Initial commit.

complexity.go

@@ -0,0 +1,79 @@
+// Complexity performs some rudimentary entropy calculations based on ASCII bi-grams, which can be
+// used to censor potential secret keys or passwords in text, and to estimate password strength.
+//
+// The matrix it uses only covers printable ASCII text, other runes are all interpreted as being \x127.
+//
+// The package isn't well tested, and the source data the matrix tabe
+
+package complexity
+
+import (
+	"bufio"
+	"io"
+	"unicode/utf8"
+)
+
+type state struct {
+	prev    rune
+	count   int64
+	entropy float64
+}
+
+func runeIdx(r rune) int {
+	if r < firstRune || r > lastRune {
+		return sz - 1
+	}
+
+	return int(r - firstRune)
+}
+
+func (s *state) put(next rune) {
+	s.entropy += float64(matrix[runeIdx(s.prev)*sz+runeIdx(next)])
+	s.count++
+	s.prev = next
+}
+
+func runeReader(r io.RuneReader) (float32, int64, error) {
+	s := state{prev: lastRune}
+	for {
+		r, _, err := r.ReadRune()
+		if err != nil {
+			if err == io.EOF {
+				err = nil
+			}
+			return float32(s.entropy), s.count, err
+		}
+
+		s.put(r)
+	}
+}
+
+func Reader(r io.Reader) (float32, int64, error) {
+	if r, ok := r.(io.RuneReader); ok {
+		return runeReader(r)
+	}
+
+	return runeReader(bufio.NewReader(r))
+}
+
+func Bytes(text []byte) (float32, int) {
+	s := state{prev: lastRune}
+
+	for i := 0; i != len(text); {
+		r, sz := utf8.DecodeRune(text[i:])
+		i += sz
+		s.put(r)
+	}
+
+	return float32(s.entropy), int(s.count)
+}
+
+func String(text string) (float32, int) {
+	s := state{prev: lastRune}
+
+	for _, r := range text {
+		s.put(r)
+	}
+
+	return float32(s.entropy), int(s.count)
+}