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!! e6 Y& d% W& h( }0 q* T U
- #!/usr/bin/env python
+ x! d: X9 C* \' a - # -*- encoding: utf-8 -*- L. b3 x+ t7 ^: `& p B
- # Created on 2019-05-05 21:43:112 W9 C; W+ }3 f- n
- # Project: XiaoShuo
: n% y/ {7 ] u, T - I4 j4 T0 ]; l3 x8 Z) _
- from pyspider.libs.base_handler import *& ~- ]$ {9 J9 [% D* O5 r
- import pymysql% s( |0 P2 F9 C. P3 N$ I
- import random% {3 C# |4 H% @
- import datetime
8 I6 Z" C% ~4 I7 f3 B! Y - import urllib2,HTMLParser,re* d# Q( o$ J+ f9 |
- import os
, r- N/ n5 ~- l1 i& i - import sys
0 ?9 i% C' E9 {) M7 ]9 @ - import re
0 |+ @" S& T4 `8 \! y- i - import codecs, I3 _+ S8 q! I& h( l
- import requests# N% k% x2 F1 S' N* b
- import json
. }% X) X* M$ b- {' ]! @. h - + u9 `- [) r. z8 F8 M) p
- class Handler(BaseHandler):3 N2 \* S2 F9 u9 O
- global Datos
0 f4 ~9 ?/ ], l4 w d - global P_dir $ s2 o/ {6 D) _3 i$ z6 K: P0 } O
- P_dir = '/Tools/Debug/' #采集时候图片保持到本地的路径8 p9 _* i! p3 u: ]/ X8 G' V
- global Datos
8 P' J: _2 i6 A$ k8 b9 H - Datos = {}
6 l; R8 `* M2 l! \ - headers= {0 |7 L, f$ i( N F; n; }
- 'Accept':'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',0 N" r: ?/ F$ d% [, Q6 |/ I- u
- 'Accept-Encoding':'gzip, deflate, sdch',# o! c8 m! O7 l: B
- 'Accept-Language':'zh-CN,zh;q=0.8',) s" ]- u$ e9 x* F* z
- 'Cache-Control':'max-age=0',6 Y6 [7 G, C* o7 C" ~
- 'Connection':'keep-alive',1 c5 y& i: m# K0 h/ X# j Q
- 'User-Agent':'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.101 Safari/537.36'$ K* ~( O& B' U* r
- }
8 k9 B1 ?* y% e1 Q! h5 H - crawl_config = {- A- [( Q) a! Q4 \ [2 q: C
- 'headers' : headers,4 x k3 k( n* ]) Y2 V
- 'timeout' : 300
1 R. m7 W3 f+ w$ P4 v; O - }8 w' F' T7 @' u, m! b+ {
- def add_BookFile(self,Bookname, BookIDs, img, Locaimg, Book_Dates):
+ U. k5 H" y5 z2 i - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")! ?2 ` U1 j! l6 [4 c% M) u
- try:2 K( N. Q0 E, y% s2 Q# u. n8 y' {
- cursor = db.cursor()
8 L2 ~% N; {8 l. z - #注意此处字符串的占位符要加双引号"%s"1 w3 `% M. C' [" e; M' [
- sql = 'insert into BookFile(Bookname, BookID, img, Locaimg, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, BookIDs, img, Locaimg, Book_Dates);
: y9 ]$ ^2 O) j- I - # print(sql)
1 X3 z4 p' M7 J; t5 ~% S7 l7 g7 x - cursor.execute(sql)
% N2 A% A! j: U* o. A -
; ]! [& t2 h2 G% f - #qid = cursor.lastrowid) _2 b5 U5 x" q. C
- #print(qid)
/ k( F- h* K5 ]/ ?. D3 D- \, { -
" ^2 K, K, y9 @; ^9 @ - db.commit()
6 m q* {( s: i - except Exception as err:
