Word Count

Counting things is what maps are for, and a word count is the smallest program that shows the whole pattern: tokenise, normalise, tally, then sort on the way out.

That last step is not decoration. Go randomises the order of for k := range m on purpose, so a function that returns "the top three words" straight out of a range loop returns a different answer each run. The map holds the counts; a sorted slice is how you hand them to anyone else.

counts := map[string]int{}
for _, w := range strings.Fields(text) {
	counts[w]++     // a new key starts at zero, so no branch is needed
}

Sorting by two keys at once - count descending, then word ascending to break ties - is the ordinary shape for this, and a sort.Slice comparison expresses it directly:

sort.Slice(pairs, func(i, j int) bool {
	if pairs[i].Count != pairs[j].Count {
		return pairs[i].Count > pairs[j].Count
	}
	return pairs[i].Word < pairs[j].Word
})

Leave the tie-break out and the equal counts come back in whatever order the map handed them over, which is to say a different one every time.

Task

Fill in the five functions in main.go. Leave the Pair type and the Punctuation constant alone - the tests use both.

1. Count(text) splits on whitespace, lowercases each token, trims Punctuation off both ends, skips anything left empty, and tallies the rest.
2. Unique(text) reports how many distinct words there are.
3. SortedPairs(counts) returns every entry as a Pair, count descending then word ascending, without disturbing the map it was given.
4. TopN(counts, n) takes the first n of those, clamping n into range.
5. Report(counts, n) renders them as "word: count\n" lines.

Hints

• strings.Fields splits on runs of any whitespace and drops empty tokens, so it is a better fit here than strings.Split(text, " ").
• strings.Trim(s, Punctuation) removes any of those characters from both ends and leaves the middle alone, which is why don't survives intact.
• A token that trims down to "" was punctuation only. Skip it before counting, or you will get an entry under the empty string.
• Unique is one line once Count works.
• Report can build its string with a strings.Builder and fmt.Fprintf(&b, ...), or with += if you prefer - the output is short.