Python将字典的平面列表转换为层次树

时间:2019-02-06 12:08:53

标签: python list dictionary tree hierarchy

我在尝试转换以下列表时遇到了麻烦:

lst = [
    {"id": 0, "job": "CEO", "ManagerID": 0, "name": "John Smith"},
    {"id": 1, "job": "Medical Manager", "ManagerID": 0, "name": "Medic 1"},
    {"id": 2, "job": "Medical Assist", "ManagerID": 1, "name": "Medic 2"},
    {"id": 3, "job": "ICT Manager", "ManagerID": 0, "name": "ICT 1"},
    {"id": 4, "job": "ICT Assist", "ManagerID": 3, "name": "ICT 2"},
    {"id": 5, "job": "ICT Junior", "ManagerID": 4, "name": "ICT 3"}
]

转换为

output = [
    {"id": 0, "job": "CEO", "ManagerID": 0, "name": "John Smith", "children" : [
        { "id":1, "job": "Medical Manager", "name": "Medic 1", "children" : [
            {"id": 2, "job": "Medical Assist", "name": "Medic 2"}
            ]
        },
        {"id": 3, "job": "ICT Manager", "name": "ICT 1", "children":[
            {"id": 4, "job": "ICT Assist", "name": "ICT 2", "children" : [
                {"id": 5, "job": "ICT Junior", "name": "ICT 3"}
            ]}
        ]}
    ],
}]

所有其他分支都存在一个根节点(ManagerID = 0)。

我尝试改编another question中的代码,但无法生成此所需格式

我一直在使用的代码如下,但这仍然会重复父节点

classes = [] #everyones id
for item in lst:
    name = item['id']
    if name not in classes:
        classes.append(name)

treenodes = {}
root_node = None

for item in lst: # Create  tree nodes
    item['children'] = []
    name = item['id']
    treenodes[name] = item
    parent = item['ManagerID']
    if parent not in classes: # parent is root node, create
        if parent not in treenodes:
            node = {}
            node['ManagerID'] = 0 #set manager to root
            node['children'] = []
            node['id'] = parent
            root_node = node
            treenodes[parent] = node

# Connect parents and children
for item in lst: # Create  tree nodes
    parent = item['ManagerID']
    parent_node = treenodes[parent]
    parent_node['children'].append(item)

output = treenodes

任何帮助都将不胜感激。

2 个答案:

答案 0 :(得分:1)

您的代码实际上正在运行,但是您需要输入treenodes[0](CEO)。 treenodes中其余的键值对仅用于簿记,以便轻松找到给定员工条目的给定经理。

如果您不能指望0作为根节点的ID,那么您可以利用CEO被标记为自我管理这一事实;根节点是管理员ID指向自己的ID的节点。更为常见的情况是,根节点根本没有父ID。

您还将CEO添加到了他们自己的children列表中(CEO的经理ID是他们自己的ID),因此您在树中有一个递归引用。

您找到的代码不是最清晰或最有效的。我会从id到复制对象建立一个字典(因此您原来的lst字典是不变的),然后遍历该结构并将条目添加到其经理ID条目中。我正在使用“根节点自引用”规则(因此管理器ID等于他们自己的ID):

employees = {}
managers = set()
root_id = None
for emp in lst:
    id, mid = emp['id'], emp['ManagerID']
    # create a copy of emp, and add a "children" list
    employees[id] = {**emp, 'children': []}
    managers.add(mid)
    if id == mid:
        # the root of the tree references itself as the manager
        root_id = id

# add empty manager entries for missing manager IDs, reporting to root ID.
for id in managers - employees.keys():
    employees[id] = {
        'id': id, 'ManagerID': root_id, 'children': [],
        'job': None, 'name': None
    }

for id, emp in employees.items():
    manager = employees[emp.pop('ManagerID')]
    if id != root_id:  # don't add the root to anything
        manager['children'].append(emp)

output = employees[root_id]

上面的代码使用一个集合来跟踪看到过哪些经理ID,因此您可以轻松添加缺少的经理条目(在这种情况下,报告给CEO)。

您的输入将产生:

{'id': 0, 'job': 'CEO', 'name': 'John Smith', 'children':
    [{'id': 1, 'job': 'Medical Manager', 'name': 'Medic 1', 'children':
        [{'id': 2, 'job': 'Medical Assist', 'name': 'Medic 2', 'children': []}],
     },
     {'id': 3, 'job': 'ICT Manager', 'name': 'ICT 1', 'children':
        [{'id': 4, 'job': 'ICT Assist', 'name': 'ICT 2', 'children':
            [{'id': 5, 'job': 'ICT Junior', 'name': 'ICT 3', 'children': []}]
         }]
     }]
}

答案 1 :(得分:1)

这是用于构建层次结构的递归版本。

递归版本

from pprint import pprint


def to_lookup(employees):
    employee_lookup = dict()
    for employee in employees:
        if employee["id"] != employee["ManagerID"]:
            manager_id = employee["ManagerID"]
            children = employee_lookup.get(manager_id)
            if not children:
                children = employee_lookup[manager_id] = list()
            children.append(employee.copy())
        else:
            manager = employee.copy()
    return manager, employee_lookup


def build_hierarchy(manager, employee_lookup):
    employees = employee_lookup.get(manager["id"], list())
    for employee in employees:
        build_hierarchy(employee, employee_lookup)
    if employees:
        manager['children'] = employees
    return manager


employees = [
    {"id": 0, "job": "CEO", "ManagerID": 0, "name": "John Smith"},
    {"id": 1, "job": "Medical Manager", "ManagerID": 0, "name": "Medic 1"},
    {"id": 2, "job": "Medical Assist", "ManagerID": 1, "name": "Medic 2"},
    {"id": 3, "job": "ICT Manager", "ManagerID": 0, "name": "ICT 1"},
    {"id": 4, "job": "ICT Assist", "ManagerID": 3, "name": "ICT 2"},
    {"id": 5, "job": "ICT Junior", "ManagerID": 4, "name": "ICT 3"}
]

manager, employee_lookup = to_lookup(employees)
hierarchy = build_hierarchy(manager, employee_lookup)
pprint(hierarchy)

输出

{'ManagerID': 0,
 'children': [{'ManagerID': 0,
               'children': [{'ManagerID': 1,
                             'id': 2,
                             'job': 'Medical Assist',
                             'name': 'Medic 2'}],
               'id': 1,
               'job': 'Medical Manager',
               'name': 'Medic 1'},
              {'ManagerID': 0,
               'children': [{'ManagerID': 3,
                             'children': [{'ManagerID': 4,
                                           'id': 5,
                                           'job': 'ICT Junior',
                                           'name': 'ICT 3'}],
                             'id': 4,
                             'job': 'ICT Assist',
                             'name': 'ICT 2'}],
               'id': 3,
               'job': 'ICT Manager',
               'name': 'ICT 1'}],
 'id': 0,
 'job': 'CEO',
 'name': 'John Smith'}

性能测试

hierarchy_size = 2000000

employees = [
    {"id": 0, "ManagerID": 0},
]
for idx in range(1, hierarchy_size):
    manager_id = random.randint(0, idx - 1)
    employees.append({"id": idx, "ManagerID": manager_id})

start = datetime.datetime.now()

manager, employee_lookup = to_lookup(employees)
hierarchy = build_hierarchy(manager, employee_lookup)

print(datetime.datetime.now() - start)