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!0 W, S# B( e+ k7 S( j3 [' y
- #!/usr/bin/env python6 Y; y" g" x t" u' P9 i
- # -*- encoding: utf-8 -*-
3 T# h! x( @3 V2 r7 X% Q- z - # Created on 2019-05-05 21:43:11
$ x3 u2 }8 c9 f, s8 w2 K" S - # Project: XiaoShuo
8 g) l# d+ D! x$ ` -
- ^+ }) U% C4 r1 ]' u" k - from pyspider.libs.base_handler import *
4 b6 q. p1 ~! g+ Q4 Z3 {; z - import pymysql
" M7 h+ b q5 A* u - import random
6 @: S3 ?% n! U0 I: | - import datetime3 I: v# ?2 @! ?* _# \, p ~+ s! X
- import urllib2,HTMLParser,re) [% a7 U8 _0 D( \
- import os7 k$ m0 ^" p3 w2 ?* p) F1 t. A
- import sys6 n" H4 h1 Q6 E+ [$ X; j
- import re
" [9 h( A' h& F# b. l - import codecs
% A' }/ O p( |2 {3 F - import requests
, a- T" d) G8 h7 B5 W - import json
+ d# Y }! d) [8 R7 E6 s! } - 9 r7 v, R7 _& Y4 S9 I8 B
- class Handler(BaseHandler):6 T- T) N$ q5 G1 |3 u: l
- global Datos
3 ?+ w: P' H4 F - global P_dir Y; d( Q: c+ p/ c# O N
- P_dir = '/Tools/Debug/' #采集时候图片保持到本地的路径
: y( R9 {/ s- u$ X% ^, [ - global Datos) \9 r2 ~( J$ w. D4 x5 o* {
- Datos = {}
0 @+ o5 C& ^& d2 g g. w- Z - headers= {# C# s8 |/ ]( p) }6 L. q' N0 z
- 'Accept':'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',1 U$ L+ L/ F0 Y4 h3 j6 \
- 'Accept-Encoding':'gzip, deflate, sdch',* @* G( }* s8 B0 J, {$ ]) B
- 'Accept-Language':'zh-CN,zh;q=0.8',5 C- |/ c y' G; ?! h
- 'Cache-Control':'max-age=0',+ Z+ E& v' ]9 H, P: R1 ~ x
- 'Connection':'keep-alive',
% k" a2 l# j. }: {! ^ - 'User-Agent':'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.101 Safari/537.36'
, w- L2 X$ \' Y% h% u - }
9 C- [6 D: w3 v# h7 t5 R - crawl_config = {
+ k# _! o+ t6 q8 U; P, X$ F - 'headers' : headers,9 Q z7 V1 ^) X2 V j
- 'timeout' : 300
6 a# k1 e8 M$ q( x q# A - }
) p3 q, r' Q# E% B - def add_BookFile(self,Bookname, BookIDs, img, Locaimg, Book_Dates):
7 i& K+ @8 u# o6 _ K! N - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")
; g1 w- J4 b# l l$ F- C5 W3 O - try:* v. @) {' ]) A$ n
- cursor = db.cursor(). d0 x3 k& T/ Y9 T7 N
- #注意此处字符串的占位符要加双引号"%s"
5 F6 V5 o0 v$ j+ S - sql = 'insert into BookFile(Bookname, BookID, img, Locaimg, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, BookIDs, img, Locaimg, Book_Dates);
' O7 w* \, ^; f - # print(sql)
2 V: i8 ]* L: T% A - cursor.execute(sql)
7 P5 ?4 L& \6 {' V3 x' d - 3 F# [+ I* d+ a1 W
- #qid = cursor.lastrowid
3 V' D' D3 W4 K7 I+ h - #print(qid). y5 x+ e1 l. b. [
- ) |/ n& c$ \9 _# q- a8 N9 G1 E& e d8 R
- db.commit()
' L- t' ~" z% q# L7 t - except Exception as err:( P2 f2 Q% x# x, D) |3 z! M
- print("Error %s for execute sql: %s" % (err, sql))! J' w: {) d" m8 K
- db.rollback()
! ~7 k/ z1 K% ]5 N6 ? - def add_comment(self,Bookname, Booktitle, BookID, Titleid, Book_Date):# ? h# \5 G. O: d- J7 ^6 Y/ I; h
- db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")
y: `( {$ K7 ? q - try:3 H' K5 o$ X4 J% ~
- cursor = db.cursor()
1 ]6 D* I% T3 s3 j$ D+ x - #注意此处字符串的占位符要加双引号"%s"4 t9 j( b, S v3 i4 ^0 S) A
- sql = 'insert into BookTitle(Bookname, Booktitle, BookID, Titleid, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, Booktitle, BookID, Titleid, Book_Date);+ x9 A2 _: Z9 n# M9 J- |
- # print(sql)
1 l2 a) C5 d7 d. a% d/ k. K - cursor.execute(sql)/ ]; O! o/ ?' c. K! e
- / j8 `/ h9 |% a5 C% E
- #qid = cursor.lastrowid
- _* z0 P5 _; X9 p8 X5 m% n9 q; s - #print(qid)
$ p# u! @2 B2 k - 2 F2 q# ^6 A$ k3 j. e' M g
- db.commit()
9 J) y! |- c5 u8 V - except Exception as err:
' I7 I! K& U6 o! y1 o. ` - print("Error %s for execute sql: %s" % (err, sql))& Q ~* \* C3 n6 z! M* S2 l
- db.rollback()
* O7 f W( a1 i6 W - def add_question(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date):
- \# v" h: _' f$ M! z0 } - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")3 m# p4 w; [. k4 f6 B6 t/ l, K
- try:
/ a0 _3 H7 u- Y: b5 B - cursor = db.cursor()
* W2 G- x" ]* m - #注意此处字符串的占位符要加双引号"%s"& y% u0 V. g" n9 k8 |
- 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);! a9 M1 A7 y- X$ j# \
- print(sql)
0 K5 y6 ], H9 M& n - cursor.execute(sql): g0 n5 B9 f, L/ ^) `6 @
- print(cursor.lastrowid)/ c: W& e7 `) X" f/ g) |
- db.commit()
# m- m" r& j5 G: M - except Exception as err:- A( G0 j( [; P$ D1 H0 t" T+ c
- # except:3 F) Q3 S1 K3 l' ~* ]
- # print('Failed')
+ W6 g" X5 g$ `; l& V& R+ D - print("Error %s for execute sql: %s" % (err, sql))+ e8 N7 G# F/ B* B+ J' Q1 `/ x
- db.rollback()
8 E8 L# m9 w' H7 [1 o - 5 j7 [0 u: W7 @5 _- V1 r; w3 g
- def add_locoy(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Palabras,Book_img,Booktitle,BookConte,abover): 9 m( n+ t) b. @. x: x8 ~' t
- reload(sys)! O) y8 y/ G$ x, _: [, g
- sys.setdefaultencoding("gbk")
: b$ u" V: B5 Q& U - locoy_url = 'http://www.******.net/locoy/?my=book' #697火车头发接口地址
# d+ y, o9 l, W7 v } - locoy_data = {5 g: I/ v3 ]0 y
- 'my_u':'用户名', #后台用户名
8 _2 e! ~3 F7 q4 n3 x2 l - 'my_p':'密码', #后台密码5 i) c5 j% R/ t9 K- C3 `
- 'subject_669977_net':Bookname.encode('gbk', 'ignore'),2 z% v( z. y7 ~& j/ G+ N
- 'caid':Cater_Name.encode('gbk', 'ignore'),
6 y4 K5 q$ h: t2 A - 'title_669977_net':Booktitle.encode('gbk', 'ignore'),
$ y$ i/ i0 C s$ m X - 'article':BookConte.encode('gbk', 'ignore'),. w; P6 @- C. @7 ~% r" u
- 'author':Book_author.encode('gbk', 'ignore'),
( ^' D- {! ]: \9 o+ B' ^ - 'ready_1':Book_Palabras.encode('gbk', 'ignore'),
8 ^" g0 {+ x8 v2 \" |* u( N - 'thumb':Book_img,$ F) d7 ]) b% @
- 'content':Book_Introduction.encode('gbk', 'ignore'),
* q" h! @& U4 n. I5 C - 'abover':abover.encode('gbk', 'ignore')
" H, g& c' R2 e - }" `2 f3 [' i# Z) }1 Y) }+ g
- res = requests.post(locoy_url, data=locoy_data)
- D' i, h% T! \3 K7 h - print res.text: {0 S' n5 C( g2 Z" A: h& k
- print res.content. f J( f8 H7 x H- V
- # print Dsd+ l2 d1 T9 f/ G
- return res/ `; E: k8 I0 Y
-
* Y( X$ [2 t* M5 @ - def __init__(self):# V" u1 O8 V8 y; @" r5 D9 \
- self.base_url1 = 'https://www.****.cc/'0 B& V" @" H* j5 R% r) n
- self.base_url2 = '/') g# o$ V" R& U# N- `
- self.CaterId = []8 o7 k, C4 f9 R3 N" Q3 T
- self.CaterIds = ['xuanhuan', 'wuxia', 'yanqing', 'lishi', 'kehuan', 'kongbu', 'nvsheng']& M g [ H, R4 j4 v* M+ L$ \2 h
- self.page_num = 11 L! m2 n+ J7 `) ]- P5 m
- self.total_num = 200 7 G7 I- `& }, H' I P4 B- Q L
- 3 E& `; I8 Y. ?) C+ v
- @every(minutes=8 * 60)2 a) h5 K4 F8 }% o" j3 E) c
- def on_start(self):
2 g- w! ^; T" V3 ^$ v: f - global Cater_Name; u1 @) P+ z& o
- Cater_Name = []1 t. K- o) I3 Q9 t% V* @
- while self.page_num <= self.total_num:
4 q* I. [; g o4 x' U/ W7 f4 Z: L - for self.CaterId in self.CaterIds:
, C9 S7 I: _1 @ - if self.CaterId == 'xuanhuan':7 C4 c n$ v4 ]* i
- Cater_Name = '玄幻'
+ e D) J# {6 t; r$ q6 ^( E1 a - if self.CaterId == 'wuxia':* ^& X2 u R! s1 @
- Cater_Name = '武侠'
, z/ d! b' i* {! n/ y1 {* C/ \; p - if self.CaterId == 'lishi':
( l# J; m# h5 m5 p; r" Q' p - Cater_Name = '历史' # T* W2 ?- r7 ]4 u8 K
- if self.CaterId == 'yanqing':
: U' G: Z0 ~/ f2 B) E( T - Cater_Name = '都市' / ? b8 R3 F& G) T7 T6 b3 T6 g# C
- if self.CaterId == 'nvsheng':
5 {% h/ ]5 ?7 q7 |9 I3 X Z6 C - Cater_Name = '都市' 5 H. n$ j: L1 {% s3 j
- if self.CaterId == 'kehuan':
( D$ k, p) ^ d$ n& g# M - Cater_Name = '科幻' ! H! y9 }: y! K+ Y5 I* l
- if self.CaterId == 'kongbu':! d& n7 ~7 x! [
- Cater_Name = '游戏'
% J" N7 O6 B7 u" o; z; W3 p - print self.CaterId5 V' u2 D' U; O) c3 }0 F
- url = self.base_url1 + str(self.CaterId) + self.base_url2 + str(self.page_num) + "/" 6 e U/ T5 ~2 G4 ^: Q/ f
- self.crawl(url, callback=self.list_Caterg,save=Cater_Name)- r# `* d) N) I& W
- self.page_num += 1 ! b, m6 y) r0 x/ G B, Q* v& a
-
3 p* h- w0 W, Y: b; `' b - def list_Caterg(self, response):
# f$ f1 G z& `+ { E" z - Cater_Name = response.save
6 [" o& w# x7 d - for each in response.doc('.pic-list a[href^="http"]').items():. Q! L5 D+ g- a) l7 C( q
- self.crawl(each.attr.href, callback=self.list_Caterg_detail,save=Cater_Name), L7 B7 A V/ X1 C8 i3 d% v
-
# R& q; ^# |, R& r2 \4 Y - def list_Caterg_detail(self, response): M8 [* O8 p( u+ C# K- @( C
- Cater_Name = response.save
' m: z; k0 J6 ~/ b% ^+ z$ A - # print Cater_Name! j. u1 v; k0 ]0 k; D
- Bookname = response.doc('h1').text()
, _& \ ?( o2 j0 f9 R2 _& h - print Bookname
?# U5 l- P& P* s - Book_author = response.doc('.authorname > a').text()
: J+ q* q- K' z6 w) u& b. Z C, I% q - # print Book_author
4 e: u1 F6 V: o; P3 G - Book_Introduction = response.doc('.book-intro > div').text()
5 W7 J, T+ _) d, d# O9 A) S - # print Book_Introduction
) D2 i- q2 y$ V+ x7 x - Book_Synopsis = response.doc('b').eq(1).text()4 j( g6 F/ [7 a5 z8 p
- # print Book_Synopsis
2 _& a! I# `, u' a4 Q: h - Book_Palabras = response.doc('.booktitle p').text().split(' ')[1].split('|')[0]
7 a0 I3 G, I3 J4 y& P3 d - # print Book_Palabras
5 }( j t0 g4 ^- P - BookIDs = response.url.split("xiaoshuo/")[-1].split("/")[0] #小说ID
; s# G, M& ?$ J" e" s6 u/ g - # print BookIDs' i: Z( b: g/ I3 B
- Book_Dates = str(datetime.datetime.now())
) N, u/ Z! d0 X - for imgs in response.doc('.bigpic > img[src^="http"]').items():: O. N; b* t0 q$ _7 f
- img = imgs.attr.src6 c" l! S1 ~+ B: G5 c7 y& M9 W
- print img% Z! H" G8 K" Q" [1 ?+ D; V
- #小说封面下载7 q8 b( L( l8 ]/ P! X- ?$ O$ V
- extension = self.getExtension(img)) T- Y/ {- y! r- u: ^% o/ L
- name = self.getname(img)9 n2 o# E @5 l+ q
- file_name = name + "." + extension8 C- W$ U: j" D
- imgDir = P_dir + name
- W$ r# A2 P. q' ?8 m - Locaimg = imgDir + "/" + file_name3 a$ Y+ q0 D8 y: ^% j1 h
- print Locaimg
& J6 B- s$ }9 L0 x0 x' i+ u - if(self.download(P_dir, imgDir, file_name, img)): #这2行可注译,图片下载到本地6 a4 c" l: S& P. u8 x3 ^
- print('attachment url is ' + img) #) n" i$ X) Z7 Z: V# q# c% z4 R$ Q1 M
- Datos = {
+ w3 m- x" k3 Q: ~ - "Cater_Name":Cater_Name,
2 a& a( K2 }& _( G: ]" x/ Y - "Book_author":Book_author,! {1 s7 j3 a- K/ [
- "Book_Introduction":Book_Introduction,- r/ `8 c# e2 B- K
- "Book_Synopsis":Book_Synopsis,
, l, T! h% S0 K2 H* m - "Book_Palabras":Book_Palabras,: c' B' j' h" M& ]2 `! c+ v
- "img":img,
+ [4 H: Q1 Q. C4 f: l5 I) M( t: e - }, p C# x. g8 E% Q
- 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. t" h, Z R5 _6 Q! T
