Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- {
- "name": "jalopnik.com",
- "base_url": "",
- "start_urls":[
- "http://www.jalopnik.com/"
- ],
- "rules":[
- {
- "link_extractor":{
- "allow": "jalopnik.com/.*-\\d+",
- "deny": null,
- "allow_domains":null,
- "deny_extensions":null,
- "restrict_xpaths": null,
- "restrict_css": null,
- "tags": null,
- "attrs": null
- },
- "follow": true,
- "callback": "parse_article"
- },
- {
- "link_extractor":{
- "allow": "jalopnik.com",
- "deny":null,
- "allow_domains": "jalopnik.com",
- "deny_extensions":null,
- "restrict_xpaths": null,
- "restrict_css": null,
- "tags": null,
- "attrs": null
- },
- "follow": true
- }
- ],
- "document_selectors":{
- "title": {
- "xpath": [
- "//h1[contains(@class, 'title')]/*[string()]"
- ]
- },
- "text": {
- "xpath": [
- "//div[contains(@class, 'post-content')]/*[string()]"
- ]
- }
- }
- }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement