Lesson 10 · Bonus at word boundaries

Bonus at word boundaries

A query that lands on the start of a word or path segment feels like a much better hit than one landing mid-word. Today you reward matches that fall on a boundary.

The goal

Add a bonus for a matched character that begins a word - the character right after a separator.

Start here - the target

TO DO

Scenario: Word-boundary bonus

Givena +8 bonus for any matched character whose preceding candidate character is a separator (one of space, slash, underscore, hyphen, or dot)

Whenscore is called with candidate "a/bc" and positions [2], then with candidate "abc" and positions [2]

Thencandidate "a/bc" with positions [2] scores 22 (16, minus 2 for the leading gap, plus 8 because index 2 follows the slash separator), and candidate "abc" with positions [2] scores 14 (16, minus 2 for the leading gap, with no boundary bonus)

Background

Where a match sits within a word matters as much as how tight it is. Typing main and landing on the m that begins the main segment of src/main.go is a far better hit than landing on an m in the middle of some word. A boundary is the position right after a separator - a space, slash, underscore, hyphen, or dot - the spots where a new word or path segment starts. A matched character on a boundary earns +8.

This bonus stacks with everything from the earlier lessons; it is added per matched position, independent of the gap logic. One deliberate choice: index 0 is not treated as a boundary here - a match at the very start already benefits from a zero leading gap, so rewarding it again would double-count. Keeping start-of-string out of the boundary rule is what lets the earlier leading-gap lesson and this one compose cleanly, each owning its own effect.

Make it work

// A boundary is a matched char preceded by a separator. Add the bonus
// for every matched position that qualifies, regardless of gap.
const scoreBoundary = 8
func isBoundary(candidate string, p int) bool {
  if p == 0 { return false }        // start-of-string handled by leading gap
  return isSeparator(candidate[p-1]) // space / _ - .
}
// after adding the per-char reward, for every k:
if isBoundary(candidate, p) { s += scoreBoundary }

CheckpointDONE

Matches at the start of a word or path segment now score higher. Commit and stop here.