- for each in response.doc('div[class="bookbtn-txt"] a[class="catalogbtn"]').items():: w* b4 _. |0 l( t% b: F4 [2 G7 y
- self.crawl(each.attr.href, callback=self.index_page,save=Datos)
. Y3 ~7 X. j, Y/ l* q9 n: N -
' _ R* u( ^9 e0 Z( V- r - @config(age=8 * 60 * 60)
$ U* Z0 [8 |, J! ?/ U. H. z - def index_page(self, response):
$ ~2 P$ u5 \; U$ N - Datos = {. S5 w6 N9 J. P5 B9 t Y
- "Cater_Name":response.save['Cater_Name'],
0 @# A' ~% h* J - "Book_author":response.save['Book_author'],
; D0 O; S' B9 V- U. C - "Book_Introduction":response.save['Book_Introduction'],
+ t, U6 H$ B; g. c - "Book_Synopsis":response.save['Book_Synopsis'],, T( l, {0 p1 j2 Z# m! z& q) q
- "Book_Palabras":response.save['Book_Palabras'],7 T4 `- U8 P5 s
- "img":response.save['img'],: m( W. P" z1 G: `+ B* N# j! b
- }
# T- y T9 x2 C: o: m - for each in response.doc('.chapter-list li:first-child a[href^="http"]').items():. G$ [# o' A1 [/ [5 M, `
- # for each in response.doc('.chapter-list a[href^="http"]').items(): % t: _7 J: U1 I3 Q5 ^) W
- self.crawl(each.attr.href, callback=self.detail_page,save=Datos)
2 a8 B" e3 R+ D% P) { - @config(priority=2)
7 j/ u6 k$ f! ~ - @catch_status_code_error/ y1 i/ O5 H- [5 W6 L6 w
- def detail_page(self, response): ( i/ Q$ r0 d7 n- A
- NewRe1 = u'哈书'
: V- v: d& U9 `; z: P! z - NewRe2 = u'huhjsd.CC'
% T) U d6 f4 N - NewRe3 = r'^\\n\\n'
$ {$ g" V% c2 h: e- @# k - NewRe5 = u'小说网'- _& e9 S/ R; K9 o$ Z2 G
- NewRe6 = u'fgdfgf'
) g8 X8 P; W) N1 R0 Z. k1 ~ - NewRe7 = u'fgfgf'
- D% N0 n+ s1 @; {7 `9 G/ ^ - NewRe8 = u'ffhgf'
5 |# R5 j/ P ~' U - NewRe4 = r'[\f\t\v+\.\{\(\)\}\!\/_,$%^*(+"\')]+|[+——()?【】“”!,。?、~@#¥%……&*()]+'' W* A$ z t8 R; ]: M i
- ReC1 = u'静思'& ` g" A ?/ {2 h) }
- ReC2 = u'aghgf.com'. Z! p8 g" k/ @3 E! s' x/ ? W
- ReC3 = u'aghgfh.com'
! J' q k3 q7 T* I, j# a7 P - ReC4 = u''
" D* v: l# S* e. Y% N( n. b - ReC5 = u'文学网' Y+ l+ r. f3 A, L7 }* B( L
- ReC6 = r'<BR>'
0 q7 O( g8 F( Q7 o& \9 i* k8 @5 I5 v - Bookname = response.doc('.readlocation a').eq(2).text() #小说名称
, H3 b# o9 N F7 S& z' N - print Bookname
0 G$ n: p. Q& C/ {! w \5 d9 i - Cater_Name = response.save['Cater_Name'] # 小说分类0 o1 Y5 C% e+ F2 a" ?$ J1 n1 E1 e+ L
- Book_author = response.save['Book_author'] #小说作者 n0 q6 @% s8 {" q, q) T& \
- Book_Introduction1 = response.save['Book_Introduction'] #小说简介1 [/ D% F1 o- F3 J: M) T5 U
- Book_Synopsis = response.save['Book_Synopsis'] #最近更新
+ O% R8 _" c6 \9 L* q - Book_Palabras = response.save['Book_Palabras'] #小说字数+ B' ?$ Y" `) K
