feat(core): 添加大化县政府网采购公告数据表和相关功能

- 创建 dahuagov_announcements 表用于存储大化县政府网采购公告
- 添加相关索引以提高查询性能
- 实现 save_dahuagov_announcements、get_new_dahuagov_announcements
  和 mark_dahuagov_announcements_sent 方法
- 修改统计查询以包含大化县公告数据
- 更新内容哈希检查逻辑以支持新表

feat(cron): 集成大化县政府网采购公告爬取功能

- 导入大化县政府网爬虫模块
- 修改定时任务流程以同时爬取广西政府采购网和大化县政府网
- 对不同来源公告采用不同处理策略:
  - 广西政府采购网:关键词筛选后推送
  - 大化县政府网:全部推送,不过滤关键词
- 分别处理和统计两个来源的公告数据
- 实现独立的通知发送和状态更新机制

feat(notification): 优化企业微信通知显示大化县来源标识

- 为不同来源公告添加前缀标识(【大化县政府网】或【广西政府采购网】)
- 根据公告来源动态调整通知标题:
  - 单一来源显示具体来源
  - 双来源显示"双源监控"标识
- 改进通知卡片的来源区分度,便于用户识别公告来源
```
This commit is contained in:
2026-05-09 10:55:00 +08:00
parent bd8df98e9a
commit c548a8b5bd
17 changed files with 128615 additions and 22108 deletions
+157 -1
View File
@@ -264,12 +264,32 @@ class DatabaseManager:
FOREIGN KEY (source_code) REFERENCES announcement_sources(code)
);
-- 大化县政府网采购公告表(全部推送,不筛选)
CREATE TABLE IF NOT EXISTS dahuagov_announcements (
id SERIAL PRIMARY KEY,
title VARCHAR(500) NOT NULL,
publish_date TIMESTAMP NOT NULL,
purchase_name VARCHAR(200),
content_url TEXT,
source_code VARCHAR(50) NOT NULL DEFAULT 'dahuagov',
source_name VARCHAR(100) NOT NULL DEFAULT '大化县政府网采购公告',
announcement_type VARCHAR(50) NOT NULL DEFAULT 'purchase',
crawled_at TIMESTAMP,
created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
content_hash VARCHAR(32) UNIQUE,
is_new BOOLEAN DEFAULT TRUE
);
-- 创建索引
CREATE INDEX IF NOT EXISTS idx_announcements_publish_date ON announcements(publish_date DESC);
CREATE INDEX IF NOT EXISTS idx_announcements_source_code ON announcements(source_code);
CREATE INDEX IF NOT EXISTS idx_announcements_content_hash ON announcements(content_hash);
CREATE INDEX IF NOT EXISTS idx_announcements_created_at ON announcements(created_at DESC);
CREATE INDEX IF NOT EXISTS idx_crawl_results_crawled_at ON crawl_results(crawled_at DESC);
CREATE INDEX IF NOT EXISTS idx_dahuagov_publish_date ON dahuagov_announcements(publish_date DESC);
CREATE INDEX IF NOT EXISTS idx_dahuagov_content_hash ON dahuagov_announcements(content_hash);
CREATE INDEX IF NOT EXISTS idx_dahuagov_created_at ON dahuagov_announcements(created_at DESC);
-- 创建更新时间触发器
CREATE OR REPLACE FUNCTION update_updated_at_column()
@@ -704,6 +724,13 @@ class DatabaseManager:
COUNT(DISTINCT source_code) as sources,
MAX(crawled_at) as last_crawl
FROM manual_announcements
UNION ALL
SELECT
'dahuagov_announcements' as table_name,
COUNT(*) as count,
COUNT(DISTINCT source_code) as sources,
MAX(crawled_at) as last_crawl
FROM dahuagov_announcements
"""
cursor.execute(detail_sql)
@@ -742,12 +769,14 @@ class DatabaseManager:
SELECT content_hash FROM auto_announcements WHERE content_hash = %s
UNION ALL
SELECT content_hash FROM manual_announcements WHERE content_hash = %s
UNION ALL
SELECT content_hash FROM dahuagov_announcements WHERE content_hash = %s
) as combined_check LIMIT 1
"""
try:
with get_db_cursor() as cursor:
cursor.execute(sql, (content_hash, content_hash, content_hash))
