Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- ---
- config:
- debug: 2
- agent: Firefox
- do:
- - link_add:
- url: http://store.americanapparel.net
- - link_add:
- url: http://store.americanapparel.net/en/factory-store/
- - walk:
- to: links
- do:
- - find:
- path: .cd-primary-nav a
- do:
- - parse:
- attr: href
- - normalize:
- routine: url
- - link_add:
- pool: main
- - walk:
- to: links
- pool: main
- do:
- - find:
- path: .product > a
- do:
- - parse:
- attr: href
- - normalize:
- routine: url
- - link_add:
- pool: sub
- - walk:
- to: links
- pool: sub
- do:
- - sleep: 3
- - find:
- path: .pdp
- do:
- - variable_clear: allli
- - variable_clear: descr
- - variable_clear: li
- - variable_clear: id
- - variable_clear: views
- - variable_clear: color
- - variable_clear: imgnum
- - variable_clear: imgxl
- - variable_clear: viewsnum
- - variable_clear: stp
- - object_new: product
- - eval:
- routine: js
- body: '(function (){var d = new Date(); return d.toISOString()})();'
- - object_field_set:
- object: product
- field: date
- - static_get: url
- - object_field_set:
- object: product
- field: url
- - find:
- in: doc
- path: head meta[name="description"]
- do:
- - parse:
- attr: content
- - space_dedupe
- - trim
- - to_block
- - node_replace:
- path: br
- with: "\n"
- - split:
- context: text
- delimiter: \n+
- - find:
- path: div.splitted
- slice: 0
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: description
- - find:
- path: .product-style
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: sku
- - find:
- path: .price
- do:
- - find:
- path: .red-text
- do:
- - parse:
- filter:
- - (\d+\.?\d*)
- - if:
- match: (\d)
- do:
- - object_field_set:
- object: product
- field: price
- type: float
- - register_set: USD
- - object_field_set:
- object: product
- field: currency
- - register_set: 1
- - variable_set: stp
- - find:
- path: span[data-test="test"]
- do:
- - variable_get: stp
- - if:
- match: (1)
- else:
- - parse:
- filter:
- - (\d+\.?\d*)
- - if:
- match: (\d)
- do:
- - object_field_set:
- object: product
- field: price
- type: float
- - register_set: USD
- - object_field_set:
- object: product
- field: currency
- - find:
- path: .product-name
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: name
- - find:
- path: .main-img
- do:
- - parse:
- attr: src
- - object_field_set:
- object: product
- field: images
- joinby: "|"
- - find:
- path: .logo
- slice: 0
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: brand
- - find:
- path: .breadcrumbs a
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: category
- joinby: "|"
- - find:
- path: '.product-details > input#skuVarData'
- do:
- - parse:
- attr: value
- - normalize:
- routine: replace_substring
- args:
- \s+\/\s+: _
- - normalize:
- routine: json2xml
- - to_block
- - find:
- path: body_safe > name
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: name
- - find:
- path: colors
- do:
- - find:
- path: zoomimage
- do:
- - parse:
- filter:
- - \s*(.+)\?
- - variable_set: imgxl
- - register_set: <%imgxl%>?$ProductZoom$
- - object_field_set:
- object: product
- field: images
- joinby: "|"
- - find:
- path: name
- do:
- - parse
- - space_dedupe
- - trim
- - object_field_set:
- object: product
- field: variations
- joinby: "|"
- - object_save:
- name: product
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement