ujO: d Z ddlZddlZddlZddlZddlZddlZddl Z ddl
Z
ddl m Z dZdZdZdZdZd Zej( j+ d
d Zej( j+ d d
ZdZ ej2 dej4 ej6 z Z ddddZ dZ e ej@ D ]k \ Z!Z"e"jG d r e$e"jK dd d Z 4e"dk( s:e!dz e'ej@ k sS e$ej@ e!dz Z m e dk\ re ndZ(d Z)d Z*d Z+d Z,d Z-d Z.d Z/d Z0e1d k( r e0 yy# e
$ r dZ Y 1w xY w# e&$ r Y w xY w# e&$ r Y w xY w)!u
crawl_jzyjt_gsgg.py - 巨正源股份-公司公告
http://www.jzyjt.cn/gsgg/list.html
CMS: 万户网络 wanhu (Powered By wanhu)
列表: div.TextList001203 ul.ul > li
分页: /gsgg/list_page_{N}.html (首页 list.html; 超出页数 404, 站点实际仅1页)
详情: div.RichInfo005 > div.news_info
标题: header.title > h1 (列表 span.dot 已含完整标题, 无需详情回退)
日期: header.title > p > span "发布日期：YYYY-MM-DD"
正文: news_info 内 header 之后到 div.info_back 之前 (Word 导出嵌套 span 结构)
附件: 附件名 (内嵌正文段落)
N)
BeautifulSoupzhttp://www.jzyjt.cnz"http://www.jzyjt.cn/gsgg/list.htmlz.http://www.jzyjt.cn/gsgg/list_page_{page}.htmlu 巨正源股份-公司公告u 广东zcrawl_jzyjt_gsgg.pyDB_PATHz/mnt/data/search.db
JSONL_PATH utf-8z}]*href='(/gsgg/info_itemid_\d+\.html)'[^>]*>\s*\s*([^<]+)zkMozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0 Safari/537.36z?text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8zzh-CN,zh;q=0.9)z
User-AgentAcceptzAccept-Language z--pages==z--pagesc 2 | syt j | } | j dd j dd j dd j dd } t j dd| } t j dd| } | j S ) Nr u u u u u3 ^\s*(?:·|·|\u00b7)?\s*(?: |\u00a0)?\s*\s+ )html_libunescapereplaceresubstrip)ts %/root/gov_crawler/crawl_jzyjt_gsgg.py clean_titler C s ! A (B ''"5==hKSST\^`aA
Er1MA
vsA A 779 c |dk( rt }nt j | }| j |t dd }d|_ |j S )Nr )page Fheaderstimeoutverifyr ) LIST_URL_P1 LIST_URL_PNformatgetHEADERSencodingtext)sessionr urlrs r
fetch_listr) M sG qy d + C"U CAAJ 66Mr c H g }t j | D ] }|j d }|j d }t |j d }|rt | dk rPt
j j t | }|j |||f |S )Nr )
LIST_LINK_REfinditergroupr lenurllibparseurljoinr append) html_textitemsmhrefdatetitleabs_urls r
parse_listr= W s E
"
"9
- -wwqzwwqzAGGAJ'E
Q ,,&&{D9
gud+,- Lr c d}d}d}t rt | d }|j dd }|r|j d }|s|j d }|r t |j dd
}|j d t j d
}|r7t j d|j } | r| j d }dD ]( }
|j |
D ] } | j * dj d |j D }|s)t j d| } | r| j d }|||fS )uT 万户: h1 标题 + 发布日期 + news_info 正文(剔除 header/info_back/info_pn)r html.parserdiv news_info)class_zheader.title h1h1r
Tr spanu 发布日期)stringz(\d{4}-\d{2}-\d{2})r )z header.titlez
div.info_backz div.info_pnc 3 2 K | ] }t | y w)N)str).0cs r z parse_detail..} s "Aa3q6"As u( 发布日期[:：]\s*(\d{4}-\d{2}-\d{2}))
r find
select_oner get_textr compilesearchr0 select decomposejoincontents) r6 page_urlr; r: content_htmlsoupinforC mddmselels r parse_detailr] d s. E
