def load(self):
tree_json = loadJson(tree_path)
for _id in tree_json:
node_data = tree_json[_id]
node = self.get(_id)
node.parent = self.get(node_data['parent'])
node.childs = [
self.get(child_id) for child_id in node_data['childs']
]
node.child_num = node_data['child_num']
if node.parent is None:
self.root_node = node
def load(self):
tree_json = loadJson(cat_tree_path)
for _id in tree_json:
node_data = tree_json[_id]
node = self.get(_id)
node.parent = self.get(node_data['parent'])
node.childs = [self.get(n_id) for n_id in node_data['childs']]
# if node.parent is not None:
# node.parent.childs.append(node)
node.potery_ids = node_data['potery_ids']
node.is_leave = len(node.potery_ids) > 0
if node.parent is None:
self.root_node = node
def loads(self):
files = os.listdir(poem_dir) #列出文件夹下所有的目录与文件
count = 0
for file in files:
path = os.path.join(poem_dir, file)
if os.path.isfile(path):
file = loadJson(path)
for potery_id in file:
file[potery_id]['id'] = +potery_id
self.createPotery(file[potery_id])
count += 1
if count % 100000 == 0:
print(count)
self.poteries = sorted(self.poteries, key=lambda potery: potery.id)
print('加载完成')
def getImpIds(num=10000):
global imp_ids
if imp_ids is not None:
return list(imp_ids)
ids = getAllIds()
potery_rank = loadJson(potery_rank_path)
def getRank(_id):
if _id in potery_rank:
return potery_rank[_id]
else:
return -999
ids = sorted(ids, key=lambda _id: getRank(_id), reverse=True)
imp_ids = ids[0:num]
return list(imp_ids)
if node.parent is None:
self.root_node = node
def getNodeAtLevel(self, level):
nodes = [self.root_node]
for index in range(level):
children = []
for node in nodes:
children += node.childs
nodes = children
return nodes
imp_ids = getImpIds(imp_ids_num)
cat2poteries = loadJson(cat_path)
nodeCompany = NodeCompany()
nodeCompany.load()
# test = nodeCompany.get(831)
# print(test.id)
# print(test.parent.id)
# print(test.vec3d, test.getChildIds())
# print(len(nodeCompany.root_node.getAllPoteries()), len(getAllIds()))
def getCenterVec(vecs):
vecs = np.array(vecs)
# size = len(vecs[0].tolist())
import json
import os
import copy
from commonFunction import writeJson, loadJson
poem_dir = './data/poem/'
potery_rank_path = './data/potery_rank.json'
p_id2rank = loadJson(potery_rank_path)
class Comments:
def __init__(self, json_object, sentence):
# print(json_object['Content'])
try:
self.content = json.loads(json_object['Content'])
except:
self.content = json_object['Content']
self.index = json_object['Index']
self.Type = json_object["Type"] #WordDictInJson是json以外其他全是text
self.sentence = sentence
class Author:
def __init__(self, author_name):
self.name = author_name
class Sentence:
def __init__(self, json_object, potery, index):
if 'Content' in json_object: