252 lines
6.7 KiB
Python
252 lines
6.7 KiB
Python
#!/usr/bin/env python3
|
||
"""
|
||
码头作业日志管理工具
|
||
从 Confluence 获取交接班日志并保存到数据库
|
||
"""
|
||
import argparse
|
||
import sys
|
||
import os
|
||
from datetime import datetime
|
||
|
||
from src.confluence import ConfluenceClient
|
||
from src.extractor import HTMLTextExtractor
|
||
from src.parser import HandoverLogParser
|
||
from src.database import DailyLogsDatabase
|
||
from src.report import DailyReportGenerator
|
||
|
||
# 加载环境变量
|
||
from dotenv import load_dotenv
|
||
load_dotenv()
|
||
|
||
# 配置（从环境变量读取）
|
||
CONF_BASE_URL = os.getenv('CONFLUENCE_BASE_URL')
|
||
CONF_TOKEN = os.getenv('CONFLUENCE_TOKEN')
|
||
CONF_CONTENT_ID = os.getenv('CONFLUENCE_CONTENT_ID')
|
||
|
||
DEBUG_DIR = 'debug'
|
||
|
||
|
||
def ensure_debug_dir():
|
||
"""确保debug目录存在"""
|
||
if not os.path.exists(DEBUG_DIR):
|
||
os.makedirs(DEBUG_DIR)
|
||
|
||
|
||
def get_timestamp():
|
||
"""获取时间戳用于文件名"""
|
||
return datetime.now().strftime('%Y%m%d_%H%M%S')
|
||
|
||
|
||
def fetch_html():
|
||
"""获取HTML内容"""
|
||
if not CONF_BASE_URL or not CONF_TOKEN or not CONF_CONTENT_ID:
|
||
print('错误：未配置 Confluence 信息，请检查 .env 文件')
|
||
sys.exit(1)
|
||
|
||
print('正在从 Confluence 获取 HTML 内容...')
|
||
client = ConfluenceClient(CONF_BASE_URL, CONF_TOKEN)
|
||
html = client.get_html(CONF_CONTENT_ID)
|
||
if not html:
|
||
print('错误：未获取到 HTML 内容')
|
||
sys.exit(1)
|
||
print(f'获取成功，共 {len(html)} 字符')
|
||
return html
|
||
|
||
|
||
def extract_text(html):
|
||
"""提取布局文本"""
|
||
print('正在提取布局文本...')
|
||
extractor = HTMLTextExtractor()
|
||
layout_text = extractor.extract(html)
|
||
print(f'提取完成，共 {len(layout_text)} 字符')
|
||
return layout_text
|
||
|
||
|
||
def save_debug_file(content, suffix=''):
|
||
"""保存调试文件到debug目录"""
|
||
ensure_debug_dir()
|
||
filename = f'layout_output{suffix}.txt' if suffix else 'layout_output.txt'
|
||
filepath = os.path.join(DEBUG_DIR, filename)
|
||
with open(filepath, 'w', encoding='utf-8') as f:
|
||
f.write(content)
|
||
print(f'已保存到 {filepath}')
|
||
return filepath
|
||
|
||
|
||
def parse_logs(text):
|
||
"""解析日志数据"""
|
||
print('正在解析日志数据...')
|
||
parser = HandoverLogParser()
|
||
logs = parser.parse(text)
|
||
print(f'解析到 {len(logs)} 条记录')
|
||
return logs
|
||
|
||
|
||
def save_to_db(logs):
|
||
"""保存到数据库"""
|
||
if not logs:
|
||
print('没有记录可保存')
|
||
return 0
|
||
|
||
db = DailyLogsDatabase()
|
||
count = db.insert_many([log.to_dict() for log in logs])
|
||
print(f'已保存 {count} 条记录到数据库')
|
||
|
||
stats = db.get_stats()
|
||
print(f'\n数据库统计:')
|
||
print(f' 总记录: {stats["total"]}')
|
||
print(f' 船次: {len(stats["ships"])}')
|
||
print(f' 日期范围: {stats["date_range"]["start"]} ~ {stats["date_range"]["end"]}')
|
||
|
||
db.close()
|
||
return count
|
||
|
||
|
||
def add_unaccounted(year_month, teu, note=''):
|
||
"""添加未统计数据"""
|
||
db = DailyLogsDatabase()
|
||
result = db.insert_unaccounted(year_month, teu, note)
|
||
if result:
|
||
print(f'已添加 {year_month} 月未统计数据: {teu}TEU')
|
||
else:
|
||
print('添加失败')
|
||
db.close()
|
||
|
||
|
||
def show_stats(date):
|
||
"""显示指定日期的统计"""
|
||
g = DailyReportGenerator()
|
||
g.print_report(date)
|
||
g.close()
|
||
|
||
|
||
def run_fetch():
|
||
"""执行：获取HTML并提取文本"""
|
||
html = fetch_html()
|
||
text = extract_text(html)
|
||
save_debug_file(text)
|
||
return text
|
||
|
||
|
||
def run_fetch_and_save():
|
||
"""执行：获取、提取、解析、保存到数据库"""
|
||
text = run_fetch()
|
||
logs = parse_logs(text)
|
||
save_to_db(logs)
|
||
|
||
|
||
def run_fetch_save_debug():
|
||
"""执行：获取、提取、保存到debug目录"""
|
||
html = fetch_html()
|
||
text = extract_text(html)
|
||
suffix = f'_{get_timestamp()}'
|
||
save_debug_file(text, suffix)
|
||
return text
|
||
|
||
|
||
def run_report(date=None):
|
||
"""执行：生成日报"""
|
||
if not date:
|
||
date = datetime.now().strftime('%Y-%m-%d')
|
||
show_stats(date)
|
||
|
||
|
||
def run_parser_test():
|
||
"""执行：解析测试"""
|
||
ensure_debug_file_path = os.path.join(DEBUG_DIR, 'layout_output.txt')
|
||
if os.path.exists('layout_output.txt'):
|
||
filepath = 'layout_output.txt'
|
||
elif os.path.exists(ensure_debug_file_path):
|
||
filepath = ensure_debug_file_path
|
||
else:
|
||
print('未找到 layout_output.txt 文件')
|
||
return
|
||
|
||
print(f'使用文件: {filepath}')
|
||
with open(filepath, 'r', encoding='utf-8') as f:
|
||
text = f.read()
|
||
|
||
parser = HandoverLogParser()
|
||
logs = parser.parse(text)
|
||
print(f'解析到 {len(logs)} 条记录')
|
||
for log in logs[:5]:
|
||
print(f' {log.date} {log.shift} {log.ship_name}: {log.teu}TEU')
|
||
|
||
|
||
# 功能映射
|
||
FUNCTIONS = {
|
||
'fetch': run_fetch,
|
||
'fetch-save': run_fetch_and_save,
|
||
'fetch-debug': run_fetch_save_debug,
|
||
'report': lambda: run_report(),
|
||
'report-today': lambda: run_report(datetime.now().strftime('%Y-%m-%d')),
|
||
'parse-test': run_parser_test,
|
||
'stats': lambda: show_stats(datetime.now().strftime('%Y-%m-%d')),
|
||
}
|
||
|
||
|
||
def main():
|
||
parser = argparse.ArgumentParser(
|
||
description='码头作业日志管理工具',
|
||
formatter_class=argparse.RawDescriptionHelpFormatter,
|
||
epilog='''
|
||
可选功能:
|
||
fetch 获取HTML并提取文本（保存到debug目录）
|
||
fetch-save 获取、提取、解析并保存到数据库
|
||
fetch-debug 获取、提取并保存带时间戳的debug文件
|
||
report 生成日报（默认今天）
|
||
report-today 生成今日日报
|
||
parse-test 解析测试（使用已有的layout_output.txt）
|
||
stats 显示今日统计
|
||
|
||
示例:
|
||
python3 main.py fetch
|
||
python3 main.py fetch-save
|
||
python3 main.py report 2025-12-28
|
||
python3 main.py parse-test
|
||
'''
|
||
)
|
||
parser.add_argument(
|
||
'function',
|
||
nargs='?',
|
||
default='fetch-save',
|
||
choices=list(FUNCTIONS.keys()),
|
||
help='要执行的功能 (默认: fetch-save)'
|
||
)
|
||
parser.add_argument(
|
||
'date',
|
||
nargs='?',
|
||
help='日期 (格式: YYYY-MM-DD)，用于 report 功能'
|
||
)
|
||
parser.add_argument(
|
||
'--unaccounted',
|
||
'-u',
|
||
metavar='TEU',
|
||
type=int,
|
||
help='添加未统计数据（需同时指定月份，如 -u 118 2025-12）'
|
||
)
|
||
parser.add_argument(
|
||
'--month',
|
||
'-m',
|
||
metavar='YEAR-MONTH',
|
||
help='指定月份（与 --unaccounted 配合使用）'
|
||
)
|
||
|
||
args = parser.parse_args()
|
||
|
||
# 添加未统计数据
|
||
if args.unaccounted:
|
||
year_month = args.month or datetime.now().strftime('%Y-%m')
|
||
add_unaccounted(year_month, args.unaccounted)
|
||
return
|
||
|
||
# 执行功能
|
||
if args.function == 'report' and args.date:
|
||
run_report(args.date)
|
||
else:
|
||
FUNCTIONS[args.function]()
|
||
|
||
|
||
if __name__ == '__main__':
|
||
main()
|