" in data}
        id_to_identifier_node = {data["value"]: node for node, data in graph.nodes(data=True) if ("value" in data and data['label'] == 'identifier')}

        for i in nodes_with_reference:
            ref_id=i[2]
            if ref_id in id_to_node.keys():
                directly_referenced_nodes.append([i[0], id_to_node[ref_id]])
            elif ref_id in id_to_identifier_node.keys():
                indirectly_referenced_nodes.append([i[0], id_to_identifier_node[ref_id]])
            #else:
            #    print("KEY ERROR: Key neither in to_node nor in to_identifier_node", i)

        for i in indirectly_referenced_nodes:
            node_from=list(graph.predecessors(i[0]))[0]
            node_to=list(graph.predecessors(i[1]))[0]
            ref_type=graph.nodes[i[0]]['label']
            graph.add_edge(node_from, node_to, edge_type='reference', reference_type=ref_type)
                
    else:
        
        #for node, data in graph.nodes(data=True):
        #    if "id" in data:
        #        if not "resourceType" in data:
        #            print("FAILS AT: ", data, node)
        id_to_node = {data["resourceType"]+'/'+data["id"]: node for node, data in graph.nodes(data=True) if ("id" in data and "resourceType" in data)}

        for i in nodes_with_reference:
            
            ref_id=i[1]
            if ref_id in id_to_node.keys():
                directly_referenced_nodes.append([i[0], id_to_node[ref_id]])
            else:
                dummy_references.append([i[0], ref_id])              

    for i in directly_referenced_nodes:
        node_from=list(graph.predecessors(i[0]))[0]
        node_to=i[1]
        ref_type=graph.nodes[i[0]]['label']
        graph.add_edge(node_from, node_to, edge_type='reference', reference_type=ref_type)

    for i in dummy_references:
        #print(i)
        node_to='dummy_' + i[1]
        graph.add_node(node_to, label='dummy', unique_id=i[1])
        node_from=list(graph.predecessors(i[0]))[0]
        ref_type=graph.nodes[i[0]]['label']
        graph.add_edge(node_from, node_to, edge_type='reference', reference_type=ref_type)
        
    #graph.remove_nodes_from([i[0] for i in nodes_with_reference])
    
    graph.remove_nodes_from([i[0] for i in directly_referenced_nodes])
    graph.remove_nodes_from([i[0] for i in indirectly_referenced_nodes])
    graph.remove_nodes_from([i[0] for i in dummy_references])
    
    nodes_to_remove = [n for n, attr in graph.nodes(data=True) if attr.get('label') in ['root', 'entry', 'request']]
    
    graph.remove_nodes_from(nodes_to_remove)
    
    #graph.remove_nodes_from(list(nx.isolates(graph)))