- Bookurl = response.url #小说网址
( g2 L1 X5 c# O) V) ^+ i; Y - Booktitle = response.doc('.article-title').text() #章节名称' j$ V1 A' P9 ?2 I0 j, {) e
- BookID = response.doc('.readset-r span').text() #小说ID) z' o* [, p, b. |# P5 r
- BookConte1 = response.doc('.article-con').text() #小说章节内容
" ~, t* ~* i5 L( U$ M) P - 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)7 c8 w2 a/ r1 k# b3 S
- Book_Date = str(datetime.datetime.now()) # 采集时间
5 f# B3 ]4 k5 K# Z" V - BookConte2 = BookConte1.replace(NewRe1 , ReC1)8 k" }) b! p7 R: @+ i
- BookConte3 = BookConte2.replace(NewRe2 , ReC2)( A. J8 v/ j7 V1 {
- BookConte5 = BookConte3.replace(NewRe5 , ReC5)
6 Q' ]+ }2 v, _7 V7 y0 l - BookConte6 = BookConte5.replace(NewRe6 , ReC2)2 E2 L- D5 N1 ?, [& z
- BookConte7 = BookConte6.replace(NewRe7 , ReC2)
# k* M9 j, T% b8 p4 P' B - BookConte8 = BookConte7.replace(NewRe3 , ReC6)2 k4 z2 ]& F- v6 R& w |
- BookConte4 = re.sub(NewRe4 , ReC4 , BookConte8)
+ S, T1 |3 k/ P/ ?" i# s' O" ? - BookConte = BookConte4.replace("\n\n","<br>")
8 b, U6 s. O1 j7 A* ? - print BookConte( ]( e$ X0 u& k% N1 _& u" d- @
- Book_Introduction2 = Book_Introduction1.replace(NewRe1 , ReC1)
# j( @% x q" p - Book_Introduction3 = Book_Introduction2.replace(NewRe2 , ReC2)
8 Q$ H# Z9 U$ b+ n$ u1 n - Book_Introduction4 = Book_Introduction3.replace(NewRe3 , ReC3)
+ U, W- B- S5 @0 R2 X - Book_Introduction = re.sub(NewRe4 , ReC4 , Book_Introduction4)
' H" ?2 l1 }9 ?' R* @& d - Titleid = response.url.split(BookID + "/")[-1].split("/")[0]
, k8 ]" G* K* {6 y# F4 s2 E9 d1 ] - Book_img = response.save['img'], #小说图片9 @/ i: q* m% D9 k3 ~1 g
-
y9 ~: X1 m- v0 X - #insert into MySQL 小说入库0 Z2 V9 Q, s. k/ Y& J
- 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; V* X3 Y8 `8 _
- 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
4 u, I+ J, V1 \ - #post提交发布$ I* E( C+ u. h: W
- 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* l3 X1 ~+ M V1 `5 u5 X
- Datos = {
: J6 M1 ~; x$ o9 Q% B - "Cater_Name":response.save['Cater_Name'],. a: m7 f S1 a& N# E# I
- "Book_author":response.save['Book_author'],
' t- P7 c2 l) o1 k9 W2 @ U( j! B( D - "Book_Introduction":response.save['Book_Introduction'],0 h! t* ?6 s* e- d
- "Book_Synopsis":response.save['Book_Synopsis'],6 |& A4 M- j$ C! |; ?" J( D
- "Book_Palabras":response.save['Book_Palabras'],
3 B7 ?/ U/ I7 C6 T - "img":response.save['img'],+ `' Y: K0 [( p9 J0 m* x
- }0 m& I3 W" R, n5 q
- for each in response.doc('.articlebtn > a:nth-child(4)[href*="/xiaoshuo"]').items():! [" Q9 h4 q$ h7 C
