278 lines
20 KiB
Python
278 lines
20 KiB
Python
import openpyxl # 添加缺少的导入
|
||
import json
|
||
import os
|
||
import traceback
|
||
|
||
from Config.Config import EXCEL_PATH
|
||
from Util.AreaUtil import query_area_info
|
||
|
||
def main():
|
||
"""主函数：提取学校面积数据"""
|
||
# 创建数据保存目录
|
||
DATA_DIR = os.path.join(os.path.dirname(os.path.dirname(__file__)), 'Data')
|
||
os.makedirs(DATA_DIR, exist_ok=True)
|
||
JSON_PATH = os.path.join(DATA_DIR, 'SchoolArea.json') # 修改为学校面积的JSON路径
|
||
|
||
file_name = EXCEL_PATH
|
||
area_data = []
|
||
name_conversion_errors = [] # 记录转换失败的名称
|
||
conversion_records = [] # 定义转换记录变量
|
||
|
||
workbook = None # 初始化workbook变量，确保在finally块中可以访问
|
||
|
||
try:
|
||
# 加载工作簿并选择学校面积、教学辅助房面积Sheet
|
||
workbook = openpyxl.load_workbook(file_name, data_only=True)
|
||
if '学校面积、教学辅助房面积' not in workbook.sheetnames:
|
||
print("❌ 错误：未找到'学校面积、教学辅助房面积'Sheet")
|
||
return
|
||
sheet = workbook['学校面积、教学辅助房面积']
|
||
|
||
# 定义数据列范围与英文属性映射
|
||
data_columns = {
|
||
# 学前教育（每年份8列：学校占地面积[总计/城区/镇区/乡村]，教学及辅助用房[总计/城区/镇区/乡村]）
|
||
'preschool_area': {
|
||
'years': [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024],
|
||
'columns': [
|
||
{'year': 2015, 'school_total': 'D', 'school_urban': 'E', 'school_town': 'F', 'school_rural': 'G', 'teaching_total': 'H', 'teaching_urban': 'I', 'teaching_town': 'J', 'teaching_rural': 'K'},
|
||
{'year': 2016, 'school_total': 'L', 'school_urban': 'M', 'school_town': 'N', 'school_rural': 'O', 'teaching_total': 'P', 'teaching_urban': 'Q', 'teaching_town': 'R', 'teaching_rural': 'S'},
|
||
{'year': 2017, 'school_total': 'T', 'school_urban': 'U', 'school_town': 'V', 'school_rural': 'W', 'teaching_total': 'X', 'teaching_urban': 'Y', 'teaching_town': 'Z', 'teaching_rural': 'AA'},
|
||
{'year': 2018, 'school_total': 'AB', 'school_urban': 'AC', 'school_town': 'AD', 'school_rural': 'AE', 'teaching_total': 'AF', 'teaching_urban': 'AG', 'teaching_town': 'AH', 'teaching_rural': 'AI'},
|
||
{'year': 2019, 'school_total': 'AJ', 'school_urban': 'AK', 'school_town': 'AL', 'school_rural': 'AM', 'teaching_total': 'AN', 'teaching_urban': 'AO', 'teaching_town': 'AP', 'teaching_rural': 'AQ'},
|
||
{'year': 2020, 'school_total': 'AR', 'school_urban': 'AS', 'school_town': 'AT', 'school_rural': 'AU', 'teaching_total': 'AV', 'teaching_urban': 'AW', 'teaching_town': 'AX', 'teaching_rural': 'AY'},
|
||
{'year': 2021, 'school_total': 'AZ', 'school_urban': 'BA', 'school_town': 'BB', 'school_rural': 'BC', 'teaching_total': 'BD', 'teaching_urban': 'BE', 'teaching_town': 'BF', 'teaching_rural': 'BG'},
|
||
{'year': 2022, 'school_total': 'BH', 'school_urban': 'BI', 'school_town': 'BJ', 'school_rural': 'BK', 'teaching_total': 'BL', 'teaching_urban': 'BM', 'teaching_town': 'BN', 'teaching_rural': 'BO'},
|
||
{'year': 2023, 'school_total': 'BP', 'school_urban': 'BQ', 'school_town': 'BR', 'school_rural': 'BS', 'teaching_total': 'BT', 'teaching_urban': 'BU', 'teaching_town': 'BV', 'teaching_rural': 'BW'},
|
||
{'year': 2024, 'school_total': 'BX', 'school_urban': 'BY', 'school_town': 'BZ', 'school_rural': 'CA', 'teaching_total': 'CB', 'teaching_urban': 'CC', 'teaching_town': 'CD', 'teaching_rural': 'CE'}
