]*href="([^"]*\.(pdf|doc|docx|xls|xlsx|zip|rar))"[^>]*>([^<]+)z//zhttps:)namer2 r7 z\s+r) , z z
z
t } |d xs d d d }|d xs d d d! }|j! d"|d |d ||d# xs d |t d$f |j" dkD }|j% |r,|dz
}| r t
d|dz d%|d d d& d
|d# dd n|dz
}|j) t+ j, d( t
d d) d t
d*| d+| d,| d |dkD rBt } |j! d-t f |j% |j) t
d/| d0d y # t& $ r }t
d'| d | dz
} Y d }~d }~ww xY w# |j) w xY w# t& $ r}t
d.| d Y d }~d }~ww xY w# |j) w xY w)1NFr r )1 incrementalz
--incrementalTz--pagesu 增量u 全量
z2==================================================)flushu 🏠 z ()z DB: u
📄 列表页 r z: u [WARN] 获取失败u [WARN] 无文章链接u 发现 u
篇文章r2 r: z [u ] ✗ ( z...rW r8 i rX rU z}INSERT OR IGNORE INTO gov_raw (title, page_url, content, publish_date, summary, site_name, tags) VALUES (?, ?, ?, ?, ?, ?, ?)r; u 环评阶段u ] ✅ 2 z
[ERROR] g333333?uZ ──────────────────────────────u ✅ 新增: u | 跳过: u | 错误: zINSERT OR IGNORE INTO gov_search(rowid, title, site_name, summary, rowid_col) SELECT id, title, site_name, substr(content, 1, 500), id FROM gov_raw WHERE site_name=?u [WARN] FTS同步:u 📤 完成！共 u
条新数据)listsysargvr/ int TOTAL_PAGESprint SITE_NAMEr r* r" r5 rO enumeraterf r execute
total_changescommitr closer, r- )ri pages_overrideargsiarg total_pagesmode new_count
skip_count error_countr! r2 sessionrG rH idxitemdetailr rW rX changedr\ s r mainr s KN D A
c$i-1g 7 7 K
I
!a%#d)"3 ac^N
FA Q c$i- $ &aK "8D Bvh-t$ E)BtfA
&d3 GG9
T* VHT"IJKaq ) - 4
!$q
Ru=TJ'-
)$ 7 4 ,T :
#e* Z0="5) IC#DKg ?FCE7&g s);(RPRT[]fhvw
,,q0
NI&$ CE7& "1E0FbPV HXXY Zbfg !OJ
JJsO= - ^
Bxj/& L ;zl+k]
S[_`1}z LL13<,
KKM
JJL yk
7tD5
!
1#&d3q
!
" 9 's+4 8 8 9
JJLsU ?BK=4'L= = L%L L( L%%L((L:= M MM# M M# #M5__main__)__doc__r< r r
osrp r, r DOMAINr rD ru getenvr r rs r r r" r5 rO rf r __name__ r r r s
, + + 1 ( *
"))K!2
3o0
, .8JZ^EB zF r