- self.crawl(each.attr.href, callback=self.detail_page,save=Datos) 3 o8 F: {$ k1 q) ?. W4 y0 W
- return {
6 Q) [; Q- a$ y6 B6 R1 C - "Cater_Name":Cater_Name,
0 t7 c9 i' k5 R/ ^) j( r9 S! ?( w8 v - "Bookname":Bookname,
) Y/ c7 l5 i1 s, q+ R7 ? - "Book_author":Book_author,; i8 R {& O$ n. k, S0 _
- "Book_Introduction":Book_Introduction,
, c7 v, Q& h. I" S ^ - "Book_Synopsis":Book_Synopsis,: u `% J @9 N) Q+ W
- "Book_Palabras":Book_Palabras,
$ R- ~# z- E1 L, H: A - "Book_img":Book_img,5 C$ ]7 ?2 S7 N n) `# N
- "Bookurl": response.url,
+ p8 R# m; B4 X% S) ^& w/ W* N7 r8 ? - "Booktitle": Booktitle, s- S) y! f( k d. X5 m4 W
- "BookID": BookID,
) G8 @3 h8 |3 n7 U8 `1 F- }1 U - "BookConte": BookConte,
; B# _5 b% b; ]# T8 q - "Titleid": Titleid,0 h" s' H0 Z: w4 d& X8 h
- "abover":abover,
$ j3 }. g$ Z6 M7 e7 q - # "Book_Date" = str(datetime.datetime.now()),% K" K& F+ [3 q
- }
) L% L- R* u# c" e - def download(self, P_dir, imgDir, file_name, Book_img):
. m$ \1 Z! c7 z/ C* @ - if not os.path.exists(imgDir):
' o9 F& e# I/ w8 I - os.makedirs(imgDir)
$ B* W3 N6 G5 R% i" \3 r - file = imgDir + "/" + file_name, X! S6 B. @, Q% d" J1 Y
- # print file
* ? u$ |4 z7 n' f1 f3 p - f = open(file, 'wb+') ?' V& v9 h" R
- imag = requests.get(Book_img)
" q! l. B, a; u8 q' C& { - f.write(imag.content)5 ]- O! m8 v& j4 Y7 R
- f.close()
" e0 R# Q9 | @. ^$ K - #保存图片前
- w+ I+ `) x) j. M: \# S - def save_imgs(self,response):
) Z, M; u8 G' a) ~# @2 R1 ] - content = response.content0 }+ v9 p4 ^- } f1 o+ P ]
- file_name = response.save["file_name"]
" M# N& V7 |+ Y% B$ V) i - imgDir = response.save["imgDir"]
' Z% b; t, Q( G# W% [- N+ n2 j - file_path = imgDir + file_name. C6 }9 \& E0 m' H; x. \
- self.save_img(content,imgDir,file_path)( ?2 K, u5 L3 z* R& z9 f1 r; k
- #保存图片4 u+ m Q' z; _6 \
- def save_img(self,content,imgDir,path):
7 q" Q) a( Z9 Y' @ m1 ~ - if not os.path.exists(imgDir):
/ }1 \; u0 ~ y3 j2 s C - os.makedirs(imgDir)& w4 l+ P" W5 S- w! \$ d$ S" u
- f = open(path,"wb" )) R+ F% G1 A% S o$ W
- f.write(content)* q1 w9 ]! C: k$ H
- f.close()
7 y9 w- Q. ?+ J; n: r - #获取url后缀名
- ?" N# |! T0 h8 Y - def getExtension(self,url):
( u8 F1 M1 Q, [" m - extension = url.split(".")[-1]
' ]$ S5 r2 v( @ - return extension
: E+ i+ B; b( W - / r4 F9 q: k- Z+ P# b+ k
- #获取图片名
8 U( H4 B. t# {, r* w: R( j - def getname(self,url):
# x2 ^3 t4 O, g' \' } - name=url.split("/")[-1].split(".")[0]
( E+ D9 {; K/ B - return name
复制代码 5 ]- ~/ e# r3 Y0 d. r* ?
) O+ ?! |& f" C, J. x4 q+ r+ r |