jdk/src/share/npt/utf.c
changeset 24991 e10a57d6e9d0
parent 24990 c9ca62c6a329
parent 24989 7eefcdab7678
child 24998 0de5415b1d2a
--- a/jdk/src/share/npt/utf.c	Thu Jun 19 11:22:18 2014 -0700
+++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
@@ -1,500 +0,0 @@
-/*
- * Copyright (c) 2004, 2013, Oracle and/or its affiliates. All rights reserved.
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * This code is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License version 2 only, as
- * published by the Free Software Foundation.  Oracle designates this
- * particular file as subject to the "Classpath" exception as provided
- * by Oracle in the LICENSE file that accompanied this code.
- *
- * This code is distributed in the hope that it will be useful, but WITHOUT
- * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
- * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
- * version 2 for more details (a copy is included in the LICENSE file that
- * accompanied this code).
- *
- * You should have received a copy of the GNU General Public License version
- * 2 along with this work; if not, write to the Free Software Foundation,
- * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
- *
- * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
- * or visit www.oracle.com if you need additional information or have any
- * questions.
- */
-
-/* Misc functions for conversion of Unicode and UTF-8 and platform encoding */
-
-#include <stdio.h>
-#include <stddef.h>
-#include <stdlib.h>
-#include <stdarg.h>
-#include <string.h>
-#include <ctype.h>
-
-#include "jni.h"
-
-#include "utf.h"
-
-/*
- * Error handler
- */
-void
-utfError(char *file, int line, char *message)
-{
-    (void)fprintf(stderr, "UTF ERROR [\"%s\":%d]: %s\n", file, line, message);
-    abort();
-}
-
-/*
- * Convert UTF-8 to UTF-16
- *    Returns length or -1 if output overflows.
- */
-int JNICALL
-utf8ToUtf16(struct UtfInst *ui, jbyte *utf8, int len, unsigned short *output, int outputMaxLen)
-{
-    int outputLen;
-    int i;
-
-    UTF_ASSERT(utf8);
-    UTF_ASSERT(len>=0);
-    UTF_ASSERT(output);
-    UTF_ASSERT(outputMaxLen>0);
-
-    i = 0;
-    outputLen = 0;
-    while ( i<len ) {
-        unsigned code, x, y, z;
-
-        if ( outputLen >= outputMaxLen ) {
-            return -1;
-        }
-        x = (unsigned char)utf8[i++];
-        code = x;
-        if ( (x & 0xE0)==0xE0 ) {
-            y = (unsigned char)utf8[i++];
-            z = (unsigned char)utf8[i++];
-            code = ((x & 0xF)<<12) + ((y & 0x3F)<<6) + (z & 0x3F);
-        } else if ( (x & 0xC0)==0xC0 ) {
-            y = (unsigned char)utf8[i++];
-            code = ((x & 0x1F)<<6) + (y & 0x3F);
-        }
-        output[outputLen++] = code;
-    }
-    return outputLen;
-}
-
-/*
- * Convert UTF-16 to UTF-8 Modified
- *    Returns length or -1 if output overflows.
- */
-int JNICALL
-utf16ToUtf8m(struct UtfInst *ui, unsigned short *utf16, int len, jbyte *output, int outputMaxLen)
-{
-    int i;
-    int outputLen;
-
-    UTF_ASSERT(utf16);
-    UTF_ASSERT(len>=0);
-    UTF_ASSERT(output);
-    UTF_ASSERT(outputMaxLen>0);
-
-    outputLen = 0;
-    for (i = 0; i < len; i++) {
-        unsigned code;
-
-        code = utf16[i];
-        if ( code >= 0x0001 && code <= 0x007F ) {
-            if ( outputLen + 1 >= outputMaxLen ) {
-                return -1;
-            }
-            output[outputLen++] = code;
-        } else if ( code == 0 || ( code >= 0x0080 && code <= 0x07FF ) ) {
-            if ( outputLen + 2 >= outputMaxLen ) {
-                return -1;
-            }
-            output[outputLen++] = ((code>>6) & 0x1F) | 0xC0;
-            output[outputLen++] = (code & 0x3F) | 0x80;
-        } else if ( code >= 0x0800 && code <= 0xFFFF ) {
-            if ( outputLen + 3 >= outputMaxLen ) {
-                return -1;
-            }
-            output[outputLen++] = ((code>>12) & 0x0F) | 0xE0;
-            output[outputLen++] = ((code>>6) & 0x3F) | 0x80;
-            output[outputLen++] = (code & 0x3F) | 0x80;
-        }
-    }
-    output[outputLen] = 0;
-    return outputLen;
-}
-
-int JNICALL
-utf16ToUtf8s(struct UtfInst *ui, unsigned short *utf16, int len, jbyte *output, int outputMaxLen)
-{
-    return -1; /* FIXUP */
-}
-
-/* Determine length of this Standard UTF-8 in Modified UTF-8.