0 K- W4 u- B7 `1 n. F# I - print("Error %s for execute sql: %s" % (err, sql))5 n( v: @) e# n) O% j I1 D
- db.rollback()- n! i9 Z8 q* a( {! u( e
- def add_comment(self,Bookname, Booktitle, BookID, Titleid, Book_Date):
% t; _! M" S. a8 s' c - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")+ Q: U( s d( c+ E# l1 F
- try:( Y; n/ l5 E5 } g5 b* X7 j
- cursor = db.cursor()
% p/ p& x- C3 E |3 }8 X% t - #注意此处字符串的占位符要加双引号"%s"
( t; n7 ]: L2 p0 F - sql = 'insert into BookTitle(Bookname, Booktitle, BookID, Titleid, Book_Date) values ("%s","%s","%s","%s","%s")' % (Bookname, Booktitle, BookID, Titleid, Book_Date);" O* S( B2 o* k
- # print(sql)' Y" |' u; {2 k' y* k2 i7 F) k
- cursor.execute(sql)& l0 ~/ I) h$ u# D0 k
- * n" v( j- S: J/ U
- #qid = cursor.lastrowid
& G1 z8 R! c: ?4 Y+ q* U - #print(qid)
8 U1 H$ G+ n- l% V+ t3 E; ~! y -
1 |. C; I" w- K- p" d - db.commit()5 ] U7 d' f* u
- except Exception as err:
- L) B9 P, u$ c. ]4 N! q$ l* l - print("Error %s for execute sql: %s" % (err, sql))" ^ \1 W# U2 r9 U
- db.rollback()% l' m3 }1 c6 M* |0 E
- def add_question(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Synopsis,Book_Palabras,Bookurl,Booktitle,BookID,BookConte,Titleid,abover,Book_Date):
6 ~% A+ J* a7 k8 {( W% b - db = pymysql.connect(host="localhost",user="数据库用户名",password="密码",db="数据库名",charset="utf8")% m. k% y! y- ~
- try:
) D) a0 [& ]% S+ p# \ - cursor = db.cursor()
! T$ | @% D$ O7 C; q - #注意此处字符串的占位符要加双引号"%s"3 e0 k$ @2 ~+ P; k# z1 e( d
- 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);& \( N- h& M3 D& Z3 C" p+ K
- print(sql). p- s; I* |& A7 b- f
- cursor.execute(sql)/ v+ e3 b% x8 a0 q
- print(cursor.lastrowid)" S: J6 g9 J0 Y1 }
- db.commit(): S: m( x2 `# b1 Z9 d' A
- except Exception as err:
( r9 L9 j9 E+ L1 C$ P, _ - # except:% c6 v7 ]& q8 `0 W. K3 K
- # print('Failed')' z8 _' ~3 u: \) e7 U) t
- print("Error %s for execute sql: %s" % (err, sql))% R/ ]+ a& [. l u6 G1 A% y
- db.rollback()4 l: d! f1 W$ r( e9 ^# o1 e
-
2 X) F7 _0 r" ^% D; Y - def add_locoy(self,Bookname,Cater_Name,Book_author,Book_Introduction,Book_Palabras,Book_img,Booktitle,BookConte,abover): % O7 W+ g6 V# t9 E
- reload(sys)
, r4 I. Z5 s6 n7 m8 |1 p1 l1 E9 e0 V - sys.setdefaultencoding("gbk")
+ ~9 P3 M% X4 t u( k4 V9 D5 t! [ - locoy_url = 'http://www.******.net/locoy/?my=book' #697火车头发接口地址3 {7 Q! ?" O3 S l6 {- |
- locoy_data = {1 g- ~( K2 {2 @) Z
- 'my_u':'用户名', #后台用户名
; a+ h* g8 ~4 `- N! Z/ }* M - 'my_p':'密码', #后台密码
. [, p" ~2 ?9 w: U& V+ Y - 'subject_669977_net':Bookname.encode('gbk', 'ignore'),, ^- x4 v: b6 E
- 'caid':Cater_Name.encode('gbk', 'ignore'),
7 d! W% y Z6 e3 u x1 n: v# `8 G - 'title_669977_net':Booktitle.encode('gbk', 'ignore'),8 f: a" T' Z4 g' O1 t
- 'article':BookConte.encode('gbk', 'ignore'),
9 y+ p( R( r& J+ d+ c$ d) t; l - 'author':Book_author.encode('gbk', 'ignore'),' |( Q" ~* o2 a! t3 `% H: X0 O
- 'ready_1':Book_Palabras.encode('gbk', 'ignore'),4 C, c2 e0 h @4 q$ Z. w- ^1 T7 P
- 'thumb':Book_img,
7 G7 K7 S/ c* p7 k# g" q - 'content':Book_Introduction.encode('gbk', 'ignore'),
/ E2 S7 c- \6 M) v ~ - 'abover':abover.encode('gbk', 'ignore')
/ M7 |% t$ l( s6 B+ M9 T; U! Y - }
: J `. M6 |& V - res = requests.post(locoy_url, data=locoy_data)7 l8 q& `7 c. h% J
- print res.text, \# X4 `9 n- L" f
- print res.content5 T2 v9 W- y6 ?) S, x" V
- # print Dsd
) M# @ u4 L6 Y- e" @, b% | - return res* {/ s* S5 b! n5 P `0 k0 r6 b/ T2 Z
- 2 l3 F, k2 E9 I3 i' Y$ O" R
- def __init__(self):( G7 t( w2 b8 g% L/ R6 K2 Z$ [! t! m
- self.base_url1 = 'https://www.****.cc/'' A: V# @+ L1 M; A% Q
- self.base_url2 = '/') B* B. [3 j/ r$ f l$ L& r
- self.CaterId = []: S8 Q9 H$ q3 d) {& [# v! U) f
- self.CaterIds = ['xuanhuan', 'wuxia', 'yanqing', 'lishi', 'kehuan', 'kongbu', 'nvsheng']
' y' ^7 }6 e) S. |8 M - self.page_num = 1
3 t# y8 J/ ]6 f1 W% R - self.total_num = 200 ( R& ` f0 ?" F" b
- / F$ e- ]+ {# Y5 d
- @every(minutes=8 * 60)3 L7 l$ X4 k" X8 t, w
- def on_start(self):% D1 L( S0 @. b( y# N+ a
- global Cater_Name
* p9 \9 |5 W2 m% n1 g8 E b$ G; L( w- j - Cater_Name = []* A! h; e$ _# C# b
- while self.page_num <= self.total_num: 8 r" g; L+ E4 ?! n( s# A, _6 |
- for self.CaterId in self.CaterIds:
8 _' c% e9 P$ J& V& b2 P - if self.CaterId == 'xuanhuan':
, R: W, u8 @0 ? - Cater_Name = '玄幻'
4 E, _8 a1 m1 n4 Q+ ] - if self.CaterId == 'wuxia':
# ]1 E" a# D8 k& y+ @3 s - Cater_Name = '武侠'
$ A! ^+ D- s6 V q. } - if self.CaterId == 'lishi':6 F9 U+ g. a( Q/ c/ q# I. [
- Cater_Name = '历史'
$ A, Y3 v. X+ c" p% C - if self.CaterId == 'yanqing':
4 e0 o$ N0 F) }; J0 f5 J9 u - Cater_Name = '都市'
7 H; i+ P ?$ x( J - if self.CaterId == 'nvsheng':
( I0 T4 |) G: k - Cater_Name = '都市' # F' e3 h2 r; s' W
- if self.CaterId == 'kehuan':9 W5 N0 L) k1 O2 g) T
- Cater_Name = '科幻'
) [* p! v' C2 N, S) m4 ? - if self.CaterId == 'kongbu':% O& F/ y- ^! h: r; @* K
- Cater_Name = '游戏'
* \) Q. a- `* ?/ m) N% M/ s& x- g - print self.CaterId
5 o* k8 R! U% Y( b! x3 [# O4 O - url = self.base_url1 + str(self.CaterId) + self.base_url2 + str(self.page_num) + "/"
, i: v' \! J0 J$ `9 a - self.crawl(url, callback=self.list_Caterg,save=Cater_Name)1 ~' {2 j1 }$ T
- self.page_num += 1 : t- l, c& E0 r# `+ j4 C: C
-
- Y @5 V5 O' \; D: b - def list_Caterg(self, response):
: l! p. g; Z! [0 d/ ]2 w - Cater_Name = response.save
! _' H4 e$ `! P0 }1 e - for each in response.doc('.pic-list a[href^="http"]').items():
, L. ^/ D# N }# ^- {( a) w - self.crawl(each.attr.href, callback=self.list_Caterg_detail,save=Cater_Name). n7 R; Y' h+ a; M
- & ~3 ^) m0 I& c0 w. I6 c
- def list_Caterg_detail(self, response):% k% D' p* H5 u+ m2 ]& B3 E$ u
- Cater_Name = response.save
. `% }1 C, ^* K2 |$ x: S- V1 d - # print Cater_Name
* g- {0 r' t! E1 r ?! s5 D( i6 j - Bookname = response.doc('h1').text()5 j5 R9 x6 J. E5 n# M
- print Bookname* j% t3 w% q' o9 f/ [& l2 G) k2 {
- Book_author = response.doc('.authorname > a').text()( l& ]# q$ [8 | Q8 a
- # print Book_author
& w! \# t: b* v& }/ i; O6 v/ s - Book_Introduction = response.doc('.book-intro > div').text()
. c( H/ W- t9 f F - # print Book_Introduction
: @* d$ t' i+ f. G - Book_Synopsis = response.doc('b').eq(1).text()$ M8 m0 I4 S: A) @- A- r, t8 f% T4 u
- # print Book_Synopsis
! f/ A" F! |$ I% n& P+ e - Book_Palabras = response.doc('.booktitle p').text().split(' ')[1].split('|')[0]
! s1 A k9 q: v - # print Book_Palabras1 R& O( P5 p! z+ b% @9 C! c3 g
- BookIDs = response.url.split("xiaoshuo/")[-1].split("/")[0] #小说ID
7 Z( Z0 g: B8 {/ z8 q' g7 e - # print BookIDs5 |9 U! k& ]3 g" g/ j! C: m: t7 t7 z
- Book_Dates = str(datetime.datetime.now()) + G: r9 J% \7 j9 f' x a; l
- for imgs in response.doc('.bigpic > img[src^="http"]').items():
8 A3 _4 y& `7 f! u' a - img = imgs.attr.src
8 R( y5 w, C% t) j9 ?# c" a- x - print img' e% h! w( W9 E% o7 v9 v" Q& t
- #小说封面下载
* P' B) B0 }, g/ x8 M, l" a - extension = self.getExtension(img)
: H. A0 d- ~, [ z - name = self.getname(img)& e3 y" S9 u& Y8 z
- file_name = name + "." + extension
! Z& x& g" B, d: y. [ - imgDir = P_dir + name
' ~# `5 {$ C' h, y! W - Locaimg = imgDir + "/" + file_name6 J( i$ h: h5 _5 c# n
- print Locaimg5 A2 h& B! O% R7 z" k9 T
- if(self.download(P_dir, imgDir, file_name, img)): #这2行可注译,图片下载到本地
$ s. q2 o+ K5 D& A - print('attachment url is ' + img) #
2 S$ w9 h3 y# n' r - Datos = {% {, I( F8 S# Q/ U4 o5 J+ [( a. I
- "Cater_Name":Cater_Name,
: n" k. f9 P2 F+ U; {/ O - "Book_author":Book_author,
& w# f, Z; J1 o! E9 Q0 U8 O - "Book_Introduction":Book_Introduction,
$ b9 H1 G9 a3 k! p7 J+ l; M: U - "Book_Synopsis":Book_Synopsis,
- x( A9 K+ e% V- o4 Z0 I - "Book_Palabras":Book_Palabras,' j2 i# H/ `- Q* d9 [3 r
- "img":img, D' H8 `! T& F& {7 ?
- }3 t- S7 K9 i% 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/ O, f3 n- Y* }; K6 o: m
- for each in response.doc('div[class="bookbtn-txt"] a[class="catalogbtn"]').items():) f+ Q8 [& m$ T
- self.crawl(each.attr.href, callback=self.index_page,save=Datos)
+ f0 f2 ?# ?& s% X - 1 W6 k4 R$ n7 H# ?, c+ F6 {
- @config(age=8 * 60 * 60) " [* r7 L& m$ D
- def index_page(self, response): ( v. U" Z% m# t5 m
- Datos = {" @6 V( y( U/ h) g C. V1 S
- "Cater_Name":response.save['Cater_Name'],/ O: }6 F6 d* {8 n" y/ s
- "Book_author":response.save['Book_author'],9 G5 i) c; |6 m& E. W
- "Book_Introduction":response.save['Book_Introduction'],
1 ^* [) J- @' M" l( R% a/ M, e" j. [' n - "Book_Synopsis":response.save['Book_Synopsis'],
7 K- |& C! X* L - "Book_Palabras":response.save['Book_Palabras'],7 F2 k6 K- h" I7 a1 m1 ?
- "img":response.save['img'],# K5 K' }9 f4 H$ W
- }, w1 v$ w. P9 k s0 A
- for each in response.doc('.chapter-list li:first-child a[href^="http"]').items():1 ]6 d7 Z; y* d1 m! y* E3 F) N
- # for each in response.doc('.chapter-list a[href^="http"]').items():
1 U6 k& _0 Z4 e+ M - self.crawl(each.attr.href, callback=self.detail_page,save=Datos)) ^2 s( Q) E. R0 I, V! V
- @config(priority=2)8 {$ @$ ~; ^4 a8 E( ?3 M
- @catch_status_code_error
0 \; X, w3 h3 i( b - def detail_page(self, response): * w7 M5 Q9 G4 S% |& [5 y. T v
- NewRe1 = u'哈书'
; Z0 f5 A, \1 j9 y' S - NewRe2 = u'huhjsd.CC'
8 k# `$ A3 O0 S6 O - NewRe3 = r'^\\n\\n'6 n1 S) q! D) N4 o/ U& o
- NewRe5 = u'小说网'1 `' Q# N# G2 f- P0 _" ~
- NewRe6 = u'fgdfgf'( r; }1 L& Q. R* f& u
- NewRe7 = u'fgfgf'8 a4 y! w; O, V1 ?& x- D9 h
- NewRe8 = u'ffhgf'
% k* Q. ^( S3 j- H - NewRe4 = r'[\f\t\v+\.\{\(\)\}\!\/_,$%^*(+"\')]+|[+——()?【】“”!,。?、~@#¥%……&*()]+'
9 I3 x" M W6 R. G& q v2 F4 f - ReC1 = u'静思'
5 X8 Y* ` G8 P! w. i5 h - ReC2 = u'aghgf.com'2 u7 d1 B( `% x9 ], ]
- ReC3 = u'aghgfh.com') n) a2 k1 q' w0 c5 M! w$ Y ]
- ReC4 = u''
! a+ Z f! l& t9 {7 n - ReC5 = u'文学网'+ Y( H7 a8 L! K' g0 Y0 |& @% _6 C
- ReC6 = r'<BR>'* l4 J$ Y7 J% q) _
- Bookname = response.doc('.readlocation a').eq(2).text() #小说名称
( ~* F0 m% D" Y9 P% Y$ y2 G D# }* d - print Bookname3 M' \- I! u7 x0 ?
