Python + pyspider某小说站的爬虫,入数据库,火车头发布,资源下载到本地,另可写爬虫!/ t5 K( u0 B! _% w+ S
- #!/usr/bin/env python
$ X" D7 t) V1 _; \' `# H( F - # -*- encoding: utf-8 -*-5 y" a8 \9 E% X% H2 A
- # Created on 2019-05-05 21:43:115 Y# M" [4 R" H5 a5 L# d$ P( d
- # Project: XiaoShuo
% x' I; A$ |0 ~ -
6 b% ?0 x* Z) l" l8 U1 q+ L) [9 v - from pyspider.libs.base_handler import *
! M0 v9 ~$ ?( ^8 m, l- P; T" G. l) r - import pymysql! H& M' S6 ^1 h, E, h" G+ z
- import random& g/ F: d) p0 c: v/ r
- import datetime" F. x) O7 m. Y8 k7 e I
- import urllib2,HTMLParser,re2 n, p3 B% H. ?6 F! U
- import os
) i, X& K, p$ y$ | - import sys; }+ D5 w- q: z9 d" o* a4 ?
- import re+ ~; U! ^4 E, w( u( K; g) T, s
- import codecs
% R; g, \( @4 m) q# V$ g7 X. u - import requests
- n0 e0 S8 F/ L( q - import json7 f6 _: Z; g8 T$ R! g
-
, h7 g3 j7 }: R - class Handler(BaseHandler):& m9 n; W) ^) j5 f! l6 ^- F: Y9 l- I: X9 ]
- global Datos
8 m1 K5 r+ `; e - global P_dir ' Z. S! k; L6 t7 `+ A ^. y
- P_dir = '/Tools/Debug/' #采集时候图片保持到本地的路径
# P$ q) b! _: u o+ ^4 {( C7 h - global Datos
8 N+ c- P" M+ }( p! m1 h - Datos = {}: F, }3 O3 I; a0 S. F! a" k+ Y; s
- headers= {
$ V s) T ]$ t. o; S/ [ [4 f. @ - 'Accept':'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
- S& i) v8 @: x! f - 'Accept-Encoding':'gzip, deflate, sdch',4 x2 K1 t% n; W3 P% h
- 'Accept-Language':'zh-CN,zh;q=0.8',
1 \9 S: C2 S+ ^! U; o: y @ - 'Cache-Control':'max-age=0',3 C' ]* p2 o( t4 e. h
- 'Connection':'keep-alive',
9 H8 g$ I: ]. A7 _4 R) | - 'User-Agent':'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.101 Safari/537.36'
$ j3 v2 b# w& v0 Q* K9 G9 x, r7 q - }
* R V1 }$ V; d: q0 ^ - crawl_config = {, e8 s( P- I2 G( `$ T& ?+ Y, z9 _
- 'headers' : headers,! [2 y \# N8 Z9 g
- 'timeout' : 3009 g' Y9 M$ b% T$ J7 C( P! {
- }/ f9 ` p, B2 a& x% e% J
- def add_BookFile(self,Bookname, BookIDs, img, Locaimg, Book_Dates):
1 x M, H p- Z. g/ u - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")4 D$ l7 x& V6 t6 D
- try:+ Q5 @) L) y1 }8 Q: a8 }
- cursor = db.cursor()5 J# l' ?* A+ [% ~
- #注意此处字符串的占位符要加双引号"%s"
& Z* Z6 z: R6 q5 ~! g - sql = 'insert into BookFile(Bookname, BookID, img, Locaimg, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, BookIDs, img, Locaimg, Book_Dates);
' M$ T- S' F! R* A) {" d; U - # print(sql)$ G- L. `+ H4 t5 F/ s% j
- cursor.execute(sql)
. `4 a; W$ U# C - . x* V! j+ R6 W. o' ], b x# f
- #qid = cursor.lastrowid/ w2 B; E, ~8 G5 ~# V7 U/ Y \
- #print(qid)) x! J& L7 F, }2 [! l. a6 i, i4 O7 A
- 8 }* Y* p+ t. y( F+ ]
