159 lines
5.2 KiB
Python
159 lines
5.2 KiB
Python
#!/usr/bin/env python
|
||
# -*- coding: utf-8 -*-
|
||
|
||
"""
|
||
测试TopicParser组件
|
||
|
||
该脚本演示如何独立使用topic_parser模块解析JSON格式的选题数据。
|
||
"""
|
||
|
||
import os
|
||
import sys
|
||
import json
|
||
import time
|
||
from pathlib import Path
|
||
|
||
# 将项目根目录添加到PATH
|
||
project_root = str(Path(__file__).parent.parent.absolute())
|
||
if project_root not in sys.path:
|
||
sys.path.append(project_root)
|
||
|
||
from core.topic_parser import TopicParser
|
||
|
||
def get_test_json():
|
||
"""返回测试用的选题JSON数据"""
|
||
return '''
|
||
[
|
||
{
|
||
"index": 1,
|
||
"date": "2024-05-30",
|
||
"logic": "这是一个著名的景点，有丰富的历史文化底蕴，适合文化和历史爱好者。",
|
||
"object": "泰宁古城",
|
||
"product": "文化体验套票.txt",
|
||
"product_logic": "泰宁古城有多处明清古建筑，游客可以通过体验套票更全面了解当地历史文化。",
|
||
"style": "人文历史.txt",
|
||
"style_logic": "古城拥有800多年历史，古建筑保存完好，适合人文历史风格。",
|
||
"target_audience": "家庭游客.txt",
|
||
"target_audience_logic": "古城内有适合家庭参观的展览和互动活动，老人孩子都能找到感兴趣的内容。"
|
||
},
|
||
{
|
||
"index": 2,
|
||
"date": "2024-05-30",
|
||
"logic": "这是一个自然景观与历史文化相结合的景点，适合各类游客。",
|
||
"object": "甘露岩寺",
|
||
"product": "亲子研学.txt",
|
||
"product_logic": "甘露岩寺建筑结构特殊，可作为建筑研学的典型案例。",
|
||
"style": "摄影指南.txt",
|
||
"style_logic": "寺庙依山而建，视角独特，适合摄影爱好者拍摄。",
|
||
"target_audience": "亲子游.txt",
|
||
"target_audience_logic": "有趣的建筑结构和历史故事适合亲子游客探索学习。"
|
||
}
|
||
]
|
||
'''
|
||
|
||
def get_invalid_json():
|
||
"""返回格式不正确的JSON数据进行测试"""
|
||
return '''
|
||
{
|
||
"topics": [
|
||
{"name": "泰宁古城", "description": "这是一个测试"},
|
||
{"name": "甘露岩寺", "description": "这是另一个测试"}
|
||
]
|
||
}
|
||
'''
|
||
|
||
def get_malformed_json():
|
||
"""返回格式错误的JSON字符串"""
|
||
return '''
|
||
[
|
||
{
|
||
"index": 1,
|
||
"date": "2024-05-30",
|
||
"object": "泰宁古城"
|
||
},
|
||
{
|
||
"index": 2
|
||
"date": "2024-05-30", // 缺少逗号
|
||
"object": "甘露岩寺"
|
||
}
|
||
]
|
||
'''
|
||
|
||
def main():
|
||
"""主函数"""
|
||
print("=== 测试 TopicParser 组件 ===")
|
||
|
||
# 测试1: 解析有效的JSON
|
||
print("\n测试1: 解析有效的JSON数据")
|
||
start_time = time.time()
|
||
|
||
valid_json = get_test_json()
|
||
print(f"输入JSON数据 ({len(valid_json)} 字节):\n{valid_json[:200]}...")
|
||
|
||
try:
|
||
topics = TopicParser.parse_topics(valid_json)
|
||
elapsed = time.time() - start_time
|
||
|
||
if topics:
|
||
print(f"成功解析! 解析到 {len(topics)} 个选题，耗时: {elapsed:.4f}秒")
|
||
print("\n解析结果:")
|
||
for i, topic in enumerate(topics):
|
||
print(f"\n选题 {i+1}:")
|
||
print(f" 对象: {topic.get('object', 'N/A')}")
|
||
print(f" 样式: {topic.get('style', 'N/A')}")
|
||
print(f" 目标受众: {topic.get('target_audience', 'N/A')}")
|
||
else:
|
||
print("解析失败: 未返回选题列表")
|
||
except Exception as e:
|
||
print(f"解析出错: {e}")
|
||
|
||
# 测试2: 解析无效结构的JSON
|
||
print("\n测试2: 解析结构无效的JSON数据")
|
||
invalid_json = get_invalid_json()
|
||
print(f"输入无效结构JSON:\n{invalid_json}")
|
||
|
||
try:
|
||
topics = TopicParser.parse_topics(invalid_json)
|
||
if topics:
|
||
print(f"注意: 成功解析了 {len(topics)} 个选题（预期应该失败）")
|
||
else:
|
||
print("测试通过: 正确地未能解析无效结构")
|
||
except Exception as e:
|
||
print(f"捕获到异常 (预期行为): {e}")
|
||
|
||
# 测试3: 格式错误的JSON
|
||
print("\n测试3: 解析格式错误的JSON")
|
||
malformed_json = get_malformed_json()
|
||
print(f"输入格式错误JSON:\n{malformed_json}")
|
||
|
||
try:
|
||
topics = TopicParser.parse_topics(malformed_json)
|
||
if topics:
|
||
print(f"注意: 成功解析了 {len(topics)} 个选题（预期应该失败）")
|
||
else:
|
||
print("测试通过: 正确地未能解析格式错误JSON")
|
||
except Exception as e:
|
||
print(f"捕获到异常 (预期行为): {e}")
|
||
|
||
# 测试4: 直接解析字典对象
|
||
print("\n测试4: 直接解析Python字典对象")
|
||
|
||
try:
|
||
# 将JSON字符串转换为Python对象
|
||
topic_list = json.loads(get_test_json())
|
||
|
||
# 测试直接传递Python对象
|
||
topics = TopicParser.parse_topics(topic_list)
|
||
|
||
if topics:
|
||
print(f"成功解析! 解析到 {len(topics)} 个选题")
|
||
print("测试通过: 可以接受Python对象作为输入")
|
||
else:
|
||
print("解析失败: 未返回选题列表")
|
||
except Exception as e:
|
||
print(f"解析出错: {e}")
|
||
|
||
print("\n=== 测试完成 ===")
|
||
|
||
if __name__ == "__main__":
|
||
main() |