From 37bc9cbae15629826a43132bf5eeaf6432d28f94 Mon Sep 17 00:00:00 2001
From: Ian Gulliver <git@flamingcow.io>
Date: Fri, 29 Dec 2023 20:48:12 -0700
Subject: [PATCH] genseed

---
 genseed/genseed.go | 80 ++++++++++++++++++++++++++++++++++++++++++++++
 heap/heap.go       | 28 ++++++++++++++++
 2 files changed, 108 insertions(+)
 create mode 100644 genseed/genseed.go

diff --git a/genseed/genseed.go b/genseed/genseed.go
new file mode 100644
index 0000000..d7f32ed
--- /dev/null
+++ b/genseed/genseed.go
@@ -0,0 +1,80 @@
+package main
+
+import (
+	"bufio"
+	"log"
+	"os"
+
+	"github.com/samber/lo"
+	"github.com/securemesh/coding"
+	"github.com/securemesh/coding/heap"
+	"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)
+	}))
+
+	log.Printf("default=%d", totalLength(heap.NewHeap(), samples))
+	log.Printf("chat=%d", totalLength(seeds.ChatHeap(), samples))
+
+	opt := optimize(heap.NewHeap(), samples)
+	log.Printf("opt=%d [%s]", totalLength(opt, samples), opt)
+}
+
+func optimize(h *heap.Heap, samples [][]byte) *heap.Heap {
+	for true {
+		better := optimize2(h, samples)
+		if better == nil {
+			return h
+		}
+		h = better
+		log.Printf("\titer=%d [%s]", totalLength(h, samples), h)
+	}
+
+	return h
+}
+
+func optimize2(baseHeap *heap.Heap, samples [][]byte) *heap.Heap {
+	var best *heap.Heap = nil
+	bestScore := totalLength(baseHeap, samples)
+
+	for i := 0; i < 256; i++ {
+		h := baseHeap.Clone()
+		h.IncrementSymbol(byte(i))
+		score := totalLength(h, samples)
+		if score < bestScore {
+			best = h
+			bestScore = score
+		}
+	}
+
+	return best
+}
+
+func totalLength(heap *heap.Heap, samples [][]byte) int {
+	return lo.SumBy(samples, func(sample []byte) int {
+		return len(coding.Encode(heap.Clone(), sample))
+	})
+}
+
+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
+}
diff --git a/heap/heap.go b/heap/heap.go
index c07a31e..3553f87 100644
--- a/heap/heap.go
+++ b/heap/heap.go
@@ -1,7 +1,11 @@
 package heap
 
 import (
+	"fmt"
 	"maps"
+	"slices"
+	"sort"
+	"strings"
 )
 
 type node struct {
@@ -55,6 +59,30 @@ func (h *Heap) IncrementSymbol(symbol byte) int {
 	return nodeIndex
 }
 
+func (h Heap) String() string {
+	nodes := []node{}
+
+	for _, node := range h.nodes {
+		if node.count == 0 {
+			continue
+		}
+
+		nodes = append(nodes, node)
+	}
+
+	slices.SortStableFunc(nodes, func(a, b node) int { return b.count - a.count })
+
+	strs := []string{}
+
+	for _, node := range nodes {
+		strs = append(strs, fmt.Sprintf("{%#U}=%d", node.symbol, node.count))
+	}
+
+	sort.Strings(strs)
+
+	return strings.Join(strs, ", ")
+}
+
 func (h Heap) parentIndex(nodeIndex int) int {
 	return (nodeIndex - 1) / 2
 }