2024-10-16 20:18:55 +08:00
|
|
|
|
import json
|
2024-10-15 20:57:58 +08:00
|
|
|
|
import re
|
|
|
|
|
|
|
2024-10-16 20:18:55 +08:00
|
|
|
|
combined_data = {
|
|
|
|
|
|
"基础信息": {
|
|
|
|
|
|
"招标人/代理信息": {
|
|
|
|
|
|
"招标人": "广水市公安局",
|
|
|
|
|
|
"招标人联系方式": {
|
|
|
|
|
|
"名称": "广水市公安局",
|
|
|
|
|
|
"地址": "广水市应山办事处应十大道 189号",
|
|
|
|
|
|
"联系电话": "未知"
|
|
|
|
|
|
},
|
|
|
|
|
|
"招标代理机构": "湖北楚振捷工程项目管理有限公司",
|
|
|
|
|
|
"招标代理机构联系方式": {
|
|
|
|
|
|
"名称": "湖北楚振捷工程项目管理有限公司",
|
|
|
|
|
|
"地址": "广水市永阳一路 41号",
|
|
|
|
|
|
"联系方式": "0722-6256088"
|
|
|
|
|
|
},
|
|
|
|
|
|
"项目联系方式":{
|
|
|
|
|
|
"名称":"张三",
|
|
|
|
|
|
"联系电话":"11"
|
|
|
|
|
|
}
|
|
|
|
|
|
},
|
|
|
|
|
|
"项目信息": {
|
|
|
|
|
|
"项目名称": "广水市公安局视频会议高清化改造项目",
|
|
|
|
|
|
"项目编号": "HBCZJ-2021-CS12",
|
|
|
|
|
|
"项目概况": "广水市公安局视频会议高清化改造项目",
|
|
|
|
|
|
"项目基本情况": {
|
|
|
|
|
|
"项目编号": "HBCZJ-2021-CS12",
|
|
|
|
|
|
"采购计划备案号": "2021-04-000399",
|
|
|
|
|
|
"项目名称": "广水市公安局视频会议高清化改造项目",
|
|
|
|
|
|
"采购方式": "竞争性磋商",
|
|
|
|
|
|
"预算金额": "192万元",
|
|
|
|
|
|
"最高限价": "192万元",
|
|
|
|
|
|
"采购需求": {
|
|
|
|
|
|
"项目概况及内容": "广水市公安局视频会议高清化改造项目。",
|
|
|
|
|
|
"招标范围": "广水市公安局视频会议高清化改造项目，具体内容见磋商文件第三章。",
|
|
|
|
|
|
"项目地点": "具体以合同约定为准。"
|
|
|
|
|
|
},
|
|
|
|
|
|
"合同履行期限": "以合同签订为准。",
|
|
|
|
|
|
"本项目（是/否）接受联合体投标": "否",
|
|
|
|
|
|
"是否可采购进口产品": "否"
|
|
|
|
|
|
},
|
|
|
|
|
|
"招标控制价": "192万元",
|
|
|
|
|
|
"投标竞争下浮率": "未知",
|
|
|
|
|
|
"是否允许分包": "未知",
|
|
|
|
|
|
"是否接受联合体投标": "否"
|
|
|
|
|
|
},
|
|
|
|
|
|
"采购要求": {
|
|
|
|
|
|
"技术要求": "未提供",
|
|
|
|
|
|
"商务要求": "未提供",
|
|
|
|
|
|
"服务要求": "未提供",
|
|
|
|
|
|
"其他要求": "未提供"
|
|
|
|
|
|
},
|
|
|
|
|
|
"关键时间/内容": {
|
|
|
|
|
|
"投标文件递交截止日期": "2021年 6月 18日 15点 00分",
|
|
|
|
|
|
"开标时间":"111",
|
|
|
|
|
|
"开标地点":"哈哈",
|
|
|
|
|
|
"澄清招标文件的截止时间": "未知",
|
|
|
|
|
|
"投标有效期": "90日历天",
|
|
|
|
|
|
"信息公示媒介": "中国湖北政府采购网：http://www.ccgp-hubei.gov.cn, 中国广水网：http://www.zggsw.gov.cn/",
|
|
|
|
|
|
"投标文件递交地点": "广水市公共资源交易中心五楼 501号开标室"
|
|
|
|
|
|
|
|
|
|
|
|
},
|
|
|
|
|
|
"保证金相关": {
|
|
|
|
|
|
"质量保证金": "未知",
|
|
|
|
|
|
"履约保证金": "不提交",
|
|
|
|
|
|
"退还投标保证金": "/",
|
|
|
|
|
|
"投标保证金": {
|
|
|
|
|
|
"价格":"100，000，000.00 元"
|
|
|
|
|
|
}
|
|
|
|
|
|
},
|
|
|
|
|
|
"其他信息": {
|
|
|
|
|
|
"是否退还投标文件": "否",
|
|
|
|
|
|
"投标费用承担": "供应商应承担所有与准备和参加磋商有关的费用，不论磋商的结果如何，采购人和采购代理机构均无义务和责任承担这些费用。",
|
|
|
|
|
|
"招标代理服务费": {
|
|
|
|
|
|
"支付人": "成交供应商",
|
|
|
|
|
|
"支付标准": "根据国家发展与改革委员会办公厅发改办价格【2003】857 号文的规定，经协商由成交供应商按国家发展和改革委员发改价格【2011】534号文规定货物类取费标准向采购代理机构支付招标代理服务费（包含“招标代理费、评标会务费、评标费”），如本项目各包服务费不足叁仟元则供应商按叁仟元支付服务费。",
|
|
|
|
|
|
"支付方式": "成交服务费由成交供应商在领取成交通知书的同时向代理机构支付，可使用现金或电汇办理",
|
|
|
|
|
|
"支付时间": "领取成交通知书的同时"
|
|
|
|
|
|
},
|
|
|
|
|
|
"偏离": {
|
|
|
|
|
|
"偏离项要求": "供应商需在响应文件中提供《采购需求响应、偏离说明表/导读表》，具体格式见第六章响应文件格式中的49页。供应商需对采购需求中的各项技术参数和服务要求进行逐项响应，明确表明无偏离、正偏离或负偏离的情况。对于负偏离项，需详细说明偏离的具体内容及原因。",
|
|
|
|
|
|
"偏离项内容": "未知"
|
|
|
|
|
|
},
|
|
|
|
|
|
"投标预备会": "不召开",
|
|
|
|
|
|
"踏勘现场": "不组织"
|
|
|
|
|
|
}
|
|
|
|
|
|
},
|
|
|
|
|
|
"资格审查": { },
|
|
|
|
|
|
"商务评分": { },
|
|
|
|
|
|
"技术评分": { },
|
|
|
|
|
|
"zhes":"11111"
|
|
|
|
|
|
# 其他键值对
|
|
|
|
|
|
}
|
|
|
|
|
|
includes = ["基础信息", "资格审查", "商务评分", "技术评分", "无效标与废标项", "投标文件要求", "开评定标流程"]
|
|
|
|
|
|
|
|
|
|
|
|
def post_processing(combined_data, includes):
|
|
|
|
|
|
# 初始化结果字典，预设'其他'分类为空字典
|
|
|
|
|
|
processed_data = {"其他": {}}
|
|
|
|
|
|
|
|
|
|
|
|
# 初始化提取的信息字典
|
|
|
|
|
|
extracted_info = {}
|
|
|
|
|
|
|
|
|
|
|
|
# 定义一个辅助函数用于获取嵌套字典中的值
|
|
|
|
|
|
def get_nested(dic, keys, default=None):
|
|
|
|
|
|
for key in keys:
|
|
|
|
|
|
if isinstance(dic, dict):
|
|
|
|
|
|
dic = dic.get(key, default)
|
|
|
|
|
|
else:
|
|
|
|
|
|
return default
|
|
|
|
|
|
return dic
|
|
|
|
|
|
|
|
|
|
|
|
# 定义一个辅助函数用于递归查找包含特定子字符串的键
|
|
|
|
|
|
def find_keys_containing(dic, substring):
|
|
|
|
|
|
found_values = []
|
|
|
|
|
|
if isinstance(dic, dict):
|
|
|
|
|
|
for key, value in dic.items():
|
|
|
|
|
|
if substring in key:
|
|
|
|
|
|
found_values.append(value)
|
|
|
|
|
|
if isinstance(value, dict):
|
|
|
|
|
|
found_values.extend(find_keys_containing(value, substring))
|
|
|
|
|
|
elif isinstance(value, list):
|
|
|
|
|
|
for item in value:
|
|
|
|
|
|
if isinstance(item, dict):
|
|
|
|
|
|
found_values.extend(find_keys_containing(item, substring))
|
|
|
|
|
|
return found_values
|
|
|
|
|
|
|
|
|
|
|
|
# 定义一个辅助函数用于根据候选键列表提取值（部分匹配）
|
|
|
|
|
|
def extract_field(contact_info, candidate_keys):
|
|
|
|
|
|
for candidate in candidate_keys:
|
|
|
|
|
|
for key, value in contact_info.items():
|
|
|
|
|
|
if candidate in key and value not in ["未知", ""]:
|
|
|
|
|
|
return value
|
|
|
|
|
|
return ""
|
|
|
|
|
|
|
|
|
|
|
|
# 定义一个辅助函数用于提取 '投标保证金'
|
|
|
|
|
|
def extract_bid_bond(guarantee_info):
|
|
|
|
|
|
# 定义投标保证金的候选键
|
|
|
|
|
|
bid_bond_candidates = ["投标保证金", "磋商保证金"]
|
|
|
|
|
|
|
|
|
|
|
|
# 第一步：查找包含 "投标保证金" 或 "磋商保证金" 的键
|
|
|
|
|
|
for candidate in bid_bond_candidates:
|
|
|
|
|
|
for key, value in guarantee_info.items():
|
|
|
|
|
|
if candidate in key:
|
|
|
|
|
|
if isinstance(value, dict):
|
|
|
|
|
|
# 在嵌套字典中查找包含 "金额" 的键
|
|
|
|
|
|
for sub_key, sub_value in value.items():
|
|
|
|
|
|
if "金额" in sub_key and sub_value not in ["未知", ""]:
|
|
|
|
|
|
return sub_value
|
|
|
|
|
|
elif isinstance(value, str):
|
|
|
|
|
|
if "金额" in key and value not in ["未知", ""]:
|
|
|
|
|
|
return value
|
|
|
|
|
|
else:
|
|
|
|
|
|
# 如果 value 既不是 dict 也不是 str，忽略
|
2024-10-15 20:57:58 +08:00
|
|
|
|
continue
|
2024-10-16 20:18:55 +08:00
|
|
|
|
|
|
|
|
|
|
# 第二步：如果没有找到包含 "金额" 的键，尝试在所有键值中查找符合模式的值
|
|
|
|
|
|
amount_pattern = re.compile(r'(?:\d{1,3}(?:[,，]\d{3})*(?:\.\d+)?|\d+(?:\.\d+)?|[\u4e00-\u9fff]+(?:\.\d+)?)\s*(?:元|万元)')
|
|
|
|
|
|
for key, value in guarantee_info.items():
|
|
|
|
|
|
if isinstance(value, str):
|
|
|
|
|
|
match = amount_pattern.search(value)
|
|
|
|
|
|
if match:
|
|
|
|
|
|
return match.group()
|
|
|
|
|
|
elif isinstance(value, dict):
|
|
|
|
|
|
# 递归查找嵌套字典中的金额
|
|
|
|
|
|
found_amount = extract_bid_bond(value)
|
|
|
|
|
|
if found_amount:
|
|
|
|
|
|
return found_amount
|
|
|
|
|
|
# 如果都没有找到，则返回空字符串
|
|
|
|
|
|
return ""
|
|
|
|
|
|
|
|
|
|
|
|
# 如果 '基础信息' 在 includes 中，则进行字段提取
|
|
|
|
|
|
if "基础信息" in includes:
|
|
|
|
|
|
base_info = combined_data.get("基础信息", {})
|
|
|
|
|
|
|
|
|
|
|
|
# 定义所需字段的映射关系，暂时不包含'联系人'和'联系电话'以及'招标项目地点'
|
|
|
|
|
|
mapping = {
|
|
|
|
|
|
"招标项目名称": [["项目信息", "项目名称"], ["项目信息", "工程名称"]],
|
|
|
|
|
|
"招标项目编号": [["项目信息", "项目编号"], ["项目信息", "招标编号"]],
|
|
|
|
|
|
"开标时间": [["关键时间/内容", "开标时间"]],
|
|
|
|
|
|
"报名截止日期": [["关键时间/内容", "投标文件递交截止日期"]],
|
|
|
|
|
|
"招标项目预算": [["项目信息", "招标控制价"]],
|
|
|
|
|
|
"招标单位名称": [["招标人/代理信息", "招标人"]],
|
|
|
|
|
|
"招标公告地址": [["关键时间/内容", "信息公示媒介"], ["关键时间/内容", "评标结果公示媒介"]]
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
# 提取并映射字段
|
|
|
|
|
|
for new_key, paths in mapping.items():
|
|
|
|
|
|
value = None
|
|
|
|
|
|
for path in paths:
|
|
|
|
|
|
value = get_nested(base_info, path)
|
|
|
|
|
|
if value:
|
|
|
|
|
|
break
|
|
|
|
|
|
extracted_info[new_key] = value if value else ""
|
|
|
|
|
|
|
|
|
|
|
|
# 特殊处理 '招标项目地点'
|
|
|
|
|
|
# 在 '项目信息' 下查找包含 "地点" 的键
|
|
|
|
|
|
project_info = base_info.get("项目信息", {})
|
|
|
|
|
|
location_candidates = find_keys_containing(project_info, "地点")
|
|
|
|
|
|
if location_candidates:
|
|
|
|
|
|
# 选择第一个找到的地点
|
|
|
|
|
|
extracted_info["招标项目地点"] = location_candidates[0]
|
2024-10-15 20:57:58 +08:00
|
|
|
|
else:
|
2024-10-16 20:18:55 +08:00
|
|
|
|
extracted_info["招标项目地点"] = ""
|
|
|
|
|
|
|
|
|
|
|
|
# 特殊处理 '联系人' 和 '联系电话'
|
|
|
|
|
|
# 提取 '项目联系方式'
|
|
|
|
|
|
project_contact = get_nested(base_info, ["招标人/代理信息", "项目联系方式"], {})
|
|
|
|
|
|
|
|
|
|
|
|
# 提取 '招标人联系方式'
|
|
|
|
|
|
bidder_contact = get_nested(base_info, ["招标人/代理信息", "招标人联系方式"], {})
|
2024-10-15 20:57:58 +08:00
|
|
|
|
|
2024-10-16 20:18:55 +08:00
|
|
|
|
# 定义候选键列表，按优先级排序
|
|
|
|
|
|
name_candidates = ["名称", "联系人", "招标"]
|
|
|
|
|
|
phone_candidates = ["电话", "手机", "联系方式"]
|
|
|
|
|
|
|
|
|
|
|
|
# 提取 '联系人'
|
|
|
|
|
|
contact_names = [project_contact, bidder_contact]
|
|
|
|
|
|
contact_name = ""
|
|
|
|
|
|
for contact in contact_names:
|
|
|
|
|
|
extracted_name = extract_field(contact, name_candidates)
|
|
|
|
|
|
if extracted_name:
|
|
|
|
|
|
contact_name = extracted_name
|
|
|
|
|
|
break
|
|
|
|
|
|
extracted_info["联系人"] = contact_name
|
|
|
|
|
|
|
|
|
|
|
|
# 提取 '联系电话'
|
|
|
|
|
|
contact_phones = [project_contact, bidder_contact]
|
|
|
|
|
|
contact_phone = ""
|
|
|
|
|
|
for contact in contact_phones:
|
|
|
|
|
|
extracted_phone = extract_field(contact, phone_candidates)
|
|
|
|
|
|
if extracted_phone:
|
|
|
|
|
|
contact_phone = extracted_phone
|
|
|
|
|
|
break
|
|
|
|
|
|
extracted_info["联系电话"] = contact_phone
|
|
|
|
|
|
|
|
|
|
|
|
# 特殊处理 '投标保证金'
|
|
|
|
|
|
# 提取 '保证金相关'
|
|
|
|
|
|
guarantee_info = get_nested(base_info, ["保证金相关"], {})
|
|
|
|
|
|
extracted_info["投标保证金"] = extract_bid_bond(guarantee_info)
|
|
|
|
|
|
|
|
|
|
|
|
# 遍历原始字典的每一个键值对
|
|
|
|
|
|
for key, value in combined_data.items():
|
|
|
|
|
|
if key in includes:
|
|
|
|
|
|
if key == "基础信息":
|
|
|
|
|
|
# 已经处理 '基础信息'，保留在处理后的数据中
|
|
|
|
|
|
processed_data[key] = value
|
|
|
|
|
|
else:
|
|
|
|
|
|
# 直接保留包含在 includes 列表中的键值对
|
|
|
|
|
|
processed_data[key] = value
|
|
|
|
|
|
else:
|
|
|
|
|
|
# 将不在 includes 列表中的键值对加入到 '其他' 分类中
|
|
|
|
|
|
processed_data["其他"][key] = value
|
2024-10-15 20:57:58 +08:00
|
|
|
|
|
2024-10-16 20:18:55 +08:00
|
|
|
|
# 如果 '其他' 分类没有任何内容，可以选择删除这个键
|
|
|
|
|
|
if not processed_data["其他"]:
|
|
|
|
|
|
del processed_data["其他"]
|
2024-10-15 20:57:58 +08:00
|
|
|
|
|
2024-10-16 20:18:55 +08:00
|
|
|
|
return processed_data, extracted_info
|
2024-10-10 21:03:02 +08:00
|
|
|
|
|
|
|
|
|
|
|
2024-10-16 20:18:55 +08:00
|
|
|
|
res1,res2=post_processing(combined_data,includes)
|
|
|
|
|
|
print(json.dumps(res2,ensure_ascii=False,indent=4))
|