- *    Validation is done of the basic UTF encoding rules, returns
- *    length (no change) when errors are detected in the UTF encoding.
- *
- *    Note: Accepts Modified UTF-8 also, no verification on the
- *          correctness of Standard UTF-8 is done. e,g, 0xC080 input is ok.
- */
-int JNICALL
-utf8sToUtf8mLength(struct UtfInst *ui, jbyte *string, int length)
-{
-    int newLength;
-    int i;
-
-    newLength = 0;
-    for ( i = 0 ; i < length ; i++ ) {
-        unsigned byte;
-
-        byte = (unsigned char)string[i];
-        if ( (byte & 0x80) == 0 ) { /* 1byte encoding */
-            newLength++;
-            if ( byte == 0 ) {
-                newLength++; /* We gain one byte in length on NULL bytes */
-            }
-        } else if ( (byte & 0xE0) == 0xC0 ) { /* 2byte encoding */
-            /* Check encoding of following bytes */
-            if ( (i+1) >= length || (string[i+1] & 0xC0) != 0x80 ) {
-                break; /* Error condition */
-            }
-            i++; /* Skip next byte */
-            newLength += 2;
-        } else if ( (byte & 0xF0) == 0xE0 ) { /* 3byte encoding */
-            /* Check encoding of following bytes */
-            if ( (i+2) >= length || (string[i+1] & 0xC0) != 0x80
-                                 || (string[i+2] & 0xC0) != 0x80 ) {
-                break; /* Error condition */
-            }
-            i += 2; /* Skip next two bytes */
-            newLength += 3;
-        } else if ( (byte & 0xF8) == 0xF0 ) { /* 4byte encoding */
-            /* Check encoding of following bytes */
-            if ( (i+3) >= length || (string[i+1] & 0xC0) != 0x80
-                                 || (string[i+2] & 0xC0) != 0x80
-                                 || (string[i+3] & 0xC0) != 0x80 ) {
-                break; /* Error condition */
-            }
-            i += 3; /* Skip next 3 bytes */
-            newLength += 6; /* 4byte encoding turns into 2 3byte ones */
-        } else {
-            break; /* Error condition */
-        }
-    }
-    if ( i != length ) {
-        /* Error in finding new length, return old length so no conversion */
-        /* FIXUP: ERROR_MESSAGE? */
-        return length;
-    }
-    return newLength;
-}
-
-/* Convert Standard UTF-8 to Modified UTF-8.
- *    Assumes the UTF-8 encoding was validated by utf8mLength() above.
- *
- *    Note: Accepts Modified UTF-8 also, no verification on the
- *          correctness of Standard UTF-8 is done. e,g, 0xC080 input is ok.
- */
-void JNICALL
-utf8sToUtf8m(struct UtfInst *ui, jbyte *string, int length, jbyte *newString, int newLength)
-{
-    int i;
-    int j;
-
-    j = 0;
-    for ( i = 0 ; i < length ; i++ ) {
-        unsigned byte1;
-
-        byte1 = (unsigned char)string[i];
-
-        /* NULL bytes and bytes starting with 11110xxx are special */
-        if ( (byte1 & 0x80) == 0 ) { /* 1byte encoding */
-            if ( byte1 == 0 ) {
-                /* Bits out: 11000000 10000000 */
-                newString[j++] = (jbyte)0xC0;
-                newString[j++] = (jbyte)0x80;
-            } else {
-                /* Single byte */
-                newString[j++] = byte1;
-            }
-        } else if ( (byte1 & 0xE0) == 0xC0 ) { /* 2byte encoding */
-            newString[j++] = byte1;
-            newString[j++] = string[++i];
-        } else if ( (byte1 & 0xF0) == 0xE0 ) { /* 3byte encoding */
-            newString[j++] = byte1;
-            newString[j++] = string[++i];
-            newString[j++] = string[++i];
-        } else if ( (byte1 & 0xF8) == 0xF0 ) { /* 4byte encoding */
-            /* Beginning of 4byte encoding, turn into 2 3byte encodings */
-            unsigned byte2, byte3, byte4, u21;
-
-            /* Bits in: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
-            byte2 = (unsigned char)string[++i];
-            byte3 = (unsigned char)string[++i];
-            byte4 = (unsigned char)string[++i];
-            /* Reconstruct full 21bit value */
-            u21  = (byte1 & 0x07) << 18;
-            u21 += (byte2 & 0x3F) << 12;
-            u21 += (byte3 & 0x3F) << 6;
-            u21 += (byte4 & 0x3F);
-            /* Bits out: 11101101 1010xxxx 10xxxxxx */
-            newString[j++] = (jbyte)0xED;
-            newString[j++] = (jbyte)(0xA0 + (((u21 >> 16) - 1) & 0x0F));
-            newString[j++] = (jbyte)(0x80 + ((u21 >> 10) & 0x3F));
-            /* Bits out: 11101101 1011xxxx 10xxxxxx */
-            newString[j++] = (jbyte)0xED;
-            newString[j++] = (jbyte)(0xB0 + ((u21 >>  6) & 0x0F));
-            newString[j++] = byte4;
-        }
-    }
-    UTF_ASSERT(i==length);
-    UTF_ASSERT(j==newLength);
-    newString[j] = (jbyte)0;
-}
-
-/* Given a Modified UTF-8 string, calculate the Standard UTF-8 length.
