Top Earners in Each Department

Bright Harbor's compensation review comes round every quarter, and HR wants the same summary each time: for every department, who are its best-paid people? How many names they want back changes with what the review is for — one apiece for a board slide, a handful for a pay-band audit — so the count is decided on the day and the report has to take it as it comes.

The obvious version of this report — order everyone in the company by salary and take the top few — is the one they had before, and it was useless. The departments are wildly different sizes and pay on completely different scales, so a company-wide list was Engineering and Data all the way down, and Marketing never appeared on it once. The best-paid people in Marketing are still the best-paid people in Marketing, however modest the numbers look next to Engineering's.

employees_df — one row per current employee. salary is annual pay in dollars, and manager is blank for people who report to nobody.

Input

The DataFrame above is already built for you. One line of input arrives: a whole number, how many names HR wants from each department this quarter. It is at least 1.

Task: Print a dict keyed by department name, with the departments in alphabetical order. Each value is a list of the names of that department's highest-paid people, as many as were asked for, better-paid first. A department holding fewer people than that gets a correspondingly shorter list — every name it has, in order — rather than being padded or dropped.

Example: if two names had been asked for, the printed dict would look like {'Legal': ['Bigger Earner', 'Smaller Earner'], 'Facilities': ['Only Person There']}.