Public Schools in Illinois digger config

---
config:
    debug: 2
    agent: "Firefox"
do:
- walk:
    to: 'http://www.illinoisschools.com/schools/public.aspx'
    do:
    - find:
        path: 'a.linkSmallclean'
        do:
        - parse:
            attr: href
        - walk:
            to: value
            do:
            - find:
                path: div#AllSchoolData
                do:
                - find:
                    path: tr>td:nth-of-type(2)>b>a
                    do:
                    - parse:
                        attr: href
                    - walk:
                        to: value
                        do:
                        - find:
                            path: td[valign="top"]:has(a:contains('Get Directions'))
                            do:
                            - split:
                                context: html
                                delimiter: '<br/>'
                            - object_new: school
                            - find:
                                path: div.element_0
                                do:
                                - parse
                                - object_field_set:
                                    object: school
                                    field: name
                            - find:
                                path: div.element_1
                                do:
                                - parse
                                - object_field_set:
                                    object: school
                                    field: street
                            - find:
                                path: div.element_2
                                do:
                                - parse:
                                    filter: \s*([^,]+),
                                - object_field_set:
                                    object: school
                                    field: city
                            - register_set: IL
                            - object_field_set:
                                object: school
                                field: state
                            - find:
                                path: div.element_2
                                do:
                                - parse:
                                    filter: (\d+)
                                - object_field_set:
                                    object: school
                                    field: zip
                            - find:
                                path: div.element_4
                                do:
                                - parse:
                                    filter: '(.+)\s+County,\s+Illinois'
                                - object_field_set:
                                    object: school
                                    field: county
                            - find:
                                path: div.element_5
                                do:
                                - parse
                                - object_field_set:
                                    object: school
                                    field: phone
                            - object_save:
                                name: school