Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- ---
- config:
- debug: 2
- agent: Firefox
- do:
- - link_add:
- pool: main
- url:
- - https://www.adidashardware.com/training
- - https://www.adidashardware.com/recovery
- - https://www.adidashardware.com/strength
- - https://www.adidashardware.com/yoga
- - https://www.adidashardware.com/benches
- - https://www.adidashardware.com/cardio
- - walk:
- to: links
- pool: main
- do:
- - find:
- path: script:matches(Wix Stores)
- do:
- - parse
- - space_dedupe
- - trim
- - filter:
- args:
- - var\s*rendererModel\s*=\s*(.+)\s*;\s*var\s*publicModel
- - normalize:
- routine: json2xml
- - to_block
- - find:
- path: clientspecmap > *:hasChild(appdefinitionname:matches(Wix Stores)) instance
- do:
- - parse
- - variable_set: instance
- - find:
- path: body
- do:
- - static_get: url
- - normalize:
- routine: replace_substring
- args:
- \s*$: ?_escaped_fragment_=
- - walk:
- to: value
- do:
- - find:
- path: meta[property="og:title"]
- do:
- - parse:
- attr: content
- filter:
- - \s*\|\s*(.+)
- - normalize:
- routine: lower
- - variable_set: cats
- - find:
- path: li[itemprop="itemListElement"] h3[itemprop="name"] > a
- do:
- - variable_clear: uri
- - parse:
- attr: href
- - normalize:
- routine: url
- - variable_set: uri
- - variable_clear: ur
- - filter:
- args:
- - product-page\/(.+)
- - variable_set: ur
- - walk:
- to: https://ecom.wix.com/storefront/product/<%ur%>?instance=<%instance%>&locale=en
- do:
- - object_new: product
- - find:
- path: head
- do:
- - variable_get: cats
- - object_field_set:
- object: product
- field: categories
- - register_set: Adidas
- - object_field_set:
- object: product
- field: brand
- - eval:
- routine: js
- body: '(function (){var d = new Date(); return d.toISOString()})();'
- - object_field_set:
- object: product
- field: date
- - variable_get: uri
- - object_field_set:
- object: product
- field: url
- - find:
- path: script:matches(eCom.eComAppConfig)
- do:
- - parse
- - space_dedupe
- - trim
- - filter:
- args:
- - eCom\.eComAppConfig\(\'productPageApp\'\,\s*(.+),\s*\'\/\/
- - normalize:
- routine: json2xml
- - to_block
- - find:
- path: body_safe
- do:
- - find:
- path: product > description
- do:
- - parse
- - space_dedupe
- - trim
- - normalize:
- routine: replace_substring
- args:
- \<p\>\s*\<\/p\>: ''
- - register_set: <div><%register%></div>
- - to_block
- - find:
- path: div
- do:
- - attr_remove:
- selector: '*'
- - parse
- - object_field_set:
- object: product
- field: description
- - find:
- path: product > media:hasChild(mediatype:matches(PHOTO)) > url
- do:
- - parse
- - register_set: https://static.wixstatic.com/media/<%register%>
- - object_field_set:
- object: product
- field: images
- joinby: "|"
- - find:
- path: product > name
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: name
- - find:
- path: product options:hasChild(optiontype:matches(COLOR)) description
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: variations
- joinby: "|"
- - find:
- path: product > sku, product > managedproductitems > sku
- do:
- - parse
- - space_dedupe
- - trim
- - if:
- match: (\S{3})
- do:
- - object_field_set:
- object: product
- field: sku
- - find:
- path: product > prices > price
- do:
- - parse:
- filter:
- - (\d+\.?\d*)
- - object_field_set:
- object: product
- field: price
- type: float
- - register_set: GBP
- - object_field_set:
- object: product
- field: currency
- - object_save:
- name: product
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement