def __init__(self):
Spider.__init__(self,
start_urls=[
'https://www.androidperformance.com',
],
index_xpath="//article/a/@href",
article_title_xpath="//*[@class='post-title']/a/text()",
article_content_xpath="//*[@class='article-entry']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'http://www.wjdiankong.cn',
],
index_xpath="//header/h2/a/@href",
article_title_xpath='//header/h1/a/text()',
article_content_xpath="//div/article",
article_trim_xpaths=['//div[@class="article-social"]'])
def __init__(self):
Spider.__init__(self,
start_urls=[
'http://fex.baidu.com/weekly/',
],
index_xpath="//ul[@class='post-list']//a/@href",
article_title_xpath="//h1[@class='title']/text()",
article_content_xpath="//div[@class='content']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://timyang.net',
],
index_xpath="//h2/a/@href",
article_title_xpath="//*[@class='posttitle']/a/text()",
article_content_xpath="//*[@class='entry']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://www.lfhacks.com/',
],
index_xpath="//li[@class='article']/h4/a/@href",
article_title_xpath="//article[@class='article']/h1/text()",
article_content_xpath="//div[@itemprop='articleBody']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://www.stackoverflow.wiki/blog',
],
index_xpath="//h1/a/@href",
article_title_xpath="//h1[contains(@class, 'title')]/text()",
article_content_xpath="//div[@id='post-article']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://xingzx.org/',
],
index_xpath="//h2[@class='post-title']/a/@href",
article_title_xpath="//div[@class='markdown-body editormd-preview-container']/h1/text()",
article_content_xpath="//div[@class='blog-body']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://www.jdon.com/approval',
],
index_xpath="//h3[@class='vid-name']/a/@href",
article_title_xpath="//title/text()",
article_content_xpath="//div[@class='post_body_content']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=['https://github.com/polaris1119/golangweekly'],
index_xpath="//article//a[contains(text(),'угг')]/@href",
article_title_xpath='//article//h1/text()',
article_content_xpath="//article",
index_limit_count=3,
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://moelove.info/',
],
index_xpath="//h1[@class='post-title']/a/@href",
article_title_xpath="//h1[@class='post-title']/text()",
article_content_xpath="//div[@class='post-content']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://yafeilee.com/archives',
],
index_xpath="//a[@class='blog-title']/@href",
article_title_xpath="//h2[@class='blog-title']/text()",
article_content_xpath="//div[@class='content markdown']",
index_limit_count=1,
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'http://www.yinwang.org',
],
index_xpath="//ul[@class='list-group']/li/a/@href",
article_title_xpath='//h2/text()',
article_content_xpath="//div[@class='inner']",
index_limit_count=3,
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://www.coder4.com/',
],
index_xpath="//h1[@class='entry-title']/a/@href",
article_title_xpath="//h1[@class='entry-title']/text()",
article_content_xpath="//div[@class='entry-content']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'http://ued.baidu.com/case',
],
index_xpath="//ul[@class='case-article-list']/li/a/@href",
article_title_xpath="//div[@class='article-title']/text()",
article_content_xpath="//div[@class='article-content']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://github.com/aliyunfe/weekly',
],
index_xpath="//a[contains(text(),'》第')]/@href",
article_title_xpath="//h2//*[@class='final-path']/text()",
article_content_xpath="//div[@id='readme']",
index_reverse=True
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://lucifer.ren/blog/',
],
index_xpath="//h2[@class='title']/a/@href",
article_title_xpath="//h1[@class='title']/a/text()",
article_content_xpath="//div[@class='article-entry']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://github.com/trending?since=monthly',
],
index_xpath='//*[@class="Box-row"]/h1/a/@href',
article_title_xpath='//meta[@property="og:title"]/@content',
article_content_xpath='//div[@id="readme"]',
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://hiwannz.com/',
],
index_xpath="//h2[@class='post-title']/a/@href",
article_title_xpath="//h1[@class='post-title']/text()",
article_content_xpath="//div[@class='post-content']",
index_limit_count=2,
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'http://www.alloyteam.com/page/0/',
],
index_xpath="//ul[@class='articlemenu']/li/a[2]/@href",
article_title_xpath="//div[@class='title1']/a[2]/text()",
article_content_xpath="//div[@class='content_banner']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://blog.miniasp.com/',
],
index_xpath="//h2[@class='post-title']/a/@href",
article_title_xpath="//h2[@class='post-title']/a/text()",
article_content_xpath="//section[@class='post-body text']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://kernel.taobao.org',
],
index_xpath="//div[@class='article-title']/a/@href",
article_title_xpath="//h1[@class='post-title']/text()",
article_content_xpath="//div[@class='post-content']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://teddysun.com',
],
index_xpath="//header/h2/a/@href",
article_title_xpath="//h1/a/text()",
article_content_xpath="//*[@class='article-content']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://juejin.im/?sort=monthly_hottest',
],
index_xpath="//div[@class='entry-box']//a[@class='entry-link']/@href",
article_title_xpath="//h1[@class='article-title']/text()",
article_content_xpath="//div[@class='article-content']",
browser=True
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'http://yalishizhude.com/',
],
index_xpath="//h1[@class='post-title']/a/@href",
article_title_xpath="//h1[@class='post-title']/text()",
article_content_xpath="//div[@class='post-body']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://blog.huoding.com/',
],
index_xpath="//header/h1/a/@href",
article_title_xpath="//article/header/h1/text()",
article_content_xpath="//article//div[@class='entry-content']",
)
def __init__(self):
Spider.__init__(
self,
start_urls=[
'https://sofi.sh/',
],
index_xpath="//div[@class='block']/h2/a/@href",
article_title_xpath="//a[@title='page.attributes.title']/h1/text()",
article_content_xpath="//div[@itemprop='articleBody']",
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://www.cnblogs.com/pick/',
],
index_xpath='//*[@id="post_list"]/div/div/h3/a/@href',
article_title_xpath='//*[@id="cb_post_title_url"]/text()',
article_content_xpath='//*[@id="cnblogs_post_body"]',
index_limit_count=4,
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'https://github.com/dt-fe/weekly',
],
index_xpath="//td[@class='content']//a[contains(text(),'.md') and not(contains(text(),'readme.md'))]/@href",
article_title_xpath="//h2//*[@class='final-path']/text()",
article_content_xpath='//article',
index_reverse=True,
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'http://www.nginx.cn',
],
index_xpath="//div[@class='post']/h2/a/@href",
article_title_xpath="//div[@class='post']/h1/text()",
article_content_xpath="//div[@class='post']/div[@class='content']",
index_limit_count=3,
)
def __init__(self):
Spider.__init__(self,
start_urls=[
'http://gityuan.com',
],
index_xpath="//div[@class='post-preview']/a/@href",
article_title_xpath='//h1/text()',
article_content_xpath="//div[contains(@class, 'post-container')]",
index_limit_count=6,
)
