Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- {
- "name": "yardbarker",
- "base_url": "",
- "start_urls":[
- "http://www.yardbarker.com"
- ],
- "rules":[
- {
- "link_extractor":{
- "allow": "http:\/\/www.yardbarker.com\/\w+\/articles\/[\w*]*\/[\w*]*$",
- "allow_domains":null,
- "deny_extensions":null,
- "restrict_xpaths": null,
- "restrict_css": null,
- "tags": null,
- "attrs": null
- },
- "follow": true,
- "callback": "parse_article"
- },
- {
- "link_extractor":{
- "allow": null,
- "deny":null,
- "allow_domains": "www.yardbarker.com",
- "deny_extensions":null,
- "restrict_xpaths": null,
- "restrict_css": null,
- "tags": null,
- "attrs": null
- },
- "follow": true
- }
- ],
- "document_selectors":{
- "title": {
- "xpath": [
- "//div[@class='article_left_column']/div[@class='art_headline']/h1/text()"
- ]
- },
- "text": {
- "xpath": [
- "//div[@class='article_left_column']/div[@class='art_body_article']/div[contains(@class,'froala-view')]/div[@class='article_chunk']/p/text()"
- ]
- },
- "categories": {
- "xpath": [
- "//div[@id='nav']/ul[@class='first']/li[contains(@class,'preselected')]/a/text()"
- ]
- }
- }
- }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement