從python中的csv列表創建一個json樹

Question

我正在嘗試從 python 中的一個簡單表構建一個 json 層次結構。

數據如下所示：

id         parent          name
1          10              test-name-1
2          10              test-name-2
3          5               test-name-3
4          none            test-name-4
5          10              test-name-5
6          none            test-name-6
7          1               test-name-7
8          1               test-name-8
9          8               test-name-9
10         4               test-name-10

我正在尋找這樣的輸出：

{"$4":{"name":"test-name-4","children":{
      "$10":{"name":"test-name-10","children":{
            "$1":{"name":"test-name-1","children":{
                 "$7":{"name":"test-name-7","children":{}},
                 "$8":{"name":"test-name-8","children":{
                      "$9":{"name":"test-name-9","children":{}}}}}},
            "$2":{"name":"test-name-2","children":{}},
            "$5":{"name":"test-name-5","children":{
                 "$3":{"name":"test-name-3","children":{}}}}}}}},
 "$6":{"name":"test-name-6","children":"test-name-6"}}

我不知道會有多少“葉子”或“根”，或者來自 csv 的行的順序是什么。我的問題是，有沒有一種方法可以從子節點遞歸構建字典/列表取決於父母？ 如何從python中樹的“葉子”部分生成分層樹？

謝謝您的幫助！

Answer 1

我也有一個基於 2 個循環的解決方案（1 個用於緩存，1 個用於構建），沒有 JSON 編碼器，它提供了您需要的輸出：

>>> import re
>>> from collections import defaultdict
>>> parents = defaultdict(list)
>>> for i, line in enumerate(file_.split('\n')):
    if i != 0 and line.strip():
        id_, parent, name = re.findall(r'[\d\w-]+', line)
        parents[parent].append((id_, name))


>>> parents
defaultdict(<type 'list'>, {'10': [('1', 'test-name-1'), ('2', 'test-name-2'), ('5', 'test-name-5')], 'none': [('4', 'test-name-4'), ('6', 'test-name-6')], '1': [('7', 'test-name-7'), ('8', 'test-name-8')], '5': [('3', 'test-name-3')], '4': [('10', 'test-name-10')], '8': [('9', 'test-name-9')]})

好的，現在我們有了緩存，遞歸函數可以輕松構建我們想要的輸出：

>>> def build_tree(d, val):
    return {'$' + id_: {'name': name, 'children': build_tree(d, id_)} for id_, name in d[val]}

我們只需要在之前構建的 dict 上調用它，值為'none' ，即樹根：

>>> from pprint import pprint
>>> pprint(build_tree(parents, 'none'))
{'$4': {'children': {'$10': {'children': {'$1': {'children': {'$7': {'children': {},
                                                                     'name': 'test-name-7'},
                                                              '$8': {'children': {'$9': {'children': {},
                                                                                         'name': 'test-name-9'}},
                                                                     'name': 'test-name-8'}},
                                                 'name': 'test-name-1'},
                                          '$2': {'children': {},
                                                 'name': 'test-name-2'},
                                          '$5': {'children': {'$3': {'children': {},
                                                                     'name': 'test-name-3'}},
                                                 'name': 'test-name-5'}},
                             'name': 'test-name-10'}},
        'name': 'test-name-4'},
 '$6': {'children': {}, 'name': 'test-name-6'}}
>>> 

Answer 2

要將所有子節點分配給其父節點，您可以對節點列表進行兩次傳遞。 第一遍將每個節點添加到UserDict 。 在第二遍中，每個節點的父節點都保證在UserDict因此可以將節點添加到其父節點的children節點中。

要序列化為 JSON，可以使用JSONEncoder 。

#!/usr/bin/env python

import sys
import json
import UserDict

class Node(object):
    def __init__(self, nid, parent, name):
        self.nid = nid
        self.parent = parent
        self.children = []
        self.name = name

class NodeDict(UserDict.UserDict):
    def addNodes(self, nodes):
        """ Add every node as a child to its parent by doing two passes."""
        for i in (1, 2):
            for node in nodes:
                self.data[node.nid] = node
                if node.parent in self.data.keys():
                    if node.parent != "none" and
                       node not in self.data[node.parent].children:
                        self.data[node.parent].children.append(node)

class NodeJSONEncoder(json.JSONEncoder):
    def default(self, node):
        if type(node) == Node:
            return {"nid":node.nid, "name":node.name, "children":node.children}
        raise TypeError("{} is not an instance of Node".format(node))

if __name__ == "__main__":
    nodes = []

    with open(sys.argv[1]) as f:
        for row in f.readlines()[1:]:
            nid, parent, name = row.split()
            nodes.append(Node(nid, parent, name))

    nodeDict = NodeDict()
    nodeDict.addNodes(nodes)

    rootNodes = [node for nid, node in nodeDict.items()
                 if node.parent == "none"]
    for rootNode in rootNodes:
        print NodeJSONEncoder().encode(rootNode)

結果：

{"name": "test-name-4", "nid": "4", "children":[
     {"name": "test-name-10", "nid": "10", "children":[
         {"name": "test-name-1", "nid": "1", "children":[
            {"name": "test-name-7", "nid": "7", "children": []},
            {"name": "test-name-8", "nid": "8", "children":[
                {"name": "test-name-9", "nid": "9", "children": []}]}]},
         {"name": "test-name-2", "nid": "2", "children": []},
         {"name": "test-name-5", "nid": "5", "children":[
             {"name": "test-name-3", "nid": "3", "children": []}]}]}]}
{"name": "test-name-6", "nid": "6", "children": []}

Answer 3

給出的答案在 python 3.6 中對我不起作用，因為 Dict.Dict 已被棄用。 因此，我進行了一些更改以使其工作並通過讓用戶通過命令行指定 child_id、parent_id 和 child name 的列來對其進行一些概括。 請參見下文（我只是在學習，我確信這可以改進，但它適用於我的目的）。

""" Converts a CSV file with Parent/Child Hierarchy to a hierarchical JSON file for front-end processing (javascript/DS)

USAGE: csv2json.py <somefile.csv> a b c  (column nrs of a=child_id, b=parent-id, c=name(of child))

ROOT of hierarchy should contain child_id and parent_id = 'none' or blank.  name must exist """

import sys
import json
import csv
#import UserDict
from collections import UserDict

class Node(object):
    def __init__(self, child_id, parent_id, name):
        self.child_id = child_id
        self.parent_id = parent_id
        self.children = []
        self.name = name

class NodeDict(UserDict):
    def addNodes(self, nodes):
        """ Add every node as a child to its parent_id by doing two passes."""
        for i in (1, 2):
            for node in nodes:
                self.data[node.child_id] = node
                if node.parent_id in self.data.keys():
                    if (node.parent_id != "none" or node.parent_id != "") and node not in self.data[node.parent_id].children:
                        self.data[node.parent_id].children.append(node)

class NodeJSONEncoder(json.JSONEncoder):
    def default(self, node):
        if type(node) == Node:
            return {"name":node.name, "children":node.children}
            raise TypeError("{} is not an instance of Node".format(node))

if __name__ == "__main__":
    nodes = []

    with open(sys.argv[1], 'r') as f:
        reader = csv.reader(f)
        for row in reader:
            if not row[int(sys.argv[4])] : #skip if no name/label exists
                continue
            child_id, parent_id, name = row[int(sys.argv[2])] , row[int(sys.argv[3])] , row[int(sys.argv[4])]

            nodes.append(Node(child_id, parent_id, name))

    nodeDict = NodeDict()
    nodeDict.addNodes(nodes)

    rootNodes = [node for child_id, node in nodeDict.items()
                 if (node.parent_id == "none" or node.parent_id == "")]
    for rootNode in rootNodes:
        print(NodeJSONEncoder().encode(rootNode))