android 汉字转字节,安卓汉字转拼音

原来研究的小项目，现在简单整理了一下，免费贡献给大家！

package com.tool.hz2py;

import android.os.Bundle;

import android.app.Activity;

import android.view.Menu;

import android.widget.TextView;

public class MainActivity extends Activity {

protected Hz2py hz2py;

@Override

protected void onCreate(Bundle savedInstanceState) {

super.onCreate(savedInstanceState);

setContentView(R.layout.activity_main);

hz2py = new Hz2py();

TextView view = (TextView) findViewById(R.id.text);

view.setText(hz2py.hz2py("汉字转拼音"));

}

@Override

public boolean onCreateOptionsMenu(Menu menu) {

// Inflate the menu; this adds items to the action bar if it is present.

getMenuInflater().inflate(R.menu.main, menu);

return true;

}

}

jni类:Hz2py

package com.tool.hz2py;

public class Hz2py {

static {

System.loadLibrary("Hz2py");

};

public native String hz2py(String text);

}

下面是C++头文件和代码

/* DO NOT EDIT THIS FILE - it is machine generated */

#include

/* Header for class com_tool_hz2py_Hz2py */

#ifndef _Included_com_tool_hz2py_Hz2py

#define _Included_com_tool_hz2py_Hz2py

#ifdef __cplusplus

extern "C" {

#endif

/*

* Class: com_tool_hz2py_Hz2py

* Method: hz2py

* Signature: (Ljava/lang/String;)Ljava/lang/String;

*/

JNIEXPORT jstring JNICALL Java_com_tool_hz2py_Hz2py_hz2py

(JNIEnv *, jobject, jstring);

#ifdef __cplusplus

}

#endif

#endif

#include "hz2py.h"

#include

#include "com_tool_hz2py_Hz2py.h"

#define HZ2PY_UTF8_CHECK_LENGTH 20

#define HZ2PY_FILE_READ_BUF_ARRAY_SIZE 1

#define HZ2PY_INPUT_BUF_ARRAY_SIZE 1024

#define HZ2PY_OUTPUT_BUF_ARRAY_SIZE 2048

#define HZ2PY_STR_COPY(to, from, count) \

ok = 1;\

i = 0;\

_tmp = from;\

while(i < count)\

{\

if (*_tmp == '\0')\

{\

ok = 0;\

break;\

}\

_tmp ++;\

i ++;\

}\

if (ok)\

{\

i = 0;\

while(i < count)\

{\

*to = *from;\

to ++;\

from ++;\

i ++;\

}\

}\

else\

{\

if (overage_buff != NULL)\

{\

while(*from != '\0')\

{\

*overage_buff = *from;\

from ++;\

}\

}\

break;\

}

//将utf8编码的字符串中的汉字解成拼音

// in 输入

// out 输出

// first_letter_only 是否只输出拼音首字母

// polyphone_support 是否输出多音字

// add_blank 是否在拼音之间追加空格

// convert_double_char 是否转换全角字符为半角字符

// overage_buff 末尾如果有多余的不能组成完整utf8字符的字节，将写到overage_buff，传NULL将输出到out

void utf8_to_pinyin(char *in, char *out, int first_letter_only,

int polyphone_support, int add_blank, int convert_double_char,

char *overage_buff) {

int i = 0;

char *utf = in;

char *_tmp;

char *_tmp2;

char py_tmp[30] = "";

char py_tmp2[30] = "";

char *out_start_flag = out;

int uni;

int ok = 0;

while (*utf != '\0') {

if ((*utf >> 7) == 0) {

HZ2PY_STR_COPY(out, utf, 1);

//如果为一个字节加上#号分隔

*out = '#'; //用#号做为分隔符

out++;

//去掉其它的英文只留汉字

//只能搜索到汉字拼音里面字母

//out--;

//*out = ' ';

}

//两个字节

else if ((*utf & 0xE0) == 0xC0) {

HZ2PY_STR_COPY(out, utf, 2);

}

//三个字节

else if ((*utf & 0xF0) == 0xE0) {

if (*(utf + 1) != '\0' && *(utf + 2) != '\0') {

uni = (((int) (*utf & 0x0F)) << 12)

(((int) (*(utf + 1) & 0x3F)) << 6)

(*(utf + 2) & 0x3F);

if (uni > 19967 && uni < 40870) {

memset(py_tmp, '\0', 30);

memset(py_tmp2, '\0', 30);

strcpy(py_tmp, _pinyin_table_[uni - 19968]);

_tmp = py_tmp;

_tmp2 = py_tmp2;

if (first_letter_only == 1) {

*_tmp2 = *_tmp;

_tmp++;

_tmp2++;

while (*_tmp != '\0') {

if (*_tmp == '' *(_tmp - 1) == '') {

*_tmp2 = *_tmp;

_tmp2++;

}

_tmp++;

}

} else {

strcpy(py_tmp2, py_tmp);

}

_tmp2 = py_tmp2;

if (polyphone_support == 0) {

while (*_tmp2 != '\0') {

if (*_tmp2 == '') {

*_tmp2 = '\0';

break;

}

_tmp2++;

}

_tmp2 = py_tmp2;

}

strcpy(out, _tmp2);

out += strlen(_tmp2);

if (add_blank) {

*out = '#'; //用#号做为分隔符

out++;

}

utf += 3;

} else if (convert_double_char && uni > 65280 && uni < 65375) {

*out = uni - 65248;

out++;

utf += 3;

} else if (convert_double_char && uni == 12288) {

*out = 32;

out++;

utf += 3;

} else {

HZ2PY_STR_COPY(out, utf, 3);

}

} else {

HZ2PY_STR_COPY(out, utf, 3);

}

}

//四个字节

else if ((*utf & 0xF8) == 0xF0) {

HZ2PY_STR_COPY(out, utf, 4);

}

//五个字节

else if ((*utf & 0xFC) == 0xF8) {

HZ2PY_STR_COPY(out, utf, 5);

}

//六个字节

else if ((*utf & 0xFE) == 0xFC) {

HZ2PY_STR_COPY(out, utf, 6);

} else {

if (overage_buff != NULL) {

*overage_buff = *utf;

overage_buff++;

} else {

HZ2PY_STR_COPY(out, utf, 1);

}

break;

}

}

}

//判断一个字符串是否为utf8编码

int is_utf8_string(char *utf) {

int length = strlen(utf);

int check_sub = 0;

int i = 0;

if (length > HZ2PY_UTF8_CHECK_LENGTH) {

length = HZ2PY_UTF8_CHECK_LENGTH;

}

for (; i < length; i++) {

if (check_sub == 0) {

if ((utf[i] >> 7) == 0) {

continue;

} else if ((utf[i] & 0xE0) == 0xC0) {

check_sub = 1;

} else if ((utf[i] & 0xF0) == 0xE0) {

check_sub = 2;

} else if ((utf[i] & 0xF8) == 0xF0) {

check_sub = 3;

} else if ((utf[i] & 0xFC) == 0xF8) {

check_sub = 4;

} else if ((utf[i] & 0xFE) == 0xFC) {

check_sub = 5;

} else {

return 0;

}

} else {

if ((utf[i] & 0xC0) != 0x80) {

return 0;

}

check_sub--;

}

}

return 1;

}

int hztpy(const char *read_buff, char *outbuf) {

char overage_buff[7] = { 0 };

char *_tmp = NULL;

char inbuf[HZ2PY_INPUT_BUF_ARRAY_SIZE] = { 0 };

int add_blank = 1;

int polyphone_support = 1;

int first_letter_only = 0;

int convert_double_char = 0;

// first_letter_only 是否只输出拼音首字母

// polyphone_support 是否输出多音字

// add_blank 是否在拼音之间追加空格

// convert_double_char 是否转换全角字符为半角字符

// overage_buff 末尾如果有多余的不能组成完整utf8字符的字节，将写到overage_buff，传NULL将输出到out

_tmp = inbuf;

if (strlen(overage_buff)) {

strcpy(_tmp, overage_buff);

_tmp += strlen(overage_buff);

memset(overage_buff, '\0', 7);

}

strcpy(_tmp, read_buff);

if (!is_utf8_string(inbuf)) {

return -1;

}

utf8_to_pinyin(inbuf, outbuf, first_letter_only, polyphone_support,

add_blank, convert_double_char, overage_buff);

return 1;

}

JNIEXPORT jstring JNICALL Java_com_tool_hz2py_Hz2py_hz2py(JNIEnv *env,

jobject thiz, jstring text) {

const char* pText = env->GetStringUTFChars(text, 0);

char* oText = new char[512];//256中文

memset(oText,0,512);

hztpy(pText,oText);

jstring returnText = env->NewStringUTF(oText);

env->ReleaseStringUTFChars(text,pText);

delete oText;

return returnText;

}

头文件有点大，我直接上传给大家下载好了。hz2py直接编译好的.so共享库和java文件，注意，包名只能用这个，不能更改。libHz2py