cursor.execute(sql, (content_hash, content_hash, content_hash, content_hash))
return cursor.fetchone() is not None
except Exception as e:
logger.error(f"检查公告存在性失败: {str(e)}")
@@ -790,6 +819,133 @@ class DatabaseManager:
logger.error(f"获取最近公告失败: {str(e)}")
return []
@retry_on_exception(RetryConfig(max_retries=3))
def save_dahuagov_announcements(self, announcements: List[Announcement]) -> int:
"""
保存大化县政府网公告(全部推送,不筛选关键词)
Args:
announcements: 公告列表
Returns:
int: 成功保存的新公告数量
"""
if not self.config.database.enabled:
return 0
if not announcements:
return 0
# 为没有哈希的公告生成哈希
for announcement in announcements:
if not announcement.content_hash:
announcement.generate_content_hash()
sql = """
INSERT INTO dahuagov_announcements (
title, publish_date, purchase_name, content_url, source_code, source_name,
announcement_type, crawled_at, content_hash, is_new
) VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
ON CONFLICT (content_hash) DO NOTHING
"""
values = []
for announcement in announcements:
values.append((
announcement.title,
announcement.publish_date,
announcement.purchase_name,
announcement.content_url,
announcement.source_code,
announcement.source_name,
announcement.announcement_type.value,
announcement.crawled_at,
announcement.content_hash,
announcement.is_new
))
try:
with get_db_cursor() as cursor:
extras.execute_batch(cursor, sql, values)
affected_rows = cursor.rowcount
logger.info(f"保存大化县公告完成,新增 {affected_rows}")
return affected_rows
except Exception as e:
logger.error(f"保存大化县公告失败: {str(e)}")
return 0
@retry_on_exception(RetryConfig(max_retries=3))
def get_new_dahuagov_announcements(self) -> List[Announcement]:
"""
获取大化县未推送的新公告(is_new = TRUE
Returns:
List[Announcement]: 未推送的公告列表
"""
if not self.config.database.enabled:
return []
sql = """
SELECT * FROM dahuagov_announcements
WHERE is_new = TRUE
ORDER BY publish_date DESC
"""
try:
with get_db_cursor() as cursor:
cursor.execute(sql)
rows = cursor.fetchall()
announcements = []
for row in rows:
row_dict = dict(row)
row_dict['announcement_type'] = AnnouncementType(row_dict['announcement_type'])
announcements.append(Announcement.from_dict(row_dict))
return announcements
except Exception as e:
logger.error(f"获取大化县新公告失败: {str(e)}")
return []
@retry_on_exception(RetryConfig(max_retries=3))
def mark_dahuagov_announcements_sent(self, announcements: List[Announcement]) -> int:
"""
标记大化县公告已发送(is_new = FALSE
Args:
announcements: 已发送的公告列表
Returns:
int: 更新的记录数
"""
if not self.config.database.enabled:
return 0
if not announcements:
return 0
# 获取所有公告的哈希值
hashes = [ann.content_hash for ann in announcements if ann.content_hash]
if not hashes:
return 0
sql = """
UPDATE dahuagov_announcements
SET is_new = FALSE, updated_at = CURRENT_TIMESTAMP
WHERE content_hash = ANY(%s)
"""
try:
with get_db_cursor() as cursor:
cursor.execute(sql, (hashes,))
affected_rows = cursor.rowcount
logger.info(f"标记大化县公告已发送完成,更新 {affected_rows}")
return affected_rows
except Exception as e:
logger.error(f"标记大化县公告已发送失败: {str(e)}")
return 0
# 全局数据库管理器实例
_db_manager = None