本文介绍了一个用于将GB2312编码转换为Unicode编码的Java程序。该程序通过使用GBK编码作为中间步骤，实现了从GB2312到UTF-16BE和UTF-8的字符编码转换，并生成了包含GB2312编码汉字及其Unicode编码对照的手册。
这本手册里的字符与汉字编码列表由下面的程式所生成。
1
/** *//**
2
* GB2312Unicde.java
3* Copyright (c) 1997-2003 by Dr. Herong Yang
4
*/
5
import java.io.*;
6import java.nio.*;
7import java.nio.charset.*;
8class GB2312Unicde 
{
9static OutputStream out = null;
10
static char hexDigit[] = {'0', '1', '2', '3', '4', '5', '6', '7',
11
'8', '9', 'A', 'B', 'C', 'D', 'E', 'F'};
12static int b_out[] = {201,267,279,293,484,587,625,657,734,782,827,
13874,901,980,5590};
14static int e_out[] = {216,268,280,294,494,594,632,694,748,794,836,
15894,903,994,5594};
16public static void main(String[] args) {
17try {
18out = new FileOutputStream("gb2312.gb");
19writeCode();
20out.close();
21} catch (IOException e) {
22System.out.println(e.toString());
23}
24}
25public static void writeCode() throws IOException {
26boolean reserved = false;
27String name = null;
28// GB2312 is not supported by JDK. So I am using GBK.
29CharsetDecoder gbdc = Charset.forName("GBK").newDecoder();
30CharsetEncoder uxec = Charset.forName("UTF-16BE").newEncoder();
31CharsetEncoder u8ec = Charset.forName("UTF-8").newEncoder();
32ByteBuffer gbbb = null;
33ByteBuffer uxbb = null;
34ByteBuffer u8bb = null;
35CharBuffer cb = null;
36int count = 0;
37for (int i=1; i<=94; i++) {
38// Defining row settings
39if (i>=1 && i<=9) {
40reserved = false;
41name = "Graphic symbols";
42} else if (i>=10 && i<=15) {
43reserved = true;
44name = "Reserved";
45} else if (i>=16 && i<=55) {
46reserved = false;
47name = "Level 1 characters";
48} else if (i>=56 && i<=87) {
49reserved = false;
50name = "Level 2 characters";
51} else if (i>=88 && i<=94) {
52reserved = true;
53name = "Reserved";
54}
55// writing row title
56writeln();
57writeString("<p>");
58writeNumber(i);
59writeString(" Row: "+name);
60writeln();
61writeString("</p>");
62writeln();
63if (!reserved) {
64writeln();
65writeHeader();
66// looping through all characters in one row
67for (int j=1; j<=94; j++) {
68byte hi = (byte)(0xA0 + i);
69byte lo = (byte)(0xA0 + j);
70if (validGB(i,j)) {
71// getting GB, UTF-16BE, UTF-8 codes
72gbbb = ByteBuffer.wrap(new byte[]{hi,lo});
73try {
74cb = gbdc.decode(gbbb);
75uxbb = uxec.encode(cb);
76cb.rewind();
77u8bb = u8ec.encode(cb);
78} catch (CharacterCodingException e) {
79cb = null;
80uxbb = null;
81u8bb = null;
82}
83} else {
84cb = null;
85uxbb = null;
86u8bb = null;
87}
88writeNumber(i);
89writeNumber(j);
90writeString(" ");
91if (cb!=null) {
92writeByte(hi);
93writeByte(lo);
94writeString(" ");
95writeHex(hi);
96writeHex(lo);
97count++;
98} else {
99writeGBSpace();
100writeString(" null");
101}
102writeString(" ");
103writeByteBuffer(uxbb,2);
104writeString(" ");
105writeByteBuffer(u8bb,3);
106if (j%2 == 0) {
107writeln();
108} else {
109writeString(" ");
110}
111}
112writeFooter();
113}
114}
115System.out.println("Number of GB characters worte: "+count);
116}
117public static void writeln() throws IOException {
118out.write(0x0D);
119out.write(0x0A);
120}
121public static void writeByte(byte b) throws IOException {
122out.write(b & 0xFF);
123}
124public static void writeByteBuffer(ByteBuffer b, int l)
125throws IOException {
126int i = 0;
127if (b==null) {
128writeString("null");
129i = 2;
130} else {
131for (i=0; i<b.limit(); i++) writeHex(b.get(i));
132}
133for (int j=i; j<l; j++) writeString(" ");
134}
135public static void writeGBSpace() throws IOException {
136out.write(0xA1);
137out.write(0xA1);
138}
139public static void writeString(String s) throws IOException {
140if (s!=null) {
141for (int i=0; i<s.length(); i++) {
142out.write((int) (s.charAt(i) & 0xFF));
143}
144}
145}
146public static void writeNumber(int i) throws IOException {
147String s = "00" + String.valueOf(i);
148writeString(s.substring(s.length()-2,s.length()));
149}
150public static void writeHex(byte b) throws IOException {
151out.write((int) hexDigit[(b >> 4) & 0x0F]);
152out.write((int) hexDigit[b & 0x0F]);
153}
154public static void writeHeader() throws IOException {
155writeString("<pre>");
156writeln();
157writeString("Q.W. ");
158writeGBSpace();
159writeString(" GB Uni. UTF-8 ");
160writeString(" ");
161writeString("Q.W. ");
162writeGBSpace();
163writeString(" GB Uni. UTF-8 ");
164writeln();
165writeln();
166}
167public static void writeFooter() throws IOException {
168writeString("</pre>");
169writeln();
170}
171public static boolean validGB(int i,int j) {
172for (int l=0; l<b_out.length; l++) {
173if (i*100+j>=b_out[l] && i*100+j<=e_out[l]) return false;
174}
175return true;
176}
177}
/** *//**2
* GB2312Unicde.java3
* Copyright (c) 1997-2003 by Dr. Herong Yang4
*/5
import java.io.*;6
import java.nio.*;7
import java.nio.charset.*;8
class GB2312Unicde {9
static OutputStream out = null;10
static char hexDigit[] = {'0', '1', '2', '3', '4', '5', '6', '7',11
'8', '9', 'A', 'B', 'C', 'D', 'E', 'F'};12
static int b_out[] = {201,267,279,293,484,587,625,657,734,782,827,13
874,901,980,5590};14
static int e_out[] = {216,268,280,294,494,594,632,694,748,794,836,15
894,903,994,5594};16
public static void main(String[] args) {17
try {18
out = new FileOutputStream("gb2312.gb");19
writeCode();20
out.close();21
} catch (IOException e) {22
System.out.println(e.toString());23
}24
}25
public static void writeCode() throws IOException {26
boolean reserved = false;27
String name = null;28
// GB2312 is not supported by JDK. So I am using GBK.29
CharsetDecoder gbdc = Charset.forName("GBK").newDecoder();30
CharsetEncoder uxec = Charset.forName("UTF-16BE").newEncoder();31
CharsetEncoder u8ec = Charset.forName("UTF-8").newEncoder();32
ByteBuffer gbbb = null;33
ByteBuffer uxbb = null;34
ByteBuffer u8bb = null;35
CharBuffer cb = null;36
int count = 0;37
for (int i=1; i<=94; i++) {38
// Defining row settings39
if (i>=1 && i<=9) {40
reserved = false;41
name = "Graphic symbols";42
} else if (i>=10 && i<=15) {43
reserved = true;44
name = "Reserved";45
} else if (i>=16 && i<=55) {46
reserved = false;47
name = "Level 1 characters";48
} else if (i>=56 && i<=87) {49
reserved = false;50
name = "Level 2 characters";51
} else if (i>=88 && i<=94) {52
reserved = true;53
name = "Reserved";54
}55
// writing row title56
writeln();57
writeString("<p>");58
writeNumber(i);59
writeString(" Row: "+name);60
writeln();61
writeString("</p>");62
writeln();63
if (!reserved) {64
writeln();65
writeHeader();66
// looping through all characters in one row67
for (int j=1; j<=94; j++) {68
byte hi = (byte)(0xA0 + i);69
byte lo = (byte)(0xA0 + j);70
if (validGB(i,j)) {71
// getting GB, UTF-16BE, UTF-8 codes72
gbbb = ByteBuffer.wrap(new byte[]{hi,lo});73
try {74
cb = gbdc.decode(gbbb);75
uxbb = uxec.encode(cb);76
cb.rewind();77
u8bb = u8ec.encode(cb);78
} catch (CharacterCodingException e) {79
cb = null;80
uxbb = null;81
u8bb = null;82
}83
} else {84
cb = null;85
uxbb = null;86
u8bb = null;87
}88
writeNumber(i);89
writeNumber(j);90
writeString(" ");91
if (cb!=null) {92
writeByte(hi);93
writeByte(lo);94
writeString(" ");95
writeHex(hi);96
writeHex(lo);97
count++;98
} else {99
writeGBSpace();100
writeString(" null");101
}102
writeString(" ");103
writeByteBuffer(uxbb,2);104
writeString(" ");105
writeByteBuffer(u8bb,3);106
if (j%2 == 0) {107
writeln();108
} else {109
writeString(" ");110
}111
}112
writeFooter();113
}114
}115
System.out.println("Number of GB characters worte: "+count);116
}117
public static void writeln() throws IOException {118
out.write(0x0D);119
out.write(0x0A);120
}121
public static void writeByte(byte b) throws IOException {122
out.write(b & 0xFF);123
}124
public static void writeByteBuffer(ByteBuffer b, int l)125
throws IOException {126
int i = 0;127
if (b==null) {128
writeString("null");129
i = 2;130
} else {131
for (i=0; i<b.limit(); i++) writeHex(b.get(i));132
}133
for (int j=i; j<l; j++) writeString(" ");134
}135
public static void writeGBSpace() throws IOException {136
out.write(0xA1);137
out.write(0xA1);138
}139
public static void writeString(String s) throws IOException {140
if (s!=null) {141
for (int i=0; i<s.length(); i++) {142
out.write((int) (s.charAt(i) & 0xFF));143
}144
}145
}146
public static void writeNumber(int i) throws IOException {147
String s = "00" + String.valueOf(i);148
writeString(s.substring(s.length()-2,s.length()));149
}150
public static void writeHex(byte b) throws IOException {151
out.write((int) hexDigit[(b >> 4) & 0x0F]);152
out.write((int) hexDigit[b & 0x0F]);153
}154
public static void writeHeader() throws IOException {155
writeString("<pre>");156
writeln();157
writeString("Q.W. ");158
writeGBSpace();159
writeString(" GB Uni. UTF-8 ");160
writeString(" ");161
writeString("Q.W. ");162
writeGBSpace();163
writeString(" GB Uni. UTF-8 ");164
writeln();165
writeln();166
}167
public static void writeFooter() throws IOException {168
writeString("</pre>");169
writeln();170
}171
public static boolean validGB(int i,int j) {172
for (int l=0; l<b_out.length; l++) {173
if (i*100+j>=b_out[l] && i*100+j<=e_out[l]) return false;174
}175
return true;176
}177
}
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