- db.commit()
( z% b+ a, ~; Y5 {2 a3 c7 I* k - except Exception as err:
1 j1 s5 Z# @. I' T- S7 O3 [ - print("Error %s for execute sql: %s" % (err, sql)): v2 f3 V0 q* h z& ~, F c! x! c
- db.rollback() l% ~% ~8 i. E1 G" D+ V' e* @
- def add_comment(self,Bookname, Booktitle, BookID, Titleid, Book_Date):
& _7 g1 y& o; i) S) f - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")& `& n0 L: x9 U. A
- try:- u+ C; k* s. m7 w4 }" O) T
- cursor = db.cursor()
/ Y9 ^+ ] G. r. d8 j: ` - #注意此处字符串的占位符要加双引号"%s") z6 ^% p# l/ c
- sql = 'insert into BookTitle(Bookname, Booktitle, BookID, Titleid, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, Booktitle, BookID, Titleid, Book_Date);! L6 e0 k3 a5 F; A1 v
- # print(sql)- ^1 I6 k; _- L" X# g) D
- cursor.execute(sql)
. z2 p% D* z, _! {8 p" s - 0 Z& ^' A# ]% w3 @8 U, N1 I
- #qid = cursor.lastrowid9 v8 \. ]5 ?: K
- #print(qid)
' E7 O( \; M+ V - 9 Z/ V( O$ H* @4 Y8 C) ^
- db.commit()
) d& R" m' b, S5 G- W - except Exception as err:
% E6 T d* K0 N - print("Error %s for execute sql: %s" % (err, sql))
6 d# A$ {' K9 L; ?0 \ - db.rollback()( v/ h1 x3 O) `) D+ J7 s
- def add_question(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date):3 S% u0 N. F2 b3 r' p+ m! L
- db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")- k/ C1 a6 }7 y( W) i9 S
- try:6 b3 [& B' G* R2 T
- cursor = db.cursor()( f0 d1 [- p) N5 F1 J a' d* h& _
- #注意此处字符串的占位符要加双引号"%s"
3 K* A2 }* A/ R: K6 o$ y" \) o$ I [: u - sql = ' INSERT INTO BookConte (Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date) VALUES("%s","%s","%s","%s","%s","%s","%s","%s","%s","%s","%s","%s","%s")' % (Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date);* }5 e5 c; x% T5 X4 s: T
- print(sql)
5 m% h* P% V: X0 _: n5 X" Y - cursor.execute(sql)2 V! l/ [2 |0 @ l3 Z
- print(cursor.lastrowid)" t( F5 a9 {! X+ U1 `; j
- db.commit()7 y/ I, ^1 ]: S; Q
- except Exception as err:
" M: t3 w7 _% ?/ m' ]. @; ?9 X - # except:
E- U3 K+ v/ {/ o7 b - # print('Failed')" C+ \ k8 f! w4 i
- print("Error %s for execute sql: %s" % (err, sql))
) U" |/ c% ^2 t5 @ S - db.rollback()
, b# z8 [8 k5 e. K F2 ~ - ! D: v+ \; P' A; p
- def add_locoy(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Palabras,Book_img,Booktitle,BookConte,abover):
/ a! a% g: f. i2 W8 Q4 Y - reload(sys). q4 J8 S( \# ~/ T7 Z( F: b
- sys.setdefaultencoding("gbk")- i- X, v4 c/ s
- locoy_url = 'http://www.******.net/locoy/?my=book' #697火车头发接口地址0 |' m# o3 |$ e$ a
- locoy_data = {
7 _% E9 E6 ~8 e5 S8 f' ~0 O- G - 'my_u':'用户名', #后台用户名
. F, }, F1 p( ` - 'my_p':'密码', #后台密码
: T: k) w: `% i, a9 b. ~; r* A - 'subject_669977_net':Bookname.encode('gbk', 'ignore'),3 T( D4 i( U2 b) P
- 'caid':Cater_Name.encode('gbk', 'ignore'),
p' {$ D- k: Q/ [1 s - 'title_669977_net':Booktitle.encode('gbk', 'ignore'),* n/ [* F7 Z" L) N$ N
- 'article':BookConte.encode('gbk', 'ignore'),
5 V& |' q9 i" J9 m7 o" f - 'author':Book_author.encode('gbk', 'ignore'),
" h, l8 ~7 S2 O; n - 'ready_1':Book_Palabras.encode('gbk', 'ignore'),, ]$ G& W/ p, s
- 'thumb':Book_img,8 X4 b- g( f e6 b
- 'content':Book_Introduction.encode('gbk', 'ignore'),
6 |4 ~1 E( W- Z/ S$ \- u7 ~: H - 'abover':abover.encode('gbk', 'ignore') 6 |" c1 i$ v6 Q" P, G% \
- }
6 R' |5 T- W0 b7 N: b. | - res = requests.post(locoy_url, data=locoy_data)
b# T8 L% a; x - print res.text
% R! ^; b$ |: s* C; @/ V. X7 O - print res.content
! {6 k( K& Q" p5 N( Q1 {1 p - # print Dsd
' J4 N( d: S/ ^! Q - return res
; D) H) f; a) ^, _2 k$ j4 j -
6 U$ D8 N; ~6 h2 N: y* y' ^9 a# L - def __init__(self):
. ^; Y1 B& e, q" B* D h2 @ - self.base_url1 = 'https://www.****.cc/'
( G, `' H) H3 s9 w) v9 j8 I - self.base_url2 = '/'
' [5 f; U9 _. K8 N" y: h+ Z' _' Q - self.CaterId = []1 N# I: r' _) D/ S' o% b3 D0 J1 p
- self.CaterIds = ['xuanhuan', 'wuxia', 'yanqing', 'lishi', 'kehuan', 'kongbu', 'nvsheng'], A8 U8 J% }! l1 U* D9 o
- self.page_num = 1
: Q2 J2 q: F1 G+ s5 S, O2 ^ - self.total_num = 200
- N3 a1 q5 s( ]# m5 V! D - + |0 o- n3 I# q0 h$ }) m0 j4 t
- @every(minutes=8 * 60)) H) m& t% f7 b3 z! r+ e
- def on_start(self):6 c+ f$ E( a2 O- u' ~
- global Cater_Name2 C' K. a# `' n! y' n2 g
- Cater_Name = []
- O0 H2 [+ p$ Z7 C, Z7 Q* w - while self.page_num <= self.total_num: ) N$ h: X3 {- S: g3 q
- for self.CaterId in self.CaterIds:2 c1 \- u( j5 s% o {
- if self.CaterId == 'xuanhuan':
: q2 w3 @9 B3 y1 P A6 n8 }6 J& D - Cater_Name = '玄幻'
. k( d9 k1 f% J4 x - if self.CaterId == 'wuxia':% Z, }, ~8 W; [' D& M/ N" C3 [0 Y
- Cater_Name = '武侠'
+ A* F: k I7 u1 k* C+ T - if self.CaterId == 'lishi':
: m, V6 d( T* n3 U& W - Cater_Name = '历史' / _8 C$ @" S$ {9 c% z4 k
- if self.CaterId == 'yanqing':
& U, ]- C# P3 ]- H - Cater_Name = '都市' 6 u' K' R2 A5 P& B' @5 f
- if self.CaterId == 'nvsheng':& {: Z+ X3 I" c j
- Cater_Name = '都市'
' K# V1 ^: F+ R( A - if self.CaterId == 'kehuan':) U* `8 W/ k) n
- Cater_Name = '科幻' 2 r7 I$ j X E( L; I) B
- if self.CaterId == 'kongbu':
) N' Y3 _- t9 @) x h0 x) r! l! T2 F( L - Cater_Name = '游戏' + C5 M) ~7 n* N/ l% V. a/ ?
