wordset: add new type for interning word sets

wordset.go

@@ -0,0 +1,64 @@
+package codenames
+
+import (
+	"crypto/sha1"
+	"errors"
+	"fmt"
+	"io"
+	"sort"
+	"strings"
+	"sync"
+)
+
+type wordSetID [sha1.Size]byte
+
+func (i wordSetID) String() string {
+	return fmt.Sprintf("%x", i[:])
+}
+
+type WordSets struct {
+	mu   sync.Mutex
+	byID map[wordSetID][]string
+}
+
+func (ws *WordSets) init() {
+	if ws.byID == nil {
+		ws.byID = make(map[wordSetID][]string)
+	}
+}
+
+func (ws *WordSets) Canonicalize(words []string) (wordSetID, []string, error) {
+	ws.mu.Lock()
+	defer ws.mu.Unlock()
+	ws.init()
+
+	set := map[string]bool{}
+	for _, w := range words {
+		set[strings.TrimSpace(strings.ToUpper(w))] = true
+	}
+	if len(set) > 0 && len(set) < 25 {
+		return wordSetID{}, nil, errors.New("need at least 25 words")
+	}
+
+	words = words[:0]
+	for w := range set {
+		words = append(words, w)
+	}
+	sort.Strings(words)
+
+	// Calculate the word set ID, a hash of the canonicalized word set.
+	h := sha1.New()
+	for _, w := range words {
+		io.WriteString(h, w)
+		h.Write([]byte{0x00})
+	}
+	idBytes := h.Sum(nil)
+	var id wordSetID
+	copy(id[:], idBytes)
+
+	if interned, ok := ws.byID[id]; ok {
+		return id, interned, nil
+	}
+	ws.byID[id] = words
+	return id, words, nil
+}

wordset_test.go

@@ -0,0 +1,43 @@
+package codenames
+
+import (
+	"bytes"
+	"encoding/json"
+	"io/ioutil"
+	"testing"
+)
+
+func TestWordSetCanonicalize(t *testing.T) {
+	b, err := ioutil.ReadFile("frontend/words.json")
+	if err != nil {
+		t.Fatal(err)
+	}
+	var defaultWordsets map[string][]string
+	err = json.NewDecoder(bytes.NewReader(b)).Decode(&defaultWordsets)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	internedSets := map[string][]string{}
+
+	var ws WordSets
+	for name, words := range defaultWordsets {
+		id, interned, err := ws.Canonicalize(words)
+		if err != nil {
+			t.Fatal(err)
+		}
+		t.Logf("%s : %s\n", name, id)
+		internedSets[name] = interned
+	}
+
+	for name, words := range defaultWordsets {
+		words2 := append([]string{}, words...)
+		_, interned, err := ws.Canonicalize(words2)
+		if err != nil {
+			t.Fatal(err)
+		}
+		if &internedSets[name][0] != &interned[0] {
+			t.Errorf("word set %q has different slice pointer 2nd canonicalization", name)
+		}
+	}
+}