Add package identify to get language identification from Google

identify/id.go

@@ -0,0 +1,47 @@
+package identify
+
+import (
+	"os/exec"
+	"regexp"
+)
+
+
+type Indetification struct {
+	word string
+	Identifier func(*Indetification) (string, error)
+}
+
+func New(word string) Indetification {
+	return Indetification{
+		word: word,
+		Identifier: executeTransShell,
+	}
+}
+
+func executeTransShell(i *Indetification) (string, error) {
+	config := []string{
+		"-no-ansi",
+		"-id",
+		i.word,
+	}
+	outBytes, err := exec.Command("trans", config...).Output()
+	return string(outBytes), err
+}
+
+func (i Indetification) Identify() string {
+	output, err := i.Identifier(&i)
+	if err != nil {
+		panic(err)
+	}
+	return parseTransIdentifyOutput(output)
+}
+
+func parseTransIdentifyOutput(out string) string {
+	re := regexp.MustCompile(`(?m)^Code\s+(\w+)$`)
+	result := ""
+	matches := re.FindStringSubmatch(out)
+	if (len(matches) >= 2) {
+		result = matches[1]
+	}
+	return result
+}

identify/id_test.go

@@ -0,0 +1,86 @@
+package identify_test
+
+import (
+	"testing"
+	"."
+	"os"
+)
+
+
+type testData struct {
+	word string
+	identifyOutput string
+	expectedCode string
+}
+
+func TestIdentify(t *testing.T) {
+	cases := []testData{
+		testData{
+"macska",
+`Magyar
+Name                  Hungarian
+Family                Uralic
+Writing system        Latin
+Code                  hu
+ISO 639-3             hun
+SIL                   http://www-01.sil.org/iso639-3/documentation.asp?id=hun
+Glottolog             http://glottolog.org/resource/languoid/id/hung1274
+Wikipedia             http://en.wikipedia.org/wiki/Hungarian_language
+`,
+"hu",
+		},
+		testData{
+"cat",
+`English
+Name                  English
+Family                Indo-European
+Writing system        Latin
+Code                  en
+ISO 639-3             eng
+SIL                   http://www-01.sil.org/iso639-3/documentation.asp?id=eng
+Glottolog             http://glottolog.org/resource/languoid/id/stan1293
+Wikipedia             http://en.wikipedia.org/wiki/English_language
+`,
+"en",
+		},
+		testData{
+"szofisztikált",
+`Magyar
+Name                  Hungarian
+Family                Uralic
+Writing system        Latin
+Code                  hu
+ISO 639-3             hun
+SIL                   http://www-01.sil.org/iso639-3/documentation.asp?id=hun
+Glottolog             http://glottolog.org/resource/languoid/id/hung1274
+Wikipedia             http://en.wikipedia.org/wiki/Hungarian_language
+`,
+"hu",
+		},
+		testData{
+"distribute",
+`English
+Name                  English
+Family                Indo-European
+Writing system        Latin
+Code                  en
+ISO 639-3             eng
+SIL                   http://www-01.sil.org/iso639-3/documentation.asp?id=eng
+Glottolog             http://glottolog.org/resource/languoid/id/stan1293
+Wikipedia             http://en.wikipedia.org/wiki/English_language
+`,
+"en",
+		},
+	}
+	for _, data := range cases {
+		id := identify.New(data.word)
+		if _, ok := os.LookupEnv("NOMOCK"); !ok {
+			id.Identifier = func(i *identify.Indetification) (string, error) {
+				return data.identifyOutput, nil
+			}
+		}
+		if id.Identify() != data.expectedCode {
+			t.Errorf("Word '%s' should identify to '%s'!", data.word, data.expectedCode)
+		}
+	}
+}