- print self.CaterId
, l7 F9 T/ g: V/ o - url = self.base_url1 + str(self.CaterId) + self.base_url2 + str(self.page_num) + "/"
6 j' H9 `' n: d4 M, v5 _ - self.crawl(url, callback=self.list_Caterg,save=Cater_Name)7 d" w4 v- T$ B
- self.page_num += 1
& t% k, M7 I$ R m; F+ B& O - ! F& }6 x. ~5 F6 e: G1 y
- def list_Caterg(self, response):: a4 z# T5 l" y7 n2 S2 U" ^
- Cater_Name = response.save8 c* D; P7 Z& ^+ J# U
- for each in response.doc('.pic-list a[href^="http"]').items():
1 x" \' }- G& N - self.crawl(each.attr.href, callback=self.list_Caterg_detail,save=Cater_Name), e# b& y6 U* I
- ; H2 W- D. P& W4 Q
- def list_Caterg_detail(self, response):+ H; F4 K7 D$ W9 Z# P0 O- `4 u1 g( r( c
- Cater_Name = response.save& j( j+ X8 R$ N: ^
- # print Cater_Name
, d7 i* I" E. t v, U - Bookname = response.doc('h1').text()
* m) r( {8 q9 @6 b. V3 j$ W3 U - print Bookname
, u' j9 A P2 ~% \ - Book_author = response.doc('.authorname > a').text()
7 A5 V0 o5 v8 D; e, t( |. p - # print Book_author# N) I- a; L0 N8 e- `
- Book_Introduction = response.doc('.book-intro > div').text()
; q6 j: v' J& d - # print Book_Introduction1 y' h* V9 c& b2 u* v
- Book_Synopsis = response.doc('b').eq(1).text(). O9 Z6 m% e$ {; |+ x4 B/ y4 t3 Z
- # print Book_Synopsis$ k- m5 \4 d! p8 E4 `" p! i
- Book_Palabras = response.doc('.booktitle p').text().split(' ')[1].split('|')[0]; ^7 p4 S$ x6 C% ? i: \
- # print Book_Palabras0 y' `- S1 E/ R5 D! {% x% J
- BookIDs = response.url.split("xiaoshuo/")[-1].split("/")[0] #小说ID) X& P g+ n5 D# ?2 b2 A* K
- # print BookIDs( Q) ^$ K4 U" a/ q" @; b0 R* f
- Book_Dates = str(datetime.datetime.now()) # R1 s, V: O: [5 t/ @
- for imgs in response.doc('.bigpic > img[src^="http"]').items():
$ j+ d9 |% ]0 F$ R3 p' W7 \ - img = imgs.attr.src2 u2 {$ L0 ^* i7 W, ]
- print img
2 a/ l6 D, T) f0 c) ] - #小说封面下载- E Q! o9 i/ B6 L4 ^
- extension = self.getExtension(img)) b/ z6 c9 J0 F A' n) w0 h
- name = self.getname(img)
2 p+ _- {4 c- ?' y- G, ? - file_name = name + "." + extension- S; B! U1 J1 s( J3 l" M: \" C
- imgDir = P_dir + name
7 f5 |; S/ o; b - Locaimg = imgDir + "/" + file_name/ n" y; X. g6 f
- print Locaimg
" X/ y6 ]6 \5 I$ `2 j/ o0 U" ? - if(self.download(P_dir, imgDir, file_name, img)): #这2行可注译,图片下载到本地* H6 t. ?0 R7 H2 S! v6 _
- print('attachment url is ' + img) #
" N( a" X+ c% K% H0 c: j4 q - Datos = {
8 ^# ?+ e% B, g5 R/ y# F - "Cater_Name":Cater_Name,
' m9 h, Y. `3 w- y: p; H/ i - "Book_author":Book_author,1 w$ j9 V, e8 Y y) F" c
- "Book_Introduction":Book_Introduction,8 o7 h% X0 Y+ w0 f
- "Book_Synopsis":Book_Synopsis,
2 Y( U$ {9 R2 k" E2 d+ R - "Book_Palabras":Book_Palabras,
' N9 t- ?1 C4 x# Y1 U+ C - "img":img,
7 d, k S0 Y" a6 s4 e! m - }
6 y" P% G* {; t0 k8 D$ X: L8 D - self.add_BookFile(Bookname, BookIDs, img, Locaimg, Book_Dates) #这行可注译,数据库发布接口,方便其他系统的发布3 F5 h# X! Z, l' I8 f2 P' w( ?0 B
- for each in response.doc('div[class="bookbtn-txt"] a[class="catalogbtn"]').items():
1 F" f8 T7 p9 C$ i' m4 ?4 }9 C - self.crawl(each.attr.href, callback=self.index_page,save=Datos)
' ^7 S% l) O" h. J - ; \7 A. w/ v6 M' j
- @config(age=8 * 60 * 60)
. a" ~! K& R9 [7 x9 T - def index_page(self, response): * u. T( L! B, t/ ^
- Datos = {
+ I5 S/ d2 q" ~* o Z' l - "Cater_Name":response.save['Cater_Name'],
0 A* d( L' M6 x6 b - "Book_author":response.save['Book_author'],, U7 `: d$ j4 V
- "Book_Introduction":response.save['Book_Introduction'], E) M8 a, i' {( x( m- ]4 _- o
- "Book_Synopsis":response.save['Book_Synopsis'],2 g' _6 ~! f2 @$ Z2 n* p; |
- "Book_Palabras":response.save['Book_Palabras'],& m8 o a/ b( V/ w; U) H
- "img":response.save['img'],! R! x, a, w" P% P1 N5 }- H* E
- }; ] j2 i2 \, O+ G7 O z
- for each in response.doc('.chapter-list li:first-child a[href^="http"]').items():$ p3 B* A9 q1 ^6 p, W- Q
- # for each in response.doc('.chapter-list a[href^="http"]').items():
0 O! X; K5 `- ?5 u2 V# |) K - self.crawl(each.attr.href, callback=self.detail_page,save=Datos)
" a. X" }' d1 u: r/ s* D - @config(priority=2)- R& w4 h2 E3 }. x& G
- @catch_status_code_error4 h, g$ U; C9 y# P
- def detail_page(self, response):
, L3 C: ~) N8 f - NewRe1 = u'哈书'' r8 Q" L2 s. g, J0 O, b0 E/ V
- NewRe2 = u'huhjsd.CC'
" d% c# c8 t* ^% a* w/ i( _ - NewRe3 = r'^\\n\\n'+ w6 ?8 `* e$ P$ n+ L! a( P6 D
- NewRe5 = u'小说网'& y" T- r l4 `0 o; n8 Y1 T& G' I. o
- NewRe6 = u'fgdfgf'8 r$ Y& O$ g9 q$ h
- NewRe7 = u'fgfgf'
7 ^- Q& d ~) P6 V4 x3 I9 a4 r - NewRe8 = u'ffhgf'; @! p/ p) S# }! [" [
- NewRe4 = r'[\f\t\v+\.\{\(\)\}\!\/_,$%^*(+"\')]+|[+——()?【】“”!,。?、~@#¥%……&*()]+'/ Z" X$ {) h. ^( M$ ?1 q+ _
- ReC1 = u'静思'
; E* B- L' T9 y - ReC2 = u'aghgf.com' `$ C5 B# R( I4 B. V: T+ A
- ReC3 = u'aghgfh.com'
5 k1 c1 a' I0 j6 h - ReC4 = u''0 t# \% g+ d, w% C. j, O
- ReC5 = u'文学网'
; Y& {/ K2 s j3 R$ w8 S& r% p - ReC6 = r'<BR>'8 D0 i+ ?2 d" D. v+ T
- Bookname = response.doc('.readlocation a').eq(2).text() #小说名称, ]2 Q7 K8 N( Z) g, G8 T% m
- print Bookname; G2 d- \. t8 L+ y
- Cater_Name = response.save['Cater_Name'] # 小说分类6 V/ `9 {3 f3 N+ ^5 v6 `
- Book_author = response.save['Book_author'] #小说作者1 R3 r9 B. U& n
- Book_Introduction1 = response.save['Book_Introduction'] #小说简介 I8 c4 @& N& |* |+ z" @ |
- Book_Synopsis = response.save['Book_Synopsis'] #最近更新
+ ~$ F2 C7 s5 U - Book_Palabras = response.save['Book_Palabras'] #小说字数
9 e9 f5 p9 {: x - Bookurl = response.url #小说网址
