Data normalizing for phone number extraction

I need some help finishing normalizing data in following example below.I understand it would have been better to start with correct formatting, but I'm trying to migrate from this old solution to a new product and need to extract as much data as I can in correct fields.

I need help to figure out how to:

1. To get rid off any text if there is e.g. "OFFICE: Jamie 11111111" to make "OFFICE: 11111111", or "OFFICE: no phone number" to "OFFICE: " or "EMERGENCY: Jenny 1111111111" to "EMERGENCY: 1111111111". I do NOT care about "Some random text" because final goal is to extract data after HOME: OFFICE: CELL: EMERGENCY: as per earlier forum post
2. If there is any 7 digit number, I need to add 000 to beginning of number, e.g. "OFFICE: 1111111" to "OFFICE: 0001111111"
3. If there is any 3 digit long number, delete it.

HOME: 1111111111 OFFICE: no phone number
HOME: 1111111111 OFFICE: 1111111
HOME: 1111111111 OFFICE: 1111111111 EMERGENCY: 1111111111
HOME: 1111111111 OFFICE: EMERGENCY: 1111111111
HOME: OFFICE: EMERGENCY: 1111111111 CELL: 1111111111
HOME: 1111111 OFFICE: 1111111111 EMERGENCY: Jenny 1111111111 CELL: 1111111111
HOME: 1111111111 CELL: 1111111 OFFICE: 1111111111 EMERGENCY: Julie 1111111111
HOME: 1111111111 OFFICE: 1111111111 Some random text HOME: 1111111111 OFFICE: EMERGENCY:
HOME: 000 OFFICE: 1111111111 EMERGENCY: Tara 1111111111 CELL: 1111111111
HOME: 1111111111 Some random text OFFICE: 1111111111 HOME: 1111111111 OFFICE: EMERGENCY:

thank for help in advance