|
||
]
|
||
},
|
||
|
||
# 小学教育
|
||
'primary_area': {
|
||
'years': [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024],
|
||
'columns': [
|
||
{'year': 2015, 'school_total': 'CF', 'school_urban': 'CG', 'school_town': 'CH', 'school_rural': 'CI', 'teaching_total': 'CJ', 'teaching_urban': 'CK', 'teaching_town': 'CL', 'teaching_rural': 'CM'},
|
||
{'year': 2016, 'school_total': 'CN', 'school_urban': 'CO', 'school_town': 'CP', 'school_rural': 'CQ', 'teaching_total': 'CR', 'teaching_urban': 'CS', 'teaching_town': 'CT', 'teaching_rural': 'CU'},
|
||
{'year': 2017, 'school_total': 'CV', 'school_urban': 'CW', 'school_town': 'CX', 'school_rural': 'CY', 'teaching_total': 'CZ', 'teaching_urban': 'DA', 'teaching_town': 'DB', 'teaching_rural': 'DC'},
|
||
{'year': 2018, 'school_total': 'DD', 'school_urban': 'DE', 'school_town': 'DF', 'school_rural': 'DG', 'teaching_total': 'DH', 'teaching_urban': 'DI', 'teaching_town': 'DJ', 'teaching_rural': 'DK'},
|
||
{'year': 2019, 'school_total': 'DL', 'school_urban': 'DM', 'school_town': 'DN', 'school_rural': 'DO', 'teaching_total': 'DP', 'teaching_urban': 'DQ', 'teaching_town': 'DR', 'teaching_rural': 'DS'},
|
||
{'year': 2020, 'school_total': 'DT', 'school_urban': 'DU', 'school_town': 'DV', 'school_rural': 'DW', 'teaching_total': 'DX', 'teaching_urban': 'DY', 'teaching_town': 'DZ', 'teaching_rural': 'EA'},
|
||
{'year': 2021, 'school_total': 'EB', 'school_urban': 'EC', 'school_town': 'ED', 'school_rural': 'EE', 'teaching_total': 'EF', 'teaching_urban': 'EG', 'teaching_town': 'EH', 'teaching_rural': 'EI'},
|
||
{'year': 2022, 'school_total': 'EJ', 'school_urban': 'EK', 'school_town': 'EL', 'school_rural': 'EM', 'teaching_total': 'EN', 'teaching_urban': 'EO', 'teaching_town': 'EP', 'teaching_rural': 'EQ'},
|
||
{'year': 2023, 'school_total': 'ER', 'school_urban': 'ES', 'school_town': 'ET', 'school_rural': 'EU', 'teaching_total': 'EV', 'teaching_urban': 'EW', 'teaching_town': 'EX', 'teaching_rural': 'EY'},
|
||
{'year': 2024, 'school_total': 'EZ', 'school_urban': 'FA', 'school_town': 'FB', 'school_rural': 'FC', 'teaching_total': 'FD', 'teaching_urban': 'FE', 'teaching_town': 'FF', 'teaching_rural': 'FG'}
|
||
]
|
||
},
|
||
|
||
# 初中教育
|
||
'junior_high_area': {
|
||
'years': [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024],
|
||
'columns': [
|
||
{'year': 2015, 'school_total': 'FH', 'school_urban': 'FI', 'school_town': 'FJ', 'school_rural': 'FK', 'teaching_total': 'FL', 'teaching_urban': 'FM', 'teaching_town': 'FN', 'teaching_rural': 'FO'},
|
||
{'year': 2016, 'school_total': 'FP', 'school_urban': 'FQ', 'school_town': 'FR', 'school_rural': 'FS', 'teaching_total': 'FT', 'teaching_urban': 'FU', 'teaching_town': 'FV', 'teaching_rural': 'FW'},
|
||
{'year': 2017, 'school_total': 'FX', 'school_urban': 'FY', 'school_town': 'FZ', 'school_rural': 'GA', 'teaching_total': 'GB', 'teaching_urban': 'GC', 'teaching_town': 'GD', 'teaching_rural': 'GE'},
|
||