+ X6 K# J0 Q! l% j- s. W - Booktitle = response.doc('.article-title').text() #章节名称
. Q$ n* B( I% D: `) t6 M - BookID = response.doc('.readset-r span').text() #小说ID0 }: W* H5 A. F' Z
- BookConte1 = response.doc('.article-con').text() #小说章节内容1 E. ~$ ]7 a$ C, j1 `
- abover = response.doc('.article-title').text() + response.save['Book_Synopsis'] + response.save['Book_Palabras'] + response.save['Book_Introduction'] #小说状态(连载还是完成)
; B: t/ N, h5 i: v; q" T - Book_Date = str(datetime.datetime.now()) # 采集时间
# P6 x$ s/ R" A - BookConte2 = BookConte1.replace(NewRe1 , ReC1)
+ ~; g$ F& X! i - BookConte3 = BookConte2.replace(NewRe2 , ReC2)
4 n z" w. p) V5 X3 l2 d E - BookConte5 = BookConte3.replace(NewRe5 , ReC5)
/ G9 u" c+ J* J& [0 d6 R - BookConte6 = BookConte5.replace(NewRe6 , ReC2)$ B, F4 U1 k! a' N2 h, T3 B
- BookConte7 = BookConte6.replace(NewRe7 , ReC2)
0 d! I9 {1 ^2 d8 B. v6 } - BookConte8 = BookConte7.replace(NewRe3 , ReC6)# p2 t& k6 R& {2 O$ G
- BookConte4 = re.sub(NewRe4 , ReC4 , BookConte8)% c1 P7 a; W% z9 `- P
- BookConte = BookConte4.replace("\n\n","<br>")
- K7 `! y; g4 F! b ] d# } - print BookConte
) P2 ~( W- v* L, y* }7 V - Book_Introduction2 = Book_Introduction1.replace(NewRe1 , ReC1)* I! D$ U' d7 k
- Book_Introduction3 = Book_Introduction2.replace(NewRe2 , ReC2)6 s' W! Q" a# g7 a- M4 H. Y0 I
- Book_Introduction4 = Book_Introduction3.replace(NewRe3 , ReC3)
* Z& Z- M- X, D9 f( U - Book_Introduction = re.sub(NewRe4 , ReC4 , Book_Introduction4)- h' @; E$ N: c" b( f$ ` C+ X
- Titleid = response.url.split(BookID + "/")[-1].split("/")[0]
! c/ ?( J1 D1 S* p - Book_img = response.save['img'], #小说图片# u! n! l1 j( @/ m
-
3 I# A# A7 X+ L3 v - #insert into MySQL 小说入库
/ S% B0 e7 i7 E+ G. ~/ P$ |/ P - self.add_question(Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date) #这行可注译,数据库发布接口,方便其他系统的发布/ L9 l" m; @- m) ^4 ~# ^0 X5 u
- self.add_comment(Bookname, Booktitle, BookID, Titleid, Book_Date) #这行可注译,数据库发布接口,方便其他系统的发布
% O2 N( ^5 g: b; C - #post提交发布
1 M3 j, k/ N, h4 C/ _7 _. L; B5 i - self.add_locoy(Bookname,Cater_Name,Book_author,Book_Introduction,Book_Palabras,Book_img,Booktitle,BookConte,abover) #这行可注译,火车头发布接口,不需要可取消
: d; h: Z I6 I* Y4 ]+ K& L+ Q- G% a - Datos = { R. y0 Q/ d' s( ^) ~
- "Cater_Name":response.save['Cater_Name'],1 B! a9 c( ?4 d; p8 J; U
- "Book_author":response.save['Book_author'],
5 I% i- h1 ~8 T0 L9 D9 H0 M - "Book_Introduction":response.save['Book_Introduction'],* z; a& p' Y7 z/ b1 {
- "Book_Synopsis":response.save['Book_Synopsis'],
3 m2 l4 k3 o& u - "Book_Palabras":response.save['Book_Palabras'],
% Y. x0 v- A+ F3 q s: c - "img":response.save['img'],7 }1 ~+ Q$ {$ c% F, Z' D: [" i$ n
- }0 {9 y# o+ k! q
- for each in response.doc('.articlebtn > a:nth-child(4)[href*="/xiaoshuo"]').items():
; w; E! w+ ^7 P" f5 F* { - self.crawl(each.attr.href, callback=self.detail_page,save=Datos)
& H4 ~3 @4 q2 d' S - return {
* B7 s; J" R4 s! e+ d - "Cater_Name":Cater_Name,
. z! T6 P# u7 R* t! X - "Bookname":Bookname,
( z3 L# R1 @0 X# s1 ~0 p' m& g - "Book_author":Book_author,
: `0 q8 G( T8 M3 p7 R: }' ^ - "Book_Introduction":Book_Introduction,: j" D. a( s# N3 W/ n
- "Book_Synopsis":Book_Synopsis,
5 ? H/ M8 ]- U, b$ i# F5 H - "Book_Palabras":Book_Palabras,
& g* A% T5 X; j- _: p - "Book_img":Book_img,( g4 @. P# _" \8 x' H; |$ l
- "Bookurl": response.url,
4 x) p; M" F9 z4 j - "Booktitle": Booktitle,
2 o9 X. r- g: |8 C0 a - "BookID": BookID,
8 l' f6 ]) u" ^9 y, k) K2 ~* V6 W - "BookConte": BookConte,
; S& U v; g% m* F# \. a8 n - "Titleid": Titleid,3 E9 a0 j2 \5 O2 i9 D% Q
- "abover":abover,
/ b3 p/ y4 u+ s - # "Book_Date" = str(datetime.datetime.now()),
7 W4 X4 ~3 |' @& M9 G - }
2 P: ?- K% b3 k4 s, C# ] - def download(self, P_dir, imgDir, file_name, Book_img):! i! H3 M, E3 u9 [+ @
- if not os.path.exists(imgDir):
5 P9 p$ t4 H! X, }1 v0 l1 X) x - os.makedirs(imgDir)
5 c9 w& f1 @/ v# N2 w - file = imgDir + "/" + file_name
% U+ e; m5 O* N6 r' a: t, W j V - # print file
; Z* p2 c \# B6 U- |1 z% G/ a - f = open(file, 'wb+')
' L$ T: |/ x7 _3 p - imag = requests.get(Book_img)
1 r) V1 Q& C) k9 n- l5 _ - f.write(imag.content)" S+ B9 M* X% y$ q- C
- f.close()) u9 X. m$ e; L. Q/ m( Z
- #保存图片前
8 Z) ?! n% K; r/ [( b7 k: ?& \. u! V - def save_imgs(self,response):' d; i. {0 _1 b+ k0 {
- content = response.content4 ]4 }$ [* L; @; D& r' g1 s; c
- file_name = response.save["file_name"], J1 P$ M1 M' {
- imgDir = response.save["imgDir"]( G; W- X# E0 d' v0 i
- file_path = imgDir + file_name6 v, u# G: |$ Z, X8 s4 A* g$ O8 I* k
- self.save_img(content,imgDir,file_path)7 R6 u0 ~; y4 }! Z- X
- #保存图片
; o- _3 P( w% v! x3 h6 c - def save_img(self,content,imgDir,path):7 z8 B7 [- P" H7 c$ M7 h
- if not os.path.exists(imgDir): 5 P# P* t' w( I6 n" X8 f# D! G0 k
- os.makedirs(imgDir)3 g! s$ ?& N3 F, ^% q, n
- f = open(path,"wb" )
9 l, P) \3 {* |; X' }8 J - f.write(content)
/ ?2 u* Y' }; x* b8 x6 {1 `6 a, f - f.close()
% _7 d# }' K& a" z9 E9 X - #获取url后缀名9 @( O# W7 m% ]% |& o% Z
- def getExtension(self,url):
! Z6 ]. e- b, K& {8 U - extension = url.split(".")[-1]/ Z* b- Z- |: s' N6 `
- return extension
: v# }) i+ d& ^* P- f2 b7 k Q -
& w8 u& O5 E( ^- Q0 `2 C: u0 L - #获取图片名
! v0 ~8 j" U" n0 @! P - def getname(self,url):
1 i0 D6 N3 S/ [ N - name=url.split("/")[-1].split(".")[0]
Y' u0 l+ N6 p" {/ q8 `+ T5 u* _ - return name
复制代码 / E. j# y( A+ D }* \/ B- l
& X8 @& q+ k1 x: s8 P) X) w( _" c
|