diff.go

package simplecsv

import "slices"

// DiffByKey compares the data rows of s (left) and other (right), keyed by
// the values of the column with the header name key. It returns three new
// csvs:
//
//   - added: the data rows of other whose key is not present in s, with
//     other's header row
//   - removed: the data rows of s whose key is not present in other, with
//     s's header row
//   - changed: the data rows of s whose key is present in other but whose
//     full row differs, with s's header row and s's values (callers that
//     want the right-side values can join on the key)
//
// The rows keep their order of appearance in the source csv. The key values
// must be unique within each csv: a key repeated in s or in other makes the
// diff ambiguous, and it returns nil, nil, nil and false. It also returns
// nil, nil, nil and false if either csv is empty or the key is not a header
// name in either csv. A csv with only the header row is a valid input (all
// the other side's rows are then added or removed). The original csvs are
// not modified and the results share no data with them.
func (s SimpleCsv) DiffByKey(other SimpleCsv, key string) (added, removed, changed SimpleCsv, ok bool) {
	if len(s) == 0 || len(other) == 0 {
		return nil, nil, nil, false
	}
	leftPos := s.GetHeaderPosition(key)
	if leftPos == -1 {
		return nil, nil, nil, false
	}
	rightPos := other.GetHeaderPosition(key)
	if rightPos == -1 {
		return nil, nil, nil, false
	}
	leftRows := make(map[string]int, len(s)-1)
	for i := 1; i < len(s); i++ {
		if _, exists := leftRows[s[i][leftPos]]; exists {
			return nil, nil, nil, false
		}
		leftRows[s[i][leftPos]] = i
	}
	rightRows := make(map[string]int, len(other)-1)
	for i := 1; i < len(other); i++ {
		if _, exists := rightRows[other[i][rightPos]]; exists {
			return nil, nil, nil, false
		}
		rightRows[other[i][rightPos]] = i
	}
	added = SimpleCsv{copyRow(other[0])}
	removed = SimpleCsv{copyRow(s[0])}
	changed = SimpleCsv{copyRow(s[0])}
	for i := 1; i < len(other); i++ {
		if _, exists := leftRows[other[i][rightPos]]; !exists {
			added = append(added, copyRow(other[i]))
		}
	}
	for i := 1; i < len(s); i++ {
		rightIndex, exists := rightRows[s[i][leftPos]]
		if !exists {
			removed = append(removed, copyRow(s[i]))
			continue
		}
		if !slices.Equal(s[i], other[rightIndex]) {
			changed = append(changed, copyRow(s[i]))
		}
	}
	return added, removed, changed, true
}