- *   Basic validation of the UTF encoding rules is done, and length is
- *   returned (no change) when errors are detected.
- *
- *   Note: No validation is made that this is indeed Modified UTF-8 coming in.
- *
- */
-int JNICALL
-utf8mToUtf8sLength(struct UtfInst *ui, jbyte *string, int length)
-{
-    int newLength;
-    int i;
-
-    newLength = 0;
-    for ( i = 0 ; i < length ; i++ ) {
-        unsigned byte1, byte2, byte3, byte4, byte5, byte6;
-
-        byte1 = (unsigned char)string[i];
-        if ( (byte1 & 0x80) == 0 ) { /* 1byte encoding */
-            newLength++;
-        } else if ( (byte1 & 0xE0) == 0xC0 ) { /* 2byte encoding */
-            /* Check encoding of following bytes */
-            if ( (i+1) >= length || (string[i+1] & 0xC0) != 0x80 ) {
-                break; /* Error condition */
-            }
-            byte2 = (unsigned char)string[++i];
-            if ( byte1 != 0xC0 || byte2 != 0x80 ) {
-                newLength += 2; /* Normal 2byte encoding, not 0xC080 */
-            } else {
-                newLength++;    /* We will turn 0xC080 into 0 */
-            }
-        } else if ( (byte1 & 0xF0) == 0xE0 ) { /* 3byte encoding */
-            /* Check encoding of following bytes */
-            if ( (i+2) >= length || (string[i+1] & 0xC0) != 0x80
-                                 || (string[i+2] & 0xC0) != 0x80 ) {
-                break; /* Error condition */
-            }
-            byte2 = (unsigned char)string[++i];
-            byte3 = (unsigned char)string[++i];
-            newLength += 3;
-            /* Possible process a second 3byte encoding */
-            if ( (i+3) < length && byte1 == 0xED && (byte2 & 0xF0) == 0xA0 ) {
-                /* See if this is a pair of 3byte encodings */
-                byte4 = (unsigned char)string[i+1];
-                byte5 = (unsigned char)string[i+2];
-                byte6 = (unsigned char)string[i+3];
-                if ( byte4 == 0xED && (byte5 & 0xF0) == 0xB0 ) {
-                    /* Check encoding of 3rd byte */
-                    if ( (byte6 & 0xC0) != 0x80 ) {
-                        break; /* Error condition */
-                    }
-                    newLength++; /* New string will have 4byte encoding */
-                    i += 3;       /* Skip next 3 bytes */
-                }
-            }
-        } else {
-            break; /* Error condition */
-        }
-    }
-    if ( i != length ) {
-        /* Error in UTF encoding */
-        /*  FIXUP: ERROR_MESSAGE()? */
-        return length;
-    }
-    return newLength;
-}
-
-/* Convert a Modified UTF-8 string into a Standard UTF-8 string
- *   It is assumed that this string has been validated in terms of the
- *   basic UTF encoding rules by utf8Length() above.
- *
- *   Note: No validation is made that this is indeed Modified UTF-8 coming in.
- *
- */
-void JNICALL
-utf8mToUtf8s(struct UtfInst *ui, jbyte *string, int length, jbyte *newString, int newLength)
-{
-    int i;
-    int j;
-
-    j = 0;
-    for ( i = 0 ; i < length ; i++ ) {
-        unsigned byte1, byte2, byte3, byte4, byte5, byte6;
-
-        byte1 = (unsigned char)string[i];
-        if ( (byte1 & 0x80) == 0 ) { /* 1byte encoding */
-            /* Single byte */
-            newString[j++] = byte1;
-        } else if ( (byte1 & 0xE0) == 0xC0 ) { /* 2byte encoding */
-            byte2 = (unsigned char)string[++i];
-            if ( byte1 != 0xC0 || byte2 != 0x80 ) {
-                newString[j++] = byte1;
-                newString[j++] = byte2;
-            } else {
-                newString[j++] = 0;
-            }
-        } else if ( (byte1 & 0xF0) == 0xE0 ) { /* 3byte encoding */
-            byte2 = (unsigned char)string[++i];
-            byte3 = (unsigned char)string[++i];
-            if ( i+3 < length && byte1 == 0xED && (byte2 & 0xF0) == 0xA0 ) {
-                /* See if this is a pair of 3byte encodings */
-                byte4 = (unsigned char)string[i+1];
-                byte5 = (unsigned char)string[i+2];
-                byte6 = (unsigned char)string[i+3];
-                if ( byte4 == 0xED && (byte5 & 0xF0) == 0xB0 ) {
-                    unsigned u21;
-
-                    /* Bits in: 11101101 1010xxxx 10xxxxxx */
-                    /* Bits in: 11101101 1011xxxx 10xxxxxx */
-                    i += 3;
-
-                    /* Reconstruct 21 bit code */
-                    u21  = ((byte2 & 0x0F) + 1) << 16;
-                    u21 += (byte3 & 0x3F) << 10;
-                    u21 += (byte5 & 0x0F) << 6;
-                    u21 += (byte6 & 0x3F);
-
-                    /* Bits out: 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx */
-
-                    /* Convert to 4byte encoding */
-                    newString[j++] = 0xF0 + ((u21 >> 18) & 0x07);
-                    newString[j++] = 0x80 + ((u21 >> 12) & 0x3F);
-                    newString[j++] = 0x80 + ((u21 >>  6) & 0x3F);
-                    newString[j++] = 0x80 + (u21 & 0x3F);
-                    continue;
-                }
-            }
-            /* Normal 3byte encoding */
-            newString[j++] = byte1;
-            newString[j++] = byte2;
-            newString[j++] = byte3;
-        }
-    }
-    UTF_ASSERT(i==length);
-    UTF_ASSERT(j==newLength);
-    newString[j] = 0;
-}
-
-/* ================================================================= */
-
-#ifdef COMPILE_WITH_UTF_TEST  /* Test program */
-
-/*
- * Convert any byte array into a printable string.
- *    Returns length or -1 if output overflows.
- */
-static int
-bytesToPrintable(struct UtfInst *ui, char *bytes, int len, char *output, int outputMaxLen)
-{
-    int outputLen;
-    int i;
-
-    UTF_ASSERT(bytes);
-    UTF_ASSERT(len>=0);
-    UTF_ASSERT(output);
-    UTF_ASSERT(outputMaxLen>=0);
-
-    outputLen = 0;
-    for ( i=0; i<len ; i++ ) {
-        unsigned byte;
-
-        byte = bytes[i];
-        if ( byte <= 0x7f && isprint(byte) && !iscntrl(byte) ) {
-            if ( outputLen + 1 >= outputMaxLen ) {
-                return -1;
-            }
-            output[outputLen++] = (char)byte;
-        } else {
-            if ( outputLen + 4 >= outputMaxLen ) {
-                return -1;
-            }
-            (void)sprintf(output+outputLen,"\\x%02x",byte);
-            outputLen += 4;
-        }
-    }
-    output[outputLen] = 0;
-    return outputLen;
-}
-
-static void
-test(void)
-{
-    static char *strings[] = {
-                "characters",
-                "abcdefghijklmnopqrstuvwxyz",
-                "0123456789",
-                "!@#$%^&*()_+=-{}[]:;",
-                NULL };
-    int i;
-    struct UtfInst *ui;
-
-    ui = utfInitialize(NULL);
-
-    i = 0;
-    while ( strings[i] != NULL ) {
-        char *str;
-        #define MAX 1024
-        char buf0[MAX];
-        char buf1[MAX];
-        char buf2[MAX];
-        unsigned short buf3[MAX];
-        int len1;
-        int len2;
-        int len3;
-
-        str = strings[i];
-
-        (void)bytesToPrintable(ui, str, (int)strlen(str), buf0, 1024);
-
-        len1 = utf8FromPlatform(ui, str, (int)strlen(str), (jbyte*)buf1, 1024);
-
-        UTF_ASSERT(len1==(int)strlen(str));
-
-        len3 = utf8ToUtf16(ui, (jbyte*)buf1, len1, (jchar*)buf3, 1024);
-
-        UTF_ASSERT(len3==len1);
-
-        len1 = utf16ToUtf8m(ui, (jchar*)buf3, len3, (jbyte*)buf1, 1024);
-
-        UTF_ASSERT(len1==len3);
-        UTF_ASSERT(strcmp(str, buf1) == 0);
-
-        len2 = utf8ToPlatform(ui, (jbyte*)buf1, len1, buf2, 1024);
-
-        UTF_ASSERT(len2==len1);
-        UTF_ASSERT(strcmp(str, buf2) == 0);
-
-        i++;
-    }
-
-    utfTerminate(ui, NULL);
-
-}
-
-int
-main(int argc, char **argv)
-{
-    test();
-    return 0;
-}
-
-#endif