Files
coding/genseed/genseed.go

101 lines
1.8 KiB
Go
Raw Normal View History

2023-12-29 20:48:12 -07:00
package main
import (
"bufio"
"log"
"os"
"github.com/samber/lo"
"github.com/securemesh/coding"
2023-12-30 20:14:01 -07:00
"github.com/securemesh/coding/state"
2023-12-29 20:48:12 -07:00
"github.com/securemesh/coding/seeds"
)
func main() {
samples := lo.Must(loadSamples())
log.Printf("orig=%d", lo.SumBy(samples, func(sample []byte) int {
return len(sample)
}))
2023-12-30 20:14:01 -07:00
def := state.NewState()
2023-12-30 15:51:29 -07:00
log.Printf("def=%d [%s]", totalLength(def, samples), def)
2023-12-30 20:14:01 -07:00
chat := seeds.ChatState()
2023-12-30 15:51:29 -07:00
log.Printf("chat=%d [%s]", totalLength(chat, samples), chat)
2023-12-29 20:48:12 -07:00
2023-12-30 20:14:01 -07:00
opt := optimize(state.NewState(), samples)
2023-12-29 20:48:12 -07:00
log.Printf("opt=%d [%s]", totalLength(opt, samples), opt)
}
2023-12-30 20:14:01 -07:00
func optimize(st *state.State, samples [][]byte) *state.State {
2023-12-29 20:48:12 -07:00
for true {
2023-12-30 20:14:01 -07:00
better := optimize2(st, samples)
2023-12-29 20:48:12 -07:00
if better == nil {
2023-12-30 21:00:37 -07:00
return st
2023-12-29 20:48:12 -07:00
}
2023-12-30 20:14:01 -07:00
st = better
log.Printf("\titer=%d [%s]", totalLength(st, samples), st)
2023-12-29 20:48:12 -07:00
}
2023-12-30 20:14:01 -07:00
return st
2023-12-29 20:48:12 -07:00
}
2023-12-29 22:02:07 -07:00
type sampleResult struct {
2023-12-30 20:14:01 -07:00
state *state.State
2023-12-29 22:02:07 -07:00
score int
}
2023-12-30 20:14:01 -07:00
func optimize2(baseState *state.State, samples [][]byte) *state.State {
2023-12-29 22:02:07 -07:00
ch := make(chan sampleResult, 100)
for i := 0; i < 256; i++ {
2023-12-30 15:51:29 -07:00
s := byte(i)
2023-12-29 22:02:07 -07:00
go func () {
2023-12-30 20:14:01 -07:00
st := baseState.Clone()
st.IncrementSymbol(s)
2023-12-29 22:02:07 -07:00
ch <- sampleResult{
2023-12-30 20:14:01 -07:00
state: st,
score: totalLength(st, samples),
2023-12-29 22:02:07 -07:00
}
}()
}
2023-12-30 20:14:01 -07:00
var best *state.State = nil
bestScore := totalLength(baseState, samples)
2023-12-29 20:48:12 -07:00
for i := 0; i < 256; i++ {
2023-12-29 22:02:07 -07:00
res := <-ch
if res.score < bestScore {
2023-12-30 20:14:01 -07:00
best = res.state
2023-12-29 22:02:07 -07:00
bestScore = res.score
2023-12-29 20:48:12 -07:00
}
}
return best
}
2023-12-30 20:14:01 -07:00
func totalLength(st *state.State, samples [][]byte) int {
2023-12-29 20:48:12 -07:00
return lo.SumBy(samples, func(sample []byte) int {
2023-12-30 20:14:01 -07:00
return len(coding.Encode(st.Clone(), sample))
2023-12-29 20:48:12 -07:00
})
}
func loadSamples() ([][]byte, error) {
fh, err := os.Open("sms.txt")
if err != nil {
return nil, err
}
defer fh.Close()
s := bufio.NewScanner(fh)
ret := [][]byte{}
for s.Scan() {
ret = append(ret, s.Bytes())
}
return ret, nil
}