DL Y
6yy{y3 !23BYYt_#BKK4K$@A6"**_*EFBYY 5r{{}E 88A;DG
# ++c* #BLLN#
# 77"A4=="AAL
YYBI
N
88A;D $ $$r c 6 | sddg fS t j dd| t j } t j dd| t j } t j dd| t j t j z } t j dd| t j t j z } g i t j dd| } t j d d| } t ra t | d
}|j d d
D ] }|j
d sG|j
d s6t j d|j
dd t j s\|d }|j d xsD |j
d xs1 t j j |j d d }t j j | } j! ||f dt# d}d| d| d |< |j% | |j d D ] }|j d } | s|j d r+g }
| D ] } | j
d xs | j
d xs d} | s.t j j | }|
j! d| dt j j | j d d xs d d |
sd t# d}d d!j' |
z d"z |< |j% | t) | } fd#}
t j d$|
| t j t j z } t rQ t | d
}|j d D ]% }d%D ] }|j, j/ |d ' t) | } g fd&}t j d'| t j rd(nd}t j d)|| t j t j z } t j d"d*| t j } t j d+d,| t j } t j d-d| t j } g }| j d. D ]W }|j1 }|st j d/d| }t j d0d| }|j3 d1d j3 d2d3 }t j d4d3| }t j d5d| }t5 j6 | }|j3 d6d3 j3 d7d3 }t j d8d| }|s j9 D ] \ }}|j3 || } t; D ] \ }}|j3 d9| d| } |j! |j1 Z d.j' | } j9 D ] \ }}|j3 || } t; D ] \ }}|j3 d9| d| } t j d:d| }t j d;d| }t j dd,| }t j d?d.| }t j d4d3| }t= }g } |j d. D ]X }t j d@dt j d5d| }|s2||v r7|j? | | j! | Z d.j' | j1 }|| fS # t* $ r Y w xY w# t* $ r Y w xY w)ANr r z
)flagsz ]*>z]*>.*?z]*>.*?u 【字号：[^】]*】u' 发布(?:时间|日期)[:：][^<]{0,30}r? aT)r9 appendixz
data-appendixz\.(pdf|docx?|xlsx?|zip|rar)r9 rD _title?
__ATTACH____ pimgsrcoldsrcu 图片__IMG__z
z z
c 8 | j d }|j d ry| j d }t j dd| }t j
| j
}t j dd| j
}|s5t j j |j d d xs d }t j j | }j ||f d
t d }d | d
| d|< |S )Nr z javascript:r r+ <[^>]+>u ^附件[:：]\s*rc r u 附件__LINK__re rf rg rh )r0
startswithr r r r r ospathbasenamesplitr2 r3 r4 r5 r1 ) r8 r9 innertxtr< key attachments link_protectrU s r
_link_replz html_to_text.._link_repl s wwqz ??= )
ffZU+ $**,ff("c288:''""4::c?1#56B(C,,&&x6GS>*\*+2. 'y0B3%t L S
r z&]*href="([^"]+)"[^>]*>(.*?)) styleclasslangdiralignvalignwidthheightborder cellpadding cellspacingc h j | j d dt dz
dS )Nr __TBL__r re )r5 r0 r1 )r8
table_protects r _tbl_replz html_to_text.._tbl_repl s3 QWWQZ(]+A-.b11r z ]*>r z]*>.*?z
z
z(]*>z
z(?<=>)\s*[\r\n\t]+\s*z\s*[\r\n\t]+\s*(?=<)
r
z [ \t]{2,}ro z z [\s\u200b\u200c\u200d\ufeff]+r z
__LINK__\d+__z __TBL__\d+__z __IMG__\d+__z__ATTACH__\d+__z[ \t]+\nz\n{3,}r ) r r SIr find_allr" rP rN rr rs rt ru r2 r3 r4 r5 r1 replace_withrS rH Exceptionattrspopr r r r r7 enumeratesetadd) contentrU _soupr` r9 rw r< rx ri imgssrcsrj rk r{ tagattrr has_tablepartsblockbplainkvitbloutseenfinalry rz r s ` @@@r html_to_textr s% 1byff]Brtt"))Ljlmlqlqrxz|l} @B @D @D CE V9D**4*0kAEE(OkrwwGWGWX\XbXbcfXghiXjGkC $ll 2 28T BG &&~6 &s<'8&9
G[][_[_bdbfbf[fgG !'=9E ~~d+
. L .DIIMM$-.
. %jG M2 YY>AIff/G244RTRVRV;WGffWj'>Gff\4=Gff@"gUWUYUYZG E v& KKM FF+R 3 FF*B 2
IIdB ' 'c 2 FF<a (z2q)!!%(
fc*228SA7UC &&( DAq !QA
. 0FAs GA3b/3/A 0
QWWY ) , ++e
C ""$ 1kk!Q M* 03kkGA3b/3/0
&&!2s
+C
&& "c
*C
&& "c
*C
&&#R
-C
C
C
&&dC
(C
&&FC
(C
&& sC
(C 5D E
YYv
ffVR
B!:; $;
Q ++e
"
"
$C ; &&{ 6 sG A:_; ?C'_; '_; :1_; ,A/_; A_; *A` ; `` ``c
| j d|t f }|j ry d} ddlm} || }| j d|||||t t t ||f
}|j dkD ryt j dd| } t j | } t j d d
| j d d } | j d |j |t | f y
y# t
$ r Y w xY w# t j $ r=}
dt# |
v r*t% j& d t) | ||||| cY d }
~
S d }
~
ww xY w)Nz7SELECT id FROM gov_raw WHERE page_url=? AND site_name=?Fotherr )classify_industryzINSERT OR IGNORE INTO gov_raw (page_url, source_url, title, content, publish_date, site_name, group_name, script_name, has_table, industry) VALUES (?,?,?,?,?,?,?,?,?,?)ro r r r
zTINSERT OR IGNORE INTO gov_search (rowid, title, site_name, summary) VALUES (?,?,?,?)Tlockedr, )execute SITE_NAMEfetchone crawler_libr r
GROUP_NAME SCRIPT_NAMErowcountr r r r r lastrowidsqlite3OperationalErrorrH timesleep store_record) connr' r; r r: r curindustryr summaryes r r r sF
,,PSVXaRb
cC
||~ 5(/H ll
+ #ugtY
KQZ\d e
<]
z skip empty/placeholder: ( )rU r; r publish_dater site_name
group_name script_namez
JSONL + z ()z + g ?g ?wr )r$ F) ensure_asciir zJSONL written: z rows)u 新增: u 跳过: z=== z done ===)(requestsSessionr updater# r r connectr r r range_PAGESr) r printr= r1 r r r] r r r r rP r r5 r r r r r r openwritejsondumpscommitclose)r sr new_count
skip_count seen_urls
jsonl_rowsr r6 r r7 r< r; r: dhtmldtitleddaterV r r atts plain_lenno_paraokfrows r mainr 8 s AIIW D w3 01 ./IJIJa!$ 3 "1d+I 9%
dV8CJfETR
HYK
- HZL
!. D 9
%T2{ E$~aS1 >
y1#6dCa
J D Ds< ! L. M;1N. M7MM N!M??NN__main__)2__doc__sysrr r r htmlr urllib.parser2 r r bs4r ImportErrorBASE_URLr r r r r environr" r r ENCrO r r r. r#
_MAX_PAGESr argvr r` rq intru
ValueErrorr1 r r r) r= r] r r r r __name__ r r r s !
!2 > *
#
**..$9
:
ZZ^^L"
-
rzz EDD244K @O'
chh
DAq||J QWWS!_Q /0J
i AECM 1 SXXa!e_-J
"QA
%BF'R D0O3d zF _ M:
s5 E E E*E E E'&E'*E21E2