Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- ---
- config:
- agent: Firefox
- iterator:
- type: csv
- name: letter
- value: A,B
- do:
- - link_add:
- url: http://www.birmingham.ac.uk/staff/profiles/index.aspx?CurrentTab=AtoZ&StaffComplete_AtoZ_AtoZLetter=<%letter%>&StaffComplete_atozlisting_goto=1
- - walk:
- to: links
- do:
- - find:
- path: 'div.listing--items>article'
- do:
- - object_new: person
- - find:
- path: h1>a
- do:
- - parse
- - object_field_set:
- object: person
- field: name
- - parse:
- attr: href
- - normalize:
- routine: url
- - object_field_set:
- object: person
- field: url
- - find:
- path: p.staff--job-title
- do:
- - parse
- - object_field_set:
- object: person
- field: title
- - find:
- path: p.staff--school-or-department
- do:
- - parse
- - object_field_set:
- object: person
- field: department
- - find:
- path: dd:contains('+44')
- do:
- - parse
- - object_field_set:
- object: person
- field: phone
- - find:
- path: dd:contains('@')
- do:
- - parse
- - object_field_set:
- object: person
- field: email
- - object_save:
- name: person
- - find:
- path: a.pagination__controls__next
- do:
- - parse:
- attr: href
- - link_add
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement