class QidianNovelInfoItem(Item):
"""
定义继承自item的Item类
"""
novel_name = TextField(css_select='.book-info>h1>em')
author = TextField(css_select='a.writer')
# 当提取的值是属性的时候，要定义AttrField
cover = AttrField(css_select='a#bookImg>img', attr='src')
abstract = TextField(css_select='div.book-intro>p')
status = TextField(css_select='p.tag>span.blue')
novels_type = TextField(css_select='p.tag>a.red')
latest_chapter = TextField(css_select='li.update>div.detail>p.cf>a')
latest_chapter_time = TextField(css_select='div.detail>p.cf>em')
def tal_cover(self, cover):
return 'http:' + cover
def tal_status(self, status):
"""
当目标值的对象只有一个，默认将值提取出来，否则返回list，可以在这里定义一个函数进行循环提取
:param ele_tag:
:return:
"""
return '#'.join([i.text for i in status])
def tal_novels_type(self, novels_type):
return '#'.join([i.text for i in novels_type])
def tal_latest_chapter_time(self, latest_chapter_time):
return latest_chapter_time.replace(
u'今天', str(time.strftime("%Y-%m-%d ", time.localtime()))).replace(
u'昨日',
str(
time.strftime("%Y-%m-%d ",
time.localtime(time.time() - 24 * 60 * 60))))
class QidianItem(Item):
"""
定义继承自item的Item类
"""
title = TextField(css_select='.book-info>h1>em')
author = TextField(css_select='a.writer')
# 当提取的值是属性的时候，要定义AttrField
cover = AttrField(css_select='a#bookImg>img', attr='src')
abstract = TextField(css_select='div.book-intro>p')
tag = TextField(css_select='span.blue')
latest_chapter = TextField(css_select='li.update>div.detail>p.cf>a')
latest_chapter_time = TextField(css_select='div.detail>p.cf>em')
def tal_title(self, title):
"""
这里可以二次对获取的目标值进行处理，比如替换、清洗等
:param title:
:return:
"""
return title
def tal_cover(self, cover):
return 'http:' + cover
def tal_tag(self, ele_tag):
"""
当目标值的对象只有一个，默认将值提取出来，否则返回list，可以在这里定义一个函数进行循环提取
:param ele_tag:
:return:
"""
return '#'.join([i.text for i in ele_tag])
def tal_latest_chapter_time(self, latest_chapter_time):
return latest_chapter_time.replace(u'今天', str(time.strftime("%Y-%m-%d ", time.localtime()))).replace(u'昨日', str(
time.strftime("%Y-%m-%d ", time.localtime(time.time() - 24 * 60 * 60))))
class RankingItem(Item):
target_item = TextField(css_select='.rank-list')
ranking_title = TextField(css_select='h3.wrap-title')
more = AttrField(css_select='h3>a.more', attr='href')
book_list = TextField(css_select='div.book-list>ul>li')
def tal_more(self, more):
return "http:" + more
class BaiduImgItem(Item):
"""
定义继承自item的Item类
"""
img_url = AttrField(css_select='img.BDE_Image', attr='src')
def tal_img_url(self, ele_img_url):
return [i.get('src').strip() for i in ele_img_url]
class ZHNovelsItem(Item):
target_item = TextField(css_select='ul.main_con>li')
novel_url = AttrField(css_select='span.chap>a.fs14', attr='href')
novel_name = TextField(css_select='span.chap>a.fs14')
novel_author = TextField(css_select='span.author>a')
novel_author_home_url = AttrField(css_select='span.author>a', attr='href')
# def tal_novel_url(self, novel_url):
# return 'http:' + novel_url
def tal_novel_author(self, novel_author):
if novel_author:
if isinstance(novel_author, list):
novel_author = novel_author[0].text
return novel_author
else:
return ''
class DoubanSpider(Item):
target_item = TextField(css_select='div.item')
title = TextField(css_select='span.title')
cover = AttrField(css_select='div.pic>a>img', attr='src')
abstract = TextField(css_select='span.inq')
def tal_title(self, title):
if isinstance(title, str):
return title
else:
return ''.join([i.text.strip().replace('\xa0', '') for i in title])
class QidianNovelsItem(Item):
target_item = TextField(css_select='ul.all-img-list>li')
novel_url = AttrField(css_select='div.book-img-box>a', attr='href')
novel_name = TextField(css_select='div.book-mid-info>h4')
novel_author = TextField(css_select='div.book-mid-info>p.author>a.name')
novel_author_home_url = AttrField(
css_select='div.book-mid-info>p.author>a.name', attr='href')
def tal_novel_url(self, novel_url):
return 'http:' + novel_url
def tal_novel_author(self, novel_author):
if isinstance(novel_author, list):
novel_author = novel_author[0].text
return novel_author
def tal_novel_author_home_url(self, novel_author_home_url):
if isinstance(novel_author_home_url, list):
novel_author_home_url = novel_author_home_url[0].get(
'href').strip()
return 'http:' + novel_author_home_url
class ZHNovelInfoItem(Item):
"""
定义继承自item的Item类
"""
novel_name = TextField(css_select='div.main div.status h1 a')
author = TextField(css_select='div.main div.status div.booksub a')
# 当提取的值是属性的时候，要定义AttrField
cover = AttrField(css_select='div.main div.book_cover img', attr='src')
abstract = TextField(css_select='div.main div.status div.info_con p')
status = AttrField(css_select='div.main div.status h1 em', attr='title')
novels_type = TextField(css_select='div.main div.status div.booksub a')
novel_chapter_url = AttrField(css_select='div.main div.status div.book_btn span.list a', attr='href')
def tal_author(self, author):
if isinstance(author, list):
return author[0].text
else:
return author
def tal_status(self, status):
"""
当目标值的对象只有一个，默认将值提取出来，否则返回list，可以在这里定义一个函数进行循环提取
:param ele_tag:
:return:
"""
if isinstance(status, list):
return '#'.join([i.get('title').strip().replace('作品', '') for i in status])
else:
return status
def tal_novels_type(self, novels_type):
if isinstance(novels_type, list):
try:
return novels_type[1].text
except:
return ''
else:
return ''
class TestSpider(Item):
title = TextField(css_select='.book-info>h1>em')
author = TextField(css_select='a.writer')
cover = AttrField(css_select='a#bookImg>img', attr='src')
abstract = TextField(css_select='div.book-intro>p')
tag = TextField(css_select='span.blue')
latest_chapter = TextField(css_select='div.detail>p.cf>a')
latest_chapter_time = TextField(css_select='div.detail>p.cf>em')
def tal_title(self, title):
# Clean your target value
return title
def tal_cover(self, cover):
return 'http:' + cover
def tal_tag(self, ele_tag):
return '#'.join([i.text for i in ele_tag])
def tal_latest_chapter_time(self, latest_chapter_time):
return latest_chapter_time.replace('今天', str(time.strftime("%Y-%m-%d ", time.localtime())))
def test_attr_field(self):
attr_field = AttrField(css_select="p a.test_link", attr='href')
value = attr_field.extract_value(self.html)
self.assertEqual(value, "https://github.com/howie6879/talonspider")