mirror of
https://github.com/hanzoai/synth.git
synced 2026-10-30 18:20:12 +00:00
475 lines
25 KiB
Python
475 lines
25 KiB
Python
import random
|
|
import numpy as np
|
|
import re
|
|
import uuid
|
|
import json
|
|
import argparse
|
|
import pathlib
|
|
import os
|
|
from typing import List, Tuple
|
|
import copy
|
|
from games.base.game import Game
|
|
from base.data import Data
|
|
from games.tasks.time_sequence.scripts.time_sequence_verifier import TimeSequenceVerifier
|
|
from games.tasks.time_sequence.scripts.time_sequence_prompt import prompt_timeSequence
|
|
|
|
class TimeSequence(Game):
|
|
"""
|
|
TimeSequence任务类
|
|
"""
|
|
def __init__(self):
|
|
"""
|
|
初始化 time sequece 任务
|
|
"""
|
|
super().__init__("TimeSequence", TimeSequenceVerifier)
|
|
|
|
def generate(self, num_of_questions: int = 100, max_attempts: int = 100,
|
|
n: int = 3, rule_prob:int=50):
|
|
"""
|
|
生成 time sequence 的题目
|
|
|
|
@param num_of_questions: 要生成的题目数量
|
|
@param max_attempts: 每个题目的最大尝试次数
|
|
@param n: 题目中的人物数量
|
|
@param rule_prob: 为人物生成额外规则的概率
|
|
@return: 生成的题目列表
|
|
"""
|
|
if n <= 1:
|
|
raise ValueError("人数n必须大于1")
|
|
if rule_prob<0 or rule_prob>100:
|
|
raise ValueError("概率值必须在[0,1]之间")
|
|
rule_prob /= 100
|
|
game_data_list = []
|
|
generated_matrices = set()
|
|
attempts = 0
|
|
|
|
# 使用自定义的游戏生成类
|
|
generator = Generate_Task()
|
|
|
|
while len(game_data_list) < num_of_questions and attempts < max_attempts:
|
|
try:
|
|
attempts += 1
|
|
# 随机选择中文或英文提示
|
|
is_chinese = random.choice([True, False])
|
|
|
|
# 生成游戏相关数据
|
|
game_data = generator.generate_main(n=n, rule_prob=rule_prob, is_chinese=is_chinese)
|
|
|
|
# 将游戏数据转为字符串以检查重复
|
|
game_data_str = json.dumps(game_data)
|
|
if game_data_str in generated_matrices:
|
|
continue
|
|
generated_matrices.add(game_data_str)
|
|
|
|
# 生成问题描述
|
|
question = prompt_timeSequence(raw_schedule=game_data['raw_schedule'], rules_desc=game_data['rules_desc'], is_chinese=is_chinese)
|
|
|
|
# 创建游戏数据
|
|
game_data = Data(
|
|
question=question,
|
|
answer=[game_data['answers']['answer_maxLen'], game_data['answers']['answer_nums']], # 保存正确答案
|
|
metadata={
|
|
"trace_id": str(uuid.uuid4()),
|
|
"n": n,
|
|
"rule_prob": rule_prob,
|
|
"records": game_data
|
|
}
|
|
)
|
|
|
|
game_data_list.append(game_data)
|
|
|
|
except Exception as e:
|
|
print(f"生成题目时出错: {e}")
|
|
# 添加调试信息
|
|
import traceback
|
|
print(f"错误详情: {traceback.format_exc()}")
|
|
continue
|
|
|
|
if len(game_data_list) < num_of_questions:
|
|
print(f"警告: 只能生成 {len(game_data_list)}/{num_of_questions} 条游戏数据")
|
|
|
|
return game_data_list
|
|
|
|
def extract_answer(self, test_solution: str):
|
|
"""
|
|
从模型的回答中提取答案（矩阵）
|
|
|
|
@param test_solution: 模型的完整回答
|
|
@return: 提取答案列表
|
|
"""
|
|
if not test_solution:
|
|
return ""
|
|
|
|
# 尝试提取列表
|
|
matrix_pattern = r'\[.*?\]'
|
|
matrix_matches = re.findall(matrix_pattern, test_solution, re.DOTALL)
|
|
if matrix_matches:
|
|
# 使用最后一个匹配的列表
|
|
print(matrix_matches)
|
|
return matrix_matches[-1].strip()
|
|
|
|
# 如果失败，返回空字符串
|
|
return ""
|
|
|
|
class Generate_Task():
|
|
def __init__(self):
|
|
self.rules_desc, self.rules_desc_english = self.add_special_rules()
|
|
self.rule_names = list(self.rules_desc.keys())
|
|
pass
|
|
|
|
def generate_main(self, n:int=3, rule_prob:float=0.5, is_chinese=True):
|
|
"""
|
|
生成任务相关数据
|
|
|
|
@para n: 人物数量，数量越多，难度越大
|
|
@para rule_prob: 为人物添加额外规则的概率
|
|
@return: 返回人物名称、初始日程安排信息、规则限制、转换后的日程安排、最终答案
|
|
"""
|
|
final_result = {'raw_schedule':{}, 'trans_schedule':{}, 'rules':{}, 'rules_desc':{}, 'schedule_withRule':{}, 'answers':{}}
|
|
# 随机选择人名
|
|
choosed_names = self.choose_names(n=n, is_chinese=is_chinese)
|
|
# 为每个人物生成一周的日程安排，并进行转换
|
|
for name in choosed_names:
|
|
raw_days_schedules, trans_days_schedules = self.relation_intergration()
|
|
final_result['raw_schedule'][name] = raw_days_schedules
|
|
final_result['trans_schedule'][name] = trans_days_schedules
|
|
|
|
# print('----------------------------------')
|
|
# print(final_result)
|
|
# 为每个人物按照rule_prob的概率添加规则
|
|
for name in choosed_names:
|
|
if random.random()<=rule_prob:
|
|
choosed_rule = random.choice(self.rule_names)
|
|
final_result['rules'][name] = choosed_rule
|
|
final_result['rules_desc'][name] = self.rules_desc[choosed_rule] if is_chinese else self.rules_desc_english[choosed_rule]
|
|
else:
|
|
final_result['rules'][name] = ''
|
|
final_result['rules_desc'][name] = ''
|
|
# print('================================')
|
|
# 按照规则描述，更新日程安排，用于计算答案
|
|
for name in choosed_names:
|
|
new_schedule = self.add_special_rules(
|
|
days_schedules=final_result['trans_schedule'][name],
|
|
choosed_rule=final_result['rules'][name]
|
|
)
|
|
final_result['schedule_withRule'][name] = new_schedule
|
|
# print('*'*20)
|
|
# print(final_result)
|
|
|
|
# 根据规则，获取最长会议时间限制
|
|
time_limit = self.cal_maxLen_time(rules=final_result['rules'])
|
|
|
|
# 计算最终答案
|
|
answer_record, answer_nums, answer_maxLen = self.cal_answers(final_result['schedule_withRule'], time_limit=time_limit)
|
|
final_result['answers'] = {'answer_record':answer_record, 'answer_nums':answer_nums, 'answer_maxLen':answer_maxLen}
|
|
return final_result
|
|
|
|
def choose_names(self, n:int=3, is_chinese=True):
|
|
candidate_names_english = [
|
|
'Kaitlyn', 'Gerald', 'Jocelyn', 'Sara', 'Edwin', 'Savannah', 'Brian', 'Angela', 'Albert', 'Sean', 'Walter',
|
|
'Bailey', 'Helen', 'Sierra', 'Henry', 'Wayne', 'Kayla', 'Janice', 'Timothy', 'Tyler', 'Cindy', 'Julie', 'Tiffany',
|
|
'Robert', 'Chelsea', 'Joan', 'Benjamin', 'Raymond', 'Tammy', 'Kristen', 'Ashley', 'Mariah', 'Alyssa', 'Taylor',
|
|
'Thomas', 'Rose', 'Alexa', 'Rebecca', 'Sophia', 'Juan', 'Kenneth', 'Eric', 'Joe', 'Melanie', 'Ronnie', 'Leonard',
|
|
'Teresa', 'Patricia', 'Sydney', 'Paige', 'Morris', 'Katherine', 'Lauren', 'Judy', 'Stephen', 'Caleb', 'Gregory',
|
|
'Danny', 'Catherine', 'Martha', 'Abigail', 'Joshua', 'Carol', 'Christine', 'Jeremy', 'Arthur', 'Willie', 'Norman',
|
|
'Rodney', 'Larry', 'Luis', 'Casey', 'Amy', 'Gary', 'Philip', 'Melissa', 'Nicholas', 'Todd', 'Jean', 'Vanessa',
|
|
'Isabella', 'Ruth', 'Dylan', 'Michael', 'Roger', 'Frances', 'Craig', 'Glenn', 'Austin', 'Dale', 'Elizabeth', 'Allen',
|
|
'Deborah', 'Steven', 'Jack', 'Nathan', 'Terry', 'Theresa', 'Clarence', 'Manuel', 'Megan', 'Brittany', 'Johnny', 'Diane',
|
|
'Alfred', 'Harvey', 'Amanda', 'Emma', 'Morgan', 'Crystal', 'Tony', 'Faith', 'Samuel', 'Keith', 'Gabriel', 'Francis',
|
|
'Anthony', 'Victoria', 'Mark', 'Jesse', 'Makayla', 'Isaac', 'Allison', 'Rachel', 'Billy', 'Lindsey', 'Brenda', 'Rylee',
|
|
'Sabrina', 'Kylie', 'Stanley', 'Olivia', 'Jasmine', 'Edgar', 'Betty', 'Sandra', 'Floyd', 'James', 'Zoey', 'Susan', 'Jacob',
|
|
'Grace', 'Ronald', 'Justin', 'Randy', 'Marilyn', 'Vincent', 'Andrea', 'Autumn', 'Dorothy', 'Ernest', 'Jane', 'Brandon', 'Maria',
|
|
'Lisa', 'Patrick', 'William', 'Diana', 'Jimmy', 'Theodore', 'Carolyn', 'Alexander', 'Christina', 'Lloyd', 'Amber', 'Mikayla', 'Brooke',
|
|
'Jason', 'Samantha', 'Hugh', 'Aaron', 'Donna', 'Tommy', 'Dennis', 'Kyle', 'Kelsey', 'Erin', 'Charlotte', 'Jerome', 'Marissa',
|
|
'Alice', 'Leon', 'Jeffrey', 'Kathryn', 'Cory', 'Madeline', 'Alan', 'Carl', 'Kimberly', 'Joyce', 'Curtis', 'Michelle', 'Marvin',
|
|
'Jose', 'Daniel', 'Virginia', 'Shawn', 'Marie', 'Sarah', 'Karen', 'Earl', 'Brianna', 'Gloria', 'Lionel', 'Madison', 'John', 'Linda',
|
|
'Sharon', 'Anna', 'Cynthia', 'Beverly', 'Ann', 'Margaret', 'Denise', 'Bradley', 'Nathaniel', 'Michaela', 'Dustin', 'Logan', 'George',
|
|
'Frank', 'Donald', 'Glen', 'Stephanie', 'Shirley', 'Douglas', 'Bryan', 'Jordan', 'Bruce', 'Ryan', 'Eugene', 'Alexis', 'Andrew', 'Paul',
|
|
'Christopher', 'Nancy', 'Heather', 'Debra', 'Peter', 'Kelly', 'Danielle', 'Oscar', 'Joel', 'Otis', 'Gabriella', 'Lawrence', 'Jennifer',
|
|
'Evelyn', 'Lori', 'Hannah', 'Adam', 'Ethan', 'Kathleen', 'Cheryl', 'Emily', 'Barbara', 'Sophie', 'Pamela', 'Nicole', 'Martin', 'Zachary',
|
|
'Edward', 'Doris', 'Ralph', 'Mary', 'Noah', 'Christian', 'Russell', 'Joseph', 'Natalie', 'Jay', 'Laura', 'Harold', 'Julia', 'Jacqueline',
|
|
'Roy', 'Frederick', 'Judith', 'Matthew', 'Andre', 'Louis', 'Jessica', 'Kim', 'Valerie', 'Richard', 'Clifford', 'Amelia', 'Charles',
|
|
'Janet', 'Howard', 'Molly', 'Jonathan', 'Derek', 'David', 'Jerry', 'Jared', 'Bobby', 'Shane', 'Scott', 'Kevin', 'Haley'
|
|
]
|
|
candidate_names_chinese = [
|
|
"张伟", "王芳", "李娜", "刘洋", "陈强", "杨静", "黄磊", "赵敏", "周涛", "吴昊",
|
|
"徐磊", "孙浩", "胡娟", "朱丽", "高峰", "马超", "董倩", "冯伟", "程琳", "宋刚",
|
|
"何敏", "梁涛", "许倩", "邓超", "曹阳", "沈丽", "田野", "韩冰", "龚涛", "金鑫",
|
|
"薛伟", "贾磊", "戴琳", "孟凡", "方婷", "邵明", "于波", "姚辉", "熊杰", "郑浩",
|
|
"任涛", "陶亮", "卢伟", "白雪", "马丽", "钱峰", "汪静", "严磊", "戴强", "阎涛",
|
|
"谭敏", "贺丽", "罗杰", "夏阳", "石磊", "史娟", "姜涛", "左娜", "冉亮", "章伟",
|
|
"蒋倩", "莫凡", "丁浩", "易波", "覃峰", "尤静", "韦超", "庄敏", "庞磊", "武杰",
|
|
"管霞", "蓝涛", "皮娜", "封伟", "栗倩", "党超", "池娜", "辛伟", "昌磊", "支昊",
|
|
"边静", "黎涛", "湛亮", "丰杰", "候丽", "满伟", "滕静", "隆磊", "蔺倩", "温涛",
|
|
"雷浩", "谷超", "左丽", "宁峰", "沙倩", "郁磊", "鲍娜", "佟涛", "仇伟", "栾敏",
|
|
"甘磊", "那超", "殷静", "井伟", "伊倩", "仲涛", "乐丽", "虞峰", "沃超", "俞娜",
|
|
"靳伟", "岳涛", "郜磊", "阮倩", "党丽", "慕超", "荆娜", "花涛", "帅伟", "占浩",
|
|
"钟涛", "邝磊", "迟倩", "冶超", "拓静", "宫伟", "宓涛", "仝磊", "督倩", "郏丽",
|
|
"甄超", "幸娜", "门涛", "来伟", "覃磊", "化倩", "厚丽", "买超", "年娜", "伊涛",
|
|
"阴伟", "侯磊", "暴倩", "封丽", "白超", "乐娜", "全涛", "贝伟", "简磊", "车倩",
|
|
"敖丽", "满超", "勾娜", "类涛", "桑伟", "谭磊", "师倩", "詹丽", "池超", "原娜",
|
|
"东涛", "斐伟", "富磊", "漆倩", "达丽", "西超", "晁娜", "巨涛", "言伟", "普磊",
|
|
"翦倩", "湛丽", "年超", "霜娜", "桑涛", "邸伟", "米磊", "阿倩", "义丽", "秋超",
|
|
"区娜", "里涛", "满伟", "詹磊", "红倩", "豆丽", "纵超", "全娜", "闻涛", "赫伟",
|
|
"左磊", "岳倩", "理丽", "白涛", "鲁伟", "周磊", "冀倩", "东丽", "冷超", "党娜",
|
|
"钱涛", "冉伟", "印磊", "殷倩", "宇丽", "关超", "糜娜", "朱涛", "吉伟", "胡磊"
|
|
]
|
|
if is_chinese:
|
|
return random.sample(candidate_names_chinese, k=n)
|
|
return random.sample(candidate_names_english, k=n)
|
|
|
|
def generate_schedule(self):
|
|
"""
|
|
为单个人物随机生成日称安排表
|
|
|
|
从早上9点到晚上5点，每5分钟作为一个时间段进行分割。
|
|
|
|
@para name：任务名称
|
|
@return：一个长度为5的列表，代表周一到周五，每个元素是一个子列表，代表一天的日程安排
|
|
"""
|
|
def generate_day_schedule(max_sche_nums):
|
|
"""
|
|
随机生成一天的日程安排。将早上9点到晚上5点，分为96个时间段，每个时间段为5分钟。
|
|
result中的子列表，第一个元素为起始时间，第二个元素为结束时间。
|
|
"""
|
|
result = []
|
|
pre_end = 0
|
|
cur_nums=0
|
|
while(cur_nums<max_sche_nums):
|
|
weights = [1 / (i + 5) for i in range(36)]
|
|
start_time = pre_end + random.choices(range(36), weights=weights, k=1)[0]
|
|
weights = [1 / (i + 3) for i in range(24)]
|
|
duration = random.choices(range(1, 25), weights=weights, k=1)[0]
|
|
end_time = start_time + duration
|
|
if end_time>96:
|
|
break
|
|
result.append([start_time, end_time])
|
|
pre_end = end_time
|
|
return result
|
|
days_schedules = []
|
|
for i in range(5):
|
|
days_schedules.append(generate_day_schedule(random.choice([2,3,4,5])))
|
|
return days_schedules
|
|
|
|
def relation_intergration(self):
|
|
"""
|
|
日程安排生成与转换
|
|
"""
|
|
# 1. 为人物生成每天的日程安排
|
|
days_schedules = self.generate_schedule()
|
|
# 2. 为人物的每一天，随机选择日程类型为 空闲、占用
|
|
days_schedules_types = {}
|
|
for day, day_schedule in zip(['一', '二', '三', '四', '五'], days_schedules):
|
|
sche_type = random.choice(['空闲', '占用'])
|
|
days_schedules_types[day] = [sche_type, day_schedule]
|
|
|
|
raw_days_schedules_types = copy.deepcopy(days_schedules_types)
|
|
|
|
# 3. 将空闲时间，转为占用时间
|
|
days_schedules_types_trans = self.trans_free_2_booked(days_schedules_types)
|
|
return raw_days_schedules_types, days_schedules_types_trans
|
|
|
|
def trans_free_2_booked(self, days_schedules_types:dict):
|
|
"""
|
|
将空闲日程，转为占用日程，方便求解答案
|
|
"""
|
|
for key in days_schedules_types:
|
|
if days_schedules_types[key][0]=='空闲':
|
|
new_day_schedule = []
|
|
start_time = 0
|
|
for sche in days_schedules_types[key][1]:
|
|
new_day_schedule.append([start_time, sche[0]])
|
|
start_time = sche[1]
|
|
new_day_schedule.append([start_time, 96])
|
|
days_schedules_types[key][1] = new_day_schedule
|
|
days_schedules_types[key][0] = '占用'
|
|
return days_schedules_types
|
|
|
|
def add_special_rules(self, days_schedules:dict={}, choosed_rule:str=''):
|
|
"""
|
|
添加一些特殊的规则限制
|
|
"""
|
|
rules_english = {
|
|
'会议前的空闲时间限制': ' requires at least 10 minutes of free time before a meeting, and the free time must be between 9:00 AM and 5:00 PM.',
|
|
'会议后的空闲时间限制': ' requires at least 10 minutes of free time after a meeting, and the free time must be between 9:00 AM and 5:00 PM.',
|
|
'结束时间限制': ' feels tired and requires meetings to end before 4:00 PM.',
|
|
'星期三午饭时间': ' needs to have lunch between 12:30 PM and 1:00 PM on Wednesdays.',
|
|
'清空早上日程': ' can cancel the schedule between 9:00 AM and 9:45 AM each day (only those ending before 9:45 AM can be cancelled).',
|
|
'时区差异': ' is in a time zone that is one hour ahead of others.',
|
|
'时间偏好二四': ' prefers to hold short meetings on Tuesdays and Thursdays and only accepts new meetings on these two days, with a maximum duration of 1 hour.',
|
|
'30分钟短会议清除': ' can remove any meetings of 30 minutes or less from the schedule.',
|
|
'10分钟短会议清除': ' can be flexible and miss up to 10 minutes of a meeting.',
|
|
'时间偏好二五': ' prefers to hold long meetings on Tuesdays and Fridays and only accepts new meetings on these two days, but the meeting duration must not exceed 30 minutes.',
|
|
}
|
|
|
|
rules = {
|
|
'会议前的空闲时间限制': '要求在会议前至少有10分钟的空闲时间，并且空闲时间必须在上午9点到下午5点之间',
|
|
'会议后的空闲时间限制': '要求在会议结束后至少有 10 分钟的空闲时间（空闲时间必须在她上午 9 点到下午 5 点之间）',
|
|
'结束时间限制': '感到疲倦，要求会议在下午4点之前结束',
|
|
'星期三午饭时间': '星期三需要在12:30到1:00之间吃午饭',
|
|
'清空早上日程': '可以取消每天上午9:00到9:45之间的日程安排，（9:45之前结束的才可以取消）',
|
|
'时区差异': '所处的时区，比其他时区要早一个小时',
|
|
'时间偏好二四': '喜欢在周二和周四举行简短的会议，并且只在这两天接受新的会议，会议时间不超过1小时',
|
|
'30分钟短会议清除': '可以从日程安排中清除任何30分钟或更短的会议',
|
|
'10分钟短会议清除': '可以灵活缺席最多10分钟的会议',
|
|
'时间偏好二五': '喜欢在周二和周五举行长时间的会议，并且只有在这两天接受新的会议，但会议时间不得超过 30 分钟',
|
|
}
|
|
if not days_schedules:
|
|
return rules, rules_english
|
|
def change_schedule(rule:str, days_schedules:dict):
|
|
"""
|
|
根据规则，修改日程安排
|
|
"""
|
|
def remove_duration_sche(days_schedules, max_duration):
|
|
for key in days_schedules:
|
|
new_day_schedule = []
|
|
for sche in days_schedules[key][1]:
|
|
if sche[1]-sche[0]<=max_duration:
|
|
continue
|
|
new_day_schedule.append(sche)
|
|
days_schedules[key][1] = new_day_schedule
|
|
return days_schedules
|
|
|
|
if rule=='会议前的空闲时间限制':
|
|
for key in days_schedules:
|
|
for sche in days_schedules[key][1]:
|
|
sche[1] += 2
|
|
elif rule=='会议后的空闲时间限制':
|
|
for key in days_schedules:
|
|
for sche in days_schedules[key][1]:
|
|
sche[0] -= 2
|
|
elif rule=='星期三午饭时间':
|
|
days_schedules['三'][1].append([42, 48])
|
|
elif rule=='结束时间限制':
|
|
for key in days_schedules:
|
|
days_schedules[key][1].append([84, 96])
|
|
elif rule=='清空早上日程':
|
|
for key in days_schedules:
|
|
new_day_schedule = []
|
|
for sche in days_schedules[key][1]:
|
|
if sche[0]<9 and sche[1]<=9:
|
|
continue
|
|
new_day_schedule.append(sche)
|
|
days_schedules[key][1] = new_day_schedule
|
|
elif rule=='时区差异':
|
|
for key in days_schedules:
|
|
days_schedules[key][1].append([84, 96])
|
|
elif rule=='30分钟短会议清除':
|
|
days_schedules = remove_duration_sche(days_schedules, 6)
|
|
elif rule=='时间偏好二四':
|
|
days_schedules['一'][1].append([0, 96])
|
|
days_schedules['三'][1].append([0, 96])
|
|
days_schedules['五'][1].append([0, 96])
|
|
elif rule=='10分钟短会议清除':
|
|
days_schedules = remove_duration_sche(days_schedules, 2)
|
|
elif rule=='时间偏好二五':
|
|
days_schedules['一'][1].append([0, 96])
|
|
days_schedules['三'][1].append([0, 96])
|
|
days_schedules['四'][1].append([0, 96])
|
|
return days_schedules
|
|
if not choosed_rule:
|
|
return days_schedules
|
|
days_schedules_copy = copy.deepcopy(days_schedules)
|
|
days_schedules_copy = change_schedule(choosed_rule, days_schedules_copy)
|
|
return days_schedules_copy
|
|
|
|
def cal_maxLen_time(self, rules:dict):
|
|
"""
|
|
根据特殊规则的限制，获取最长会议时间
|
|
|
|
时间偏好二四：限制一小时以内
|
|
时间偏好二五：限制30分钟以内
|
|
"""
|
|
maxLen_time = 480
|
|
for name in rules:
|
|
if rules[name]=='时间偏好二四':
|
|
maxLen_time = min(maxLen_time, 60)
|
|
elif rules[name]=='时间偏好二五':
|
|
maxLen_time = min(maxLen_time, 30)
|
|
return maxLen_time
|
|
|
|
