Merge pull request #44 from shogo82148/support-nbest

support N-Best parsing

example_test.go

@@ -59,6 +59,60 @@ func ExampleMeCab_ParseLattice() {
 	// EOS
 }
 
+func ExampleMeCab_ParseLattice_nBest() {
+	options := map[string]string{}
+	if path := os.Getenv("MECABRC_PATH"); path != "" {
+		options["rcfile"] = path
+	}
+
+	tagger, err := mecab.New(options)
+	if err != nil {
+		panic(err)
+	}
+	defer tagger.Destroy()
+
+	lattice, err := mecab.NewLattice()
+	if err != nil {
+		panic(err)
+	}
+
+	lattice.SetSentence("こんにちは世界")
+	lattice.AddRequestType(mecab.RequestTypeNBest)
+	err = tagger.ParseLattice(lattice)
+	if err != nil {
+		panic(err)
+	}
+	for i := 0; i < 5; i++ {
+		fmt.Println(lattice.String())
+		if !lattice.Next() {
+			break
+		}
+	}
+	// Output:
+	// こんにちは	感動詞,*,*,*,*,*,こんにちは,コンニチハ,コンニチワ
+	// 世界	名詞,一般,*,*,*,*,世界,セカイ,セカイ
+	// EOS
+	//
+	// こんにちは	感動詞,*,*,*,*,*,こんにちは,コンニチハ,コンニチワ
+	// 世界	名詞,一般,*,*,*,*,世界,セカイ,セカイ
+	// EOS
+	//
+	// こんにちは	感動詞,*,*,*,*,*,こんにちは,コンニチハ,コンニチワ
+	// 世	名詞,一般,*,*,*,*,世,ヨ,ヨ
+	// 界	名詞,接尾,一般,*,*,*,界,カイ,カイ
+	// EOS
+	//
+	// こんにちは	感動詞,*,*,*,*,*,こんにちは,コンニチハ,コンニチワ
+	// 世	名詞,一般,*,*,*,*,世,ヨ,ヨ
+	// 界	名詞,固有名詞,地域,一般,*,*,界,サカイ,サカイ
+	// EOS
+	//
+	// こんにちは	感動詞,*,*,*,*,*,こんにちは,コンニチハ,コンニチワ
+	// 世	名詞,接尾,助数詞,*,*,*,世,セイ,セイ
+	// 界	名詞,接尾,一般,*,*,*,界,カイ,カイ
+	// EOS
+}
+
 func ExampleMeCab_ParseToNode() {
 	options := map[string]string{}
 	if path := os.Getenv("MECABRC_PATH"); path != "" {

lattice.go

@@ -9,6 +9,41 @@ import (
 	"unsafe"
 )
 
+// RequestType is a request type.
+type RequestType int
+
+const (
+	// RequestTypeOneBest is a request type for one best result.
+	RequestTypeOneBest RequestType = 1
+
+	// RequestTypeNBest is a request type for N-best results.
+	RequestTypeNBest RequestType = 2
+
+	// RequestTypePartial enables a partial parsing mode.
+	// When this flag is set, the input |sentence| needs to be written
+	// in partial parsing format.
+	RequestTypePartial RequestType = 4
+
+	// RequestTypeMarginalProb is a request type for marginal probability.
+	// Set this flag if you want to obtain marginal probabilities.
+	// Marginal probability is set in [Node.Prob].
+	// The parsing speed will get 3-5 times slower than the default mode.
+	RequestTypeMarginalProb RequestType = 8
+
+	// RequestTypeMorphsToNBest is a request type for alternative results.
+	// Set this flag if you want to obtain alternative results.
+	// Not implemented.
+	RequestTypeAlternative RequestType = 16
+
+	// RequestTypeAllMorphs is a request type for all morphs.
+	RequestTypeAllMorphs RequestType = 32
+
+	// RequestTypeAllocateSentence is a request type for allocating sentence.
+	// When this flag is set, tagger internally copies the body of passed
+	// sentence into internal buffer.
+	RequestTypeAllocateSentence RequestType = 64
+)
+
 var errLatticeNotAvailable = errors.New("mecab: lattice is not available")
 
 type lattice struct {
@@ -26,7 +61,8 @@ func newLattice(l *C.mecab_lattice_t) *lattice {
 // It is a marker that a lattice must not be copied after the first use.
 // See https://github.com/golang/go/issues/8005#issuecomment-190753527
 // for details.
-func (*lattice) Lock() {}
+func (*lattice) Lock()   {}
+func (*lattice) Unlock() {}
 
 func finalizeLattice(l *lattice) {
 	if l.lattice != nil {
@@ -123,7 +159,7 @@ func (l Lattice) SetSentence(s string) {
 	input := C.CString(s)
 	defer C.free(unsafe.Pointer(input))
 
