Initial commit

mystem.go

@@ -0,0 +1,77 @@
+package mystem
+
+/*
+#cgo LDFLAGS: -lmystem_c_binding
+#include "mystem.h"
+*/
+import "C"
+
+import (
+	"unsafe"
+)
+
+import "encoding/binary"
+
+func StringToSymbols(word string) []C.ushort {
+	var (
+		symbols []C.ushort
+	)
+	for _, char := range word {
+		symbols = append(symbols, C.ushort(char))
+	}
+	return symbols
+}
+
+func SymbolsToString(symbols *C.TSymbol, length C.int) string {
+	var (
+		runes []rune
+	)
+	bytes := C.GoBytes(unsafe.Pointer(symbols), length*2)
+	for len(bytes) != 0 {
+		letter := binary.LittleEndian.Uint16(bytes[:2])
+		runes = append(runes, rune(letter))
+		bytes = bytes[2:]
+	}
+	return string(runes)
+}
+
+type Analyses struct {
+	handle unsafe.Pointer
+}
+
+type Lemma struct {
+	handle unsafe.Pointer
+}
+
+func NewAnalyses(word string) *Analyses {
+	cWord := StringToSymbols(word)
+	cWordLength := len(cWord)
+	analyses := new(Analyses)
+	handle := C.MystemAnalyze((*C.TSymbol)(unsafe.Pointer(&cWord[0])), C.int(cWordLength))
+	analyses.handle = handle
+	return analyses
+}
+
+func (analyses *Analyses) Length() int {
+	length := C.MystemAnalysesCount(analyses.handle)
+	return (int)(length)
+}
+
+func (analyses *Analyses) Close() {
+	C.MystemDeleteAnalyses(analyses.handle)
+}
+
+func (analyses *Analyses) GetLemma(i int) *Lemma {
+	lemma := new(Lemma)
+	handle := C.MystemLemma(unsafe.Pointer(analyses.handle), (C.int)(i))
+	lemma.handle = handle
+	return lemma
+}
+
+func (lemma *Lemma) TextLength() C.int {
+	return C.MystemLemmaTextLen(lemma.handle)
+}
+
+func (lemma *Lemma) Text() string {
+	return SymbolsToString(C.MystemLemmaText(lemma.handle), lemma.TextLength())
+}

mystem.h

@@ -0,0 +1,34 @@
+typedef void MystemAnalysesHandle;
+typedef void MystemLemmaHandle;
+typedef void MystemFormsHandle;
+typedef void MystemFormHandle;
+typedef unsigned short int TSymbol;
+
+MystemAnalysesHandle* MystemAnalyze(TSymbol* word, int len);
+void MystemDeleteAnalyses(MystemAnalysesHandle* analyses);
+int MystemAnalysesCount(MystemAnalysesHandle* analyses);
+
+MystemLemmaHandle* MystemLemma(MystemAnalysesHandle* analyses, int i);
+
+TSymbol* MystemLemmaText(MystemLemmaHandle* lemma);
+int     MystemLemmaTextLen(MystemLemmaHandle* lemma);
+TSymbol* MystemLemmaForm(MystemLemmaHandle* lemma);
+int     MystemLemmaFormLen(MystemLemmaHandle* lemma);
+int     MystemLemmaQuality(MystemLemmaHandle* lemma);
+char*   MystemLemmaStemGram(MystemLemmaHandle* lemma);
+char**  MystemLemmaFlexGram(MystemLemmaHandle* lemma);
+int     MystemLemmaFlexGramNum(MystemLemmaHandle* lemma);
+int     MystemLemmaFlexLen(MystemLemmaHandle* lemma);
+int     MystemLemmaRuleId(MystemLemmaHandle* lemma);
+
+MystemFormsHandle* MystemGenerate(MystemLemmaHandle* lemma);
+void MystemDeleteForms(MystemFormsHandle* forms);
+int MystemFormsCount(MystemFormsHandle* forms);
+
+MystemFormHandle* MystemForm(MystemFormsHandle* forms, int i);
+
+TSymbol* MystemFormText(MystemFormHandle* form);
+int     MystemFormTextLen(MystemFormHandle* form);
+char*   MystemFormStemGram(MystemFormHandle* form);
+char**  MystemFormFlexGram(MystemFormHandle* form);
+int     MystemFormFlexGramNum(MystemFormHandle* form);