treefmt/format/task.go
Brian McGee ce14ee828f
feat: simplify pipeline model
For each path we determine the list of formatters that are interested in formatting it. From there, we sort
the list of formatters first by priority (lower value, higher priority) and then by name (lexicographically).

With this information we create a batch key which is based on the unique sequence of formatters. When enough paths with the same sequence is ready we apply them in order to each formatter.

By default, with no special configuration, this model guarantees that a given path will only be processed by one formatter at a time.

If a user wishes to influence the order in which formatters are applied they can use the priority field.

Signed-off-by: Brian McGee <brian@bmcgee.ie>
2024-05-26 16:52:04 +01:00

43 lines
897 B
Go

package format
import (
"cmp"
"slices"
"git.numtide.com/numtide/treefmt/walk"
)
type Task struct {
File *walk.File
Formatters []*Formatter
BatchKey string
}
func NewTask(file *walk.File, formatters []*Formatter) Task {
// sort by priority in ascending order
slices.SortFunc(formatters, func(a, b *Formatter) int {
priorityA := a.Priority()
priorityB := b.Priority()
result := priorityA - priorityB
if result == 0 {
// formatters with the same priority are sorted lexicographically to ensure a deterministic outcome
result = cmp.Compare(a.Name(), b.Name())
}
return result
})
// construct a batch key which represents the unique sequence of formatters to be applied to file
var key string
for _, f := range formatters {
key += f.name + ":"
}
key = key[:len(key)-1]
return Task{
File: file,
Formatters: formatters,
BatchKey: key,
}
}