-	C.mecab_lattice_add_request_type(l.l.lattice, 64) // MECAB_ALLOCATE_SENTENCE = 64
+	C.mecab_lattice_add_request_type(l.l.lattice, C.int(RequestTypeAllocateSentence)) // MECAB_ALLOCATE_SENTENCE = 64
 	C.mecab_lattice_set_sentence2(l.l.lattice, input, length)
 	runtime.KeepAlive(l.l)
 }
@@ -136,3 +172,41 @@ func (l Lattice) String() string {
 	runtime.KeepAlive(l.l)
 	return s
 }
+
+// Next obtains next-best result. The internal linked list structure is updated.
+// You should set [RequestTypeNBest] in advance.
+// Return false if no more results are available or [RequestType] is invalid.
+func (l Lattice) Next() bool {
+	if l.l.lattice == nil {
+		panic(errLatticeNotAvailable)
+	}
+	next := C.mecab_lattice_next(l.l.lattice) != 0
+	runtime.KeepAlive(l)
+	return next
+}
+
+// RequestType returns the request type.
+func (l Lattice) RequestType() RequestType {
+	if l.l.lattice == nil {
+		panic(errLatticeNotAvailable)
+	}
+	return RequestType(C.mecab_lattice_get_request_type(l.l.lattice))
+}
+
+// SetRequestType sets the request type.
+func (l Lattice) SetRequestType(t RequestType) {
+	if l.l.lattice == nil {
+		panic(errLatticeNotAvailable)
+	}
+	C.mecab_lattice_add_request_type(l.l.lattice, C.int(t))
+	runtime.KeepAlive(l)
+}
+
+// AddRequestType adds the request type.
+func (l Lattice) AddRequestType(t RequestType) {
+	if l.l.lattice == nil {
+		panic(errLatticeNotAvailable)
+	}
+	C.mecab_lattice_add_request_type(l.l.lattice, C.int(t))
+	runtime.KeepAlive(l)
+}

mecab.go

@@ -29,7 +29,8 @@ func newMeCab(m *C.mecab_t) *mecab {
 // It is a marker that a mecab must not be copied after the first use.
 // See https://github.com/golang/go/issues/8005#issuecomment-190753527
 // for details.
-func (*mecab) Lock() {}
+func (*mecab) Lock()   {}
+func (*mecab) Unlock() {}
 
 func finalizeMeCab(m *mecab) {
 	if m.mecab != nil {
@@ -90,7 +91,8 @@ func (m MeCab) Destroy() {
 	m.m.mecab = nil
 }
 
-// Parse parses the string and returns the result as string
+// Parse parses the string and returns the result as string.
+// Parse is not safe for concurrent use by multiple goroutines.
 func (m MeCab) Parse(s string) (string, error) {
 	if m.m.mecab == nil {
 		panic(errMeCabNotAvailable)
@@ -108,7 +110,8 @@ func (m MeCab) Parse(s string) (string, error) {
 	return C.GoString(result), nil
 }
 
-// ParseToString is alias of Parse
+// ParseToString is alias of [Parse].
+// ParseToString is not safe for concurrent use by multiple goroutines.
 func (m MeCab) ParseToString(s string) (string, error) {
 	if m.m.mecab == nil {
 		panic(errMeCabNotAvailable)
@@ -117,6 +120,8 @@ func (m MeCab) ParseToString(s string) (string, error) {
 }
 
 // ParseLattice parses the lattice and returns the result as string.
+// ParseLattice is safe for concurrent use by multiple goroutines.
+// Create a lattice for each goroutine.
 func (m MeCab) ParseLattice(lattice Lattice) error {
 	if m.m.mecab == nil {
 		panic(errMeCabNotAvailable)
@@ -128,7 +133,8 @@ func (m MeCab) ParseLattice(lattice Lattice) error {
 	return nil
 }
 
-// ParseToNode parses the string and returns the result as Node
+// ParseToNode parses the string and returns the result as [Node].
+// ParseToNode is not safe for concurrent use by multiple goroutines.
 func (m MeCab) ParseToNode(s string) (Node, error) {
 	if m.m.mecab == nil {
 		panic(errMeCabNotAvailable)
@@ -148,6 +154,7 @@ func (m MeCab) ParseToNode(s string) (Node, error) {
 	}, nil
 }
 
+// Error returns the error of MeCab.
 func (m MeCab) Error() error {
 	if m.m.mecab == nil {
 		panic(errMeCabNotAvailable)

node.go

@@ -90,7 +90,7 @@ func (node Node) Next() Node {
 	return Node{node: (*C.mecab_node_t)(node.node.next)}
 }
 
-// ENext resturns a node which ends same position
+// ENext returns a node which ends same position
 func (node Node) ENext() Node {
 	return Node{
 		node:    (*C.mecab_node_t)(node.node.enext),
@@ -99,7 +99,7 @@ func (node Node) ENext() Node {
 	}
 }
 
-// BNext resturns a node which begins same position
+// BNext returns a node which begins same position
 func (node Node) BNext() Node {
 	return Node{
 		node:    (*C.mecab_node_t)(node.node.bnext),