- Cater_Name = response.save['Cater_Name'] # 小说分类
- g! i! b( H* E8 { - Book_author = response.save['Book_author'] #小说作者
( |0 l2 K/ x8 x) B" O/ ~ - Book_Introduction1 = response.save['Book_Introduction'] #小说简介
2 ?0 I2 I" J' u7 j8 u/ g - Book_Synopsis = response.save['Book_Synopsis'] #最近更新3 ?' N; Q3 w) [8 b. f' F
- Book_Palabras = response.save['Book_Palabras'] #小说字数
2 ~; f) Q3 s/ ?& X4 \ - Bookurl = response.url #小说网址
9 I$ q9 p- h. O; p9 ]5 t n& Q- z- P - Booktitle = response.doc('.article-title').text() #章节名称3 a( J8 y( j, {7 n
- BookID = response.doc('.readset-r span').text() #小说ID
/ K9 O7 D' I6 i8 v" I - BookConte1 = response.doc('.article-con').text() #小说章节内容$ y1 v" S7 b; Y# r8 h
- 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)6 y [0 [) u/ W8 o8 J2 c* V8 ~$ a
- Book_Date = str(datetime.datetime.now()) # 采集时间; C" q6 D M1 k, @0 }; ~, j
- BookConte2 = BookConte1.replace(NewRe1 , ReC1)
. L; V- H2 ]" y4 @. ] - BookConte3 = BookConte2.replace(NewRe2 , ReC2)6 G7 r, X+ U# c. T5 a" r3 J; D
- BookConte5 = BookConte3.replace(NewRe5 , ReC5)9 ]8 B3 ?# J B" p
- BookConte6 = BookConte5.replace(NewRe6 , ReC2)
6 y# ^* N/ n4 I" M - BookConte7 = BookConte6.replace(NewRe7 , ReC2)
( N, ^6 D& \$ Q# c* v# W# @9 G) I - BookConte8 = BookConte7.replace(NewRe3 , ReC6)
/ L( Q- J% F" D( g - BookConte4 = re.sub(NewRe4 , ReC4 , BookConte8)
8 R6 q' k$ G( T5 @5 X- ]& O - BookConte = BookConte4.replace("\n\n","<br>")
" t+ n3 H G# E) L - print BookConte( t8 ?5 Y$ d0 ?) v \; | q( r; `
- Book_Introduction2 = Book_Introduction1.replace(NewRe1 , ReC1)
! E1 U- K, l. m) r$ {7 ]7 W - Book_Introduction3 = Book_Introduction2.replace(NewRe2 , ReC2)
. y! W7 C, B# q - Book_Introduction4 = Book_Introduction3.replace(NewRe3 , ReC3)
/ Z$ a& D" W+ E+ h: I - Book_Introduction = re.sub(NewRe4 , ReC4 , Book_Introduction4)) A: [4 {& o0 v" e% w! X
- Titleid = response.url.split(BookID + "/")[-1].split("/")[0] % V4 S G0 E$ y" S
- Book_img = response.save['img'], #小说图片' B, k* ~% N0 t5 W; t+ d6 r2 ] K, U
- 7 m, n! N9 s) {: O
- #insert into MySQL 小说入库, I3 T! x) K6 R: g+ _
- 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
6 E8 \0 m) U" l2 e - 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" B* E7 {9 L c+ [; X9 Y+ y! U9 ^
- #post提交发布' }7 ^& q# P) P/ b
- 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: d3 t# g0 j1 E) L9 n
- Datos = {/ J% B$ m, r$ ]. g( p5 G
- "Cater_Name":response.save['Cater_Name']," x4 F, w9 G: \- F5 R
- "Book_author":response.save['Book_author'],
) c. u7 p. o$ Q. y; w& o - "Book_Introduction":response.save['Book_Introduction']," y7 N2 ?: d8 p# E4 k
- "Book_Synopsis":response.save['Book_Synopsis'],2 C: F3 h6 o, k* Y2 J8 ?" O
- "Book_Palabras":response.save['Book_Palabras'],. g8 J2 S6 g2 ~! A) c5 A' |6 X
- "img":response.save['img'],
1 S+ q% }" B. ^/ j - }
. t" l _: R& L - for each in response.doc('.articlebtn > a:nth-child(4)[href*="/xiaoshuo"]').items():0 l7 ?$ W2 @/ s- Y' @; w3 m
- self.crawl(each.attr.href, callback=self.detail_page,save=Datos) 2 X1 r( o; i N$ i
- return {0 A7 W: k* R2 }
- "Cater_Name":Cater_Name," w9 I4 l) H! b
- "Bookname":Bookname,
. Y% p+ V0 Y7 ]9 I - "Book_author":Book_author,4 J: T; b& m" o0 Y: b
- "Book_Introduction":Book_Introduction,' h. i' }6 T3 x, g
- "Book_Synopsis":Book_Synopsis,
$ ~1 h* I0 s& m - "Book_Palabras":Book_Palabras,7 H) O9 p/ p3 F" |
- "Book_img":Book_img,
6 U0 a+ z0 _ n: x' z - "Bookurl": response.url,
7 X% p) y6 C/ Z5 K* j; A - "Booktitle": Booktitle,
* I! k" G; i1 Y* v$ [4 k$ Z/ l7 } - "BookID": BookID,! p( n1 E3 `7 P) V' c% L* E
- "BookConte": BookConte,
* S4 D/ ]1 ]% v$ s; n - "Titleid": Titleid,$ ^5 ^5 K" l& b0 B* } V
- "abover":abover,
2 P5 y7 C( u1 h - # "Book_Date" = str(datetime.datetime.now()),+ {! o2 Q1 |) t) K
- }/ N/ q3 o f- @, y
- def download(self, P_dir, imgDir, file_name, Book_img):
/ @& X& { U) R% o/ {" b s - if not os.path.exists(imgDir): 5 w! B" @9 N* G4 \6 H) ]0 j
- os.makedirs(imgDir)# _ W$ s3 V" q# k7 g
- file = imgDir + "/" + file_name& }) H/ Q% u8 F9 V |2 N
- # print file9 E. ` l' o+ L2 _- p
- f = open(file, 'wb+')3 a# W, n! `2 ^$ f7 p, E/ x
- imag = requests.get(Book_img)
: p- h$ y8 ]1 } N: n4 ~ i) P - f.write(imag.content)$ o5 h: y7 C9 g" E0 q8 ^" |
- f.close()3 l1 Q) V/ Y* z4 e7 d
- #保存图片前2 `: y$ s. x. {! Z
- def save_imgs(self,response):
, L: s+ V' y( p& h$ S& ~( q8 Y' D3 p - content = response.content
4 e% s- o3 U. |# ~. X6 ? - file_name = response.save["file_name"]
H! M8 ~& M5 z, O H7 B - imgDir = response.save["imgDir"]
- o; j: ~% ~: A* I4 Q0 R - file_path = imgDir + file_name1 q. H7 f0 {& i$ Z! q7 A
- self.save_img(content,imgDir,file_path)
! g/ N7 L9 W" t - #保存图片
2 `# |3 [$ E Q b* { - def save_img(self,content,imgDir,path):. p9 g/ `. d9 j' ?4 [
- if not os.path.exists(imgDir): 0 f" t( j R. C9 \
- os.makedirs(imgDir)
/ Z: G t" U7 | - f = open(path,"wb" )
5 x* n) j u" g, s/ N2 }( d2 b. K% o - f.write(content)+ d. r. c' a, Y' u7 e
- f.close(); G7 W1 N% d% Z
- #获取url后缀名/ y9 o; Z, |/ H
- def getExtension(self,url):
) ?( _% S; M4 P) W, L/ Q( Z - extension = url.split(".")[-1]% q. o. D& W8 c' y% e& a
- return extension % ?9 {3 z$ F! c2 y6 \* @- K( q
-
+ M1 C- y: X! S2 _( i: B2 J* P - #获取图片名
3 ~" G& @3 @- T/ F# N4 d _+ x - def getname(self,url):3 O/ o2 v8 w4 r/ N, v! i
- name=url.split("/")[-1].split(".")[0]
* Z, e* h- q6 j: a4 z/ z9 A+ X3 l( i( U - return name
复制代码
6 a# N: h. i* N! ] & h2 w6 F5 K$ N. b$ S
|