{'year': 2018, 'school_total': 'GF', 'school_urban': 'GG', 'school_town': 'GH', 'school_rural': 'GI', 'teaching_total': 'GJ', 'teaching_urban': 'GK', 'teaching_town': 'GL', 'teaching_rural': 'GM'},
|
||
{'year': 2019, 'school_total': 'GN', 'school_urban': 'GO', 'school_town': 'GP', 'school_rural': 'GQ', 'teaching_total': 'GR', 'teaching_urban': 'GS', 'teaching_town': 'GT', 'teaching_rural': 'GU'},
|
||
{'year': 2020, 'school_total': 'GV', 'school_urban': 'GW', 'school_town': 'GX', 'school_rural': 'GY', 'teaching_total': 'GZ', 'teaching_urban': 'HA', 'teaching_town': 'HB', 'teaching_rural': 'HC'},
|
||
{'year': 2021, 'school_total': 'HD', 'school_urban': 'HE', 'school_town': 'HF', 'school_rural': 'HG', 'teaching_total': 'HH', 'teaching_urban': 'HI', 'teaching_town': 'HJ', 'teaching_rural': 'HK'},
|
||
{'year': 2022, 'school_total': 'HL', 'school_urban': 'HM', 'school_town': 'HN', 'school_rural': 'HO', 'teaching_total': 'HP', 'teaching_urban': 'HQ', 'teaching_town': 'HR', 'teaching_rural': 'HS'},
|
||
{'year': 2023, 'school_total': 'HT', 'school_urban': 'HU', 'school_town': 'HV', 'school_rural': 'HW', 'teaching_total': 'HX', 'teaching_urban': 'HY', 'teaching_town': 'HZ', 'teaching_rural': 'IA'},
|
||
{'year': 2024, 'school_total': 'IB', 'school_urban': 'IC', 'school_town': 'ID', 'school_rural': 'IE', 'teaching_total': 'IF', 'teaching_urban': 'IG', 'teaching_town': 'IH', 'teaching_rural': 'II'}
|
||
]
|
||
},
|
||
|
||
# 高中教育
|
||
'senior_high_area': {
|
||
'years': [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024],
|
||
'columns': [
|
||
{'year': 2015, 'school_total': 'IJ', 'school_urban': 'IK', 'school_town': 'IL', 'school_rural': 'IM', 'teaching_total': 'IN', 'teaching_urban': 'IO', 'teaching_town': 'IP', 'teaching_rural': 'IQ'},
|
||
{'year': 2016, 'school_total': 'IR', 'school_urban': 'IS', 'school_town': 'IT', 'school_rural': 'IU', 'teaching_total': 'IV', 'teaching_urban': 'IW', 'teaching_town': 'IX', 'teaching_rural': 'IY'},
|
||
{'year': 2017, 'school_total': 'IZ', 'school_urban': 'JA', 'school_town': 'JB', 'school_rural': 'JC', 'teaching_total': 'JD', 'teaching_urban': 'JE', 'teaching_town': 'JF', 'teaching_rural': 'JG'},
|
||
{'year': 2018, 'school_total': 'JH', 'school_urban': 'JI', 'school_town': 'JJ', 'school_rural': 'JK', 'teaching_total': 'JL', 'teaching_urban': 'JM', 'teaching_town': 'JN', 'teaching_rural': 'JO'},
|
||
{'year': 2019, 'school_total': 'JP', 'school_urban': 'JQ', 'school_town': 'JR', 'school_rural': 'JS', 'teaching_total': 'JT', 'teaching_urban': 'JU', 'teaching_town': 'JV', 'teaching_rural': 'JW'},
|
||
{'year': 2020, 'school_total': 'JX', 'school_urban': 'JY', 'school_town': 'JZ', 'school_rural': 'KA', 'teaching_total': 'KB', 'teaching_urban': 'KC', 'teaching_town': 'KD', 'teaching_rural': 'KE'},
|
||
{'year': 2021, 'school_total': 'KF', 'school_urban': 'KG', 'school_town': 'KH', 'school_rural': 'KI', 'teaching_total': 'KJ', 'teaching_urban': 'KK', 'teaching_town': 'KL', 'teaching_rural': 'KM'},
|
||
{'year': 2022, 'school_total': 'KN', 'school_urban': 'KO', 'school_town': 'KP', 'school_rural': 'KQ', 'teaching_total': 'KR', 'teaching_urban': 'KS', 'teaching_town': 'KT', 'teaching_rural': 'KU'},
|
||
{'year': 2023, 'school_total': 'KV', 'school_urban': 'KW', 'school_town': 'KX', 'school_rural': 'KY', 'teaching_total': 'KZ', 'teaching_urban': 'LA', 'teaching_town': 'LB', 'teaching_rural': 'LC'},
|
||
{'year': 2024, 'school_total': 'LD', 'school_urban': 'LE', 'school_town': 'LF', 'school_rural': 'LG', 'teaching_total': 'LH', 'teaching_urban': 'LI', 'teaching_town': 'LJ', 'teaching_rural': 'LK'}
|
||
]
|
||
},
|
||
|
||
# 中职教育（特殊：每年2列）
|
||
'vocational_area': {
|
||
'years': [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024],
|
||
'columns': [
|
||
{'year': 2015, 'school_total': 'LL', 'teaching_total': 'LM'},
|
||
{'year': 2016, 'school_total': 'LN', 'teaching_total': 'LO'},
|
||
{'year': 2017, 'school_total': 'LP', 'teaching_total': 'LQ'},
|
||
{'year': 2018, 'school_total': 'LR', 'teaching_total': 'LS'},
|
||
{'year': 2019, 'school_total': 'LT', 'teaching_total': 'LU'},
|
||
{'year': 2020, 'school_total': 'LV', 'teaching_total': 'LW'},
|
||
{'year': 2021, 'school_total': 'LX', 'teaching_total': 'LY'},
|
||
{'year': 2022, 'school_total': 'LZ', 'teaching_total': 'MA'},
|
||
{'year': 2023, 'school_total': 'MB', 'teaching_total': 'MC'},
|
||
{'year': 2024, 'school_total': 'MD', 'teaching_total': 'ME'}
|
||
]
|
||
}
|
||
}
|
||
|
||
# 遍历数据行（跳过前4行表头，从第5行开始）
|
||
for row_num, row in enumerate(sheet.iter_rows(min_row=5, values_only=True), start=5):
|
||
# 区域名称从B列获取（索引1）
|
||
raw_name = row[1] if (len(row) > 1 and row[1] is not None) else '未知地区'
|
||
if not raw_name: # 跳过空行
|
||
continue
|
||
|
||
# 区域名称转换
|
||
str_raw_name = str(raw_name).strip() if raw_name is not None else '未知地区'
|
||
area_info = query_area_info(str_raw_name)
|
||
if area_info and isinstance(area_info, dict) and 'full_name' in area_info and 'area_code' in area_info:
|
||
area_name = area_info['full_name']
|
||
area_code = area_info['area_code']
|
||
if raw_name != area_name:
|
||
conversion_records.append({
|
||
'row': row_num,
|
||
'raw_name': raw_name,
|
||
'converted_name': area_name
|
||
})
|
||
else:
|
||
area_name = raw_name
|
||
area_code = 'unknown'
|
||
name_conversion_errors.append(f"行 {row_num}: '{raw_name}'")
|
||
|
||
# 循环中创建区域数据字典（重命名变量）
|
||
current_area_data = {
|
||
'area_name': area_name,
|
||
'area_code': area_code,
|
||
'raw_name': raw_name
|
||
}
|
||
|
||
# 提取各教育阶段面积数据
|
||
for stage, config in data_columns.items():
|
||
stage_data = {}
|
||
for year_config in config['columns']:
|
||
year = year_config['year']
|
||
year_data = {}
|
||
# 处理学校占地面积数据
|
||
school_cols = ['school_total', 'school_urban', 'school_town', 'school_rural']
|
||
has_school_categories = all(col in year_config for col in school_cols)
|
||
|
||
if has_school_categories:
|
||
# 处理分类学校面积数据
|
||
for col in school_cols:
|
||
col_name = year_config[col]
|
||
col_idx = openpyxl.utils.column_index_from_string(col_name) - 1
|
||
if col_idx < len(row):
|
||
value = row[col_idx]
|
||
# 数据清洗与转换
|
||
if value is None:
|
||
year_data[col] = 0
|
||
else:
|
||
str_value = str(value).strip()
|
||
if str_value in ['', '####']:
|
||
year_data[col] = 0
|
||
else:
|
||
try:
|
||
year_data[col] = float(str_value)
|
||
except ValueError:
|
||
year_data[col] = 0
|
||
else:
|
||
# 处理中职学校总面积
|
||
col_name = year_config['school_total']
|
||
col_idx = openpyxl.utils.column_index_from_string(col_name) - 1
|
||
if col_idx < len(row):
|
||
value = row[col_idx]
|
||
if value is None:
|
||
year_data['school_total'] = 0
|
||
else:
|
||
str_value = str(value).strip()
|
||
if str_value in ['', '####']:
|
||
year_data['school_total'] = 0
|
||
else:
|
||
try:
|
||
year_data['school_total'] = float(str_value)
|
||
except ValueError:
|
||
year_data['school_total'] = 0
|
||
|
||
# 处理教学辅助用房面积数据
|
||
teaching_cols = ['teaching_total', 'teaching_urban', 'teaching_town', 'teaching_rural']
|
||
has_teaching_categories = all(col in year_config for col in teaching_cols)
|
||
|
||
if has_teaching_categories:
|
||
# 处理分类教学辅助用房面积数据
|
||
for col in teaching_cols:
|
||
col_name = year_config[col]
|
||
col_idx = openpyxl.utils.column_index_from_string(col_name) - 1
|
||
if col_idx < len(row):
|
||
value = row[col_idx]
|
||
# 数据清洗与转换
|
||
if value is None:
|
||
year_data[col] = 0
|
||
else:
|
||
str_value = str(value).strip()
|
||
if str_value in ['', '####']:
|
||
year_data[col] = 0
|
||
else:
|
||
try:
|
||
year_data[col] = float(str_value)
|
||
except ValueError:
|
||
year_data[col] = 0
|
||
else:
|
||
# 处理中职教学辅助用房总面积
|
||
col_name = year_config['teaching_total']
|
||
col_idx = openpyxl.utils.column_index_from_string(col_name) - 1
|
||
if col_idx < len(row):
|
||
value = row[col_idx]
|
||
if value is None:
|
||
year_data['teaching_total'] = 0
|
||
else:
|
||
str_value = str(value).strip()
|
||
if str_value in ['', '####']:
|
||
year_data['teaching_total'] = 0
|
||
else:
|
||
try:
|
||
year_data['teaching_total'] = float(str_value)
|
||
except ValueError:
|
||
year_data['teaching_total'] = 0
|
||
|
||
stage_data[str(year)] = year_data
|
||
# 修复：将阶段数据添加到当前区域字典，而非区域列表
|
||
current_area_data[stage] = stage_data
|
||
|
||
# 将当前区域数据添加到列表
|
||
area_data.append(current_area_data)
|
||
|
||
# 保存JSON文件
|
||
with open(JSON_PATH, 'w', encoding='utf-8') as f:
|
||
json.dump(area_data, f, ensure_ascii=False, indent=2)
|
||
|
||
# 输出统计信息
|
||
print(f"✅ 学校面积数据提取完成，已保存至：{JSON_PATH}")
|
||
print(f"📊 共处理 {len(area_data)} 条地区数据")
|
||
print("\n=== 名称转换记录 ===")
|
||
if conversion_records:
|
||
for record in conversion_records:
|
||
print(f"🔄 行 {record['row']}: {record['raw_name']} → {record['converted_name']}")
|
||
print(f"📊 共检测到 {len(conversion_records)} 项名称转换")
|
||
else:
|
||
print("📝 不存在名称转换的情况")
|
||
if name_conversion_errors:
|
||
print(f"⚠️ 发现 {len(name_conversion_errors)} 个区域名称转换失败：")
|
||
for error in name_conversion_errors:
|
||
print(f" - {error}")
|
||
|
||
except FileNotFoundError:
|
||
print(f"🔴 错误：Excel文件 '{file_name}' 不存在")
|
||
except Exception as e:
|
||
print(f"🔴 处理数据时发生错误：{str(e)}{traceback.format_exc()}")
|
||
finally:
|
||
try:
|
||
if workbook:
|
||
workbook.close()
|
||
except:
|
||
pass
|
||
|
||
if __name__ == "__main__":
|
||
main()
|
||
|