def cal_answers(self, schedule_withRule:dict, time_limit:int=480):
|
|
"""
|
|
根据日程安排表，计算最终答案
|
|
|
|
限制要求：
|
|
1、会议必须在整点或半点开始；
|
|
2、有些规则中，存在最长会议时间的限制
|
|
|
|
@param schedule_withRule: 引入规则限制之后的日程安排，可以直接计算答案
|
|
@param time_limit: 最长会议时间限制
|
|
@return [最长会议时间， 可选择的选项数]
|
|
"""
|
|
|
|
record = np.zeros((96, 5))
|
|
map_dict = {'一':0, '二':1, '三':2, '四':3, '五':4}
|
|
for name in schedule_withRule.keys():
|
|
days_schedules = schedule_withRule[name]
|
|
for key in days_schedules:
|
|
map_id = map_dict[key]
|
|
for sche in days_schedules[key][1]:
|
|
for tmp in range(max(0, sche[0]), min(sche[1], 96)):
|
|
record[tmp, map_id] = 1
|
|
answer_maxLen = 0
|
|
answer_record = []
|
|
for col in range(5):
|
|
for row in range(0, 96, 6): # 以整点或半点开始查找
|
|
if record[row, col]==1:
|
|
continue
|
|
cur_end = row
|
|
length = 0
|
|
while(cur_end<96 and record[cur_end, col]==0 and length<time_limit):
|
|
cur_end += 1
|
|
length += 5
|
|
answer_record.append([row, cur_end])
|
|
answer_maxLen = max(answer_maxLen, length)
|
|
answer_nums = 0
|
|
for tmp in answer_record:
|
|
if (tmp[1]-tmp[0])*5==answer_maxLen:
|
|
answer_nums += 1
|
|
return answer_record, answer_nums, answer_maxLen
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
parser = argparse.ArgumentParser(description="Time Sequence任务生成器")
|
|
parser.add_argument("--num_of_data", type=int, default=100, help="生成的题目数量")
|
|
parser.add_argument("--max_attempts", type=int, default=1000, help="每个题目的最大尝试次数")
|
|
parser.add_argument("--n", type=int, default=3, help="人物数量")
|
|
parser.add_argument("--rule_prob", type=int, default=50, help="额外规则生成概率")
|
|
args = parser.parse_args()
|
|
|
|
# 创建数据目录
|
|
base_data_dir = pathlib.Path(__file__).parent.parent / "data"
|
|
if not base_data_dir.exists():
|
|
base_data_dir.mkdir(parents=True, exist_ok=True)
|
|
|
|
# 创建游戏实例
|
|
game = TimeSequence()
|
|
|
|
# 生成游戏数据
|
|
game_data_list = game.generate(
|
|
num_of_questions=args.num_of_data,
|
|
max_attempts=args.max_attempts,
|
|
n=args.n,
|
|
rule_prob=args.rule_prob
|
|
)
|
|
|
|
print(f"成功生成 {len(game_data_list)} 条游戏数据")
|
|
|
|
# 创建嵌套目录结构
|
|
nested_dir = base_data_dir / f"num_of_data_{args.num_of_data}" / f"n_{args.n}" / f"rule_prob_{args.rule_prob}"
|
|
if not nested_dir.exists():
|
|
nested_dir.mkdir(parents=True, exist_ok=True)
|
|
|
|
# 设置输出文件名
|
|
output_file = nested_dir / f"time_sequence_{args.num_of_data}.jsonl"
|
|
|
|
# 将数据保存到文件
|
|
try:
|
|
with open(output_file, "w", encoding="utf-8") as f:
|
|
for game_data in game_data_list:
|
|
f.write(json.dumps(game_data.to_json(), ensure_ascii=False) + "\n")
|
|
print(f"游戏数据已保存到 {output_file}")
|
|
except Exception as e:
|
|
print(f"保存数据时出错: {e}") |