initial commit

Signed-off-by: Vincent Batts <vbatts@hashbangbash.com>

main.go

@@ -0,0 +1,106 @@
+package main
+
+import (
+	"crypto/rand"
+	"flag"
+	"fmt"
+	"io/ioutil"
+	"log"
+	"os"
+	"sort"
+)
+
+var (
+	flTop      = flag.Int("top", 1000, "top-end of the iteration")
+	flCompress = flag.Bool("compress", false, "attempt to compress the rand data")
+	flCSV      = flag.Bool("csv", false, "output in csv format")
+)
+
+const (
+	c = 10
+)
+
+func main() {
+	flag.Parse()
+
+	var fh *os.File
+	if *flCompress {
+		var err error
+		fh, err = ioutil.TempFile("", "trie.things.")
+		if err != nil {
+			log.Fatal(err)
+		}
+		defer os.Remove(fh.Name())
+	}
+
+	d := map[byte]*ByteCount{} // dictionary of byte counts
+	b := make([]byte, c)
+	for i := 0; i < *flTop; i++ {
+		if _, err := rand.Read(b); err != nil {
+			log.Fatalf("error:", err)
+		}
+		for j := 0; j < c; j++ {
+			if _, ok := d[b[j]]; ok {
+				d[b[j]].C++
+			} else {
+				d[b[j]] = &ByteCount{
+					B: b[j],
+					C: 1,
+				}
+			}
+		}
+		if *flCompress {
+			// write the bytes to tmpfile for comparison and compression
+			if _, err := fh.Write(b); err != nil {
+				log.Fatal(err)
+			}
+		}
+	}
+
+	l := make([]*ByteCount, len(d))
+	i := 0
+	for _, v := range d {
+		l[i] = v
+		i++
+	}
+	// now sort by descending frequency ...
+	sort.Sort(sort.Reverse(ByCount(l)))
+	if *flCSV {
+		// print a CSV for comparison/graphing
+		fmt.Printf("byte,byte_hex,frequency\n")
+		for i := range l {
+			fmt.Printf("%d,%2.2x,%d\n", l[i].B, l[i].B, l[i].C)
+		}
+	} else {
+		// print stats
+		lLen := len(l)
+		fmt.Printf("Length: %d\n", lLen)
+		fmt.Printf("bytes counted: %d\n", *flTop*c)
+		spread := (l[0].C - l[lLen-1].C)
+		fmt.Printf("Freq Spread: %d\n", spread)
+		fmt.Printf("Spread ratio: %.4f\n", 100*float64(spread)/float64(*flTop*c))
+		fmt.Printf("Freq Max: %d\n", l[0].C)
+		fmt.Printf("Freq Min: %d\n", l[lLen-1].C)
+		fmt.Printf("Max 3 bytes: %2.2x(%d) %2.2x(%d) %2.2x(%d)\n", l[0].B, l[0].C, l[1].B, l[1].C, l[2].B, l[2].C)
+		fmt.Printf("Min 3 bytes: %2.2x(%d) %2.2x(%d) %2.2x(%d)\n", l[lLen-1].B, l[lLen-1].C, l[lLen-2].B, l[lLen-2].C, l[lLen-3].B, l[lLen-3].C)
+	}
+
+	// and produce a right leaning trie
+}
+
+type ByteCount struct {
+	B byte
+	C int64
+}
+
+type ByByte []*ByteCount
+
+func (a ByByte) Len() int           { return len(a) }
+func (a ByByte) Swap(i, j int)      { a[i], a[j] = a[j], a[i] }
+func (a ByByte) Less(i, j int) bool { return a[i].B < a[j].B }
+
+type ByCount []*ByteCount
+
+func (a ByCount) Len() int           { return len(a) }
+func (a ByCount) Swap(i, j int)      { a[i], a[j] = a[j], a[i] }
+func (a ByCount) Less(i, j int) bool { return a[i].C < a[j].C }